clp en

Transcript

1 Guidance on the Application of the CLP Criteria – July 2017 1 Version 5.0 G U I D A N C E Guidance on the Application of the CLP Criteria Guidance to Regulation (EC) No 1272/2008 on classification, labelling and packaging (CLP) of substances and mixtures Version 5.0 2017 July

2 Guidance on the Application of the CLP Criteria 2 Version 5.0 – July 2017 LEGAL NOTICE aims to assist users in complying with their obligations under the CLP This document However, users are reminded that the text of the CLP R Regulation . egulation is the only authentic legal reference and that the information in this document does not constitute legal advice. Usage of the information remains under the sole responsibility of the user. The ity with regard to the use that may be European Chemicals Agency does not accept any liabil this document. made of the information contained in Guidance on the Application of CLP Criteria ECHA Reference: 17 - G - 21 - EN - Cat.Number: ED - 02 - 17 - 754 - EN - N ISBN: 978 - 92 - 9020 - 050 - 5 10.2823/124801 DOI: Publ.date: Ju ly 2017 Language: EN © European Chemicals Agency, 201 7 If you have questions or comments in relation to this document please send them (indicating the document reference, issue date, chapter and/or page of the document to which your comment refers) using the Guidance feedback form. The feedback form can be acc essed via the ECHA Guidance website or directly via the following link: https://comments.echa.europa.eu/comments_cms/FeedbackGuidance.aspx European Chemicals Agency Mailing address: P.O. Box 400, FI - 00121 Helsinki, Finland Visiting address: Annankatu 18, Helsinki, Finland

3 Guidance on the Application of the CLP Criteria 5.0 July 2017 3 Version – DOCUMENT HISTORY Comment Date Version n.a. August 2009 First edition n.a. April 2011 Please note that change between the version published in August 2009 and that of April 2011 are not recorded in this document history. Version 2.0 Revision of the Guidance addressing content in relation to the April 2012 nd environmental criteria chapters and Annexes following the 2 Adaptation to Technical Progress to the CLP Regulation (Commission Regulation (EU) No 286/2011). The ECHA 4 Environmental Secretariat revised the Guidance Part – Annexes of the guidance document referring to and hazards the revised criteria for the long - term aquatic hazard for – Additional Part substances and mixtures and added new 5 referring to the hazard class ‘ hazardous to the ozone hazards layer ’ . As well, a number of examples have been included in the respec tive Parts and Annexes to illustrate the revisions performed. Further to this, a range of editorial corrections were Part 1 General principles for classification and proposed for – . labelling The update includes the following:  Revision of Part 1, by elimin ating and amending out of date information and restructuring the text in order to reflect the Guidance update. All green boxes in Part 4 that are impacted by the 2nd  ATP were updated. As the CLP legal text uses commas instead of dots to define numbers sma ller than 1, the green boxes now show commas as well. Revision of Part 4, by providing guidance on the  - application of the new long term aquatic hazard criteria for substances and mixtures. Section 4.1.3 Classification of substances hazardous to  c environment and section 4.1.4 Classification the aquati of mixtures hazardous to the aquatic environment were substantially revised, for example by addition of new references, as well as the new/revised examples to illustrate relevant topics in the Part 4. New Part 5 – Additional hazards was added (please note  that Part 5: Labelling was deleted from the Guidance in - recorded versions and covered via a new previous non Guidance on Labelling and Packaging in accordance with il 2011). Regulation (EC) No 1272/2008 published in Apr Most of the I.3 sub - sections in Annex I  Aquatic toxicity – were revised.

4 Guidance on the Application of the CLP Criteria 4 Version 5.0 – July 2017  – Rapid degradation the terminology was In Annex II modified.  – Metals and Inorganic Metal Most of the Annex IV Compounds was substantially modified and revised, as in sub - section IV.7 new examples were added. well as Revision of Guidance Part 3 Health Hazards, relating to specific November Version 3.0 2012 concentration limits (SCLs) for 4 hazard classes and the . inclusion of a new Annex The update includes the following: Revision of Part 3, by providing gu  idance on the setting of lower and higher SCLs for 4 health hazard classes in section 3.2.2.5 Skin Corrosion/Irritation; section 3.3.2.5 Serious Eye Damage/Eye Irritation; section 3.7.2.5 - Reproductive Toxicity and section 3.8.2.6 STOT SE, in h CLP Article 10(7); accordance wit Inclusion of a new Annex (Annex VI) providing guidance  on setting SCLs for the reproductive toxicity hazard class based on potency considerations. Version 4.0 (i) Revision of the CLP Guidance addressing content in relation November to the Part 2: Physical hazards, Part 3: Health hazards and 2013 th nd Adaptation to Technical Annex VI following the 2 and the 4 Progress to the CLP Regulation (Commission Regulation (EU) No 286/2011 of 10 March 2011 and Commission Regulation (EU) No 487/2013 of 8 May 2013). The revision includes: Numbering of chapters within CLP Guidance, Parts 2 & 3  were synchronised with corresponding chapter numbering of CLP, Annex I. nd th and 4 ATPs. Changes in the legal text due the 2  Changes in the legal text due to the 4th  ATP were highlighted in orange within all relevant green boxes. All changes are preceded by a note highlighting the changes. (To note: a corrigendum will change the colour to green of relative legal text boxes from orange when th ATP applies). the 4 In addition, the revisions to Part 2: Physical hazards include the following: Chapters ‘Pyrophoric liquids and solids’ and ‘Oxidising  liquids and solids’ were divided into four chapters: ‘Pyrophoric liquids’, ‘Pyrophoric solids’, ‘Oxidising liquids’ ising solids’ respectively. and ‘Oxid th Based on the 4 ATP the CLP Guidance Chapter 2.2  Flammable gases was extended to take into account the scope of CLP, Annex I, section 2.2 to include chemically unstable gases. th  ATP amended the criteria in CLP Annex Further, the 4 I, Section 2.3 Flammable aerosols and renamed it into

5 Guidance on the Application of the CLP Criteria 5.0 – July 2017 5 Version 2.3 Aerosols. Hence, the CLP Guidance was amended accordingly.  All chapters were rechecked and redundant and/or outdated information were deleted, reorganised and /or revised. For example, ‘Introduction’ chapters were significantly shortened, however several “examples” ‘Example for classification...’) were further sections (i.e . elaborated.  Where missing, a new sub chapter ‘Relation to other - physical hazards’ was adde d. Sub  chapter 2.0.4 ‘Physical state’ was extended with - additional information about substance/mixture form and some examples. In sub - chapter 2.1.5.2 ‘Additional labelling provisions’  within chapter 2.1 ‘Explosives’ further guidance about tion was provided. hazard communica  In sub - chapter 2.5.6.1 a new recommendation for shot hazard codes to identify the classification of gasses under pressure was added. - going  Footnotes with references to endorsed or on revisions of the GHS which have not yet been implemen ted into the CLP via a respective ATP were included in relevant sub - chapters of this guidance for information only. In addition, the major revisions to Part 3: Health hazards include the following: th ATP,  All sections: revisions to legal text for the 4 luding revisions to Precautionary Statements in the inc Tables with labelling information . th he introduction of new guidance for the 4  Section 3.1: t ATP in section 3.1.4.1 .  Sections 3.2 . 2.5 and 3.3.2.5: clarification to the recently published text (Version 3.0 ) for the setting of SCLs.  Section 3.4 (sensitisation) has been significantly re - organised to present all the information on respiratory sensitisation together, followed by the information on skin sensitisation. This is in line with how the sections esented in the CLP Regulation and in GHS are pr documents.  Section 3.4: integration of subcategories for respiratory and skin sensitisation based on potency of a substance; f semi - quantitative terms like ‘ low to clarification o moderate sensitisation rate’ and ‘hi gh or low exposure’ ; elaboration of eva uation of human data for skin l sensitisation and the addition of new examples. th Section 3.7 : t he introduction of new guidance for the 4  ATP in section3.7.4.1 and section 3.7.5.1.

6 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 6 (ii) Corrigendum of Part 1: General principles for classification and labelling and Part 4: Environmental hazards and its related - Annexes I V. The corrigendum includes the following: The list of abbreviations was updated.  Update or deletion of outdated references to Guidance  on information requirements and chemical safety assessment, Endpoint specific guidance (Chapter R.7a) - V. within Annexes I  A footnote informing the reader that with effect from 1 September 2013, Directive 98/8/EC had been repealed by Biocidal Products Regulation (EU) No 528/2012 was added.  In Part 1, Part 4 and Annexes modal verbs ‘shall’ were replaced with ‘must’ where appropriate. n  A footnote related to respiratory sensitisation and ski sensitisation in Table 1.1 was removed. A corr  - chapter 4.1.4.7.5 was ection to Example D, sub applied, namely a reference to CLP, Annex I, point (b) (ii) of Table 4.1.0 was introduced. In addition , the result of a summation method calculation was corrected. June 2015 Corrigendum to take account of the end of the transition period Version 4.1 th ATP (as foreseen in version 4.0 above) : of the 4  change the colour of relative legal text boxes from to green ; orange in Part 2, to delete section 2.2. 1 Flammable gases and  2.3.1 Flammable Aeroso (outdated text) and section ls renumber sections 2.2.2 Flammable gases (including and 2.3.2 Aerosols chemically unstable gases) accordingly; in Part 3, out dated text” in sections  to delete the “ .7.5.1 3.7.4. 1 and 3 in Reproductive Toxicity. addition errors were corrected and , minor editorial minor In reformatting was made. 2017 Version 5.0 Partial r evision of the Guidance to update the content mainly ly Ju th Adaptation to Technical Progress to the CLP following the 8 (EU) No 286/2011). Regulation (Commission Regulation Revision of few specific additional topics. The update includes the following: ut the document: o (i) Through Revision of legal references and legal text quotations.  Renumbering of some  . sections reclassification of  regarding the Deletion of sections in substances and mixtures previously classified or accordance with the DSD DPD.

7 Guidance on the Application of the CLP Criteria 5.0 – July 2017 7 Version (ii) Revision of Part 1: to pre - CLP legislation and  Deletion of reference transitional period. Addition of reference to read - across and grouping in the  co ntext of bioavailability. Removal of quotation of Article 31(3) of REACH.  Clarification about applicability of additivity principle .  Clarification about the application of mixture rules to  substances with CMR constituents.  Reduction of section 1.2.3.1 on p hysical hazards to avoid redundancy with section 2.0.4.  Revision of section 1.7 and removal of unnecessary information. Table on additional information using transport classification moved to a new Annex VII. the Par of Revision of t 2: following sections (iii) 2.1 (Explosives): replacement of new figure 2.1.3;  u pdate of label elements; addition new note 2 to table on requirement for SDSs. 2.1.2  of text on classification criteria; 2.3 (Aerosols): update - ; u pdate of section 2.3 .6 update of decision logic 2.3.1 a on the relation to transport classification.  2.14 (Oxidising solids): addition of criteria using test 0.3; update of labelling elements . (iv) Minor changes to the following sections in Part 2:  - reactive): u pdate of label elements. 2.8 (Self 2.12 (Emitting flammable gases): pdate of label  u elements.  u pdate of decision 2.15 (Organic peroxides): . 2.15.1 ; u pdate of label elements. logic (v) Revision of following sections in Part 3: in - vitro test.  3.1 (Acute toxicity): Reference to new hat harmonised ATE values will be included Indication t to CLP . Deletion of refere nce to the concept in Annex VI of relat ing the conditions of an acute inhalation test to real life. Indication that not classified components may - and, in general, clarificat ion about influence ATE components to be considered for mixture classification Indication to avoid according to the case. under classification for oral toxicity. A dditon of a new example (13) on the application of additivity methods for mixtures with components in diffe rent physical forms.  3.2 (Skin corrosion ) : S ubsection on non - testing methods updated and clarified the need to assess the relevance. Update of classification criteria. Inclusion of new figure illustrating the tiered evaluation approach. Inclusion of a new figure illustrating the relative weight

8 Guidance on the Application of the CLP Criteria 8 Version 5.0 – July 2017 of different available pieces of information to be considered when weight of Evidence (WoE) is applied. separate Replacement of the decision logic chart with ures, based on decision logics for substances and mixt chart the from GHS. Clarification about classification of mixture as Category 1 without subcategory.  C larification of the 3.3 (Serious eye damage/irritation): for further data when considerations about need alkaline/acid reserve suggest no r isk added. - testing methods results enhanced. Interpretation of non Inclusion of new figure Mentioned the use of LVET data. e tiered evaluation approach . illustrating th Inclusion of reference to new figure on hierarchy of information R added in section 3.2. eplacement of the decision logic chart with separate decision logics for substances and chart from GHS mixtures, based on the .  3.4 ( Respiratory or skin sensitisation): Deletion of the relationship between skin and respiratory sensitisation potential. Iden - human data brought in tification of non line with REACH guidance. Introduction of available non - Clarification of the test sample to be testing systems. used in human diagnostic patch testing. 3.5 (Germ cell mutagenicity): Reference to OECD TG  488 added. Ne w section on classification of substances containing CMR constituents, additives or impurities included. the following sections in Part 3: (iv) Minor changes to  3.6 (Carcinogenicity): R emoval of reference to supporting evidence for classification under DS D. included on Update of label elements. New section constituents, classification of substances containing CMR additives or impurities .  3.7 (Reproductive toxicity): New section included on classification of substances containing CMR constituents, additives or impurities .  3.8 ( STOT - SE ): Editorial corrections to the examples. (vi) Part 4 to update the terminology when Minor changes to referring to short - term (acute) and long - term (chronic) studies.

9 Guidance on the Application of the CLP Criteria 5.0 – July 2017 9 Version PREFACE document is the Guidance on the Application of the CLP Criteria. It is a comprehensive This technical and scientific document on the application of Regulation (EC) No 1272/2008 on the the classification, labelling and packaging of substances and mixtures (CLP), which replace d nces Directive 67/548/EEC (DSD) and the Dangerous Preparations Directive Dangerous Substa 1999/45/EC (DPD) in a staggered way. CLP is based on the Globally Harmonised System of Classification and Labelling of Chemicals (GHS) and is implementing the provisions of the GHS wi thin the EU. The objective of this document is to provide detailed guidance on the application of the CLP criteria for physical, health and environmental hazards. The guidance is developed to assist manufacturers , importers and downstream users i n applying the primarily classification and labelling criteria and it also includes practical examples. It is also assumed to be the , guidance on classification and labelling for Competent Authorities in the Member States (MS CA), for the Commission services and the European Chemicals Agency (ECHA). In certain chapters, like for example the ones on carcinogenicity, mutagenicity and reproductive toxicity, the guidance includes to a larger extent scientific advice on how to interpret different data used for classificat ion. This additional guidance is based on experience gained within the EU during the application of the classification criteria under Directive 67/548/EEC, and is written for the experts within the respective fields. s a REACH Implementation Project (RIP 3.6) at the This guidance document was developed a Institute for Health and Consumer Products (IHCP) of the Joint Research Centre in Ispra, with support from working groups consisting of experts on classification and labelling from EU stry. The project started in September 2007 and the different working Member States and Indu groups had meetings and continuous discussions to discuss and develop the guidance text until spring 2009. Finally all texts were consolidated and edited at the IHCP. RIP 3.6 was financi ally - supported with an administrative arrangement made with Directorate General Enterprise and Industry (currently DG Growth) . The guidance was handed over to ECHA in summer 2009. After that the guidance has been revised twice – version 2.0 in April 2012 o n the long - term aquatic hazard and version 3.0 in November 2012 in relation to the guidance chapters on setting of specific concentration limits (SCLs) for health hazards. uring 2012/2013 , further drafting work was done in close collaboration with European experts, D to take account of a range of guidance aspects (for example further guidance on the criteria for respiratory and skin sensitisation, and other health related points, as well as guidance on the criteria for chemically unstable gases and ae rosols and other phy sical hazards related changes) nd th and/or the 4 Adaptation to Technical Progress (ATP) to the CLP (Commission following the 2 1 (EU) No 286/2011 and No 487/2013 Regulation This work resulted in publication of version ). 4.0 in November 20 13 and the subsequent corrigendum version 4.1 June 2015 to update the th text following the transitional period for the 4 . ATP - In relation to labelling and packaging, a new stand alone guidance document was prepared ‘ Guidance on Labelling and Packaging in ( ’ ), accordance with Regulation (EC) No 1272/2008 warranting the deletion of Part 5 and of Annex V of the Guidance on the Application of the CLP Criteria. The Guidance on Labelling and Packaging in accordance with Regulation (EC) No 1272/2008 is published on ECHA’s guidance website, under http://guidance.echa.europa.eu/guidance_en.htm . 1 Commission Regulation (EU) No 286/2011 of 10 March 2011 and Commission Regulation (EU) No 487/2013 of 8 May 2013 amending, for the purposes of its adaptation to technical and scientific progress, Regulation (EC) No 1272/2008 of the European Parliament and of the Coun cil on classification, labelling and packaging of substances and mixtures.

10 Guidance on the Application of the CLP Criteria Version 5.0 – 10 July 2017 Both guidance documents were further updated in 2016 to address the changes due to the 8th ATP (e.g. new alternat i dising solids, changes in the classification for ive methods to classify ox skin corrosion/irritation, serious eye damage/irritation and aerosols , as well as changes in precautionary statements). th Therefore the current version of the Guidance reflects the changes made by the 8 ATP , to CLP . These changes apply from 1 February 2018. (Regulation 2016/918) in Annex I However: th  The 8 ATP may already be applied on a voluntary basis before that date. Substances and mixtures placed on the market before 1 February 2018 shall not be  th required to be relabelled and repackaged in accordance with the 8 ATP during a period of two years, i.e. before 1 February 2020.

11 Guidance on the Application of the CLP Criteria 5.0 – July 2017 11 Version Table of Contents CIPLES FOR CLASSIFIC ATION AND 1. PART 1: GENERAL PRIN ... 44 ... ... LABELLING ... ... ... 44 1.1. INTRODUCTION The objective of the guidance document ... ... 44 1.1.1. Background ... ... ... ... 45 1.1.2. Hazard classification ... ... ... 45 1.1.3. Who is responsible for the hazard classification ... ... 46 1.1.4. Which substances and mixtures should be classified ... ... 46 1.1.5. What information is needed for cl ... ... 48 1.1.6. assification Information for the classification of substances ... ... 48 1.1.6.1. 1.1.6.2. ... . 50 Information relevant for the classification of mixtures 1.1.7. Data evaluation and reaching a decision on classification ... ... 50 Classification of substances ... ... ... 50 1.1.7.1. Influence of impurities, additives or individual constituents on the 1.1.7.2. ... ... ... 51 classification of a substance Updating of hazard classifications ... ... ... 51 1.1.8. The interface between hazard classification and hazard communication ... 51 1.1.9. - 1.1.10. classification and harmonised classification, and the list The interface between self ... ... 51 of harmonised classifications ... ... .. 53 1.1.11. The Classification and Labelling Inventory (C&L Inventory) Relation of classification to other EU legislation 1.1.12. ... 54 ... 1.1.12.1. REACH 54 1.1.12.2. ... ... 54 Plant Protection Products and Biocides 1.1.12.3 . Transport legislation ... ... ... 54 1.2. THE TERMS ‘FORM OR P HYSICAL STATE’ AND THE SIGNIFICANCE OF USE’ WITH RESPECT TO CLASSIFICATION ‘REASONABLY EXPECTED ... 55 ... ACCORDING TO CLP ... ‘Form or physical state’ and ‘reasonably expected use’ ... ... 55 1.2.1. The term ‘reasonably expected use’ in relation to hazard classification ... 55 1.2.2. ... 56 1.2.3. The term ‘form or physical state’ in relation to hazard classification Physical hazards ... ... ... 56 1.2.3.1. 1.2.3.2. ... ... ... 56 Human health hazards ... ... ... 57 1.2.3.3. Environmental hazards SPECIFIC CASES REQUI RING FURTHER EVALUAT ION – LACK OF 1.3. BIOAVAILABILITY ... ... 57 ... ... ... ... 1.3.1. ... 57 Definition 1.3.2. Bioavailability ... ... ... 58 1.3.2. 1. Human health hazards ... ... ... 58 59 ... ... ... 1.3.2.2. Environmental hazards - EGORISATION (READ 1.4. ACROSS AND GROUPING) AND USE OF SUBSTANCE CAT (Q)SARS FOR CLASSIFI ... ... 60 CATION AND LABELLING 1.4.1. (Q)SAR ... ... ... ... 61 ... 1.4.2. g ... ... Groupin ... 62 1.4.3. Read - across ... ... ... ... 62 1.5. SPECIFIC CONCENTRATI ON LIMITS AND M - FACTORS ... .. 62 1.5.1. Specific concentration limits ... ... ... 62

12 Guidance on the Application of the CLP Criteria 12 Version 5.0 – July 2017 1.5.2. - factors) ... ... ... 64 Multiplying factors (M Harmonised ATE values 1.5.3. ... ... 65 ... ... ... 65 MIXTURES ... 1.6. ... ... ... 1.6.1. How to classify a mixture 65 Classification for physical hazards ... ... ... 66 1.6.2. Health and environmental hazards ... ... 1.6.3. 67 ... ... . 67 1.6.3.1. Classification derived using data on the mixture itself Bridging principles ... ... ... 68 1.6.3.2. Dilution ... ... ... 1.6.3.2.1. 68 ... ... Batching 69 1.6.3.2.2. ... Concentration of highly hazardous mixtures ... ... 69 1.6.3.2.3. 1.6.3.2.4. Interpolation within one hazard category ... ... 69 Substantially similar mixtures ... ... 70 1.6.3.2.5. Review of classification where the composition of a mixture has changed .. 71 1.6.3.2.6. Aerosols (some health hazards only) ... ... 72 1.6.3.2.7. Classification based on calculation or concentration thresholds ... 72 1.6.3.3. Classification based on c ... ... 72 1.6.3.3.1. alculation Classification based on concentration thresholds ... ... 1.6.3.3.2. 74 1.6.3.3.3. Additivity Vs. non additivity of hazards ... ... 75 ... ... 77 1.6.4. Classification of mixtures in mixtures Example: Classification of Mixture A ... ... 77 1.6.4.1. 1.6.4.2. ... ... 80 Example: Classification of Mixture B 1.7. ANNEX VII TO CLP ... ... ... 83 2. PART 2: PHYSICAL HAZ ARDS ... ... 86 2.0. ... ... ... 86 INTRODUCTION Gene ... 86 2.0.1 ral remarks about the prerequisites for classification and testing Safety ... ... ... ... 86 2.0.2 General conditions for testing ... ... ... 86 2.0.3 . ... ... ... 87 2.0.4 Physical state Quality ... ... ... ... 88 2.0.5 2.1. ... ... ... 88 EXPLOSIVES ... ... ... ... 88 2.1.1. Introduction D efinitions and general considerations for the classification of explosives ... 89 2.1.2. ... ... ... 90 Relation to other physical hazards 2.1.3. Classification of substances, mixtures or articles as explosives ... 90 2.1.4. 2.1.4.1. Identification of hazard information ... ... 90 2.1.4.2. ... ... 90 Screening procedures and waiving of testing ... ... ... 91 2.1.4.3. Classification criteria Testing and evaluation of hazard information ... ... 93 2.1.4.4. 2.1.4.5. ... ... 93 Classification procedure and decision logics 2.1.4.5.1. ... ... ... 94 Acceptance procedure 2.1.4.5.2. Assignment procedure to a division ... ... 97 2.1.5. ... ... 103 Hazard communication for explosives Pictograms, signal words, hazard statements and precautionary statements 103 2.1.5.1. 2.1.5.2. Additional labelling provisions ... ... 104 2.1.5.2.1. Packaging dependance ... ... ... 104 2.1.5.2.2. Supplemental hazard information ... ... 105 2.1.5.3. Further communication requirements ... ... 106

13 Guidance on the Application of the CLP Criteria 5.0 – July 2017 13 Version ... ... .. 106 Relation to transport classification 2.1.6. ... 107 Examples of classification for explosives 2.1.7. ... ... 107 2.1.7.1. Example of substances and mixtures fulfilling the classification criteria Example of substances and mixtures not fulfilling the classification criteria 108 2.1.7.2. .. FLAMMABLE GASES (INC STABLE GASES) ... 112 2.2. LUDING CHEMICALLY UN Introduction ... 2.2.1. ... . 112 ... 2.2.2. Definitions and general considerations for the classification of flammable gases (including chemically 112 ... unstable gases) ... Relation to other physical hazards ... .. 112 2.2.3. ... Classification of substances and mixtures as flammable gases (including chemically 2.2.4. ... unstable gases) ... 112 ... ... ... 112 2.2.4.1. Identification of hazard information Screening procedures and waiv ing of testing for gas mixtures ... 2.2.4.2. 113 2.2.4.3. Classification criteria ... ... ... 113 Testing and evaluation of hazard information ... ... 114 2.2.4.4. Decision logic ... ... ... 115 2.2.4.5. Decision logic for flammable gases ... ... 116 2.2.4.5.1. Decision logic for chemically unstable gases 2.2.4.5.2. ... ... 117 Hazard communication for flammable gases (including chemically unstable gases) 118 2.2.5. P ictograms, signal words, hazard statements and precautionary statements 118 2.2.5.1. 2.2.6. Relation to transport classification ... ... .. 119 ... ... 119 2.2.7. Example of classification for flammable gases AEROSOLS ... ... ... 121 2.3. 2.3.1. ... ... ... . 121 Introduction 2.3.2. Definitions and general considerations for the classification of aerosols ... 121 2.3.3. Relation to other physical hazards ... ... .. 121 2.3.4. ols ... ... ... 122 Classification of aeros Classification criteria ... ... ... 122 2.3.4.1. Testing and evaluation of hazard information ... ... 123 2.3.4.2. Decision logic ... ... ... 2.3.4.3. 123 4.3.1. Decision logic for aerosols ... ... 124 2.3. .. Decision logic for spray aerosols ... ... 2.3.4.3.2. 125 2.3.4.3.3. Decision logic for foam aerosols ... ... 126 ... ... 127 2.3.5. Hazard communication for aerosols Pictograms, signal words, hazard statements and precautionary statements 127 2.3.5.1. provisions 2.3.5.2. ... ... 127 Additional labelling 2.3.6. Relation to transport classification ... ... .. 128 2.3.7. Examples of classification for aerosols ... ... 128 2.3.7.1. of aerosols fulfilling the classification criteria ... 128 Examples ... 129 2.3.7.2. Examples of aerosols not fulfilling the classification criteria OXIDISING GASES ... ... ... 130 2.4. 2.4.1. ... ... ... . 130 Introduction Definitions and general considerations for the classification of oxidisin g gases ... 130 2.4.2. 2.4.3. Relation to other physical hazards ... ... .. 130 2.4.4. ... 130 Classification of substances and mixtures as oxidising gases 2.4.4.1. Identification of hazard information ... ... 130 2.4.4.2. Screening procedures and waivin g of testing ... ... 130 2.4.4.3. Classification criteria ... ... ... 130 2.4.4.4. Testing and evaluation of hazard information ... ... 131

14 Guidance on the Application of the CLP Criteria 14 Version 5.0 – July 2017 2.4.4.5. ... ... ... 131 Decision logic Hazard communication for oxidising gases 2.4.5. ... 132 ... Pictograms, signal words, hazard statements and precautionary statements 2.4.5.1. 132 ... ... .. 132 2.4.6. Relation to transport classification Example of classification for oxidising gases ... 132 2.4.7. ... Example of substances and mixtures not fulfilli .. 132 2.4.7.1. ng the classification criteria GASES UNDER PRESSURE ... ... 2.5. 134 ... ... ... ... . 134 2.5.1. Introduction D 2.5.2. efinitions and general considerations for the classification of gases under pressure ... ... ... 134 ... Definition of ‘gas’ ... ... ... 134 2.5.2.1. Definition of gases under pressure ... ... 134 2.5.2.2. Relation to other physical hazards ... ... 2.5.3. 134 .. 2.5.4. Classification of substances and mixtures as gases under pressure ... 134 Identification of hazard information ... ... 134 2.5.4.1. Classification criteria ... ... ... 135 2.5.4.2. Testing and evaluation of hazard information ... ... 2.5.4.3. 135 ... ... Decision logic 136 2.5.4.4. ... 5. Hazard communication for gases under pressure ... ... 138 2.5. 2.5.5.1. ments 138 Pictograms, signal words, hazard statements and precautionary state ... ... .. 13 9 2.5.6. Relation to transport classification Examples of classification for gases under pressure ... ... 140 2.5.7. Examples of substances and mixtures fulfilling the classification criteria ... 140 2.5.7.1. ) 2.5.7.1.1. 140 O) + 75 % (N Example mixture: 9 % (O ... ) + 16 % (N 2 2 2 2.6. FLAMMABLE LIQUIDS ... ... ... 141 2.6. 1. ... ... ... . 141 Introduction Definitions and general considerations for the classification of flammable liquids .. 141 2.6.2. Relation to other physical hazards ... ... .. 141 2.6.3. Classification of substances and mixtures as flammable liquids ... 141 2.6.4. I ... ... 141 2.6.4.1. dentification of hazard information Screening procedures and waiving of testing ... ... 141 2.6.4.2. 2.6.4.2.1. ... ... ... 141 Boiling point ... ... ... 142 2.6.4.2.2. Flash point 4.3. Classification criteria ... ... ... 142 2.6. ... ... 142 Testing and evaluation of hazard information 2.6.4.4. Testing ... ... ... 143 2.6.4.4.1. 2. ... Evaluation of hazard information ... 2.6.4.4. 144 2.6.4.5. ... ... ... 144 Decision logic ... ... 146 2.6.5. Hazard communication for flammable liquids Pictograms, signal words, hazard statements and precautionary statements 146 2.6.5.1. 2.6.5.2. ... .. 146 Additional labelling provisions for flammable liquids Re - classification of substances and mixtures classified as flammable liquids 2.6.6. according to DSD and DPD or already classified for transport ... 147 2.6.6.1. Relation to transport classification ... ... 147 2.6.7. ... ... 147 Examples of classification for flammable liquids Examples of substances and mixtures fulfilling the classification criteria ... 147 2.6.7.1. 2.6.7.1.1. Exa mple 1 ... ... ... 147 2.6.7.1.2. Example 2 ... ... ... 148 2.6.7.2. Examples of substances and mixtures not fulfilling the classification criteria . 148 2.6.7.2.1. Example 3 ... ... ... 148

15 Guidance on the Application of the CLP Criteria 5.0 – July 2017 15 Version ... ... ... .. 148 References 2.6.8. ... ... 149 FLAMMABLE SOLIDS 2.7. ... ... ... ... . 149 2.7.1. Introduction Definitions and general considerations for the classification of flammable solids 149 2.7.2. ... Relation to other physical hazards ... .. 149 2.7.3. ... Classification of substances and mixtures as flammable solids ... 2.7.4. 150 ... ... 150 2.7.4.1. Identification of hazard information Screening procedures and waiv ... ... 150 2.7.4.2. ing of testing Classification criteria ... ... ... 150 2.7.4.3. Testing and evaluation of hazard information ... ... 151 2.7.4.4. ... ... ... 151 2.7.4.5. Decision logic 2.7.5. ... ... 153 Hazard communication for flammable solids 2.7.5.1. Pictograms, signal words, hazard statements and precautionary statements 153 Relation to transport classification ... ... .. 153 2.7.6. Examples of classification for flammable solids ... ... 153 2.7.7. Example of substances and mixtures fulfilling the classification criteria ... 153 2.7.7.1. Examp les of substances and mixtures not fulfilling the classification criteria . 154 2.7.7.2. .. References ... ... ... 154 2.7.8. SELF - 2.8. AND MIXTURES ... ... 155 REACTIVE SUBSTANCES 2.8.1. Introduction ... ... ... . 155 ... - reactives lf 155 2.8.2. Definitions and general considerations for the classification of se Relation to other physical hazards ... ... 2.8.3. 156 .. 2.8.4. Classification of substances and mixtures as self - reactive .. 156 ... 2.8.4.1. Identification of hazard information ... ... 156 2.8.4.2. ... ... ... 156 Classification criteria Testing and evaluation of hazard information ... ... 158 2.8.4.3. Thermal stability tests and temperature control ... ... 158 2.8.4.3.1. Additional considerations and testing ... ... 2.8.4.3.2. 159 rations ... Additional classification conside 160 2.8.4.3.3. ... Decision logic ... ... ... 161 2.8.4.4. Hazard communication for self - reactives ... ... 163 2.8.5. 163 2.8.5.1. Pictograms, signal words, hazard statements and precautionary statements 2.8.6. Relation to trans port classificationaccording to DSD and DPD or already classified ... ... ... 164 for transport self - reactives ... ... 164 2.8.7. Examples of classification for 2.8.7.1. Examples of substances and mixtures fulfilling the classification criteria ... 164 2.9. PYROPHORIC LIQUIDS ... ... ... 168 2.9.1. ... ... ... . 168 Introduction ... 168 2.9.2. Definitions and general considerations for the classification pyrophoric liquids Relation to other physical hazards ... ... .. 168 2.9.3. 2.9.4. ... 169 Classification of substances and mixtures as pyrophoric liquids 2.9.4.1. Identification of hazard information ... ... 169 2.9.4.2. Screening procedures and waiving of testing ... ... 169 2.9.4.3. ... ... ... 169 Classification criteria Testing and evaluation of hazard information ... ... 169 2.9.4.4. 2.9.4.5. Decision logic ... ... ... 170 2.9.4.5.1. Decision logic for pyrophoric liquids ... ... 170 2.9.5. Hazard communication for pyrophoric liquids ... ... 171 2.9.5.1. Pictograms, signal words, hazard statements and precautionary statements 171

16 Guidance on the Application of the CLP Criteria 16 Version 5.0 – July 2017 2.9.6. ion to transport classification ... ... .. 171 Relat Examples of classification for pyrophoric liquids ... 172 2.9.7. ... 172 Examples of substances and mixtures fulfilling the classification criteria ... 2.9.7.1. ... ... ... 172 2.9.7.1.1. Example 1 Example 2 ... ... ... 173 2.9.7.1.2. Examples of substances and mixtures not fulfilling the classification criteria . 173 2.9.7.2. Example 3 ... ... ... 173 2.9.7.2.1. References ... ... .. ... 173 2.9.8. PYROPHORIC SOLIDS ... ... ... 174 2.10. Introduction ... ... ... . 174 2.10.1. Definitions and general considerations for the classification pyrophoric solids ... 174 2.10.2 . Relation to other physical hazards ... ... 2.10.3. 175 .. 2.10.4. Classification of substances and mixtures as pyrophoric solids ... 175 Identification of hazard information ... ... 175 2.10.4.1. Screening procedures and waiving of testing ... ... 175 2.10.4.2. Classification criteria ... ... ... 2.10.4.3. 175 Testing and evaluation of hazard information ... ... 176 2.10.4.4. Decision logic ... ... ... 176 2.10.4.5. 2.10.4.5.1. ... ... 176 Decision logic for pyrophoric solids Hazard communication for pyrophoric solids ... ... 177 2.10.5. Pictograms, signal words, hazard statements and precautionary statements 177 2.10.5.1. Relation to transport classification ... ... 2.10.6. 177 .. 2.10.7. Examples of classification for pyrophoric solids ... ... 178 Examples of substances and mixtures fulfilling the classification criteria ... 178 2.10.7.1. 2.10.7.1.1. Example 1 ... ... ... 178 2.10.7.1.2. ... ... ... 178 Example 2 Examples of substances and mixtures not fulfilling the classification criteria . 178 2.10.7.2. Example 3 ... ... ... 178 2.10.7.2.1. Example 4 ... ... ... 179 2.10.7.2.2. .. ... ... ... 179 2.10.8. References SE LF - HEATING SUBSTANCES A ND MIXTURES ... ... 180 2.11. 2.11.1. ... ... ... . 180 Introduction Definitions and general considerations for the classification of self - heating 2.11.2 . ... ... 180 substances and mixtures ... Relation to ot her physical hazards ... ... .. 180 2.11.3. ... heating substances and mixtures ... - 180 Classification of self 2.11.4. Identification of hazard information ... ... 180 2.11.4.1. 2.11.4.2. Screening procedures and waiving of testing ... ... 181 2.11.4.3. ... ... ... 181 Classification criteria ... ... 182 2.11.4.4. Testing and evaluation of hazard information General remarks ... ... ... 182 2.11.4.4.1. 2.11.4.4.2. ... ... ... 182 Sample preparation 2.11.4.4.3. Criteria and evaluation ... ... ... 182 2.11.4.5. Decision logic ... ... ... 183 2.11.4.6. ... ... ... 184 Exemption 2.11.5. Hazard communication for self - heating substances and mixtures ... 186 2.11.5.1. Pictograms, signal words, hazard statements and precautionary statements 186 2.11.6. Relation to transport classification ... ... .. 186 Examples of classification for self - heating substances and mixtures ... 187 2.11.7. 2.11.7.1. Examples of substances and mixtures fulfilling the classification criteria ... 187

17 Guidance on the Application of the CLP Criteria 5.0 – July 2017 17 Version . 187 Examples of substances and mixtures not fulfilling the classification criteria 2.11.7.2. 188 ... .. ... References ... 2.11.8. RES WHICH, IN CONTAC 2.12. WATER, EMIT SUBSTANCES AND MIXTU T WITH ... ... ... 189 FLAMMABLE GASES Introduction ... ... ... . 189 2.12.1. Definitions and general considerations for the classification of substances and 2.12.2. ... 189 mixtures which, in contact with water, emit flammable gases Relation to other physical hazards ... .. 189 ... 2.12.3. Classification of substances and mixtures which, in contact with water, emit 2.12.4. ... ... ... 190 flammable gases Identification of hazard information 2.12.4.1. ... ... 190 ... ... 191 2.12.4.2. Screening procedures and waiving of testing Classification criteria ... ... ... 2.12.4.3. 191 2.12.4.4. Testing and evaluation of hazard information ... ... 191 Testing procedure ... ... ... 191 2.12.4.4.1. Evaluation of hazard information ... ... 193 2.12.4.4.2. Decision logic ... ... ... 2.12.4.5. 193 water, 2.12.5. Hazard communication for substances and mixtures which, in contact with ... ... 195 emit flammable gases ... Pictograms, signal words, hazard statements and precautionary statements 2.12.5.1. and mixtures ... ... for substances .. 195 ... ... 195 2.12.5.2. Additional labelling provisions Relation to transport classification ... ... .. 196 2.12.6. Examples of classification for substances and mixtures which, in contact with water, 2.12.7. ... ... 196 emit flammable gases ... 2.12.7.1. fulfilling the classification criteria ... 196 Example of a substance 2.12.7.1.1. Example 1 ... ... ... 196 2.12.7.2. ... 197 Example of a substance not fulfilling the classification criteria Example 2 ... ... ... 197 2.12.7.2.1. References ... ... .. ... 197 2.12.8. OX IDISING LIQUIDS ... ... 2.13. 198 ... ... ... ... Introduction 198 2.13.1. . Definitions and general considerations for the classification of oxidising liquids ... 198 2.13.2. 2.13.3. ... ... .. 198 Relation to other physical hazards tion of substances and mixtures as oxidising liquids ... 199 2.13.4. Classifica Identification of hazard information ... ... 199 2.13.4.1. 2.13.4.1.1. ... ... 199 Screening procedures and waiving of testing 2.13.4.2. Classification criteria ... ... ... 200 2.13.4.3. Testing and evaluation of hazard information ... ... 201 2.13.4.4. ... ... ... 201 Decision logic ... ... 203 2.13.4.5. Hazard communication for oxidising liquids Pictograms, signal words, hazard statements and precautionary statements 2.13.4.5.1. ... ... ... 203 2.13.5. Relation to transport classification ... ... .. 203 2.13.6. Examples of classification for oxidising liquids ... ... 204 2.13.6.1. ... 204 Examples of substances and mixtures fulfilling the classification criteria 2.13.6.2. Examples of substances and mixtures not fulfilling the classification criteria . 204 2.13.7. Reference ... ... ... ... 204 2.14. OXIDISING SOLIDS ... ... ... 205 Introduction ... ... ... . 205 2.14.1.

18 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 18 2.14.2. ... 205 Definitions and general considerations for the classification of oxidising solids ... ... .. 206 14.3. Relation to other physical hazards 2. Classification of substances and mixtures as oxidising solids ... 2.14.4. 2 06 Identification of hazard information ... ... 206 2.14.4.1. Screening procedures and waiving of testing ... ... 206 2.14.4.1.1. cation criteria 2.14.4.2. ... ... ... 207 Classifi ... ... 209 2.14.4.3. Testing and evaluation of hazard information Decision logic ... ... ... 209 2.14.4.4. 14.4.5. Hazard communication for oxidising solids 2. ... ... 211 Pictograms, signal words, hazard statements and precautionary statements 2.14.4.5.1. ... ... 211 ... 2.14.5. Relation to transport classification ... ... .. 211 2.14.6. ... ... 212 Examples of classification for oxidising solids .1. Examples of substances and mixtures fulfilling the classification criteria ... 212 2.14.6 Examples of substances and mixtures not fulfilling the classification criteria . 212 2.14.6.2. Reference ... ... ... ... 212 2.14.7. ORGA NIC PEROXIDES ... ... ... 213 2.15. 2.15.1. Introduction ... ... ... . 213 2.15.2. .. 213 Definitions and general considerations for the classification of organic peroxides 2.15.3. ... ... .. 213 Relation to other physical hazards 2.15.4. Classification of substances and mixtures as organic peroxides ... 214 2.15.4.1. Identification of hazard information ... ... 214 2.15.4.2. Classification criteria ... ... ... 214 2.15.4.3. Testing and evaluation of hazard information ... ... 216 2.15.4.3.1. ... ... 216 Thermal stability tests and temperature control Additional considerations and testing 2.15.4.3.2. ... 217 ... Additional classification ... ... 217 2.15.4.3.3. considerations Decision logic ... ... 2.15.4.4. 218 ... ... ... 220 2.15.5. Hazard communication for organic peroxides Pictograms, signal words, hazard statements and precautionary statements 220 2.15.5.1. Additional labelling provisions for organic peroxides ... .. 221 2.15.5.2. 2.15.6. ... ... .. 221 Relation to transport classification Examples of classification for organic peroxides ... ... 221 2.15.7. Examples of substances and mixtures fulfilling the classification criteria ... 221 2.15.7.1. ... ... ... 224 Additional remarks 2.15.7.2. CORROSIVE TO METALS ... ... ... 225 2.16. Introduction ... ... 2.16.1. ... . 225 2.16.2 . Definitions and general considerations for the classification of substances and ... ... 226 mixtures corrosive to metals ... .. Relati ... ... 226 2.16.3. on to other physical hazards Classification of substances and mixtures as corrosive to metals ... 2.16.4. 226 2.16.4.1. Identification of hazard information ... ... 226 2.16.4.2. Screening procedures a nd waiving of testing ... ... 227 2.16.4.3. Classification criteria ... ... ... 227 2.16.4.4. ... ... 228 Testing and evaluation of hazard information 2.16.4.4.1. General considerations ... ... ... 228 2.16.4.4.2. Additional notes on best practice for testing ... ... 230 2.16.4.5. Decision logic ... ... ... 232 2.16.5. Hazard communication for substances and mixtures corrosive to metals ... 233 2.16.5.1. Pictograms, signal words, hazard statements and precautionary statements 233

19 Guidance on the Application of the CLP Criteria 5.0 – July 2017 19 Version ... ... .. 234 Relation to transport classification 2.16.6. 234 Examples of classification for substances and mixtures corrosive to metals ... 2.16.7. metal specimen plates after exposure to a corrosive mixture ... 235 2.16.7.1. Example of ... ... .. ... 235 References 2.16.8. PART 3: HEALTH HAZAR DS ... ... 236 3. ACUTE 3.1. ... ... ... 236 TOXICITY Definitions and general considerations for acute toxicity ... ... 236 3.1.1. Classification of substances for acute toxicity ... ... 236 3.1.2. Identification of hazard information ... ... 236 3.1.2.1. ... ... 236 3.1.2.1.1. Identification of human data Identification of non - human data ... ... 3.1.2.1.2. 237 3.1.2.2. Classification criteria ... ... ... 237 Harmonised ATE values ... ... ... 239 3.1.2.2.1. Minimum classification ... ... ... 239 3.1.2.2.2. Evaluation of hazard information ... ... 3.1.2.3. 239 ... ... Evaluation of human data 239 3.1.2.3.1. .. Evaluation of non - human data ... ... 240 3.1.2.3.2. 3.1.2.3.3. ... ... ... 243 Weight of evidence ... ... ... 243 3.1.2.4. Decision on classification Setting of specific concentration limits ... ... 243 3.1.2.5. 3.1.2.6. ... ... 243 Decision logic for classification of substances 3.1.3. Classification of mixtures for acute toxicity ... ... 245 3.1.3.1. General considerations for classi fication ... ... 245 3.1.3.2. ... ... 245 Identification of hazard information Classification criteria ... ... ... 245 3.1.3.3. When data are available for the complete mixture ... 245 3.1.3.3.1. When data are not available for the complete mixture: bridging principles 246 3.1.3.3.2. When data are available for all ingredients ... ... 246 3.1.3.3.3. ... ... 247 3.1.3.3.4. Special case for acute inhalation toxicity When data are not available for all ingredients ... ... 249 3.1.3.3.5. Ingredients that should be taken into account for the purpose of 3.1.3.3.6. ... ... 252 classification ... 252 Non classified components ... . - 3.1.3.3.7. ... Generic concentration limits for substances triggering classification of 3.1.3.4. mixtures ... ... 253 ... 3.1.3.5. Decision on classification ... ... ... 253 3.1.3.6. Decision logic for classification of mixtures ... ... 253 3. 1.4. Hazard communication in the form of labelling for acute toxicity ... 255 3.1.4.1. precautionary statements 255 Pictograms, signal words, hazard statements and ... ... 257 3.1.4.2. Additional labelling provisions Examples of classification for acute toxicity ... ... 260 3.1.5. 3.1.5.1. ... 26 0 Examples of substances fulfilling the criteria for classification 3.1.5.1.1. Example 1: Methanol ... ... ... 260 3.1.5.1.2. Example 2: N,N - Dimethylaniline ... ... 260 3.1.5.1.3. ... ... ... 261 Example 3 3.1.5.1.4. Example 4 ... ... ... 262 3.1.5.1.5. Example 5 ... ... ... 262 3.1.5.1.6. Example 6 ... ... ... 262 3.1.5. Example 7: 2,3 - Dichloropropene ... ... 263 1.7.

20 Guidance on the Application of the CLP Criteria 20 Version 5.0 – July 2017 3.1.5.1.8. ... ... ... 264 Example 8 Example 9 3.1.5.1.9. ... ... 264 ... fulfilling the criteria for classification ... 265 Examples of substances not 3.1.5.2. ... ... ... 3.1.5.2.1. Example 10 265 Example of mixtures fulfilling the criteria for classification ... 266 3.1.5.3. Example 11 ... ... ... 266 3.1.5.3.1. Example 12a 3.1.5.3.2. ... ... ... 267 ... 267 5.4. Examples of mixtures not fulfilling the criteria for classification 3.1. Example 12b ... ... ... 267 3.1.5.4.1. 3.1.5.5. Example of the application of the additivity method for mixtures for acute forms (gas, inhalation toxicity with ingredient substances in different physical vapour, mist or dust). ... ... 268 ... Example 13 ... ... ... 268 3.1.5.5.1. 3.1.6. ... ... ... .. 270 References SKIN CORROSION/IRRIT ... ... 271 3.2. ATION Definitions for classification for skin corrosion/irritation ... ... 271 3.2.1. Classification of substances for skin corrosion/irritation ... ... 3.2.2. 271 Identification of hazard information ... ... 271 3.2.2.1. Identification of human data ... ... 271 3.2.2.1.1. 3.2.2.1.2. ... ... 271 Identification of non human data - chemical properties ... 271 3.2.2.1.2.1. Consideration of physico pH and acid/alkaline reserve ... ... 272 3.2.2.1.2.2. Non - testing methods: (Q)SARs and expert systems ... 272 3.2.2.1.2.3. ... in vitro methods 3.2.2.1.2.4. ... 273 Testing methods: 3.2.2.1.2.5. Testing methods: In vivo data ... ... 274 3.2.2.2. ... ... ... 275 Classification criteria Evaluation of hazard information ... ... 276 3.2.2.3. Evaluation of human data ... ... .. 280 3.2.2.3.1. Evaluation of non human data ... ... 280 3.2.2.3.2. ... ... ... 280 3.2.2.3.2.1. In vitro data In vivo data ... ... ... 280 3.2.2.3.2.2. 2.2.3.3. 3. ... ... ... 282 Weight of evidence Decision on classification ... ... ... 284 3.2.2.4. Setting of specific concentration limits ... ... 284 3.2.2.5. ... ... 286 Decision logic for classification of substances 3.2.2.6. Classification of mixtures for skin cor rosion/irritation ... ... 287 3.2.3. 3.2.3.1. Identification of hazard information ... ... 287 3.2.3.2. ... ... 288 Classification criteria for mixtures ... 288 3.2.3.2.1. When data are available for the complete mixture Mixtures with extreme pH ... ... 288 3.2.3.2.1.1. 3.2.3.2.2. mixture: bridging principles 290 When data are not available for the complete 290 3.2.3.2.3. .. When data are available for all ingredients or only for some ingredients 3.2.3.2.3.1. Ingredients that should be taken into account for the purpose of classification ... ... ... 290 3.2.3.2.3.2. ... ... 290 The additivity approach is applicable 291 3.2.3.2.3.3. ... ... The additivity approach is not applicable 3.2.3.3. Generic concentration limits for substances triggering classification of mixtures ... ... ... 292 3.2.3.3.1. When the additivity approach is applicable ... ... 292 When the additivity approach is not applicable ... ... 293 3.2.3.3.2. 3.2.3.4. Decision logic for classification of mixtures ... ... 293

21 Guidance on the Application of the CLP Criteria 5.0 – July 2017 21 Version ... 296 Hazard communication in form of labelling for skin corrosion/irritation 3.2.4. Pictograms, signal words, hazard statements and precautionary statements 3.2.4.1. 296 ... ... 297 3.2.4.2. Additional labelling provisions ... ... 297 3.2.5. Examples of classification for skin corrosion/irritation ... 297 3.2.5.1. Examples of substances fulfilling the criteria for classification Example 1: Standard test according to OECD TG 404 with three animals . 297 3.2.5.1.1. Example 2: Test carried out with one animal with a test substance which is 3.2.5.1.2. ... 298 ... suspected as corrosive ... Example 3: ... 298 3.2.5.1.3. Test carried out with more than three animals Examples of mixtu res fulfilling the criteria for classification ... 299 3.2.5.2. ... 299 3.2.5.2.1. Example 4: Mixture without extreme pH, with ingredients with SCLs Example 5: Mixture without extreme pH, and non applicability of the 3.2.5.2.2. - ... ... 299 additivity approach ... 3.2.5.3. ... 300 Examples of mixtures not fulfilling the criteria for classification 3.2.5.3.1. Example 6: Mixture without extreme pH, with ingredients with SCLs ... 300 References ... ... ... .. 301 3.2.6. SERIOUS EYE DAMAGE/E ... ... 302 YE IRRITATION 3.3. Definitions for classification for serious eye damage/eye irritation ... 3.3.1. 302 Classification of substances fo r serious eye damage/eye irritation ... 302 3.3.2. Identification of hazard information ... ... 302 3.3.2.1. Identification of human data ... ... 302 3.3.2.1.1. ... ... 302 3.3.2.1.2. Identification of non human data Consideration of physico - chemical properties ... ... 302 3.3.2.1.3. 3.3.2.1.4. ... ... 302 pH and the acid/alkaline reserve 3.3.2.1.5. Non - testing methods: (Q)SARs and expert systems ... 303 3.3.2.1.5.1. Testing methods: in vitro methods ... ... 303 3.3.2.1.5.2. ... ... 304 Testing methods: In vivo methods Classification criteria ... ... ... 305 3.3.2.2. Evaluation of hazard information ... ... 306 3.3.2.3. Evaluation of human data ... ... .. 3.3.2.3.1. 309 - human data ... Evaluation of non 309 3.3.2.3.2. ... Ex vivo/ in vitro data 3.3.2.3.2.1. ... .. 309 ... 3.3.2.3.2.2. In vivo data ... ... ... 310 ... ... ... 312 3.3.2.3.3. Weight of evidence Decision on classification ... 3.3.2.4. ... 312 ... 3.3.2.5. Setting of specific concentration limits ... ... 312 3.3.2.6. Decision logic for classification of substances ... ... 314 3.3.3. eye damage/eye irritation ... 315 Classification of mixtures for serious ... ... 315 3.3.3.1. Identification of hazard information Classification criteria for mixtures ... ... 316 3.3.3.2. 3.3.3.2.1. ... 316 When data are available for the complete mixture 3.3.3.2.1.1. ... 317 Mixtures with extreme pH ... 3.3.3.2.2. When data are not available for the complete mixture: bridging principles 318 3.3.3.2.3. When data are available for all ingredients or only for some ingredients of the mixtu re ... ... ... 318 Ingredients that should be taken into account for the purpose of 3.3.3.2.3.1. ... classification ... 318 ... 3.3.3.2.3.2. The additivity approach is applicable ... ... 318 3.3.3.2.3.3. The additivity approach is not applicable ... ... 319 3.3.3.3. Generic concentration limits for substances triggering classification of mixtures ... ... ... 320

22 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 22 3.3.3.3.1. ... ... 320 When the additivity approach is applicable When the additivity approach is not applicable ... ... 321 3.3.3.3.2. Decision logic for classification of mixtures ... ... 321 3.3.3.4. Hazard communication in form of l .. 323 3.3.4. abelling for serious eye damage/eye irritation Pictograms, signal words, hazard statements and precautionary stateme nts 323 3.3.4.1. Examples of classification for serious eye damage/eye irritation ... 3.3.5. 324 3.3.5.1. Examples of substances fulfilling the criteria for classification ... 324 . 324 3.3.5.1.1. Example 1: Standard test according to OECD TG 405 with three animals Example 2: Test carried out with more than 3 rabbits ... 326 3.3.5.1.2. xamples of mixtures fulfilling the criteria for classification 3.3.5.2. ... 328 E Example 3: Application of the additivity approach for mixtures containing 3.3.5.2.1. ... ingredients without SCLs ... 328 ... 3.3.5.2.2. Example 4: Application of the additivity approach for mixtures containing ingredients which may have SCLs ... 329 ... 3.3.5.2.3. Example 5: Application of the additivity approach for mixtures containing ingredients which may have SCLs ... ... 329 References ... ... ... .. 330 3.3.6. 3.4. RESPIRATORY OR SKIN SENSITISATION ... ... 331 Definitions and general considerations for respiratory or skin sensitisation ... 331 3.4.1. 3.4.2. Classification of subst ances for sensitisation ... ... 331 3.4.2.1. ... 331 Classification of substances for respiratory sensitisation ... ... 3.4.2.1.1. 331 Identification of hazard information 3.4.2.1.1.1. Identification of human data ... ... 331 3.4.2.1.1.2. Identification of non human data ... ... 332 3.4.2.1.2. Classification criteria for substances ... ... 332 Evaluation of hazard information ... ... 333 3.4.2.1.3. 3.4.2.1.3.1. Human data ... ... ... 333 3.4.2.1.3.2. ... ... ... 334 Non human data Decision on classification ... ... ... 334 3.4.2.1.4. Setting of specific concentration limits ... ... 334 3.4.2.1.5. Decision logic for classification of substances ... ... 335 3.4.2.1.6. skin sensitisation ... ... 336 3.4.2.2. Classification of substances for Identification of hazard information ... ... 336 3.4.2.2.1. 3.4.2.2.1.1. ... ... 336 Identification of human data ... ... 336 3.4.2.2.1.2. Identification of non human data Classification criteria for substances ... ... 336 3.4.2.2.2. ... ... 338 Evaluation of hazard information 3.4.2.2.3. Human data ... ... ... 338 3.4.2.2.3.1. ... 2.2.3.2. Non human data ... 341 ... 3.4. 3.4.2.2.3.2.1. ... ... 343 Mouse Local Lymph Node Assay Guinea Pig Maximisation Test (GPMT, OECD TG 406) 343 3.4.2.2.3.3. ... Buehler assay (OECD TG 406) ... ... 344 3.4.2.2.3.4. Non - 3.4.2.2.3.5. ... ... 344 guideline skin sensitisation tests 3.4.2.2.3.6. Animal test methods conducted for purposes othe r than sensitisation ... ... 344 ... 3.4.2.2.3.7. Weight of evidence ... ... ... 344 3.4.2.2.4. ... ... ... 346 Decision on classification Setting of specific concentration limits ... ... 346 3.4.2.2.5. 3.4.2.2.6. Decision logic for classification of substances ... ... 349 3.4.3. Classification of mixtures for respiratory or skin sensitisation ... 350 3.4.3.1. Identification of hazard information for respiratory sensitisation ... 350 3.4.3.2. Identification of hazard information for skin sensitisation ... 350

23 Guidance on the Application of the CLP Criteria 5.0 – July 2017 23 Version ... ... 350 Classification criteria for mixtures 3.4.3.3. are available for all ingredients or only for some ingredients .. 351 When data 3.4.3.3.1. ... 353 3.4.3.3.2. When data are available for the complete mixture 353 3.4.3.3.3. When data are not available for the complete mixture: Bridging Principles ... Decision logic for classification of mixtures 353 3.4.3.4. ... Decision logic for classification of mixtures for respiratory sensitisation ... 354 3.4.3.4.1. H 3.4.4. ... 356 azard communication for respiratory or skin sensitisation Pictograms, signal words, hazard statements and precautiona 356 3.4.4.1. ry statements Additional labelling provisions ... ... 357 3.4.4.2. Examples of classification for skin sensitisation ... ... 357 3.4.5. Example of substances and mixtures fulfilling the criteria for classification for 3.4.5.1. ... ... skin sensitisation 357 ... 3.4.5.1.1. ... ... ... 357 Example 1 3.4.5.1.2. Example 2 ... ... ... 357 Example 3 ... ... ... 357 3.4.5.1.3. ... ... ... 1.4. 357 3.4.5. Example 4 Example 5 ... ... ... 358 3.4.5.1.5. Example 6 ... ... ... 358 3.4.5.1.6. Example 7 ... ... ... 358 3.4.5.1.7. ... 1.8. ... ... Example 8 358 3.4.5. Example of substances or mixtures not fulfilling the criteria for classification 3.4.5.2. on ... for skin sensitisati ... ... 359 3.4.5.2.1. Example 9 ... ... ... 359 3.4.5.2.2. Example 10 ... ... ... 359 3.4.5. 3. Examples of substances fulfilling the criteria for classification for respiratory ... ... 3 59 sensitisation ... Example 11 ... ... 359 3.4.5.3.1. ... Example 12 ... ... ... 359 3.4.5.3.2. References ... ... ... .. 360 3.4.6. GERM CELL MUTAGENICI TY ... ... ... 362 3.5. 3.5.1. 362 Definitions and general considerations for classification for germ cell mutagenicity 3.5.2. Classification of substances for germ cell mutagenicity ... ... 363 ... ... 363 3.5.2.1. Identification of hazard information Identification of human data ... ... 363 3.5.2.1.1. 3.5.2.1.2. ... ... 363 Identification of non human data 3.5.2.2. Classification criteria for substances ... ... 364 3.5.2.3. Evaluation of hazard information ... ... 365 3.5.2.3.1. ... ... .. 365 Evaluation of human data ... ... 365 3.5.2.3.2. Evaluation of non human data Decision on classification ... ... ... 365 3.5.2.4. Classification of substances containing CMR constituents, additives or 3.5.2.5. ... ... 367 impurities ... 3.5.2.6. Setting of specific concentration limits ... ... 368 3.5.2.7. Decision logic for classification of substances ... ... 369 3.5.3. ... ... 370 Classification of mixtures for germ cell mutagenicity 3.5.3.1. Classification criteria for mix tures ... ... 370 3.5.3.1.1. When data are available for the complete mixture ... 370 3.5.3.1.2. When data are not available for the complete mixture: bridging principles 370 3.5.3.2. Generic concentration limits for substances triggering classification of ... ... ... 370 mixtures

24 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 24 3.5.3.3. ... ... 371 Decision logic for classification of mixtures Hazard communication in form of l ... 374 3.5.4. abelling for germ cell mutagenicity Pictograms, signal words, hazard statements and precautionary statements 374 3.5.4.1. Additional labelling provisions ... ... 374 3.5.4.2. ... ... ... 376 3.6. CARCINOGENICITY Definitions and general considerations for classification for carcinogenicity ... 376 3.6.1. 3.6.2. Classification of substances for carcinogenicity ... ... 376 ntification of hazard information ... ... 376 3.6.2.1. Ide Classification criteria for substances ... ... 376 3.6.2.2. 3.6.2.3. ... ... 378 Evaluation of hazard information 3.6.2.3.1. Specific considerations for classification ... ... 378 3.6.2.3.2. Additional considerations for classification ... ... 380 3.6.2.3.3. ... ... 387 Consideration of mutagenicity ... ... ... 387 3.6.2.3.4. Non testing data Decision on classification ... ... ... 388 3.6.2.4. 3.6.2.5. Classification of substances containing CMR constituents ... 388 3.6.2.6. Setting of specific concentration limits ... ... 389 3.6.2. 7. Decision logic for classification of substances ... ... 389 3.6.3. ... ... 390 Classification of mixtures for carcinogenicity 3.6.3.1. Classification criteria for mixtures ... ... 390 3.6.3.1.1. When data are available for all ingredients or only for some ingredients .. 390 3.6.3.1.2. When data are available for the complete mixture ... 390 3.6.3.1.3. When data are not available for the complete mixture: bridging principles 391 3.6.3.2. Decision logic for classification of mixt ures ... ... 391 3.6.4. Hazard communication in form of labelling for carcinogenicity ... 393 3.6.4.1. 393 Pictograms, signal words, hazard statements and precautionary statements Additional labelling provisions ... ... 394 3.6.4.2. Some additional considerations for re - classification ... ... 394 3.6.4.3. Examples of classification for carcinogenic ... ... 394 3.6.5. ity .. ... ... ... 394 3.6.6. References REPRODUCTIVE TOXICIT Y ... ... ... 398 3.7. 3.7.1. ... 398 Definitions and general considerations for reproductive toxic ity Special considerations on effects on or via lactation ... .. 399 3.7.1.1. ... ... 399 3.7.2. Classification of substances for reproductive toxicity Identification of hazard information ... ... 399 3.7.2.1. ... ... 399 Identification of human data 3.7.2.1.1. Identification of non human data ... ... 3.7.2.1.2. 399 3.7.2.2. Classification criteria ... ... ... 399 3.7.2.2.1. ... .. 400 Classification in the presence of parental toxicity 400 3.7.2.2.1.1. Effects to be considered in the presence of marked systemic effects Relevance of specific effects in the parent ... ... 401 3.7.2.2.1.2. Substances causing effects on or via lactation ... ... 403 3.7.2.2.2. 3.7.2.3. Evaluation of hazard information ... ... 404 3.7.2.3.1. Use of data from standard repeat dose tests ... ... 404 3.7.2.3.2. ... ... ... 404 Study design Evaluation of evidence relating to effects on or via lactation ... 405 3.7.2.3.3. 3.7.2.4. Decision on classification ... ... ... 406 3.7.2.5. Classification of substances containing CMR constituents ... 406 Setting of specific concentration limits ... ... 407 3.7.2.6. 3.7.2.6.1. Procedure ... ... ... 407

25 Guidance on the Application of the CLP Criteria 5.0 – July 2017 25 Version ... 408 Cases where potency evaluation is difficult or unfeasible 3.7.2.6.2. 408 ... ... value Determination of the ED 3.7.2.6.3. 10 ... ... 408 3.7.2.6.3.1. Determination in practice Quantal or non - ... ... 409 3.7.2.6.3.2. parametric data data ... ... 409 3.7.2.6.3.3. Continuous or parametric Data combining incidence and magnitude 3.7.2.6.3.4. ... 410 ... ... ... ... 410 3.7.2.6.3.5. Specific data types Provisional evaluation of the potency classification ... 411 3.7.2.6.4. Modifying factors ... ... ... 411 3.7.2.6.5. Type of effect / severity ... ... 412 3.7.2.6.5.1. ... ... ... 412 3.7.2.6.5.2. Data availability Dose - response relationship ... 3.7.2.6.5.3. 413 ... 3.7.2.6.5.4. Mode or mechanism of action ... ... 41 3 Toxicokinetics ... ... ... 413 3.7.2.6.5.5. Bio ... ... accumulation of substances 413 3.7.2.6.5.6. - Assigning specific concentration limits (SCLs) ... ... 414 3.7.2.6.6. Assigning two SCLs to a substance ... ... 415 3.7.2.6.6.1. ... ... 416 3.7.2.7. Decision logic for classification of substances Classification of mixtures for reproductive toxicity ... ... 3.7.3. 417 3.7.3.1. Classificatio n criteria for mixtures ... ... 417 idual ingredients ... 418 3.7.3.1.1. When data are available for the indiv When data are available for the complete mixture ... 418 3.7.3.1.2. 418 3.7.3.1.3. When data are not available for the complete mixture: bridging principles 419 ... ... 3.7.3.2. Decision logic for classification of mixtures 3.7.4. Hazard communication in form of labelling for reproductive toxicity ... 422 3.7.4.1. Pictograms, signal words, hazard statements and precautionary statements 422 3.7.4.2. ... ... 424 Additional labelling provisions Examples ... ... ... ... 425 3.7.5. Examples of the determination of SCLs ... ... 425 3.7.5.1. Example 1 ... ... ... 425 3.7.5.1.1. Example 2 (developmental part only) ... ... 426 3.7.5.1.2. ... ... 429 3.7.5.1.3. Example 3 (limited to developmental toxicity) Example 4 ... ... ... 431 3.7.5.1.4. 3.8. SPECIFIC TARGET ORGA N TOXICITY – SINGLE EXPOSURE (STO T - SE) ... 433 ... SE ... - 433 3.8.1. Definitions and general considerations for STOT Classification of substances for STOT - SE ... ... 434 3.8.2. 3.8.2.1. Identification of hazard information ... ... 434 Identification of human data ... ... 434 3.8.2.1.1. 3.8.2.1.2. Identification of non human data ... ... 434 3.8.2.2. ... ... 435 Classification criteria for Categories 1 and 2 Guidance values ... ... ... 437 3.8.2.2.1. Classification criteria for Category 3: Transient target organ effects ... 438 3.8.2.3. Evaluation of hazard information on STOT - SE for substances 3.8.2.4. 439 ... 3.8.2.4.1. Evaluation of human data ... ... .. 439 Evaluation of non human data ... ... 441 3.8.2.4.2. 3.8.2.4.3. Evaluation of non - testing and in vitro data ... ... 443 3.8.2.4.4. ... ... ... 443 Conversions 3.8.2.4.5. Weight of evidence ... ... ... 443 3.8.2.5. Decision on classification of substances ... ... 444 3.8.2.6. Setting of specific concentration limits for STOT - SE ... .. 444 3.8.2.7. Decision logic for classification of substances ... ... 446

26 Guidance on the Application of the CLP Criteria 26 Version 5.0 – July 2017 3.8.3. for STOT - SE ... ... 448 Classification of mixtures Identification of hazard information ... 448 3.8.3.1. ... ... 448 Classification criteria for mixtures ... 3.8.3.2. ... 3.8.3.2.1. When data are available for the complete mixture 448 When data are not available for the complete mixture: bridging principles 449 3.8.3.2.2. 3.8.3.2.3. When data are available for all ingredients or only for some ingredients of ... ... ... 449 the mixture Components of a mixture that should be taken into account for the purpose 3.8.3.2.4. ... of classification 449 ... ... nces triggering classification of Generic concentration limits for substa 3.8.3.3. SE ... ... ... 449 mixtures for STOT - Decision logic for classification of mixtures 3.8.3.4. ... 451 ... ... Hazard communication in form of labelling for STOT SE ... 454 3.8.4. - Pictograms, signal words, hazard statements and precautionary statements 454 3.8.4.1. Additional labelling provisions ... ... 455 3.8.4.2. Examples of classification for STOT SE ... ... 455 3.8.5. - Examples of substances fulfilling the criteria for classification ... 455 3.8.5.1. Example 1: Methanol ... ... ... 3.8.5.1.1. 455 Example 2: Tricresyl phosphate ... ... 456 3.8.5.1.2. Example 3: Sulfur dioxide ... ... .. 456 3.8.5.1.3. 3.8.5.1.4. ... ... ... 457 Example 4: Toluene ... 457 3.8.5.2. Examples of substances not fulfilling the criteria for classification Example 5: ABC ... ... ... 457 3.8.5.2.1. Dimethylaniline - 3.8.5.2.2. ... ... 458 Example 6: N,N - N TOXICITY – REPEATED EXPOSURE (S TOT 3.9. RE) ... 459 SPECIFIC TARGET ORGA 3.9.1. Definitions and general considerations for STOT - RE ... ... 459 3.9.2. RE ... ... 460 Classification of substances for STOT - Identification of hazard information ... ... 460 3.9.2.1. Identification of human data ... ... 460 3.9.2.1.1. Identification of non human data ... ... 460 3.9.2.1.2. ... ... 461 3.9.2.2. Classification criteria for substances Evaluation of hazard information ... ... 464 3.9.2.3. 3.9.2.3.1. ... ... .. 465 Evaluation of human data Evaluation of non human data ... ... 465 3.9.2.3.2. Conversions ... ... ... 467 3.9.2.3.3. ... 2.3.4. ... ... Weight of evidence 468 3.9. Decision on classification ... ... ... 469 3.9.2.4. ... Additional considerations ... ... 3.9.2.5. 470 3.9.2.5.1. ... ... 470 Irritating/corrosive substances ... ... ... 470 3.9.2.5.2. Hematotoxicity Mechanisms not relevant to humans (CLP Annex I, 3.9.2.8.1. (e)) ... 473 3.9.2.5.3. 3.9.2.5.4. ... . 474 Adaptive responses (CLP Annex I, 3.9.2.8.1. (d)) 3.9.2.5.5. Post - observation periods in 28 day and 90 day studies ... 474 3.9.2.6. Setting of specific concentration limits ... ... 474 3.9.2.7. ... ... 476 Decision logic for classification of substances ... - RE 3.9.3. ... 477 Classification of mixtures for STOT 3.9.3.1. Identification of hazard information ... ... 477 3.9.3.2. Classification criteria for mixtures ... ... 477 When data are available for the complete mixture ... ... 477 3.9.3.3. 3.9.3.3.1. When data are not available for the complete mixture: bridging principles 477

27 Guidance on the Application of the CLP Criteria 5.0 – July 2017 27 Version When data are available for all ingredients or only for some ingredients of 3.9.3.3.2. ... ... 477 ... the mixture Components of a mixture that should be taken into account for the purpose 3.9.3.3.3. 478 of classification ... ... ... 3.9.3.4. Generic concentration limits for substances triggering classification of ... mixtures 478 ... ... Decision logic for classification of mixtures ... 478 3.9.3.5. ... Hazard communication in form of labelling for STOT - RE ... ... 3.9.4. 480 statements 480 3.9.4.1. Pictograms, signal words, hazard statements and precautionary Additional labelling provisions ... ... 481 3.9.4.2. Examples of classification for STOT - RE ... ... 481 3.9.5. Examples of s 3.9.5.1. ... 481 ubstances fulfilling the criteria for classification Example 1: Hydroxylamine / Hydroxylamonium salts (CAS no. 7803 - 49 - 8) 3.9.5.1.1. ... ... 481 ... 3.9.5.1.2. Example 2: But - 2 - yn - 1,4 - diol (EC No 203 - 788 - 6; CAS No 110 - 65 - 6) ... 483 Example 3: XYZ ... ... ... 485 3.9.5.1.3. Examples of substances not fulfilling the criteria for classification ... 487 3.9.5.2. Chlorinated Paraffins) = Alkanes, C , 3.9.5.2.1. Example 4: MCCPs (Medium Chain 14 - 17 - (EC No 287 - 477 - 0; CAS No 85535 - 85 - 9) ... .. 487 Chloro 3.9.5.3. Examples of mixtures fulfilli ng the criteria for classification ... 489 ... ... ... 489 3.9.5.3.1. Example 5 Example 6 ... ... 3.9.5.3.2. 489 ... 3.9.5.3.3. Example 7 ... ... 489 ... 3.9.5. 3.4. Example 8 ... ... ... 490 3.9.5.4. ... 490 Example of mixtures not fulfilling the criteria for classification Example 9 ... ... ... 490 3.9.5.4.1. References ... ... .. ... 491 3.9.6. PART 4: ENVIRONMENTA L HAZARDS ... 4. 492 ... HAZARDOUS TO THE AQU ATIC ENVIRONMENT ... ... 492 4.1. Introduction ... ... ... . 492 4.1.1. 4.1.2. Scope ... ... ... ... 492 ... 493 4.1.3. Classification of substances hazardous to the aquatic environment Information applicable for classification of substances hazardous to the 4.1.3.1. ... ... ... 493 aquatic environment 4.1.3.1.1. ... ... 493 Substance properties used for classification 4.1.3.1.2. Information and data availability ... ... 493 4.1.3.2. Evaluation of available information ... ... 494 4.1.3.2.1. ... ... ... 494 General considerations 494 ... ... . 4.1.3.2.2. Substances difficult to test Interpretation of data for aquatic toxicity, degradation and bioaccumulation 4.1.3.2.3. 496 4.1.3.2.3.1. ... ... ... 496 Aquatic toxicity 4.1.3.2.3.2. Degradation ... ... ... 497 4.1.3.2.3.3. Bioaccumulation ... ... ... 500 4.1.3.2.4. ... 501 Using weight of evidence in evaluations in the context of C&L 4.1.3.2.4.1. General aspects of weight of evidence ... ... 501 4.1.3.2.4.2. Guidance on WoE for data deficient substances ... 502 4.1.3.2.4.3. Guidance on WoE for substances for which more than one valid piece of data is available for a given data element ... 502 Outliers ... ... ... 503 4.1.3.2.4.4.

28 Guidance on the Application of the CLP Criteria 28 Version 5.0 – July 2017 4.1.3.2.4.5. ... ... 503 Weight of evidence in degradation Weight of evidence in bioaccumulation ... 503 4.1.3.2.4.6. ... ... ... 503 4.1.3.3. Classification categories and criteria Outline of the core classification system ... ... 503 4.1.3.3.1. The ‘safety net’ ... ... ... 507 4.1.3.3.2. Setting an M ... factor for highly toxic substances ... 508 4.1.3.3.3. - Decision on classification: examples for substances ... .. 509 4.1.3.4. Example A: Hydrophilic substance, straightforward classification based on 4.1.3.4.1. ... acute and chronic toxicity data 512 ... Example B: Hydrophilic substance, straightforward classification based on 4.1.3.4.2. ... 514 ... acute data, no chronic data available 4.1.3.4.3. Example C: Moderately water soluble substance, straightforward classification based on acute data, chronic data avail able for two trophic ... levels only; combined set of QSAR data and experimental data 517 toxicity data for a trophic level Example D: Substance with several 520 4.1.3.4.4. ... Example E: ‘Safety net’ classification category Chronic 4 4.1.3.4.5. 523 ... 4.1.3.4.6. Example F: Substance difficult to test, toxicity above level of water ... ... 525 solubility ... Classification of mixtures hazardous to the aquatic environment ... 4.1.4. 528 4.1.4.1. General considerations for classification of mixtures hazardous to the aquatic ... environment ... 528 ... 4.1.4.2. Information requirements ... ... ... 529 4.1.4.3. he aquatic environment Classification criteria for mixtures hazardous to t ... 530 based on test data on the mixture as a whole ... When experimental aquatic toxicity data are not a 4.1.4.4. vailable for the complete ... ... ... 532 mixture: bridging principles When hazard data (information on toxicity or classification) 4.1.4.5. are available for ... 532 all the components of the mixture ... When hazard data (information on toxicity or classification) are available for 4.1.4.6. only some components of the mixture ... 537 ... ... ... 538 4.1.4.7. Decision on classification: examples for mixtures Example A: When classification data are available for some or all 4.1.4.7.1. components of a mixture: straightforward application of the summa tion 539 ... ... method ... 4.1.4.7.2. Example B1: When toxicity data on the mixture as a whole is available for all three trophic levels: cla ssification based on test data for the mixture .. 541 4.1.4.7.3. components is Example B2: When information on the classification of the available and toxicity data on the mixture as a whole is available for some, but not all three trophic levels: use of the summation method 543 ... Example C: When no data are available on the mixture as a whole and its 4.1.4.7.4. components, but test data are available on a similar tested mixture: use of – dilution with water the bridging principles ... ... 545 4.1.4.7.5. Example D: When test data are available for some, but not all components of the mixture: use of the additivity formula and of the summati on method 546 4.1.5. Metal and metal compounds ... ... ... 550 4.1.6. ... 550 Hazard communication for hazards to the aquatic environment 4.1.7. Re - classification of substances and mixtures classified as hazardous to the aquatic environment according to DSD/DPD ... ... 552 ... 4.1.8. References ... ... .. 553

29 Guidance on the Application of the CLP Criteria 5.0 – July 2017 29 Version ... ... 554 PART 5: ADDITIONAL H 5. AZARDS NE LAYER ... ... 554 HAZARDOUS TO THE OZO 5.1. ... ... 555 ANNEXES ... ANNEX I: AQUATIC TOX ... ... ICITY 555 I ... Introduction ... ... ... . 555 I.1 Description of tests ... ... ... 555 I.2 ... ... ... 556 2.1 I. Fish tests Acute testing ... ... ... 556 I.2.1.1 Chronic testing ... ... ... 556 I.2.1.2 ... 2 ... ... Tests with Crustaceae 556 I.2. Acute testing ... ... ... I.2.2.1 556 I.2.2.2 Chronic testing ... ... ... 556 Algae / other aquatic plant tests ... ... 556 I.2.3 5 Tests with algae ... ... ... 56 I.2.3.1 Tests with aquatic macrophytes ... ... I.2.3.2 557 ... ... Aquatic toxicity concepts 557 I.3 ... Acute toxicity ... ... ... 557 I.3.1 Chronic toxicity ... ... ... 558 I.3.2 ... ... ... 559 I.3.3 Exposure regimes Test media for algae and Lemna ... ... 559 I.3.4 - I.3.5 across and grouping) and (Q)SARs for Use of substance categorisation (read ... ... 559 classification and labelling ... I.4 bstances which are difficult to test ... ... 559 Su I.4.1 Unstable substances ... ... ... 560 I.4.2 ... ... ... 560 Poorly soluble substances Other factors contributing to concentration loss ... ... 561 I.4.3 Perturbation of the test media ... ... 561 I.4.4 Complex substances ... ... ... I.4.5 562 ... ... ... References 562 I.5 .. ANNEX II: RAPID DEGR ADATION ... ... 563 II II.1 ... ... ... . 563 Introduction ... ... 563 II.2 Interpretation of degradability data Ready biodegradability ... ... ... 563 II.2.1 II.2.1.1 ... ... 564 Concentration of test substance II.2.1.2 Time window ... ... ... 564 II.2.2 BOD /COD ... ... ... 564 5 II.2.3 Other convincing scientific evidence ... ... 564 ... ... ... 565 II.2.3.1 Aquatic simulation tests Field investigations ... ... ... 565 II.2.3.2 II.2.3.3 ... ... ... 565 Monitoring data .2.3.4 Inherent and Enhanced Ready Biodegradability tests ... 566 II II.2.3.5 Sewage treatment plant simulation tests ... ... 566 II.2.3.6 ... ... 566 Soil and sediment degradation data ... ... II.2.3.7 566 Anaerobic degradation data II.2.3.8 Hydrolysis ... ... ... 566 II.2.3.9 Photochemical degradation ... ... 567 II.2.3.10 Estimation of degradation ... ... .. 567 Volatilisation ... ... ... 567 II.2.3.11

30 Guidance on the Application of the CLP Criteria 30 Version 5.0 – July 2017 II.2.4 ... ... 568 No degradation data available General interpretation problems ... ... 568 II.3 ... ... ... 568 Complex substances ... II.3.1 ... ... ... II.3.2 568 Availability of the substance Test duration less than 28 days ... ... 568 II.3.3 Primary biodegradation ... ... ... 569 II.3.4 Conflicting results from screening tests ... ... 569 II.3.5 Variation in simulation test data ... ... 570 II.3.6 Decision scheme ... ... ... 570 II.4 References ... ... ... .. 571 II.5 ANNEX II I: BIOACCUMULATION ... ... .. 572 III Introduction ... ... ... . III.1 572 III.2 Interpretation of bioconcentration data ... ... 572 Bioconcentration factor (BCF) ... ... . 573 III.2.1 BCF in different test species ... ... 573 III.2.1.1 Use of radio - labelled substances ... ... 574 III.2.1.2 - water - partitioning coefficient (K III.2.2 ) Octanol ... ... 574 ow Experimental determination of Kow ... ... III.2.2.1 575 III.2.2.2 Use of QSARs for determination of log Kow ... ... 575 .. values III.3 575 Chemical classes that need special attention with respect to BCF and K ow Substances difficult to test ... ... ... 576 III.3.1 III.3.2 Poorly soluble and complex substances ... ... 576 High molecular weight substances ... ... 576 III.3.3 577 III.3.4 Surface - active substances (surfactants) ... ... III.3.4.1 water - partition coefficient (K Octanol ) ... ... 577 - ow ... ... ... 577 III.4 Conflicting data and lack of data Conflicting BCF data ... ... ... 577 III.4.1 ... Conflicting log K data ... ... 578 III.4.2 ow Expert judgement ... ... ... 578 III.4.3 Decision scheme ... ... ... 578 III.5 III.6 ... ... ... .. 579 References ANNEX IV: METALS AND INORGANIC METAL COMP OUNDS ... 580 IV Introduction ... ... ... . 580 IV.1 ... 582 Application of aquatic toxicity data and solubility data for classification IV.2 Interpretation of aquatic toxicity data ... ... 582 IV.2.1 Metal complexation and speciation ... ... IV.2.1.1 584 IV.2.2 ... ... 584 Interpretation of solubility data Assessment of existing data ... ... 584 IV.2.2.1 ning T/D test for assessing solubility of metal compounds ... 585 IV.2.2.2 Scree Full T/D test for assessing solubility of metals and metal compounds IV.2.2.3 585 ... IV.2.3 Comparison of aquatic toxicity data and solubility data ... 586 IV.3 Assessment of environmental transformation ... ... 586 IV.4 Bioaccumulation ... ... ... 587 IV.5 ... .. 588 Classification strategies for metals and metal compounds IV.5.1 Introduction ... ... ... 588 IV.5.2 Classificat ion strategies for metals ... ... 588 IV.5.2.1 Classification strategy for determining acute aquatic hazard for metals ... 588 IV.5.2.2 Classification strategy for determining long - term aquatic hazard for metals ... ... ... 589 IV.5.2.2.1 Approach based on available chronic toxicity reference data ... 589

31 Guidance on the Application of the CLP Criteria 5.0 – July 2017 31 Version ... ... ... 590 The surrogate approach IV.5.2.2.2 ... 593 Classification strategies for metal compounds IV.5.3 ... IV.5.3.1 Classification strategies for determining acute aquatic hazard for metal ... ... 59 3 compounds ... - term aquatic hazard for metal Classification strategy for determining long IV.5.3.2 ... ... ... 594 compounds ... 595 IV.5.3.2.1 Approach based on available chronic toxicity reference data The surrogate approach ... ... ... 596 IV.5.3.2.2 Setting M - factors for metals and inorganic metal comp ounds IV.5.4 599 ... ... ... Particle size and surface area 600 IV.5.5 . IV.5.6 Classification of mixtures of metals and metal compounds ... 602 IV.5.6.1 Classification of alloys and complex metal containing materials ... 602 IV.6 ... ... ... .. 603 References ... 604 ompounds IV.7 Decision on classification: examples for metals and metal c Example A: Soluble metal compound with acute and chronic toxicity data and IV.7.1 transformation (Me 605 (SO4) ). ... no evidence of rapid environmental 2 2 Example B: Poorly soluble metal compound with acute and chronic toxicity IV.7.2 tion/dissolution data at 7 days (low loading rate) and at 28 data, transforma days (only low and medium loading rates) and no evidence of rapid environmental transformation ... 608 ... Example C: Metal in powder and massive form with acute and chronic toxicity IV.7.3 data and Transformation/Dissolution data at 7 days (low, medium and high loading rates) and at 28 days (only the high loading rate) and no evid ence of ... ... rapid environmental transformation 612 IV.7.3.1 Explanatory note to Example C - Critical Surface Area (CSA) approach 617 ... IV.7.4 Example D: Hazard classification of a soluble metal salt: the case of rapid environmental transformation through speci ation in the water column ... 619 V OF INTERNET LINKS FO R THE USERS OF THE G UIDANCE ANNEX V: COLLECTION ... ... ... 623 ... ANNEX VI: BACKGROUND DOCUMENT TO THE GUID ANCE FOR SETTING VI SPECIFIC CONCENTRATI ON LIMITS FOR SUBSTA NCES CLASSIFIED FOR REGULATION (EC) NO 1 272/2008 624 REPRODUCTIVE TOXICIT Y ACCORDING TO ... ... Executive summary 624 VI.1 ... Introduction ... ... ... . 625 VI.2 VI.2.1 General description of the classification system for reprotoxic substances and ... ... 625 mixtures ... Description of the process for the development of a method to set SCLs for VI.2.2 ... ... 626 reproductive toxic substances VI.2.3 Considering potency in setting specific concentration limits for various health hazards ... ... ... 627 VI.2.4 Parameters for potency for reproductive toxicity ... ... 628 VI.2.4.1 Potency parameters for developmental toxicants (Muller et al , 2012) ... 628 VI.2.4.2 Potency parameters for substances with an adverse effect on sexual et al , 2012) ... 630 function and fertility (Muller ... usions on the most appropriate parameter for potency ... 632 VI.2.4.3 Concl VI.3 ... ... ... 633 Modifying factors VI.3.1 Boundaries of the potency groups ... ... 633 VI.4 Non - modifying factors ... ... ... 634 VI.4.1 Species and strains ... ... ... 634 VI.4.2 Systemic or maternal toxicity ... ... . 634 VI.4.3 Mutagenicity ... ... ... 634

32 Guidance on the Application of the CLP Criteria 32 Version 5.0 – July 2017 VI.4.4 ... ... ... 634 Volatility Poten cy groups and specific concentration limits ... ... 635 VI.5 Justification of the proposed potency boundaries and specific concentration VI.5.1 limits 635 General considerations on potency groups ... ... 635 VI.5.1.1 ... VI.5.1.1.1 ... ... 635 Legal requirements Scientific results of the database analysis ... ... 635 VI.5.1.1.2 VI.5.1.1.3 Policy related considerations and proposed method ... 636 VI.5.1.1.4 ... ... . 636 Other methods considered ... VI.5.1.2 637 Justification of the boundaries between the three potency groups VI.5.1.3 Concentration limits for Category 1 and Category 2 substances ... 641 VI.5.2 Assigning SCLs ... ... ... 642 ... VI.6 ... ... References .. 642 VII ANNEX VII: RELATION BETWEEN TRANSPORT AN D CLP CLASSIFICATION REGARDING PHYSICAL H AZARDS ... ... 643

33 Guidance on the Application of the CLP Criteria 5.0 – July 2017 33 Version of Tables Table Possibilities for setting SCL for health hazards addressed in relevant Table 1.1 ... ... sections of the guidance ... 63 ... ... ... Table 1.2 79 Ingredients in Mixture A Ingredient 'Fragrance mixture' ... ... 80 Table 1.3 Ingredients in Mixture B ... Table 1.4 ... 82 ... ... ... ... 83 Table 1.5 Ingredients ‘base powder’ Hazard classes where the translation tables in Annex VII to CLP indicate Table 1.6 ... 84 that no direct translation was possible from DSD to CLP Examples of hazards, depending on the property of the emitted gas, when Table 2.1 ... 189 substances and mixtures are in contact with water Minimum mass loss of specimens after different exposure times Table 2.2 ... (corresponding to the criterion of 6.25 mm/year) 229 .. Table 2.3 Minimum intrusion depths after exposure times (corresponding to the ... 229 criterion of localized corrosion of 6.25 mm/year) ... Examples of classified and non classified substances and mixtures in Class Table 2.4 ... ... ... . 234 2.16 Types of Human Studies ... ... ... 339 Table 3.1 Relatively high or low frequency of occurrence of skin sensitisation* ... 339 Table 3.2 Table 3.3 ... ... 340 Relatively high or low exposure * ... categorisation decision table ... - 340 Table 3.4 Sub Definition of significant skin sensitising effect Table 3.5 ... 342 ... Table 3.6 Skin Sensitisation Potency in the Mouse Local Lymph Node Assay ... 346 Table 3.7 Potency on basis of the Guinea Pig Maximisation Test ... 347 Table 3.8 ... ... 347 Potency on basis of the Buehler assay Skin sensitising potency for sub Table 3.9 stances and recommendations on ... ... 348 concentration limits ... Example of the calculation of the ED ... ... Table 3.10 409 10 Example on the calculation of the ED Table 3.11 ... 409 ... 10 Example on the calculation of the ED Table 3.12 ... 410 for testicular effects (N=10) 10 ... ... 411 Table 3.13 Boundaries of the potency groups. SCLs for substances in each potency group and classification category ... Table 3.14 415 Table 3.15 Hazard statements for reproductive toxicity: H360 and H361, and their ... specifications ... 423 ... Table 3.16 Equivalent guidance values for 28 - day and 90 - day studies ... 464 Table 3.17 ... ... ... 467 Food conversion ... ... .. 467 Table 3.18 Conversion drinking water Inclusion of route of exposure in Hazard statement ... Table 3.19 470 Table 4.1 Hazard statement Codes relevant for the hazard class Hazardous to the ... Aquatic Environment ... 552 ... Table III. 1 Examples of software programs for the estimation of log K ... 575 ow ... - factors for inorganic substances Table IV. 1 M ... 599 Table IV. 2 Acute toxicity data deemed reliable for ‘Metal’ are presented as mg/l Me .. 621 Table IV. 3 Chronic toxicity data deemed reliable for ‘Metal’ are presented as mg/l Me 621 Table VI. 1 Average values (assuming log/normal distribution) (in mg/kg bw/day) and potency differences for parameters for all developmental toxicants of the database (Muller et al, 2012) ... ... 629

34 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 34 Table VI. 2 Average values (assuming log/normal distribution) (in mg/kg bw/day) for parameters for developmental toxicants and potency differences (N=44) with all 6 parameters (Muller et al, 2012) .. 629 ... Average values (assuming log/normal distribution) (in mg/kg bw/day) Table VI. 3 and potency differences for parameters for all fertility toxicants of the ... ... ... 630 database Table VI. 4 Average values (assuming log/normal distribution) (in mg/kg bw/day) and potency differences for parameters for fertility toxicants (N=34) with ... ... 632 all 6 parameters ... Boundaries of the potency groups ... ... Table VI. 5 633 Table VI. 6 Percentages of substances in the three potency groups using the ED and 10 some of the modifying factors for different boundaries of the potency gro ups and considering the saturated vapour concentration of low potency substances ... ... ... 638 Table VI. 7 three potency groups using the ED Percentages of substances in the and 10 some of the modifying factors but not volatility for different borders of the ... ... ... 640 potency groups ... 641 Table VI. 8 SCLs for substances in each potency group and classification category Tabel VII. 1 Relation between transport and CLP classifications regarding physical hazards ... ... ... 643 Figures Table of Figure 1.1 How to classify a mixture ... ... ... 66 Figure 1.2 Application of the bridging principle: dilution for determining the acute toxicity classification of a mixture ... ... 69 Figure 1.3 Application of the bridging principle: interpolation for determining the aquatic acute hazard classification of a mixture ... ... 70 Figure 1.4 Application of the bridging principle: substantially similar mixtures for ... 71 determining the skin irritation classification of a mixture Decision logic for oxidising gases (Decision logic 2.4 of GHS) ... 131 Figure 2.1 Decision logic for gases under pressure (Decision logic 2.5 of GHS) ... 137 Figure 2.2 Amended GHS decision logic for flammable liquids to include derogations Figure 2.3 for gas oil, diesel, light heating, sustained combustibility and for phrases ... ... 145 EUH018, EUH209 and EUH209A ... 152 Decision logic for flammable solids (Decision logic 2.7 of GHS) Figure 2.4 Decision logic 2.8 for self - reactive substances and mixtures ... 162 Figure 2.5 ... .6 Decision logic for self - reactive substance example: NP, technically pure 167 Figure 2 ... Decision logic for pyrophoric liquids (Decision logic 2.9 of GHS) 170 Figure 2.7 Figure 2.8 ... 176 Decision logic for pyrophoric solids (Decision logic 2.10 of GHS) ... ... 185 Figure 2.9 Extrapolation towards large volumes Figure 2.10 temperature for charcoal ... Volume dependency of the critical 188 Figure 2.11 Decision logic for substances and mixtures which, in contact with water, emit flammable gases (Decision logic 2.12 of GHS) ... 194 Figure 2.12 Decision logic for oxidising liquids (Decision logic 2.13 of GHS) ... 202 Figure 2.13 Decision logic for oxidising solids (Decision logic 2.14 of GHS) ... 210 Figure 2.14 Decision logic 2.15 for organic peroxides ... ... 219 Figure 2.15 Potential pH (also called Pourbaix) diagram for iron in water at 25 °C, indicating stable form of the Fe element and implicitly, corrosion domains 225

35 Guidance on the Application of the CLP Criteria 5.0 – July 2017 35 Version Example of testing equipment available on the market to perform UN Test Figure 2.16 ... ... C.1 231 ... .. Decision logic for substances and mixtures corrosive to metals (Decision Figure 2.17 ... ... ... 232 logic 2.16 of GHS) Example of corroded metal plates after testing according to UN Test C.1 Figure 2.18 ... ... for a classified mixture 235 ... Tiered evaluation for skin corrosion/skin irritation ... .. 277 Figure 3.1 Figure 3.2 Simplified illustration of the relative weight of the available information ... 283 Mixture without human or animal data on skin corrosion/irritation or Figure 3.3 ...  relevant data from similar tested mixtures, pH is 2 or 289  11.5 Tiered evaluation for serious eye damage/eye irritation ... Figure 3.4 307 Figure 3.5 Mixture not classified as Skin Corr. 1 and without animal or human data on serious eye damage/eye irritation or relevant data from similar tested 2 or  mixtures, pH is 11.5  ... ... 317 Figure 3.6 Procedure for setting SCL for reproductive toxicity ... . 407 Figure 3.7 .. 442 Comparison between the NOAEL and the ED versus the guidance values .. 466 Figure 3.8 Comparison between the NOAEL and the ED versus the guidance values Classification strategy for determining acute aquatic hazard for metals ... 589 Figure IV. 1 Classification strategy for determining long - term aquatic hazard for Figure IV. 2 591 ... ... metals ... Cla ssification strategy for determining long - term aquatic hazard for Figure IV. 3 metals in absence of appropriate chronic toxicity reference and/or T/Dp data ... ... ... . 592 Figure IV. 4 Classification strategy for determining acute aquatic hazard for metal ... ... compounds ... 594 Figure IV. 5 Classification strategy for determining long - term aquatic hazard for metal compounds ... ... ... 597 Figure IV. 6 Classification strategy for determining long - term aquatic hazard for metal compounds in absence of appropriate chronic toxicity reference and/or T/Dp data ... ... ... 598

36 Guidance on the Application of the CLP Criteria 36 Version 5.0 – July 2017 LIST OF ABBREVIATIONS Standard term / Explanation Abbreviation 2 ADD Directive 75/324/EEC on aerosol dispensers European Agreement concerning the International Carriage of ADN Dangerous Goods by Inland Waterways ( Accord européen relatif au transport international des marchandises dangereuses par voie de 3 navigation intérieure ) European Agreement concerning the International Carriage of ADR Accord européen relatif au transport Dangerous Goods by Road ( 4 des marchandises dangereuses par route ) international Ammonium Nitrate Emulsion ANE ASTM American Society for the Testing of Materials Acute Toxicity Estimate ATE ATP Adaptation to Technical Progress (ATP) to the CLP Regulation BAM prüfung (Federal Institute Bundesanstalt für Materialforschung und - for Materials Research and Testing) Bioconcentration Factor BCF Bovine Corneal Opacity and Permeability test BCOP German Federal Institute for Risk Assessment BfR BfR DSS German Federal Institute for Risk Decision support system by the Assessment Biomagnification factor BMF BOD Biological Oxygen Demand BP Boiling point bw Body weight 2 Directive (75/324/EEC) of the Council on the approximation of the laws of the Member States relating to aerosol dispensers [OJ L 147, 9.6.1975, p.40]. Directive as last amended by Commission Directive 2013/10/EU [ OJ L 77, 20.03.2013, p.20]. 3 European Agreement concerning the International Carriage of Dangerous Goods by Inland Waterways, concluded at Geneva on 26 May 2000, as amended. 4 European Agreement concerning the Internatio nal Carriage of Dangerous Goods by Road, concluded at Geneva on 30 September 1957, as amended.

37 Guidance on the Application of the CLP Criteria 5.0 – July 2017 37 Version Explanation Standard term / Abbreviation Classification and Labelling C&L Competent Authority CA Converted Acute Toxicity point Estimate cATpE Regulation (EC) No 1272/2008 on classification, labelling and CLP 5 packaging of substances and mixtures CNS Central Nervous System Chemical Oxygen Demand COD Chemical Safety Assessment CSA Chemical Safety Report CSR DIN s Institut für Normung ( German Institute for Deutsche Standardisation ) DNA Deoxyribonucleic Acid DOC Dissolved Organic Carbon DPD Directive 1999/45/EC on the classification and labelling of 6 Dangerous Preparations Directive 67/548/EEC on the classification and labelling of DSD 7 Dangerous Substances EC3 Effective Concentration inducting a stimulation index of 3 in the LLNA test ECHA European Chemicals Agency, Helsinki ( https://echa.europa.eu/ ) European Centre for the Validation of Alternative Methods ECVAM ( http://ihcp.jrc.ec.europa.eu/our_labs/eurl - ecvam ) ED Effective Dose 5 Regulation (EC) No 1272/2008 of the European Parliament and Council of 16 December 2008 on classification, labelling and packaging of substances and mixtures, amending and repealing Directives 67/548/EEC and 1999/45/EC and amending Regulation (EC) No 1907/2006 [OJ L 353, 31.12.2008, p. 1]. 6 Directive 1999/45/EC of the European Parliament and of the Council of 31 May 1999 concerning the approximation of the laws , regulations and administrative provisions of the Member States relating to the classification, packaging and labelling of dangerous preparations [OJ L 200, 30.7.1999, p. 1]. 7 Council Directive 67/548/EEC of 27 June 1967 on the approximation of laws, reg ulations and administrative provisions relating to the classification, packaging and labelling of dangerous substances [OJ 196, 16.8.1967, p. 1].

38 Guidance on the Application of the CLP Criteria 38 Version 5.0 – July 2017 Standard term / Explanation Abbreviation A European Standard EN Ecotoxicity Reference Value ERV ECVAM Scientific Advisory Committee ESAC https://eurl - ecvam.jrc.ec.europa.eu/about - ecvam ) ( EUH The hazard statements carried through from DSD and DPD, which ‘ EUH ’ are not yet included in the GHS are codified as f/F Female FP Flash point General Concentration Limits GCL Globally Harmonised System of Classification and Labelling of GHS 8 Chemicals Gap junction intercellular communication GJIC GLP Good Laboratory Practice GnRH Gonadotropin - releasing hormone GPMT Guinea Pig Maximisation Test Guidance Value GV Haemoglobin Hb - CAM Hen's Egg Test on Chorio - allantoic Membrane HET or H HS ( Hazard statement statement) HSM Human skin model Ht Hematocrit IARC International Agency for Research on Cancer ( http://www.iarc.fr/ ) IATA DGR International Air Transport Association , Dangerous Goods Regulations Manual IBC Intermediate Bulk Container 8 Globally Harmonised System of Classification and Labelling of Chemicals (GHS), Fifth revised edition, United Nations, New York and Geneva, 2013.

39 Guidance on the Application of the CLP Criteria 5.0 – July 2017 39 Version Explanation Standard term / Abbreviation International Civil Aviation Organization ( Technical Instructions for ICAO TI the Safe Transport of Dangerous Goods by Air) ICE Isolated Chicken Eye http://www.iec.ch/ ) IEC International Electrotechnical Commission ( IMDG International Maritime Dangerous Goods Code Code IMO International maritime Organisation IPCS International Programme on Chemical Safety (joint programme of WHO, ILO and UNEP) & CSA Guidance on Information Requirements and Chemical Safety IR Assessment, ECHA ( http://guidance.echa.europa.eu/docs/guidance_document/informa ) tion_requirements_en.htm E Isolated Rabbit Eye IR ISO International Organisation for Standardization 9 ITDG Directive 2008/68 on the Inland Transport of Dangerous Goods Integrated Testing Strategy ITS T - octanol/water partition coefficient K he n ow LEL Lower Explosion Limit LD /LC Median (50%) lethal dose/concentration 50 50 LFL Lower Flammability Limit LLNA Local Lymph Node Assay LO (A) EL/C Lowest Observed (Adverse) Effect Level/Concentration Low Volume Eye Test LVET m/M Male MetHB Methaemoglobinaemia 9 Directive 2008/68/EC of the European Parliament and of the Council of 24 September 2008 on the inland transport of dangerous goods, implementing the European Agreement concerning the International Carriage of Dangerou s Goods by Road (ADR), the Regulations concerning the International Carriage of Dangerous Goods by Rail (RID) and the European Agreement concerning the International Carriage of Dangerous Goods by Inland Waterways (ADN) [OJ L 260, 30.9.2008, p. 13].

40 Guidance on the Application of the CLP Criteria 40 Version 5.0 – July 2017 Standard term / Explanation Abbreviation Methaemoglobin MetHb factor Multiplying factor M - Melting Point MP Member State Competent Authority MSCA MTD Maximal Tolerated Dose MW Molecular weight n.a. Not available NC No Classification NE Narcotic effect(s) NO(A)EC No Observed (Adverse) Effect Concentration NO(A)EL No Observed (Adverse) Effect Level ODS Ozone Depleting Substances Ozone Depleting Potential ODP OECD Organisation for Economic Co - operation and Development OECD Test Guideline OECD TG All Test Guidelines are available at the OECD homepage: http://www.oecd.org/document/40/0,3343,en_2649_34377_37051 368_1_1_1_1,00.html OP Oxidising Power P statement Precautionary statement (or PS) PB/PK Physiologically - based pharmacokinetic PPARα Peroxisome proliferator - activated receptor - alpha PS (or P Precautionary statement statement) (Q)SAR (Quantitative) Structure Activity Relationship

41 Guidance on the Application of the CLP Criteria 5.0 Version July 2017 41 – Explanation Standard term / Abbreviation Regulation (EC) No 1907/2006 of the European Parliament and of REACH the Council concerning the Registration, Evaluation, Authorisation 10 and Restriction of Chemicals RID Règlement concernant le transport international ferroviaire de marchandises dangereuses (Regulations concerning the 11 International Carriage of Dang erous Goods by Rail) RIP REACH Implementation Project RTI Respiratory tract irritation Self - Accelerating Decomposition Temperature SADT SCL Specific Concentration Limit Safety Data Sheet SDS SIFT Skin integrity function test SSD Species Sensitivity Distribution STOT - Specific Target Organ Toxicity - Single Exposure SE - Specific Target Organ Toxicity - Repeated Exposure STOT RE SVC Saturated Vapour Concentration The daily dose (in mg/kg bodyweight/day) inducing a tumour T25 incidence of 25 % upon lifetime exposure T95 Inhalation chamber equilibrium (attained at the time t95) T/D Transformation/Dissolution T/Dp Transformation/Dissolution Protocol Transcutaneous electrical resistance TER 10 Regu lation (EC) No 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), establishing and repealing Council Regulation (EEC) No a European Chemicals Agency, amending Directive 1999/45/EC 793/93 and Commission Regulation (EC) No 1488/94 as well as Council Directive 76/769/EEC and omission of Directives 91/155/EEC, 93/67/EEC, 93/105/EC and 2000/21/EC. [OJ L 396, 30.12.2006 p.1.] [Corrigendum: OJ L 13 6, 29.5.2007 p.3]. 11 Regulations concerning the International Carriage of Dangerous Goods by Rail, appearing as Appendix C to the Convention concerning International Carriage by Rail (COTIF) concluded at Vilnius on 3 June 1999, as amended.

42 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 42 Standard term / Explanation Abbreviation Test Guideline TG TGD Technical Guidance Document TM Test Method as listed in the Test Methods Regulation Test Methods pursuant Regulation (EC) No 440/2008 laying down test methods 12 to the REACH Regulation Regulation TOPKAT Mathematical (Q)SAR model for prediction of skin corrosion/irritation UDP Uridine 5' - diphosphate UDPG Uridine diphosphate glucuronyl UEL Upper Explosion Limit UFL Upper Flammability Limit UGT - glucuronyltransferase UDP UN United Nations - MTC The UN Manual of Tests and Criteria contains criteria, test methods UN and procedures to be used for classification of dangerous goods according to the provisions of Parts 2 and 3 of the United Nations l Recommendations on the Transport of Dangerous Goods, Mode Regulations, as well as of chemicals presenting physical hazards according to the Globally Harmonized System of Classification and Labelling of Chemicals (GHS). More information and the latest revision are available at: http://www.unece.org/trans/danger/publi/manual/manual_e.html . UN RTDG Model - UN Recommendations on the Transport of Dangerous Goods It covers all modal transport regulations (ADR, Model Regulations. Regulations RID, ADN, IMDG and ITDG). It is regularly updated and amended every two years. More information and the latest revision are available at: http://www.unece.org/trans/danger/publi/unrec/rev13/13nature_e. html UNSCEGHS (or United Nations SubCommittee of Experts on the Globally SCEGHS) Harmonised System 12 Council Regulation (EC) No 440/2008 of 30 May 2008 laying down test methods pursuant to Regulation (EC) No 1907/2006 of the European Parliament and of the Council on the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) [OJ L 142, 31.5.2 008, p. 1] [Corrigendum: OJ L 143, 3.6.2008, p. 55].

43 Guidance on the Application of the CLP Criteria 5.0 Version July 2017 43 – Explanation Standard term / Abbreviation ( http://www.unece.org/trans/danger/publi/ghs/ghs_welcome_e.ht ml ) United Nations SubCommittee of Experts on the Transport of UNSCETDG (or Dangerous Goods SCETDG) http://www.unece.org/trans/danger/danger.htm ( ) - FHSA United States Federal Hazardous Substance Act - US 40 Code of Federal Regulations 1500.41 UVCB Substances of unknown or variable composition, complex reaction products or biological materials VDI Verein Deutscher Ingenieure (The Association of German Engineers) VP Vapour Pressure WAF Water Accommodated Fraction WoE Weight of Evidence WSF Water soluble fraction NOTE s to the reader: this document , text cited from Regulation (EC) No 1272/2008 is indicated in green In boxes in italic font . This symbol highlights text to be noted.

44 Guidance on the Application of the CLP Criteria 44 Version 5.0 – July 2017 1. GENERAL PRINCIPLES F OR CLASSIFICATION AN D PART 1: LABELLING 1.1. INTRODUCTION The objective of the guidance document 1.1.1. is a comprehensive technical and scientific guidance This document on the application of classification, labelling and packaging of substances and Regulation (EC) No 1272/2008 on the 13 mixtures , hereafter referred to as CLP. 14 the Dangero us Substance Directive 67/548/EEC CLP (DSD), the Dangerous amended 15 16 Preparations Directive 1999/45/EC (DPD) and Regulation (EC) No 1907/2006 (REACH), and based on the eal DSD and DPD from 1 June 2015 (CLP Article 61). CLP wa s implemented d rep e Globall y Harmonised System of Classification and Labelling of Chemicals (UN United Nations’ GHS) without lowering the protection of human health and the environment, compared to the The implementation of GHS into classification, labelling and packaging system in DSD and DPD. CLP made by the Community to confirm its intention to contribute followed various declarations EU law. to GHS development and to implement GHS into - A core principle of CLP is self classification of a substance or mixture by the manufacturer, importer or down stream user (CLP Article 4(3) and Recital 17), which involves identification of the hazards of the substance or mixture followed by classification as a result of the comparison - of the hazard information with the criteria in CLP. This guidance will enable i ndustry to self classify chemicals and to provide appropriate hazard communication information to the target populations potentially to it . For substances of handling the substance or mixture or exposed toxic for reproduction (CMRs) and particular concern (carcinogens, mutagens, substances wide action is needed, CLP sets out a - respiratory sensitisers) or for other substances where EU system for formal harmonisation of classifications at EU level. cation, the implementation of Given that many provisions under REACH are linked to classifi REACH and CLP is interlinked and should be planned and applied in tandem. General advice on ECHA ’s Introductory Guidance on the CLP the implementation of CLP is available in the , available ECHA website ( http://echa.europa.eu/web/guest/guidance - on the Regulation - on - clp ). documents/guidance to provide his document is The objective of t detailed guidance on the application of the CLP criteria for ph ysical, health and environmental hazards. 13 Regulation (EC) No 1272/2008 of the European Parliament and of the Council on classification, labelling and packaging of substances and mixtures, amending and repealing Directives 67/548/EEC and 1999/4 . 5/EC, and amending Regulation (EC) No 1907/2006 [OJ L 353, 31.12.2008, p. 1] 14 Council Directive 67/548/EEC relating to the classification, packaging and labelling of dangerous substances, as amended [OJ 196, 16.8.1967, p. 1] . 15 Directive 1999/45/EC as of 30 July 2002 of the European Parliament and of the Council relating to the classification, packaging and labelling of dangerous preparation, as amended [OJ L 200, 30.7.1999, p.1] . 16 Regulation (EC) No 1907/2006 of the European Parliament and of the Counci l concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), establishing a European Chemicals Agency, amending Directive 1999/45/EC and repealing Council Regulation (EEC) No 793/93 and Commission Regulation (EC) No 1488/9 4 as well as Council Directive 76/769/EEC and omission of Directives 91/155/EEC, 93/67/EEC, 93/105/EC and 2000/21/EC. [OJ L 396, 30.12.2006 p.1.] [Corrigendum: OJ L 136, 29.5.2007 p.3] .

45 Guidance on the Application of the CLP Criteria 5.0 Version 45 – July 2017 Background 1.1.2. The aim of classification and labelling is to identify the hazardous properties of a substance or a mixture by applying specific classification criteria to the available hazard data, and then to provide appropriate hazard labelling and information on safety measures. The EU has had a comprehensive system for the classification and labelling of dangerous substances and mixtures for over 40 years, in the past mainly DSD and DPD. In addition, the 17 required suppliers to provide more detailed ety Data Sheet (SDS) Directive 91/155/EEC Saf information for professional users. These directives contributed to a single market in chemicals safety and h and the environment. healt in the EU, based on a high level of protection of human The GHS was developed worldwide to minimise differences between systems of different jurisdictions for classification and labelling of substances and mixtures. The GHS aims to contribute towards global efforts to provide protection f rom hazardous effects of chemicals and to facilitate trade. were developed taking into and mixtures The GHS criteria for classifying hazardous substances account existing systems for hazard classification, such as the EU supply and use system, the 18 19 Canadian hazard evaluation procedure, IMO and US Pesticide systems, GESAMP Scheme for Marine Pollutants, the ( UN / RTGD ), UN Recommendations on the Transport of Dangerous Goods , the and the US Land Transport. These systems include supply and subsequent use of chemicals sea transport of chemical substances as well as transport of chemical substances by road and rail. The harmonised criteria are therefore intended to identify hazardous chemicals in a common way for use throughout all these systems. The GHS provides a basis for an internationally uniform information system on hazardous substances and mixtures. It provides harmonised criteria for classification and hazard communication measures for different target audiences, including consumers, worke rs and ‘ building block ’ emergency responders, and in transport. It follows a approach to enable jurisdictions to adopt the system according to the needs of their law and the various target classification although the final aim of GHS is to have a full y harmonised audiences. However, system worldwide , it is recognised that differences may persist between sectors ( and labelling e.g. transport , supply and use), but should not occur within a sector globally (section 1.1.3.1.5, th UNSCEGHS, 6 revision) . was agreed by the UN Committee of Experts on the Transport of Dangerous Goods and The GHS the Globally Harmonized System of Classification and Labelling of Chemicals (CETDG/GHS). It was formally approved by the UN Economic and Social Council (UN ECOSOC) in July 2 003 and published further in 2003 after a decade of negotiations. It is updated biannually. The changes not authomatically reflected in t he CLP Regulation . The latter is adapted and in GHS are by the Commission via Adaptations to Technical Progress updated - see Article 53(1) of (ATPs CLP). Hazard classification 1.1.3. Hazard classification is a process involving the identification of information on the physical, health , environmental or other hazards of a substace or a mixture as set out in Annex I to CLP . This followed by the comparison of th e hazard information (including the severity of hazard ) is with defined criteria , in order to determine the classification of the substance or mixture. T hus , 17 Council Directive 91/155/EEC relating to defining and laying down t he detailed arrangements for the system of specific information relating to dangerous preparations and dangerous substances, as amended [OJ L 076, 22.03.1991, p. 35], repealed and replaced by Regulation (EC) No 1907/2006 as of 1 June 2007. 18 Group of Exper ts on the Scientific Aspects of Marine Environmental Protection. 19 International Maritime Organisation.

46 Guidance on the Application of the CLP Criteria 46 Version 5.0 – July 2017 u nder CLP, a manufacturer, importer or downstream user will apply the following steps to arrive classification of a substance or a mixture: at a self -  (including identification of relevant available information regarding the potential hazards severity of hazard ) of a substance or mixture; thered to assess whether it is relevant, reliable and  examination of the information ga sufficient for classification purposes; evaluation  by applying the classification criteria in Annex I, CLP of the information (data) ; and for each hazard class and differentiation t decision on whether hazard information for the he substance or mixture meets the  one or more hazard class es or differentiation s criteria for and therefore decision on the classif the substance or mixture as hazardous in relation to th ese hazard ication of (s) tiation assign ment of hazar d categories , SCL ( , M - factor (s) and classes or differen s ( s ) hazard statement I, CLP ) . according to the provisions in Annex Preliminary information on identification of relevant data is provided in section of this 1.1.6 is provided in Part B of the ECHA guidance document, while guidance on available test methods Guidance document on (Chapters Information Requirements and Chemical Safety Assessment R.2 to R.4, IR & CSA), available on the ECHA Website s ( http://echa.europa.eu/web/guest/guidance - documents/guidance - on - information - requirement - and - safety - assessment ). Chapters R.7a/b/c of the same Guidance provide more chemical - specific guidance. - detailed information and endpoint Classification according to CLP is based on intrinsic hazards, i.e. the basic properties of a or mixture substance as determined in standard tests or by other means designed to identify It should be noted that for some hazard classes the intrinsic properties of a substance hazards. always the only aspects relevant for classi fication, e.g. explosives or aerosols or mixture are not which classification is also dependent , or aspiration hazard which may not be package for As CLP is hazard - based, it does not take exposure into relevant for certain package types. classification . I t should further be noted that classification o consideration in arriving at a f substances and mixtures may be required even when placed on the market in forms that are not hazardous. E.g. ym ers or metals in massive form, alloys, mixtures containing pol , shoul d be classified according to the criteria for e.g. toxic effects by inhalation but elastomers not need to be labelled. may 1.1.4. Who is responsible for the hazard classification CLP and REACH place the responsibility for hazard classification and related provisions such as packaging, hazard communication and SDS on the suppliers of substances and mixtures. Both substances and mixtures must be classified , labelled and packaged in accordance with CLP before placing them on the market . 1.1.5. Which substances and mixtures should be c lassified Substances and mixtures placed on the market fall within the scope of classification under CLP and should be evaluated in order to reach a decision as to whether or not the criteria are met and therefore should be classified. Substances are also if they subject to classification where they are subject to registration or notification under REACH, even if they are not placed on the market. However, a number of substances and mixtures are exempted from the requirements of the CLP Regulati on as a whole (CLP Article 1):

47 Guidance on the Application of the CLP Criteria 5.0 Version July 2017 47 – 20 ); radioactive substances and mixtures (Directive 96/29/Euroatom   substances and mixtures which are subject to customs supervision, provided that they do not undergo any treatment or processing, and which are in temporary stor age, or in a exportation, or in transit; free zone or free warehouse with a view to re - non  isolated intermediates; -  substances and mixtures used in scientific experimentation, analysis or chemical research, provided they are not placed on the market and the y are used under controlled conditions in accordance with EU workplace and environmental legislation; 21 waste, as defined in Directive 2006/12/EC ; and  certain substances or mixtures in the finished state, intended for the final user:  22 medicinal products, as ,  defined in Directive 2001/83/EC 23 veterinary medicinal products, as defined in Directive 2001/82/EC ,  24  , cosmetic products, as defined in Directive 76/768/EEC 25  (active implantable medical medical devices as defined in Directive 90/385/EEC 26 devices) and 93/42/EEC (medical devices in general), which are invasive or used in direct physical contact with the human body, and in vitro diagnostic 27 ), and medical devices (Directive 98/79/EC 28  ed in Regulation 178/2002 , including when they food or feeding stuffs as defin 29 , as a are used as food additives within the scope of Directive 89/107/EEC flavouring in foodstuffs within the scope of Directive 88/388/EEC and Decision 20 Council Directive 96/29/Euratom of 13 May 1996 laying down basic safety standards for the protection [OJ L against the dangers arising from ionizing radiation of the health of workers and the general public 159, 29.6.1996, p. 1]. 21 Directive 2006/12/EC of the European Parliament and of the Council of 5 April 2006 on waste [OJ L 114, . 27.4.2006, p. 9] 22 Directive 2001/83/EC of the European Parliament an d of the Council of 6 November 2001 on the . Community code relating to medicinal products for human use [OJ L 311, 28.11.2001, p. 67] 23 Directive 2001/82/EC of the European Parliament and of the Council of 6 November 2001 on the Community code relating to veterinary medicinal products [OJ L 311, 28.11.2001, p. 1] . 24 Council Directive 76/768/EEC of 27 July 1976 on the approximation of the laws of the Member States . relating to cosmetic products [OJ L 262, 27.9.1976, p. 169] 25 Council Directive 90/385/EEC of 20 June 1990 on the approximation of the laws of the Member States . relating to active implantable medical devices [OJ L 189, 20.7.1990, p. 17] 26 Council Directive 93/42/EEC of 14 June 1993 concerning medical devices [OJ L 169, 12.7.1993, p. 1] . 27 Directiv e 98/79/EC of the European Parliament and of the Council of 27 October 1998 on in vitro diagnostic medical devices [OJ L 331, 7.12.1998, p. 1] . 28 Regulation (EC) No 178/2002 of the European Parliament and of the Council of 28 January 2002 laying down the g eneral principles and requirements of food law, establishing the European Food Safety Authority and laying down procedures in matters of food safety [OJ L 31, 1.2.2002, p. 1] . 29 Council Directive 89/107/EEC of 21 December 1988 on the approximation of the l aws of the Member States concerning food additives authorized for use in foodstuffs intended for human consumption [OJ L 40, 11.2.1989, p. 27] .

48 Guidance on the Application of the CLP Criteria 48 Version 5.0 – July 2017 30 the scope of Regulation 1999/217/EC , as an additive in feeding stuffs within 31 , and in animal nutrition within the scope of Directive (EC) 1831/2003 32 . 82/471/EEC In addition, Member States may exempt certain substances or mixtures in specific cases where necessary for the purpose of national defence. h CLP does not apply to the transport of dangerous goods by air, sea, road, rail or inland Althoug waterways (CLP Article 1(6)), the criteria for classification are normally intended to be the same in the two systems. Thus, a substance or mixture classified in a h azard class which is common to both CLP and the transport legislation will normally be classified the same in both systems. However, the transport classifications do not include all of the GHS categories, so the absence ot mean the substance or mixture should not be classified of a transport classification does n . The relation between transport and CLP classification regarding physical hazards is under CLP to this document . detailed in Annex VII 1.1.6. What information is needed for classification Information for the classification of substances 1.1.6.1. The classification of a substance is based on the relevant information available on its hazardous properties. This information can include experimental data generated in tests for physical hazards, toxicological and ecotoxi cological tests, historical human data such as accident records or epidemiological studies, or information generated in in vitro tests, (Quantitative) Structure Activity Relationships ((Q)SAR), ‘ read - across ’ , or grouping approaches. CLP does not require new testing for the purpose of classification for health or environmental hazards; testing for physical hazards is required unless adequate and reliable information is n the market for research already available (CLP Article 8(2)). However, a substance placed o and development (R&D) purposes may have been manufactured or imported in quantities that are too small to perform physical hazard testing. In these cases it would not be proportionate to request the respective manufacturer, import er or downstream user to perform the tests required in Part 2 of Annex I to CLP. Although data may be provided through the application of REACH, it should be recognised that the data set required by REACH (particularly at lower tonnages) will not necessar ily enable the nformation may also be available from other comparison with the criteria for all hazard classes. I EU legislation for which there are specific requirements for test data to be generated, such as 33 legislation on plant protection products (Regul and Directive ation (EC) No 1107/2009 30 1999/217/EC: Commission Decision of 23 February 1999 adopting a register of flavouring substances used in or o n foodstuffs drawn up in application of Regulation (EC) No 2232/96 of the European Parliament and of the Council of 28 October 1996 [OJ L 84, 27.3.1999, p. 1] . 31 Regulation (EC) No 1831/2003 of the European Parliament and of the Council of 22 September 200 3 on additives for use in animal nutrition [OJ L 268, 18.10.2003, p. 29] . 32 Council Directive 82/471/EEC of 30 June 1982 concerning certain products used in animal nutrition [OJ L 213, 21.7.1982, p. 8] . 33 Regulation (EC) No 1107/2009 of the European Parliament and of the Council of 21 October 2009 concerning the placing of plant protection products on the market repeals Council Directives 79/117/EEC and 91/414/EEC with effect from 14 June 2011. However Article 80 of Regulation (EC) No 1107/2009 specif ies that directive 91/414/EEC shall continue to apply with respect to active substances included in Annex I to that Directive for certain transitional periods.

49 Guidance on the Application of the CLP Criteria 5.0 – July 2017 49 Version 34 35 and Directive ) and on biocidal products ( Regulation (EU) No 528/2012 91/414/EEC 36 - EU programmes. Finally, the supplier may decide to conduct ), or from various non 98/8/EC new testing in order to fill data gaps, provi ded that he has exhausted all other means of generating information. Testing on animals must be avoided wherever possible and alternative methods (including testing, the use of (Q)SARs, read - across and/or grouping in vitro approaches) must always be consid ered first, provided they are scientifically validated, sufficiently adequate and reliable. case of the substance containing impurities, additives or other constituents, the In a of the substance should , similar to classification mixtures, pre ferably be b ased on available information (including test data) on the except when classifying for CMR properties or substance when evaluating the bioaccumulation and degradation properties within the ‘hazardous to the aquatic environment’ hazard class (referred to in sections 4.1. 3.3.2 and 4.1.2.9 of Annex I to . I n such cases i t is strongly recommended that the classification of the substa nce , similar CLP) and 10 A rticle s 6(3) , 6(4) to mixtures of CLP) , is based on information of known CMR ( constituent ( s ) as there is no toxicological difference between a mixture and a substance 37 containing other constituent substances In exceptional cases, data on the substance itself . and/ or might show relevant effects for classification for CMR bioaccumulation or degradation which have not been identified from the information on the constituent s tances. ubs properties These data should then be used, if available . If, for the purpose of CLP, it is required or decided to generate new data, certain test methods and quality conditions must be met. Studies must be conducted in accordance with the EU test 38 440/2008) methods validated according or other international test methods (Regulation (EC) must be to international procedures such as those of the OECD. For physical hazards new tests a relevant recognised quality system or by laboratories carried out in compliance with complying with a relevant recognised standard, and for health and environmental hazards in 39 compliance with the principles of Good Laboratory Practice (GLP ). Animal tests must comply 40 . Tests on non - human primates are prohibited for the purposes with the Directive 86/609/EEC must of CLP. Tests on humans not be performed for the purpose of CLP. However, existing data obtained from other sources, such as accident records and epidemiological and clinical studies, can be used. 34 ts on the Council Directive 91/414/EEC of 15 July 1991 concerning the placing of plant protection produc . market, as amended [OJ L 230, 19.8.91, p. 1] 35 Regulation (EU) No 528/2012 of the European Parliament and of the Council of 22 May 2012 concerning the making available on the market and use of biocidal products . It should be noted that with effe ct from 1 September 2013, Biocidal Products Regulation (EU) No 528/2012 repealed Directive 98/8/EC. 36 Directive 98/8/EC of the European Parliament and of the Council of 16 February 1998 concerning the placing of biocidal products on the market, as amended [OJ L 123, 24.4.98, p. 1] . 37 Please note that there is a case still pending before the Court of Justice on the classification of an UVCB substance based on information on its constituents: Case C - 691/15 P. 38 test methods pursuant to Regulation (EC) No Council Regulation (EC) No 440/2008 laying down 1907/2006 of the European Parliament and of the Council concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH)[OJ L 142, 31.5.2008, p. 1]. 39 More information on the GLP princi ples and related requirements is available in the Q&As section on the ECHA website at https://www.echa.europa.eu/web/guest/support/qas - support/qas . 40 Directive 86/609/EEC regarding the protection of animals used for experimental and other scientific purposes, [OJ L 358, 18.12.1986, p. 1].

50 Guidance on the Application of the CLP Criteria 50 Version 5.0 – July 2017 1.1.6.2. Information relevant for the classification of mixtures For mixtures, classificati on for physical hazards should normally be based on the results of tests carried out on the mixtures themselves (unless, as for substances, a mixture placed on the o market for R&D purposes has been manufactured or imported in quantities that are too small t perform physical hazard testing). New tests for physical hazards must be carried out in compliance with a relevant recognised quality system or by laboratories complying with a . relevant recognised standard ds, the classification should preferably be When considering health and environmental hazar except when based on information (including test data) on the mixture itself, if available, classifying for e.g. CMR effects or when evaluating the bioaccumulation and degradation properties within the ‘hazardous to the aquatic environment’ hazard class referred to in sections , classification of the mixtures must be 4.1.2.8 and 4.1.2.9 of Annex I to CLP. In these cases based on the information on the substances. for the purpose of classification and labe lling for health or environmental hazards of New tests may only be performed when the manufacturer, importer or substances and mixtures, of generating information according to A rticle downstream user has exhausted all other means of CLP . According to this article, this includes application of the general rules provided in 8 section 1 of Annex XI to REACH to possible alternative methods/approaches to which refers use existing data, w animal testing of a substance t he when required in REACH eigh t of , i.e. , provided they are e vidence, (Q)S ARs, in vitro , grouping of substances and read - across considered adequate for the purpose of classification and labelling (and . In the case of mixtures it has to be re assured that the method is relevant , reliable - multiconstituent substances) and (see specific guidance for each hazard class) . for the mixture f no in vivo test data are available on a mixture, such data should normally not be Thus, i 41 of the mixture should be u sed generated; rather, all available information on the ingredients to derive a classification . ‘ bridging principles ’ which enables suppliers to derive health or Annex I to CLP specifies environmental classifications of their mixtures based on available data on similar tested mixtures and on the ingredient substances. Annex I also provides specific rules for the classification of mixtures based on the classification of the individual substances in the mixture. 1.1.7. Data evaluation and reaching a decision on classification 1.1.7.1. Classification of substances After the available information has been assembled, a systematic evaluation of this information is nec essary in order to derive a classification. The information must be compared with the criteria for classification for each hazard class or differentiation within the hazard class. Differentiation is a distinction depending on the route of exposure or the n ature of the effects. A decision should be made as to whether the substance meets the criteria for classification. When this is the case; the classifier should assign one or more hazard categories for each relevant hazard class or differentiation. The subs tance is then assigned the appropriate hazard communication elements. In some cases the classification decision may be straightforward, requiring only an evaluation of directly whether the substance gave a positive or negative result in a specific test that can be compared with the classification criteria. In other cases, scientific judgements must be made (e.g. on dose - response relationships, equivocal results and non - standardised tests) in a weight of evidence determination when applying the criteria . Exp ert judgement may therefore be 41 Note that the term “ingredient” is used in this guidance with the same meaning of “component” to indicate a substance in amixture.

51 Guidance on the Application of the CLP Criteria 5.0 – July 2017 51 Version or the available information in a Weight needed to decide whether the results of a particular test meet the criteria laid down in Annex I. of evidence assessment 1.1.7.2. Influence of impurities, additives or individual constituents on the classification of a substance Substances may contain impurities, additives, or other constituents while still meeting the substance definition in CLP. This applies to both mono - constituent (e.g. - constituent, multi es. The classification of such impurities, additives or reaction masses) and UVCB substanc individual constituents may influence the classification of the substance, in addition to the other hazardous properties. I f data on the substance with its components are not available (or for , s CMRs c tion 1.1.6.1 ), i n principle, the same classification and labelling rules as for ee se 42 mixtures should apply also for such . substances 1.1.8. Updating of hazard classifications Updating of classifications may be necessary if , for example, new information is obtained or if the criteria in CLP are amended. When manufacturers, importers or downstream users become aware of new information or an amendment to CLP or when a change is introduced in a substance or lassification of the substance or mixture . Note mixture, they must reconsider the c that “new” here refers to information not previously considered (or even new interpretation of old data) , not necessarily newly produced data. A downstream user may use the classification derived in accordanc e with the criteria by his supplier; this does not relieve the downstream user from the obligation to share new information with the supplier to allow him to meet the requirements. ection 1.1.10 addressing changes in harmonised classifications. Please, see also S The interface between hazard classification and hazard 1.1.9. communication hazard communication elements e label including CLP provides an integrated system of on th hazard pictograms, signal words, hazard statements and precautionary statements. Provision of this information to the end user is obligatory, irrespective of conditions of use and risk. While the Chemical Safety Assessment (CSA) on a particular substance p erformed for the purpose of ‘ ’ , a situation resulting in unforeseen exposure may occur, such as REACH may indicate safe use In such a situation, workers, managers and emergency personnel will need in an accident. stance, which will be provided by the label and the information on the hazard profile of the sub SDS. These sources of information will also provide useful information to the worker on the safe handling of the chemical. It is recognised that the hazard communication needs of the various end users ma y differ. Consumers are primarily dependent on the label of a substance or a mixture as a source of hazard and precautionary information, while the requirement for provision of an SDS is es communication of key primarily applicable to professional users. Thus, the label facilitat on a substance or a mixture and additional safety advice (precautionary hazard information statements) to consumers as well as to workers . , 1.1.10. The interface between self - classification and harmonised classification, and the list of har monised classifications CLP places emphasis on self - classification by industry of the substances or mixtures they supply. In some cases, substances are subject to harmonised classification at EU level, while 42 Please note that a case is still pending before the Court of Justice on the classification of a UVCB based on information on its constituents: Case C - 691/15 P.

52 Guidance on the Application of the CLP Criteria 52 Version 5.0 – July 2017 mixtures must always be self for pesticidal and biocidal products where the - classified, except Member State competent authorities (MSCAs) decide on the classification as part of the national authorisation scheme (CLP Article 36(2)). to CLP, this classification If a substance has a harmonised classification as provided in Annex VI the minimum must always be used by a manufacturer, importer or downstream user, except for classifications . The use of the minimum classification indicated with an asterisk (*) in Table 3.1 VI. For such minimum classifications, when is explained in section 1.2.1 of Annex available data a category than the given minimum , th e more stringent exists to justify more stringent must be used . It should be noted that w here some but not all hazard classes or category a hazard class have been harmonised, the remaining hazards must differentiations within be evaluated and - classified to complete the classification ( according to CLP Article 4(3) and self /additive/constituent . e presence o f an impurity Note that th which leads to CLP Recital 17) the harmonised classification of the classification in a more severe hazard classification than substance (in Annex VI, CLP) should be taken into account in the classification of the substance. As for substances in Annex VI, the name of the substance should include ( to be put on the label (i.e. substance name followed by “containing also the name of the impurity/additive/constituent f ≥x% name o in cases where they contribute sig ni fican tly to the classification of the impurity”) substance as in the case above (see 1.1.1.4, Annex VI, CLP) ) . Under CLP, the harmonised normally aims to cover classification and labelling of substances but CLP also allow s properties of the highest concern (CMR and respiratory sensitisation) for other properties if there is a need for such a n action at EU - level. Decisions on harmonisation ough comitology (CLP harmonised classification are taken by the European Commission thr ECHA and an opinion developed by ECHA 's Risk Article 37(5)), following a proposal submitted to on the proposal (CLP Article 37(4)). Whenever a manufacturer, Assessment Committee (RAC) ch may affect a harmonised classification, importer or downstream user has new information whi he must submit a proposal for a change to the member State Competent A t h ority where the u substance is placed on the market. Substances regulated under the Biocidal Products Regulation (EU) No 528/2012 or under the Plant Protection Products Regulation (EC) No 1107/2009 will normally be subject to harmonised classification and labelling for all hazardous properties. These proposals for harmonised classification and labelling are prepared by MSCAs only (CL P Article 36(2)). However, in general proposals for harmonised classification for a particular substance to be added in Annex VI to CLP can be made by both MSCAs and by manufacturers, importers and downstream users (CLP . Only MSCAs can propose a revision of an existing harmonised classification and Article 37) labelling to ECHA . (CLP Article 37(6)) A new or revised harmonised classification of a substance set out in Annex VI to CLP must be liers may use this applied from the date specified in the respective ATP, although supp classification before that date. When a supplier decides not to apply the harmonised C&L of a substance before this date, they must identify and examine all available information for the self - classification. Thus they should take into co nsideration the opinion adopted by the ECHA Risk Assessment Committee (RAC) on the harmonised C&L for that substance. If the C&L of a substance is already harmonised in the same hazard class, compliance with the existing harmonised C&L is legally required until it is formally changed in an ATP to CLP. The new harmonised C&L may be voluntarily applied as soon as the respective ATP enters into force. At the date of applicability, as provided for in the respective ATP, the suppliers are obliged to comply with the new harmon i sed C&L. Harmonised classification and labelling of a substance provides for a high level of protection of human health and the environment, and provides legal clarity for different suppliers of the same acturers of substances, importers of substances or substance of high concern (i.e. manuf

53 Guidance on the Application of the CLP Criteria 5.0 – July 2017 53 Version mixtures, producers of specific articles, downstream users (including manufacturers of mixtures) and distributors). Part 3 of Annex VI to CLP contains the list of harmonised classifications and labellings (except precautionary statements) . All harmonised classifications previously adopted under DSD and carried over to the list of listed in Annex I to DSD were translated to CLP classifications and harmonised classifications in Annex VI to CLP also including the Notes assigned to the entries as referred to in the DSD. This was done to maintain the same level of protection under CLP as under DSD. The harmonisation of classi building fication of substances is a continuous process on all efforts already done within the EU so far to evaluate hazards of substances that caused concern. Annex VI contains a number of entries indicated with Note B. The note relates to substances (aci ds, bases, etc.) that are placed on the market in aqueous solutions. The required classification and labelling may be different at different concentrations. These entries have a se entr ies gi general designation of the following type: ‘nitric acid ... %’. The ve the classification of the substance in a water solution above the G SCL. The G CLs or SCLs are applied as CL or usual in the classification of any mixture containing the substance. Thus, the concentration of the undiluted substance is compared with the SCL , as appropriate . For example, when GC L or diluted 75% phosphoric acid is added to a mixture to make up 10% of the mixture, the final concentration of phosphoric acid in the final mixture is 7.5%. As the SCL for for this substance skin and eye irritation is 10%, the final mixture does not require classification for these hazard classes based on phosphoric acid. The presence of Note B specifies that the supplier of an on aqueous solution of such a substance must state the percentage concentration of the soluti on the label. Note that the pure substance, i.e. not in water solution, may have different hazards. If there is , a classification would need to be derived no entry in Annex VI covering the anhydrous form contains water , based on available information. As the human body it is likely that the hazards of the aquatic solution still apply. Additional hazards may however occur , for example, hydrogen cyanide is F lam. liq.1 when it is pure but not in solution. 1.1.11. ntory) The Classification and Labelling Inventory (C&L Inve are required to notify ECHA Manufacturers and importers of the classification and labelling of as such or in a mixture (above a certain hazardous substance(s) placed on the market concentration leading to the classification of the mixture) and of substances subject to registration in accordance with the REACH Regulation. ECHA will then include the information in classification and labelling inventory in the form of a database. Substances require the notification within one month after their placing on the market. T here is no need to notify the substance if the same information has already been submitted as part of a registration under REACH by the same actor, as the classification and labelling, when part of the registration package, will au tomatically be added to the C&L Inventory (CLP Article 40(1)). Further guidance on what should be included in a notification and how to do it is available on the ECHA website http://echa.europa.eu/web/guest/regulations/clp/cl - inventory/notification - to - the - cl - inventory . ECHA s certain information from the C&L Inventory publicly available on its website, make ation, labelling and any relevant specific including the substance name, the classific - factor(s). It is indicated concentration limit or M in the Inventory if there is a harmonised classification for the entry, or if it is an agreed entry between manufacturers or importers. M ultiple notifications of the same substance can be submitted by different manufacturers or importers, with potential differences in the classifications . Notifiers and registrants are notified required to make every effort to come to an agreed entry. The information in the C&L Inv entory comes from registrati o ns and C&L notifications. This information has not been reviewed or verified by the Agency or any other authority.

54 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 54 1.1.12. Relation of classification to other EU legislation network of EU legislation relies on classification in one w ay or the other (see section 22 of the A for a detailed list of the laws concerned). Th i s Introductory Guidance on the CLP Regulation downstream legislation include s laws protecting consumers and workers, as well as rules on transport, biocides, pesticides , cosmetics and waste. Therefore, apart from the important hazard communication the label and in the SDS, the re are significant downstream on consequences of classification in that it also has a direct effect on risk manage me nt measures under REACH and other legislation . 1.1.12.1. REACH Classification plays a key role in REACH; it must be included in the registration dossier for a substance and it triggers certain provisions such as the performance of an exposure assessment and risk characterisation as part of the CSA and the obligation to provide an SDS. Classification of a substance as mutagenic, carcinogenic or toxic to reproduction (CMR) may also lead to restrictions and the need to apply for authorisations ((EC) No 1907/2006). 1.1.12.2. Plant Protection Products and Biocides or biocidal products containing them products Active substances as well as any plant protection be classified in accordance with the CLP Regulation. must t should be noted that with effect from 14 June 2011, Regarding plant protection products, i Directive 91/414/EEC has been repealed by Regulation (EC) 1107/2009 , which concerns their . This means that references to the repealed Directive must now be placing on the market new Regulation construed as references to the new Regulation. Nevertheless, Article 80 of the specifies that Directive 91/414/EEC must continue to apply with respect to active substances . included in Annex I to that Directive for certain transitional periods t should be noted that with effect from 1 Sep tember 2013, Regarding biocidal products, i Directive 98/8/EC has been repealed by Regulation (EU) 528/2012 , which concerns ther making available on the market and use. This means that references to the repealed Directive must now be construed as references to the new Regulation. Nevert heless, Articles 89 95 of the new – Regulation specifies the transitional measures which must continue to apply. In relation to classification, the new Regulation , bring about some changes, e.g. certain s classifications (e.g. CMR, Cat. 1A and 1B) may now p reclude approval of the respective substance as an active substance, safener, or synergist in plant protection products or biocidal products . 1.1.12.3. Transport legislation Many of the GHS criteria (by hazard class) are already implemented through the UN Model Regu lations for Transport of Dangerous Goods and related legal instruments (ADR, RID, ADN, IMDG Code and ICAO TI). Available transport classifications can be a source of information for the classification and labelling of substances and mixtures under CLP, especially for physical hazards, see also Section 2 of this document.

55 Guidance on the Application of the CLP Criteria 5.0 – July 2017 55 Version THE TERMS ‘ FORM OR PHYSI CAL STATE ’ 1.2. THE SIGNIFICANCE OF REASONABLY EXPECTED USE ’ WITH RESPECT TO AND ‘ DING TO CLP CLASSIFICATION ACCOR ‘ ‘ reasonably expected use’ 1.2.1. Form or physical state’ and form or physical state’ and ‘ reasonably expected use’ in the following CLP refers to the terms ‘ Articles: Artic Manufacturers, importers and downstream users of a substance shall identify the le 5(1) e purposes of determining whether relevant available information for th the substance entails a hazard as set out in Annex I physical, health or environmental [ ] ... The i nformation shall relate to the forms or physical states in which the substance is placed on the market and in which it can reasonably be expected to be used. The information shall relate to the forms or physical states in which the mixture Article 6(1) placed on the market and, when relevant, in which it can reasonably be expected to be is used. Tests that are carried out for the purposes of this Regulation shall be carried out Article 8(6) e(s) in which the substance on the substance or on the mixture in the form(s) or physical stat or mixture is placed on the market and in which it can reasonably be expected to be used. Article 9(5) When evaluating the available information for the purposes of classification, the manufacturers, importers and downstream use rs shall consider the forms and physical states in which the substance or mixture is placedon the market and in which it can be reasonably be expected to be used. ive of hazard classification is to identify the intrinsic physical, health and The object environmental hazards of substances and mixtures taking into account all uses that can be reasonably expected. In this context, the intention of the UN GHS should be kept in mind: (subsection 1.3.2.2.1) uses the term ‘ hazard classification ’ The GHS to indicate that only the intrinsic hazardous properties of substances or mixtures are considered. ' and The following guidance is intended to clarify the references to 'reasonably expected use 'form or physical state' in this context. 1.2.2. The term ‘ reasonably expected use’ in relation to hazard classification Hazard classification is based on the intrinsic properties of a substance or mixture and does not take into account exposure. Reasonably expected use summarises all physical forms and states of a substance or mixture that may occur during intended use or reasonably foreseeable conditions of misuse. or mixture is as follows: Reasonably expected use of a substance Any process, including prod uction, handling, maintenance, storage, transport or disposal.   All technical operations/manufacturing activities like e.g. spraying, filing, and sawing .  Any putative consumer contact through e.g. do - it - yourself or household chemicals.  All professional an d non - professional uses including reasonably foreseeable accidental exposure , but not abuse such as criminal or suicidal uses.

56 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 56 Reasonably expected use is also related to any consumer disposal or any work in which a d to be used irrespective of its present limited use or substance or mixture is used, or intende use pattern. Thus, use should not be mixed up with usage category. The term ‘form or physical state’ in relation to hazard classification 1.2.3. s taken into account differently in Depending on different prerequisites, form or physical state i the practice of testing and classification for physical, health, and environmental hazards which is described in the following paragraphs. It should be noted that in some cases a substance may autooxidise ( in contact wit h air ) or decompose to a more hazardous form. This may warrant classification of the substance even though it in itself is not or is less hazardous. A case - by - case evaluation should be done considering available hazard information on humans or animals and / or the rate and extent of case evaluation should also consider how the autoxidation or decomposition . The case - by - substance can be reasonably expected to be used. Physical hazards 1.2.3.1. different physical Different forms or physical states of a substance or mixture may result in properties and hazards with possible consequences for the hazard classification of a substance or mixture. Putative forms comprise properties such as crystal structure, particle size, ity or tablet form). Examples of physical homogeneity (e.g. emulsions) and texture (e.g. viscos state factors are: surface treatment (e.g. coating), state of aggregation, moisture content, residual solvent, activation or stabilisation. d physical state. If the The classification of a substance or mixture relates to the tested form an form and / or physical state is changed it has to be evaluated whether this might affect the classification and whether re testing is necessary. For example, a hazardous phase separation - conditions of storage, or a solid substance may may occur due to a temperature change under be molten to bring it into the liquid phase (e.g. for pumping). General considerations The test sample should be representative for the substance or mixture placed on the market. This is especially important in case of small 'batch' production. Mixtures might for example contain inert components which, if they are over - represented in the test sample, will lead to incorrect hazard classification. Specific requirements of certain test methods or the classification of physical hazards have specific requirements Some test methods f regarding the form / particle size of the sample to be tested. In these cases, the specific requirements of the test methods prevail. Examples of tests which have specific requirements re garding the form/particle size of the sample to be tested include those used to determine the classification of explosives and of substances which in contact with water emit flammable gases. In other test methods, there are no specific requirements regardi ng the particle size but it is stated explicitly that the particle size may have a significant effect on the test result. Therefore, these properties should be mentioned in the test report (i.e. testing of oxidising solids). Section provide further details about the relevance of the physical state for testing 2.0.4 purposes. 1.2.3.2. Human health hazards Also for human health, different forms (e.g. particle sizes, coating) or physical states may result in different hazardous properties of a substance or mixture in use. However, due to test complexity, not every form or physical state can be tested for each health hazard. In general, testing should be performed on the smallest available particle size and the default approach is

57 Guidance on the Application of the CLP Criteria 5.0 – July 2017 57 Version to test for different routes of exposure (oral, dermal, inhalation). Again, due to test complexity, mostly the data for only one exp osure route are available. In general, the assumption is made that the testing conditions of valid animal assays reflect the be used for classification. Moreover, it is assumed that hazards to man and these data must s takes into account all the potential hazards which are classification for human health hazard likely to be faced for all forms or physical states in which the substance is placed on the market and can reasonably be expected to be used . It is assumed that it comprises putative accidental ures. This approach generally, but not necessarily comprehensively, covers the whole expos range of intrinsic properties of a substance or mixture: in some cases, substances or mixtures have to be transformed into specific forms not mirroring ‘real life’ exposur es in order that an - animal test can be performed. As a consequence, the results of such tests may have to be evaluated taking into account any limitations due to the fact that the specific form of the tested epresent that to which human exposure may substance or mixture does not or not perfectly r occur during intended, known, or reasonably expected use. Such evaluation has to be performed according to the state of the scientific and technical knowledge. The burden of proof ce or mixture on the market. is on the person placing a substan Environmental hazards 1.2.3.3. The environmental hazard classification is principally concerned with the aquatic environment and the basis of the identification of hazard is the aquatic toxicity of the substance or mixture, ion on the degradation and bioaccumulation behaviour. and informat The system of classification is designed to ensure that a single classification applies to a substance. In general it takes no account of the specific form since this can vary and is not intrinsic to th e substance. The form in which the substance is placed on the market is taken into account when deciding what label to apply and various derogations from labelling exist, e.g. for t and the metals in the massive form. In the massive form the hazard may not be presen substance need not be labelled. The SDS will, however, indicate the classification and intrinsic hazardous properties to warn the user that subsequent transformation of the substance may produce the hazardous form. For aquatic hazard classificat ion, organic substances are generally tested in the dissolved form. - component substances and metals and their Exceptions to this approach include complex, multi compounds. Examples of alternative approaches include the use of Water Accommodated Fractions ( WAF) for complex, multi - component substances where the toxicity cut - off is related to the loading, and a test strategy for metals and their compounds in which the specific form (i.e. particle size) used for testing is standardised and forms or physical sta tes are not further taken into account. 1.3. SPECIFIC CASES REQUI RING FURTHER EVALUAT ION – LACK OF BIOAVAILABILITY 1.3.1. Definition Bioavailability is t he rate and extent to which a substance can be taken up by an organism and is available for metabolism or interact ion with biologically significant receptors. Bioavailability (biological availability) involves both release from a medium (if present) and absorption by an organism (IPCS 2004).

58 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 58 1.3.2. Bioavailability 12 Article Specific cases requiring further evaluation 9, the following properties Where, as a result of the evaluation carried out pursuant to Article or effects are identified, manufacturers, importers and downstream users shall take them into account for the purposes of classification: [...] (b) conclusive sc ientific experimental data show that the substance or mixture is not biologically available and those data have been ascertained to be adequate and reliable; [...] – the observation of In general, bioavailability is not explicitly evaluated in hazard classification systemic toxicity implicitly demonstrates a degree of bioavailability. On the other hand, when no toxicity is demonstrated in a test, this may be a result of either lack of intrinsic toxicity of the e test system employed. Nevertheless, as indicated in substance or lack of bioavailability in th Article 12 (b) of CLP there may be cases where a specific evaluation of bioavailability is warranted. Bioavalibility may also need to be considered for grouping and read - across. In general terms, for a substance or mixture to have an effect on a biological or environmental system, there must be some degree of bioavailability. Therefore, it follows that a substance or mixture need normally not be classified when it can be shown by conclusive experimental data from internationally acceptable test methods, e.g. from the Test Method Regulation (EC) No 440/2008, that the substance or a substance in a mixture is not biologically available (UN GHS other components in a 1.3.2.4.5.1). A non bioavailable substance may, however, react wi th e.g. mixture to transform to soluble available forms. The rate and extent at which this process, transformation ’ known as for the purposes of the classification guidance, takes place can vary ‘ extensively between different substanc es, and can be an important factor in determining the IV , Section IV.1 of this document). Note that a appropriate hazard category (see Annex nce which is inert and insoluble may still pose a hazard requiring classification , e.g. substa Further, it is important to note that bioavailability is not limited to systemic asbestos fibers. r local effects like irritation bioavailability but also includes local bioavailability for example fo and sensitisation. bioavailability of a - substance or a When considering the non mixture, the evaluation should be based on data for all relevant constituents of a substance or ingredients of the mixture. Further, consider potential interaction of the ingredients that could influence the one should bioavailability of the mixture as such or one of its components. and/ or Bioavailability considerations are only relevant with respect to classification for health environmental haza rds and not for physical hazards. 1.3.2.1. Human health hazards The assumption is that all substances and mixtures are considered to be bioavailable to some extent. However, there are a few specific cases in which bioavailability may have an influence on hazard cla ssification. For instance in the case of some metals and polymers, the nature of the physical form (metals in solid form) and the molecular size (polymers are very large molecules), or their physico - chemical properties may limit absorption. Where a supplie r proposes derogation from hazard classification on the basis of bioavailability, he has to provide adequate and robust data to support the conclusion of lack of bioavailability. It is possible that a substance is bioavailable by one route but not another (e.g. absorbed following inhalation but not absorbed through the skin). In such cases the lack of bioavailability may derogate classification for the relevant route.

59 Guidance on the Application of the CLP Criteria 5.0 – July 2017 59 Version d a In general, a prediction of lower bioavailability must be supported by robust evidence an be applied. weight of evidence determination using expert judgment must Information on bioavailability is usually obtained from adequate, reliable, and conclusive toxicokinetic studies for all relevant routes of exposure and all relevant forms or physi cal states where the substance and/or metabolite(s) of the substance have been quantified in body fluids At present (2016) , in vitro tests for release of moieties in biological fluids and/or target organs. are being developed, but not yet been agreed by OECD. It should be noted that concluding have that there is lack of or reduced bioavailability has a high burden of evidence and needs to be supported by robust data and expert evaluation. substance in mixtures is normally assumed if there are in Bioavailability of a substance or a vitro studies available which show the solubility of a substance or mixture in body fluids or artificial simulated body fluids. Furthermore, conclusions on bioavailability of a substance or a siderations of the physical properties of a substance or derived mixture may be based on con from Structural Activity Relationships (SAR). Note also that bioavailability is not limited to solubility, local bioavailability and the uptake of (nano)particles also has to be taken into ac count. Further, a substance or mixture can be transformed, e.g. by gastric fluid so that the substance absorbed may differ from the substance delivered. In certain exceptional circumstances it may be possible that a substance on its own or in a mixture can be considered to be non - in vitro data, e.g. from skin absorption bioavailable, based on either appropriate consideration of the substance, if the models, SAR considerations or the physical properties of respective requirements described above have been ta ken into account in an adequate analysis. Environmental hazards 1.3.2.2. The hazard classification for the aquatic environment is based on the three elements aquatic toxicity, bioaccumulation and degradation. The measurement of toxicity to aquatic organisms and its use within a hazard classification system introduces a number of compounding problems. The substance is not dosed directly into the organism but rather into water in which anism will the organism lives. While this reflects more accurately the manner in which the org receive the dose in the environment, it does not allow the direct control of the dose which is an important part of much mammalian toxicity testing. The dose is limited by the bioavailability of the substance, the maximum dose being determined b y the level of water solubility. It is usually assumed that toxic effects are only measured following exposure to the dissolved fraction, i.e. organisms are exposed to substances dissolved in water. It is assumed that the substances will either be absorbed by the organisms through passive diffusion or taken up actively by a specific mechanism. Bioavailability may, therefore, vary between different organisms. In the case of bioaccumulation, oral exposure could also be considered for K . Further guidance of the impact of bioavailability caused by the substances with high Log ow size of the molecule and how this is considered for aquatic hazard classification can be found in Annex III to this document. In general, there are no specific environmental test methods developed to measure biological availability of substances or mixtures. This aspect is built into the testing methodology for toxicity and if adverse effects are identified the substance should b e classified accordingly. Substances which lack bioavailability would not be absorbed by the exposed organisms and therefore due to lack of toxic effects these substances would not be classified, unless they are known to degrade or transform to hazardous p roducts. For example see the strategy for metals classification in Annex IV to this document.

60 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 60 1.4. USE OF SUBSTANCE CAT EGORISATION ( READ - A CROSS AND UPING ) AND (Q)SARS FOR CLASSIFICATION A GRO ND LABELLING Article 5(1) Manufacturers, importers and downstream users of a substance shall identify the relevant available information for the purposes of determining whether the substance entails a physical, health or environmental hazard as set out in Annex I, and, in particular, the following: [...] (c) any other information generated in accordance with section 1 of Annex XI to Regulation (EC) No 1907/2006; Article 6(1) Manufacturers, importers and downstream users of a mixture shall identify the relevant available information on the mixture itself or the substances contained in it for the purposes of determining whether the mixture entails a physical, health or environmental rticular, the following: hazard as set out in Annex I, and, in pa [...] (c) any other information generated in accordance with section 1 of Annex XI to Regulation (EC) No 1907/2006 for the mixture itself or the substances contained in it; Article 9(1) Manufacturers, importers and downstream users of a substance or a mixture shall evaluate the information identified in accordance with Chapter 1 of this Title by applying to it the criteria for classification for each hazard class or differentiation in Parts 2 to 5 of azards associated with the substance or mixture Annex I, so as to ascertain the h Article 9(3) Where the criteria cannot be applied directly to available identified information, manufacturers, importers and downstream users shall carry out an evaluation by applying a weight of evidence de termination using expert judgement in accordance with section 1.1.1 of Annex I to this Regulation, weighing all available information having a bearing on the determination of the hazards of the substance or the mixture, and in accordance with section 1.2 o f Annex XI to Regulation (EC) No 1907/2006. Article 13 If the evaluation undertaken pursuant to Article 9 and Article 12 shows that the hazards associated with the substance or mixture meet the criteria for classification in one or more hazard classes or d ifferentiations in Parts 2 to 5 of Annex I, manufacturers, importers and downstream users shall classify the substance or mixture in relation to the relevant hazard class or classes or differentiations by assigning the following: tegories for each relevant hazard class or differentiation; (a) one or more hazard ca (b) subject to Article 21, one or more hazard statements corresponding to each hazard category assigned in accordance with (a). used instead of testing when Section 1 of Annex XI to REACH provides a list of data that can be standard data are missing. This Annex specifies the conditions under which results of (Q)SARs, - across and grouping may be used in order to fulfil the information requirements under read information for the purpose of classification of REACH and refers to the adequacy of the e.g. that results of (Q)SARs may be used instead of testing when the substances. It states ‘ the substance falls within the applicability (Q)SAR models have been scientifically validated, result ’ , the ‘ ’ s are adequate for the purpose of classification and labelling domain and ‘ adequate and reliable documentation of the applied method is provided . Results generated by read - ’ , across and grouping may according to the same principles , be used for classification and labelling if they are ‘ adequate for classification and labelling ’ , ‘ have adequate and reliable coverage of the key parameters addressed in the corresponding test method , ‘ cover an ’ adequate od ’ , and ‘ exposure duration comparable to or longer than the corresponding test meth and reliable documentation of the applied method ’ is provided.

61 Guidance on the Application of the CLP Criteria 5.0 – July 2017 61 Version determination using expert judgement has According to CLP Article 9(3), a weight of evidence to be applied where the criteria cannot be applied directly to the available data. This described in CLP Annex I, 1.1.1. determination is further I t is important to note that most of the criteria for classification are directly related to specific - across and grouping should be test methods. Thus, th e adequacy of results of (Q)SARs, read evaluated against the criteria taking into account that normally the individual method attempts - across and to estimate the same hazard as the criterion. Nevertheless, when grouping, read (Q)SARs are being used alone or as a part of the basis for classification, it is normally necessary to do so employing weight of evidence and expert judgement in order to be able to apply the criteria to the information leading to a decision on the classif ication when the criteria are met A rticle 13, CLP) ( . to the consideration of any information that the CLP Annex I, 1.1.1.3 refers is relevant for determination of a hazard including the category approach. The latter encompasses grouping - across to and read help in a weight of evidence determination which is needed when the application of the criteria is not straightforward and cannot be applied directly to the available A information ( rticle 9(1)(3), recital (33)). I : Annex . A weight of evidence dete rmination means that all available information 1.1.1.3 bearing on the determination of hazard is considered together, such as the results of suitable in vitro tests, relevant animal data, information from the application of the category approach (grouping, read - ac ross), (Q)SAR results, human experience such as occupational data and data from accident databases, epidemiological and clinical studies and well documented case reports and observations. The quality and consistency of the data shall be eight. Information on substances or mixtures related to the substance or given appropriate w mixture being classified shall be considered as appropriate, as well as site of action and mechanism or mode of action study results. Both positive and negative results shall be bled together in a single weight of evidence determination. assem CSA, Chapter R.6 provides extensive advice on the use of (Q)SARs and grouping of IR & across, for developing the data set for hazard substances including guidance on read - evaluation. Guidance on the use of (Q)SAR and grouping for specific hazard classes is given in IR CSA, Chapter R.7. & In general, read across, grouping and use of (Q)SARs as the sole information elements to - - chemical properties is not recommended, since obtain data on basic physical reliable data should normally be available or is easily obtainable through testing. However, there may occasionally be practical problems with testing of substances for physical - chemical properties, especially for UVCBs where the properties may be depende nt on the variable composition. Therefore, the appropriateness of using read - - across, categorisation and (Q)SARs for physical This should also be the case chemical assessment should be considered on a case by case basis. e evaluation of health and environmental hazards in order when such data are considered for th to apply the criteria for classification. Given the availability of extensive guidance only a brief overview of each approach is presented below. For classification of mixtures see Section 1.6 of this document. 1.4.1. (Q)SAR Structure Activity Relationships and Quantitative Structure Activity Relationships, collectively referred to as (Q)SARs, are defined in IR&CSA , Chapter R.6.1.1 as theoretical models that can be used to predict in a qualitative or quantitative manner the ph ysico - chemical, biological (e.g. toxicological) or environmental fate properties of compounds from knowledge of their chemical structure.

62 Guidance on the Application of the CLP Criteria 62 Version 5.0 – July 2017 It should be noted that the use of (Q)SAR results requires the user to be sufficiently skilled to understand the appl icability of the selected (Q)SAR and to interpret the results in terms of reliability and adequacy for the purpose of classification and labelling. Extensive guidance on the use of (Q)SAR for hazard identification is given in , Chapter IR&CSA R.6.1. Guidance on the use of (Q)SARs for classification and labelling is also given in IR&CSA , Chapter R.6.1.4.2. This guidance is directly applicable to CLP. It should be noted that the (Q)SAR approach is not directly applicable to inorganic subs tances. Grouping 1.4.2. Guidance on grouping of substances for the purpose of hazard evaluation is given in IR&CSA , - Chapter R.6.2. Annex XI to REACH opens the possibility of evaluating substances not on a one - . A substance category is a group of by one basis, but by grouping substances in categories - chemical, human health, environmental and/or environmental fate substances whose physico properties are expected to be similar or to follow a regular pattern as a result of structural similarity. The use of grouping for hazard evaluation in the grouping approach means that not every Read - cross by interpolation can be used to fill substance needs to be tested for every hazard. data gaps, as well as trend analysis and (Q)SAR, and in addition the overall data for that categ ory must prove adequate to support the hazard assessment. - groups within a category of substances, e.g. when In some cases it is necessary to create sub there is a consistent trend within a group with regard to the potency of an effect which may justify di fferent classifications or setting of SCLs (see also IR&CSA , R.6.2.1.2). 1.4.3. Read - across Read - across is the use of hazard specific information for one substance (‘source’) to predict the e similar physico - same hazard for another substance (‘target’), which is considered to hav chemical, human health, environmental fate and/or (eco)toxicological properties. This can be based on structural similarity with a parent substance or its transformation products, and their bioavailability, bioaccessiblity, or known physi co - chemical properties such as water solubility. For certain substances without test data, the formation of common significant metabolites or information on metabolites of tested substances or information from precursors, may be IR&CS valuable information ( , Chapter R.6.2.5.2 and OECD 2004). For any hazard, read - across A may be performed in a qualitative or quantitative manner. Extensive guidance on the use of read - across is given in IR&CSA , Chapter R.6.2.2.1. Specific guidance for certain types of substance s such as reaction products and multi - constituent substances, complex substances, isomers, metals and metal compounds and other inorganic compounds is given in , Chapter R.6.2.5. IR&CSA 1.5. SPECIFIC CONCENTRATI ON LIMITS AND M - FACTORS 1.5.1. Specific concentration limi ts Article 10(1) Specific concentration limits and generic concentration limits are limits assigned to a substance indicating a threshold at or above which the presence of that substance in another substance or in a mixture as an identified impurity, addi tive or individual constituent leads to the classification of the substance or mixture as hazardous. Specific concentration limits shall be set by the manufacturer, importer or downstream user where adequate and reliable scientific information shows that t he hazard of a substance is evident when the substance is present at a level below the concentrations set for any hazard

63 Guidance on the Application of the CLP Criteria 5.0 – July 2017 63 Version class in Part 2 of Annex I or below the generic concentration limits set for any hazard class in Parts 3, 4 and 5 of Annex I. In except ional circumstances specific concentration limits may be set by the manufacturer, importer or downstream user where he has adequate, reliable and conclusive scientific information that a hazard of a substance classified as hazardous is not evident at a lev el above the concentrations set for the relevant hazard class in Part 2 of Annex I or above the generic concentration limits set for the relevant hazard class in Parts 3, 4 and 5 of that Annex. concentration limits shall not be set for Article 10(3) Notwithstanding paragraph 1, specific harmonised hazard classes or differentiations for substances included in Part 3 of Annex VI. The specific concentration limit (SCL) concept allows a fine tuning of the contribution of certain potency of the substances, hazardous substances to the classification of mixtures based on the as well as a classification of other substances containing these substances as impurities, generally additives or individual constituents. The SCL concept is h only applicable to healt normally hazards. For physical hazards, classification must be established on the basis of test data for the respective mixture, where applicable. The procedure of derivation of SCLs is different for every health hazard class and therefore guidance on ho c hapters of the different health w to set SCLs is provided in the respective hazard classes. A general overview on the applicability of SCLs and guidance availability for Table 1 1 below. setting SCLs for health hazards is illustrated by . s should take precedence over the generic concentration limits (GCLs) given in the relevant SCL health hazard sections of Annex I to CLP. In case specific concentration limits have been set in Annex VI to CLP, these must be applied. Moreover, manufacturers, importers or downstream users may not set their own SCLs for hazards subject to harmonised classifications in Annex VI to CLP. However, if not included in Annex VI and adequate and reliable data exist a hazard class is a hazard below the GCL , SCLs must be set by a manufacturer, importer or downstream showing user in accordance with CLP and be available in the C&L Inventory. SCLs should be communicated via the SDS. 1 . P ossibilities for setting SCL for health hazards addressed in relevant sections of the 1 Table guidance Higher SCLs than Lower SCL GCL (in exceptional Guidance Hazard class Category than GCL circumstances) not applicable Acute toxicity not applicable all not necessary Skin corrosion/ 3.2 all yes yes available in Section irritation Serious eye damage/ 3.3 all yes yes available in Section eye irritation Respiratory 3.4 see Section all * yes * yes sensitisation * currently not available;

64 Guidance on the Application of the CLP Criteria 64 Version 5.0 – July 2017 Higher SCLs than Lower SCL Category Hazard class GCL (in exceptional Guidance than GCL circumstances) Section 3.4 available in Skin sensitisation all yes yes * *currently not available see Section 3.5 Germ cell yes * yes * all mutagenicity *currently not available available in yes yes Section 3.6 all Carcinogenicity 3.7 Section available in Reproductive yes all yes toxicity Annex IV and in - SE 1 STOT yes no available in Section 3.8 2 no no see Section 3.8 3 yes yes available in Section 3.8 - 1 yes no available in Section 3.9 STOT RE 2 no no see 3.9 Section not 1 Aspiration hazard not appropriate not necessary appropriate 1.5.2. Multiplying factors (M - factors) Article 10(2) M - factors for substances classified as hazardous for the aquatic environment, acute category 1 or chronic category 1, shall be established by manufacturers, importers and downstream users. Article 10(4) - factors shall not be set for harmonised hazard Notwithstanding paragraph 2, M classes or differentiations for substances included in Part 3 of Annex VI for which an M - factor is given in that Part. - factor is not given in Part 3 of Annex VI for substances classified as However, where an M hazardous to the aquatic environment, acute category 1 or chronic category 1, an M - factor based on available data for the substance shall be set by the manufacturer, importer or downstream user. When a mixture including the substance is classified by the manufacturer, importer or downstream user using the summation method, this M - fa ctor shall be used. For the hazard class Hazardous to the Aquatic Environment ’ , SCLs are not applicable. Instead ‘ the M - factors concept is used. The M - factors are used in the application of the summation method for classification of mixtures containing s ubstances that are classified as very toxic. The concept of M - factors has been established to give an increased weight to very toxic substances when classifying mixtures. M - factors are only applicable to the concentration of a substance classified as hazar dous to the aquatic environment (categories Acute 1 and Chronic 1) and are used to derive by the summation method the classification of a mixture in which the substance is present. They are,

65 Guidance on the Application of the CLP Criteria 5.0 – July 2017 65 Version - ng established already when however, substance specific and it is important that they are bei classifying substances. n how to establish the M - factor see Section 4.1.3.3.3 of this document. For further guidance o factors should have been established in accordance with Article 10 of CLP and be available in - M the C&L Inventory. For the harmonised classifications in Annex VI to CLP, M must be set by the - factors - factor provided, in accordance is no M manufacturer, importer or downstream user in case there with CLP Article 10(4). 1.5.3. Harmonised ATE values Acute Toxicity Estimates ( ATE may be included in annex VI of CLP. From 2016 harmonised ) have ATEs are one way o f These values to be used, just as any other harmonised item. Annex I to CLP , 3.1.2.1) . expressing acute toxicity (see 1.6. MIXTURES 1.6.1. How to classify a mixture The classification of mixtures under CLP is for the same hazards as for substances. As a general on the mixture as a whole rule and as is the case with substances, available relevant data should primarily be used to determine classification where applicable, also considering the validity and suitability of the used test method, with regard to testing mixtures in general and the specific mixture of concern. Not a ll the test methods relevant for substances may be suitable for (all) mixtures and for this reason care has to be taken. Note that for skin If sensitisation, care has to be taken so that the doses used do not render the results unreliable. , further approaches to mixture classification may be applied. When this cannot be done CMR hazards and biodegrad ation and bioaccumulati on properties , classification of the evaluating mixture should always be based on the ingredient substance s according to Article 6(3) and (4) if data on a mixture show CMR for these particular hazard classes. However, properties even in absence of data on possible CMR ingredientes , the mixture has to be classified appropriat e ly following Article 6(3) . It is important to choose the most appropriate method to determine the classification for a mixture for each hazard class, differentiation or category. The method will depend on whether zards and on the type and the mixture is being assessed for physical, health or environmental ha 1.2.3 of this document on form or quality of information that is available (see also Section physical state). It is impo rtant to get a clear picture on which substances and mixtures are contained in a mixture. Basic information on substances would include the substance identity, its classification and any assigned SCLs or M - factors, and concentration in the mixture and, whe re relevant, details of any impurities and additives including their identity, classification and concentration. Where an ingredient in a mixture is itself a mixture, it is necessary to get information on the ingredient substances of that mixture together with their concentrations, classifications and any applied SCLs or M - factors. Useful sources for such information are the SDS from the supplier of the substance or the mixture, and the C&L Inventory provided by ECHA, which also includes the harmonised clas sifications of substances listed in Annex VI to CLP. Also data from registration dossiers are a valuable source of information. It should be noted that an SDS should also be provided in some cases when the mixture does not meet the criteria for classificat ion but certain specific criteria are met (see Article 31(3) of REACH).

66 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 66 Further dialogue with the supplier may be necessary to obtain additional information. For example on compositional information for the mixture supplied. follows the sequence displayed in The classification of mixtures 1 . 1 , for each hazard class Figure independently (except for CMR and when evaluating biodegrada tion and bioaccumulati on properties : ) Figure 1 1 H ow to classify a mixture . There is a mixture to classify All available information should be gathered Are available test data for the mixture sufficient for classification? Yes (CLP Article 9 (2) - (3)) Classify the mixture for the (For physical hazards: consider relevant hazard whether new testing needs to be performed. Consult the criteria.) No Classify the Is there data available Is it possible to Yes mixture for Yes on similar tested apply any of the the relevant mixtures and individual bridging principles? hazard hazardous ingredients? No No Use the known or derived hazard data on the individual ingredients to classify Yes Are hazard data available the mixture for the relevant hazard, for all or some using the methods in each section of ingredients? CLP Annex I, Part 3, Part 4 and Part 5 No Unable to classify the mixture go back to – ingredient suppliers to obtain additional information Note: The principles for using expert judgement and weight of evidence determination (CLP Article 9(3) and (4)) and Annex I, section 1.1.1.) should be taken . into account Classification for physical hazards 1.6.2. The majority of the physical hazards of mixtures should be determined through testing based on the methods or standards referred to in CLP Annex I, Part 2. In few cases, the classification of a mixtures c , if sufficient appropriate data are available an also be derived through a calculation

67 Guidance on the Application of the CLP Criteria 5.0 – July 2017 67 Version see CLP Annex I 2.2.4.1 and ISO 10156 for flammable gases, CLP Annex I 2.4.4 and ISO ( s ) . 10156 for oxidizing gases and CLP Annex I, 2.6.4.2 and 2.6.4.3 for flammable liquid Test methods for physical hazards are referred to in each physical hazard class chapter of CLP. Most of these test methods can be found in the UN Manual of Tests and Criteria, see the website . A few of these test methods http://www.unece.org/trans/danger/publi/manual/manual_e.html are contained in standards which are also referred to in CLP (see particularly flammable gases, ult, based on other methods or standards oxidizing gases and flammable liquids). When test res (which are not referred to in CLP ) are available , then these data may still be used, provided they are adequate for the purpose of hazard determination. Expert judgement is necessary t o whether there is suf conclude ficient documentation to assess the suitability of the test used, and whether the test was carried out using an acceptable level of quality assurance and thus on the . adequacy of such data for the purposes of classification according to CLP Please note th at in practice the physical hazards of a substance or mixture may differ from those shown by tests, e.g. in case of certain ammonium - nitrate based compounds (explosive / - oxidising properties) and certain halogenated hydrocarbons (flammable properties). Suc h experience must be taken into account for the purpose of classification (CLP Article 12(a)). The information available or generated must be checked to determine if it is directly comparable to the respective hazard criteria and if it is, then it can be u sed to derive the classification immediately. Where the criteria cannot be directly applied to the available data, expert judgement should be used for the evaluation of the available information in a weight of LP Annex I, 1.1.1.). evidence determination (CLP Article 9(3) and C 1.6.3. Health and environmental hazards For the purpose of classification for health or environmental hazards, for each hazard check whether or not there is information:  on the mixture itself; on similar tested mixtures and ingredient subst  ances; or on the classification of ingredient substances and their concentrations in the mixture.  As pointed out in the introduction to this chapter, the supplier should be contacted if it is considered that the information on the substances or mixtures s upplied is not sufficient for classification purposes. The information available on the hazard under consideration, will determine if the mixture should be classified using the approaches below in the following sequence (CLP Article 9): Classification deri ved using data on the mixture itself (see a. 1.6.3.1 of this Section document), by applying the substance criteria of Annex I to CLP; b. Classification based on the application of bridging principles (see Section 1.6.3.2 of this document), which make use of test data on similar tested mixtures and ingredient substances; and including SCLs and M - c. Classification based on calculation or on concentration thresholds, factors. Classification derived using data on the mixture itself 1.6.3.1. Classification derived using data on the mixture itself, by applying the substance criteria of Annex I to CLP, is applicable for all hazards , except: CMR hazards (see CLP Article 6(3)), bioaccumulation and biodegradation properties within the evaluation of the ‘hazardous to the aquatic environment’ hazard class referred to in sections 4.1.2.8 and 4.1.2.9 of Annex I to CLP (see CLP Article 6(4)).

68 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 68 Article 6(3) For the evalua tion of mixtures pursuant to Chapter 2 of this Title in relation to the ‘germ cell mutagenicity’, ‘carcinogenicity’ and ‘reproductive toxicity’ hazard classes referred to in sections 3.5.3.1, 3.6.3.1 and 3.7.3.1 of Annex I, the manufacturer, importer or do wnstream user shall only use the relevant available information referred to in paragraph 1 for the substances in the mixture. Further, in cases where the available test data on the mixture itself demonstrate germ cell mutagenic, carcinogenic or toxic to re production effects which have not been identified from the information on the individual substances, those data shall also be taken into account. Article 6(4) For the evaluation of mixtures pursuant to Chapter 2 of this Title in relation to the ‘biodegradation and bioaccumulation’ properties within the ‘hazardous to the aquatic environment’ hazard class referred to in sections 4.1.2.8 and 4.1.2.9 of Annex I, the manufacturer, importer or downstream user shall only use the relevant available infor mation referred to in paragraph 1 for the substances in the mixture. Where the criteria cannot be directly applied to the available data, expert judgement should be used for the evaluation of the available information in a weight of evidence determination (CLP Article 9(3) and CLP Annex I, 1.1.1). Note that the test method used must be suitable for the mixture tested. If data from test methods other than those indicated in Article 8(3) are used, a comparison with the methods indicated in that article has t on the o be made to verify the effect evaluation of the information. 1.6.3.2. Bridging principles In the case of a classification for health or environmental hazards, relevant information on the ufficient data on similar mixture itself may not always be available. However, where there are s tested mixtures and individual hazardous ingredient substances, CLP allows bridging principles to be used to classify the mixture (CLP Annex I, 1.1.3). be Only one bridging principle could s with the exception of Aerosols, where a mixture applied in the evaluation of a hazard clas based on another bridging principle is used in an aerosol container . However, different classified to different hazard classes. bridging principles may apply To apply these bridging principles certain condi tions should be considered for their application . T he conditions are summarised below. It is necessary to consult Annex I of CLP, Part 3 for health hazards and Part 4 for environmental hazards, before undertaking any of these assessments. In case it is not possible to classify the mixture by applying bridging principles and a weight of evidence determination using expert judgement by applying the criteria in Annex I to test results of a mixture, then the mixture should be classified using the other methods described in CLP Annex I, Parts 3 and 4. 1.6.3.2.1. Dilution Where the tested mixture is diluted with a substance (diluent) that has an equivalent or lower hazard category than the least hazardous original ingredient substance, then it can be assumed that the respect ive hazard of the new mixture is equivalent to that of the original tested mixture. The application of dilution for determining the classification of a mixture is illustrated by Figure 1 . 2 .

69 Guidance on the Application of the CLP Criteria 5.0 July 2017 69 Version – 1 . 2 Figure Application of the bridging principle: dilution for determining the acute toxicity classification of a mixture Diluent B (classification known) Mixture C (A+B) Mixture A (tested) (not tested) Mixture A, which has been classified as acute toxic category 2 based on test data, is Example: subsequently diluted with diluent B to form mixture C. If diluent B has an equivalent or lower ixture A and is not acute toxicity classification than the least acutely toxic ingredient in m expected to affect the hazard classification of other ingredients, then mixture C may be also classify mixture C, thus classified as acutely toxic category 2. However, this approach may over - the supplier may choose to apply the additivi ty formula described in CLP Annex I, 3.1.3.6 (see Section 1.6.3.3.1 of this document). ed a relevant ingredient. Note that also the diluent of the tested mixture is consider Consider using this particular bridging principle also when, for example, diluting an irritant mixture with water,   diluting an irritant mixture with a non - classified ingredient, or classifi - diluting a corrosive mixture with a non  ed or irritant ingredient. 1.6.4.1 of this document. In case a mixture is diluted with another mixture, see Section Within the ‘hazardous to the aq uatic environment’ hazard class, i f a mixture is formed by diluting another classified mixture or substance with water or other totally non - toxic material, the toxicity of the mixture can also be calculated from the original mixture or substance (see Section of this document). secti on 4.1.3.4.3 of Annex I to CLP and mixture example C in 4.1.4.7 Batching 1.6.3.2.2. mixture is produced under a controlled process, then it can be tested Where a batch of a ssumed that the hazards of each new batch are equivalent to those of previous batches. This a method must not be used where there is reason to believe that the composition may vary significantly , affecting the hazard classification. 1.6.3.2.3. Concentration of highly hazardous mixtures category, an - Where a tested mixture is already classified in the highest hazard category or sub stances that are untested mixture which contains a higher concentration of those ingredient sub in that category or sub - category should also be classified in the highest hazard category or sub - category (CLP Annex I, 1.1.3.3). 1.6.3.2.4. category hazard Interpolation within one Assume there are three mixtures (A, B and C) which contain identic al hazardous components. If mixtures A and B have been tested and are in the same hazard category, and mixture C is not

70 Guidance on the Application of the CLP Criteria 70 5.0 – July 2017 Version tested and has concentrations of those hazardous components intermediate to the concentrations in mixtures A and B, then mixture C is as sumed to be in the same hazard category as A and B. The application of interpolation for determining the classification of a Figure 1 3 (CLP Annex I, 1.1.3.4). mixture is illustrated by . 1 . 3 Application of the bridging principle: interpolation for determining the aquatic acute Figure hazard classification of a mixture 30% 90% 70% 10% Mixture B Mixture A (Aquatic Acute 1) (Aquatic Acute 1) 60% 40% 30% ≤ conc. ≤ 90% 10% ≤ conc. ≤ 70% Mixture C (Interpolate as Aquatic Acute 1) Substantially similar mixtures 1.6.3.2.5. the two Two mixtures contain an identical ingredient at the same concentration. Each of mixtures contains an additional ingredient which is not identical with each other; however they are present in equivalent concentrations and the hazard category of these two ingredients is the ard classification of the other same and neither of them is expected to affect the haz . ingredient If one of the mixtures is classified based on test data it may be assumed that the hazard category of the other mixture is the same. The application of substantially similar mixtures for determining the classific ation of a mixture is illustrated by . Figure (CLP Annex I, 1.1.3.5). 4 1

71 Guidance on the Application of the CLP Criteria 5.0 – July 2017 71 Version Figure . 4 Application of the bridging principle: substantially similar mixtures for determining 1 lassification of a mixture the skin irritation c Ingredient A Ingredient C Ingredient B Ingredient B 10% 10% 90% Mixture P Mixture Q (tested) (not tested) 2) (Skin Irrit. Example: If the Ingredient C has the same hazard category and the same potency as Ingredient A, then Mixture Q can be classified as Skin Irrit. 2 like Mixture P. Potency may be expressed by, for example, differences in the specific concentration limits of Ingredie nts A and C. This method should not be applied where the irritancy of Ingredient C differs from that of Ingredient A. 1.6.3.2.6. Review of classification where the composition of a mixture has changed Article 15(2) Where the manufacturer, importer or downstream user introduces a change to a mixture that has been classified as hazardous, that manufacturer, importer or downstream user shall carry out a new evaluation in accordance with this Chapter where the change is either of the following: (a) a change in the composition of the initial concentration of one or more of the hazardous constituents in concentrations at or above the limits in Table 1.2 of Part 1 of Annex I; (b) [...] ion where the composition of a mixture has changed Review of classificat 1.1.3.6 Annex I: The following variations in initial concentration are defined for the application of Article 15(2)(a): Table 1.2 Bridging Principle for changes in the composition of a mixture Initial concentration range of the Permitted variation in initial concentration of the constituent constituent ± 30 % ≤ 2,5 % 2,5 < C ≤ 10 % ± 20 % 10 < C ≤ 25 % ± 10 % 100 % ≤ ± 5 % 25 < C NOTE: The guidance below explaining Table 1.2 in the green box relates to a change in the composition of mixtures already classified as hazardous. A change in the composition of - hazardous mixtures may result in concentration thresholds being reached an d a need non

72 Guidance on the Application of the CLP Criteria 72 Version 5.0 – July 2017 to classify the changed mixture as hazardous. Where the manufacturer, importer or classified for a specific hazard, that downstream user introduces a change to a mixture not a new manufacturer, importer or downstream user must therefore always carry out evaluation for that hazard in accordance with Chapter 2 of Title II to CLP (see Article 15(1) of CLP). change in the composition of When a manufacturer, importer or downstream user introduces a the initial concentration of one or more of the hazardous constituents of a mixture classified as , that manufacturer, importer or downstream user must carry out a new evaluation , if hazardous the change in concentrations is at or above the limits in Table 1.2 of Part 1 of Annex I to CLP. the variations of the initial concentrations of the constituents lie within the However, where permitted variation, manufacturer, importer or downstream user does not need to carry out a new evaluation and may use the current classification of the mixture. example is to illustrate what is meant by the permitted variations in Table 1.2. The following Example : Mixture A is classified as hazardous based on the initial concentration of two constituents, substance A and substance B. The initial concentrations in the mixture hazardous of substance A and substance B are 2 % and 12 %, respectively. The permitted variation according to T able 1.2 is for substance A ± 30 % of the initial concentration and for substance B tration in the mixture may for ± 10 % of the initial concentration. This means that the concen substance A vary between 1.4 % and 2.6 % and for substance B between 10.8 %, % and 13.2 without having to carry out a new evaluation in accordance with Chapter 2 of Title II to CLP:  Substance A: 2  ±0.3 = ±0.6 1.4 – 2.6 Substance B:  ±0.1 = ±1.2  10.8 – 13.2 12 Aerosols (some health hazards only) 1.6.3.2.7. - A mixture in aerosol form is considered to have the same classification as the non aerosolised form of a mixture, provided that the propellant used does not affect these haza rds upon - spraying and data demonstrating that the aerosolised form is not more hazardous than the non aerosolised form is available (see CLP Annex I, 1.1.3.7.). 1.6.3.3. Classification based on calculation or concentration thresholds In most cases, test data on the mixture itself or similar mixtures will not be available, therefore bridging principles and weight of evidence determination using expert judgement for all of the necessary health and environmental hazard assessments may not be applied. In these cases, cl assification must be based on calculation or on concentration thresholds referring to the classified substances present in the mixture. In the case where one or more mixtures are added to another mixture, the same requirement applies: it is necessary to kn ow all ingredient substances, their hazard classifications and their concentrations to be able to derive a correct hazard classification of the final mixture. For further details see Section 1.6.4 of this document. 1.6.3.3.1. Classification based on calculation More detailed guidance on the selection of the most appropriate method is provided in the specific section for each hazard class. An example is the hazard class acute toxicity where a calculation formula is used which is based on acute toxicity estimates and concentrations, and a modified formula for determining the classification of a mixture containing substances of unknown acute toxicity.

73 Guidance on the Application of the CLP Criteria 5.0 Version July 2017 73 – : Annex I 3.1.3.6.1. [...] The ATE of the mixture is determined by calculation from the ATE values for all relevant ingredients according to the following formula for Oral, Dermal or Inhalation Toxicity: C 100 i   ATE ATE n i mix where: = concentration of ingredient i ( % w/w or % v/v) C i i = the individual ingredient from 1 to n n = the number of ingredients = Acute Toxicity Estimate of ingredient i. ATE i Annex I: acute 3.1.3.6.2.3. If the total concentration of the ingredient(s) with unknown 10 % then the formula presented in section 3.1.3.6.1 shall be used. If the total ≤ toxicity is concentration of the ingredient(s) with unknown toxicity is > 10 %, the formula presented in ercentage of the unknown section 3.1.3.6.1 shall be corrected to adjust for the total p ingredient(s) as follows: (   C 100 %) 10 if C  unknown i   ATE ATE n mix i Section For more information on the CLP calculation formulae for this hazard, please see 3.1.3.3.3 of this document. Another example is provided by hazard class ‘ hazardous to the aquatic environment ’ , namely the additivity formula: 4.1.3.5.2. Annex I: Mixtures can be made of a combination of both components that are classified (as Acute Category 1 and/or Chronic Category 1, 2, 3 or 4) and others for which adequate toxicity test data are available. When adequate toxicity data are available for more one component in the mixture, the combined toxicity of those components is calculated than using the following additivity formulas(a) and (b), depending on the nature of the toxicity data: toxicity: (a) Based on acute aquatic C C  i i   L(E)C L(E)C η 50m 50i where: C = concentration of component i (weight percentage) i L ( E ) C = (mg/l) LC for component i or EC 50 50i 50 η = number of components L ( E ) C = L(E)C of the part of the mixture with test data 50 50m The calculated toxicity may be used to assign that portion of the mixture a short - term ( acute ) hazard category which is then subsequently used in applying the summation method; (b) Based on chronic aquatic toxicity:

74 Guidance on the Application of the CLP Criteria 74 Version 5.0 – July 2017 C  C C C   i j j i     x NOEC NOEC 0,1 Eq NOEC n n i j m Where: = concentration of component i (weight percentage) covering the rapidly degrada ble C i components C = concentration of component i (weight percentage) covering the non - rapidly degradable j components = NOEC (or other recognised measures for chronic toxicity) for component i covering NOEC i the rapidly degradable components, in mg/l; = NOEC (or other recognised measures for chronic toxicity) for component i covering NOEC j the non rapidly degradable components, in mg/l; - n = number of components, and I and j are running from 1 ton; EqNOEC data; = Equivalent NOEC of the part of the mixture with test m [...] NOTE: T he full use of this approach requires access to the whole aquatic toxicity data set and the necessary knowledge to select the best and most appropriate data. CLP has limited the use of the additivity formulae to those circumstances where the substance hazard category is not known, although the acute and/or chronic toxicity data are available. With the aquatic toxicity data at hand the ingredient substance classification and M - factor(s) which then could could easily be gained by a direct comparison with the substance criteria, be fed straight into the summation method. It will therefore usually not be necessary to use the additivity formulae. For more information on the CLP calculation formulae for this hazard please see 4.1.4.3 Section of this document. Class ification based on concentration thresholds 1.6.3.3.2. Generic concentration thresholds For most hazard classes or differentiations, classification based on concentration thresholds may be applicable. CLP distinguishes between two different kinds of generic concentra tion thresholds: Generic cut - off values: these values are the minimum concentrations for a substance to  for classification purposes. These substances are also referred to be taken into account Sections 3.1 , 3.2 and 3.3 ). When a as relevant ingredients in some hazard classes (see - off value it classified substance is present in a concentration above the generic cut contributes to the mixture classification even if it does not trigger classification of the mixture directly. The generic cut - off values are defined for some hazard classes and categories only and are listed in Table 1.1 of Annex I to CLP; (GCL) : these values are the minimum concentrations for a  Generic concentration limits trigger the classification of a mixture if exceeded by the individual substance which concentration or the sum of concentrations of relevant substances (where the individual substance concentrations can b e ‘added’ to each other in a straight forward way); they are set out in parts 2 - 5 of Annex I for those hazard classes where they apply. Generic concentration thresholds are generic for a hazard class, differentiation or category. The difference between a generic cut - off value and a generic concentration limit is demonstrated through the example of the skin irritation hazard: while Table 1.1 of Annex I to CLP defines the - off value to be 1 % for a skin irritant substance which is present in a mix ture, Table generic cut 3.2.3 of Annex I to CLP shows that a GCL of the skin irritant substance above or equal to the concentration limit of 10% triggers classification of the mixture for skin irritation. However, at  1 % and below 10 %, the substance may still contr ibute to the classification of the mixture as the concentration would be taken into account if other skin skin irritant . This because

75 Guidance on the Application of the CLP Criteria 5.0 – July 2017 75 Version corrosive/irritant substances are present in the mixture below the relevant generic as provided by the summation in CLP If additivity ap concentration limits. plies, classification Annex I, Table 3.2.3 may be applicable, i.e.:  (10 Skin Corrosive Categories 1A, 1B, 1C) + Skin Irritant Category 2 should be ≥ 10 % Specific concentration thresholds ‘ Specific Concentration Limits ’ In contrast to generic thresho (SCLs) and/or specific cut - off lds, for individual substances values may be established : SCLs are described in section 1.5.1 of this document and where they have been  43 o CLP and/or in the C&L established they are included in Table 3.1 of Annex VI t ‘ Inventory (CLP Article 42). For ’ the Multiplying hazardous to the aquatic environment 44 factors) concept factors (M is used instead of SCLs, see section 1.5.2 of this guidance. - - factors included in Tables 3.1 must be used where applicable and, for SCLs and M factors notified to - the C&L Inventory classifications not included in Annex VI, SCLs and M can be considered and used where applicable. Cut -  off values that may be different from the generic values and that are to be used in speci fic cases are given in 1.1.2.2.2(a) and (b) of Annex I to CLP. For example - - off concerning aquatic hazard, for a substance with an established M factor, the cut - off value divided by the M value is always the generic cut factor; hence, (0.1/M) % (see - 1.1.2.2 .2(b) and 4.1.3.1 of Annex I to CLP). 1.6.3.3.3. Additivity Vs . non additivity of hazards For some hazard classes additivity concepts are normally the not applicable. In these cases, substance or mixture contains two substances each general approach is that if a present at a of below the GCL defined for that hazard class or differentiation, even if the sum concentration is above this limit, the mixture will not be classified, as far as no the substances' concentrations lower SCL has been set. is normal ly not applied for the following hazard classes: Additivity a. skin and respiratory sensitisation ; b. germ cell mutagenicity; carcinogenicity; c. d. reproductive toxicity; e. specific target organ toxicity, single and repeated exposure, categories 1 and 2; f. on in certain cases (see CLP Annex I, 3.2.3.3.4); and skin corrosion/irritati certain cases (see CLP Annex I, 3.3.3.3.4). g. serious eye damage/eye irritation in However, in certain cases for these hazard classes may be scientifically justified. additivity Expert judgement is needed . 43 Please note that Table 3.2 of Annex VI to CLP is deleted from 1 June 2017 by Commission Regula tion (EU) 2016/1179 (9th ATP) amending CLP. 44 M - factors are used to derive, by means of the summation method, the classification of a mixture in which the substance is present for which the M - factor has been established. For further guidance on how to est ablish and use M - factors see sections 4.1.3.3.2 and 4.1.4.5, respectively.

76 Guidance on the Application of the CLP Criteria 76 Version 5.0 – July 2017 If the mode of action (MoA) is the same , additivity can reasonably be of two substances anticoagulant . Examples of cases where additivity applies is assumed reprotoxicity of (a group of substances affecting the same enzyme in the same way) , reprotoxicity rodenticides n sensitisation by nickel substances and carcinogenicity of substances releasing boron ions, ski For t he lat and mutagenicity of formaldehyde releasers er group of substances there are . t 45 in A nnex notes releasable formaldehyde from different VI stating that the levels of whether the substances have a components of a mixture must be added. This applies regardless , whether the purpose of the substance is to act as a harmonised classification or not . formaldehyde releaser or not and it in cludes formaldehyde itself the MoA is different there may be some cases where it is deemed appropriate to assume When , In other cases , there may be no cause for additivity. additive or synergistic effects. - For STOT SE 2 additivity may be assumed for substances with the same target organ, RE 1 and s are similar. Again , especially if the MoA in other cases there may be no reason to assume additivity. Additivity is used for the following hazard classes or differentiations: a. Acute toxicity (according to specific formula) ; b. skin corrosion/irritation (besides the cases mentioned in CLP Annex I, 3.2.3.3.4); serious eye damage/eye irritation (besides the cases mentioned in CLP Annex I, c. 3.3.3.3.4); d. specific target organ toxicity, single exposure Category 3 (respiratory tract irritation); h. specific target organ toxicity, single exposure Category 3 (narcotic effects); aspiration hazard (plus consideration of viscosity of the final mixture) ; e. short - term ( acute ) and long - term (chronic) aquatic toxicity and f. Hazardous for the ozone layer. g. In these cases, as well as in the specific cases described above when additivity may be scientifically justified, if the sum of the concentrations of one or several substances classified for the same hazard class /category in the mixture equals or exceeds the GCL set out for this hazard class/category, the mixture must be classified for that hazard. For substances that have an SCL or M factor(s), these should be taken into account when applying the summation method s. The - method described in section 3.2.3.2.3.2 can be used when one or more substances in a mixture have SCLs.  1 then the mixture needs If the sum of (ConcA / clA) + (ConcB / clB) + ... + (ConcZ / clZ) is to be classified for the hazard class in questio n. Where ConcA = the concentration of substance A in the mixture; clA = the concentration limit (either specific or generic) for substance A; ConcB = the concentration of substance B in the mixture; 45 The 10th ATP added the following notes in Annex I to CLP: “Note 8: The classification as a carcinogen need not apply if it can be shown that the maximum theoretical concentration of releasable formaldehyde, irrespective of the source, in the mixture as placed on the market is less than 0,1%.” “Note 9: The classification as a mutagen need not apply if it can be shown that the maximum theoretical concentration of releas able formaldehyde, irrespective of the source, in the mixture as placed on the market is less than 1%.”

77 Guidance on the Application of the CLP Criteria 5.0 – July 2017 77 Version (either specific or generic) for substance B; etc. clB = the concentration limit An example is provided for the hazard class serious eye damage /eye irritation: i n case there are only substances classified as eye irritation Category 2 present in a mixture, then their sum must be equal to or exceed the generic concentration limit of 10 % in order for the mixture to . Note that only relevant substances (i.e. for eye be classified in Category 2 as well irritants, - off value cut should be summed up and contribute to m ixture above the generic of 1%) classification. Further guidance on the application of SCLs when using the summation method to conclusions on skin corrosion / irritation or serious eye damage/eye irritation hazards can derive Sections be found in and 3.3 of this document. 3.2 1.6.4. Classification of mixtures in mixtures For physical hazards, an adequate hazard classification is generally derived by testing. To determine the classification of a mixture for health or environmental hazards using the component substances, including their additivity or summation methods, information on all the individual hazard classification and concentration, is generally required. In the case where one or more mixtures a re added to another mixture, the same requirement applies: it is generally component substances, their hazard classifications and their necessary to know all concentrations to be able to derive a correct hazard classification of the final mixture. It is ge nerally not possible to derive the correct hazard classification for the final mixture by using only the hazard classification(s) of the mixtures that were combined to make it. For example, a mixture containing 1% of a Carc. Cat. 1B substance would be clas sified as Carc. Cat. 1B. Taking 1% of this mixture into another mixture would lead to a concentration of the ingredient causing the carcinogenic classification of 0.01%, i.e. below the GCL. The same situation may occur also for substances classified due to an impurity. However, there is one exception. I the acute toxicity estimate (ATE) of a mixture is known f (either actual or derived), this value can be used to derive a correct classification for acute toxicity if this mixture is added to another mixture. Thus, it is very important that suppliers of mixtures communicate the necessary information listed above on component substances (including their individual hazard classification and concentration) down the supply chain, in the SDS, to enable a co rrect classification to normally be established by downstream users formulating new mixtures from their products. However, the information provided in the SDS may not be sufficient, for example where only a concentration range is quoted for a particular substance o r where the mixture contains other which triggers substances classified as hazardous but which are present below the concentration the SDS. Thus further dialogue with the supplier of the obligation to indicate the substance in to obtain additional information on the constituent substances to the mixture may be necessary ensure correct classification and labelling of the new mixture. In situations, where tested mixtures are added to other tested or untested mixtures, an y be derived by taking account of the test data as well as adequate hazard classification can onl the knowledge on all ingredient substances, their hazard classifications, and their concentrations in these mixtures. Such an approach is a case - by - case analysis and requires expert judgement. 1.6.4.1. ple: Classification of Mixture A Exam Note that the example only addresses health hazards. For compositional details see 1 . 2 Table and Table 1 . 3 below. Mixture A is a water solution containing a surfactant, a thickening agend dye and a frag ance r mixture. Classification of components an d composition of the frag r ance mixture are known. No test data are available on Mixture A and it is not possible to apply bridging principles due to lack of data on similar tested mixtures. Therefore it is necessary to identify the ingredients in Mixture A (including their % w/w and classification).

78 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 78 Mixture A does not contain any ingredients classified as a respiratory sensitiser, CMR, STOT or aspiration hazard. Therefore it is possible to conclude that Mixture A will not be classified as se particular hazard classes. hazardous for the Acute toxicity point 3.1.3.3(b), there are two options to calculate the acute As indicated in CLP Annex I, toxicity of Mixture A: (i) treat the 'fragrance mixture' as an ingredient when calculating the ATE for Mixture A, or (i i) break the 'fragrance mixture' down into its component ingredients and only take over the relevant ingredients (CLP Annex I, 3.1.3.3(a) and 3.1.3.6.1) into the calculation for the ATE of Mixture A. Following option (i) it is first necessary to calculat e ATE of the 'fragrance mixture' (see Table mix . ) taking into account 'FM component 1' and 'FM component 2' (other components can be 1 3 their LD excluded as values are > 2000 mg/kg): 50 C 100 i    ATE ATE n i mix 100   ATE mix C i  ATE n i 100 1597   mg/kg ATE mix 35 . 2 0 17 .  1230 500 for the 'fragrance mixture' can then be included in the calculation of the ATE for The ATE mix mix Mixture A: 100 13300   ATE mg/kg mix . 8 . 5 0 0  1800 1597 Following option (ii) it is only necessary to include 'FM compone nt 1' from the 'fragrance mixture' (present in Mixture A at 1.76 %), as 'FM component 2' is present in a concentration < 1%). Calculation of the ATE for Mixture A according to option (ii): mix 100 17200   mg/kg ATE mix 0 76 8 . 1 .  1800 1230 Both options indicate that the calculated of Mixture A is > 2000 mg/kg thus mixture A is ATE mix not classified as hazardous for acute toxicity by the oral route. value) was available for the NOTE: If an acute oral toxicity test (i.e. an actual LD 50 fragrance mixture, then this should be used in the calculation for the ATE of Mixture A. Skin corrosion/irritation Work out the actual levels of the 'fragrance mixture' ingredients in Mixture A and carry out the summation met hod (CLP Annex I, Table 3.2.3) using the relevant ingredients.

79 Guidance on the Application of the CLP Criteria 5.0 – July 2017 79 Version Mixture A does not contain any ingredient classified as Skin Corr. 1A, B or C. Therefore Mixture A is not classified as Skin Corr. 1A, B or C. The 'fragrance mixture' contains ingredients class ified as Skin Irrit. 2, but these are all present generic cut - in Mixture A at concentrations < 1 % and can be disregarded ( off values to be taken into account, CLP Annex I, Table 1.1). Mixture A does also contain 8 % of the ' anionic classified as Skin Irrit. 2, but as the concentration of the ' surfactant' < 10 % anionic surfactant' , Mixture A is not classified as Skin Irrit. 2. (GCL, CLP Annex I, Table 3.2.3) Serious eye damage/eye irritation Work out the actual levels of the 'fragrance mixture' ingredients in Mixture A and carry out the summation method (CLP Annex I, Table 3.3.3) using the relevant ingredients: Mixture A contains 8 % of an ingredient classified thus Mixture A must also be as Eye Dam. 1, classified as ( i.e. the relevant ingredient is present in a concentration above the Eye Dam. 1 3 %). The 'fragrance mixture' also contains an ingredient classified as Eye Dam. 1, but GCL of this is present in Mixture A at a concentration < 1 % and can disregarded. Skin sensitisation ntains four ingredients classified as skin sensitisers (cat 1) The 'fragrance mixture' co but their actual levels in Mixture A are 1 % thus Mixture A is not classified as a skin belowthe GCL of sensitiser. However, the four skin sensitiser ingredients are present above 0.1 %, thus a EUH208 (CLP Annex II, 2.8) would be required on the label for dditional labelling information Mixture A. In summary, mixture A is classified as Eye Dam.1 and additional labelling information is needed on the label. EUH208 — ‘Contains (name of sensitising s ubstance). May produce an allergic reaction’ . 1 2 Ingredients in Mixture A Table . % w/w Oral LD (rat) Ingredient Classification 50 1800 mg/kg Anionic surfactant 8.00 Acute Tox. 4 (oral) Eye Dam. 1 Skin Irrit. 2 Thickening agent 0.80 > 5000 mg/kg Not classified Dye > 5000 mg/kg Not classified 0.05 Fragrance mixture 5.00 not tested Acute Tox. 4 (inhalation, oral) (see list of ingredients below) Skin Sens. 1 Eye Dam. 1 Skin Irrit. 2 Aquatic Chronic 2 Water 86.15 Not classified Total: 100.00

80 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 80 Table . 3 Ingredient ' Fragrance mixture' 1 % w/w % in Mixture A Oral LD Ingredient Classification (rat) 50 1.76 FM component 1 35.20 Acute Tox. 4 1230 mg/kg (inhalation, oral) FM component 2 17.00 0.85 not available Acute Tox. 4 (oral) Skin Sens. 1 (use cATpE 500) 0.8 FM component 3 3600 mg/kg 16.00 Skin Sens. 1 Skin Irrit. 2 FM component 4 13.40 0.67 3100 mg/kg Skin Sens. 1 > 2000 mg/kg FM component 5 0.35 7.00 Eye Dam. 1 Aquatic Chronic 2 FM component 6 6.00 0.3 4400 mg/kg Flam. Liq. 3 Skin Sens. 1 Skin Irrit. 2 Aquatic Chronic 1 > 5000 mg/kg FM component 7 Not classified 2.80 0.14 2.60 0.13 > 5000 mg/kg Aquatic Chronic 1 FM component 8 5.00 100.00 Total: Example: Classification of Mixture B 1.6.4.2. Note that the example only addresses health hazards. Mixture B is a powder form detergent co ntaining a base powder, silicates, carbonate and including the %w/w and c compositional details lassification of the inorganic processing aid. The 5 ingredients Table 1 . 4 and Table 1 . are provided in below. No test data are available on Mixture B and it is not possible t o apply bridging principles due to lack of data on similar tested mixtures. Mixture B does not contain any ingredients classified as a skin sensitiser, CMR or aspiration hazard. Therefore it is possible to conclude that Mixture A will not be classified as hazardous for these particular hazard classes. Acute toxicity As indicated in CLP Annex I, 3.1.3.3(b), there are two options to calculate acute toxicity of Mixture B: (i) treat the 'base powder' as an ingredient when calculating the ATE for Mixture B, or (ii) break the 'base powder' down into its component ingredients and only take over the relevant ingredients (CLP Annex I, 3.1. 3.3(a) and 3.1.3.6.1) into the calculation for the ATE of Mixture B. Following option (i) it is first necessary to calculate th e ATE of the 'base powder' taking into mix account the non - ionic surfactant (other components can be excluded as LD values are > 2000 50 mg/kg):

81 Guidance on the Application of the CLP Criteria 5.0 July 2017 81 – Version C 100 i    ATE ATE n i mix 100 ATE   mix C i  ATE n i 100  ATE 2778 mg/kg  mix . 18 0     500   for the 'base powder' can then be used for the calculation of the ATE for Mixture B: The ATE mix mix 100 mg/kg  ATE  2860 mix . 18 8 0 . 20 0 . 0   2778 770 1800 Following option (ii) it is only necessary to include the non ionic surfactant from the 'base - powder' (present in Mixture B at 3.6%). Other ingredients in the 'base powder' can be excluded as LD them. The calculation of the ATE for Mixture B applying option > 2000 mg/kg for all of mix 50 (ii): 100  mg/kg ATE  2860 mix . . 8 6 . 3 0 0 18   500 1800 770 Both options indicate that the calculated ATE of Mixture B is > 2000 mg/kg. Therefore Mixture mix B is not classified as hazardous for acute toxicity by the oral route. NOTE: If an acute oral toxicity test (i.e. an actual LD value) was available for the 'base 50 powder' then this should be used in the calculation for the ATE of Mixture B. Skin corrosion/irritation considered to apply. Additvity is ctual levels of the 'base powder' ingredients in Work out the a Mixture B and carry out the summation method (CLP Annex I, Table 3.2.3) using the relevant ingredients: Mixture B does not contain any ingredients classified as Skin Corr. 1A, B or C thus Mixture B is not classified as Skin Corr. 1A, B or C. Mixture B does however contain 23 % ingredients classified as Skin Irrit. 2 (11% silicates, 8% anionic surfactant and 4% anionic surfactant from the 'base powder') , as the content of classified ing redients are > 10% also Mixture B is classified as Skin Irrit. 2 .

82 Guidance on the Application of the CLP Criteria 82 Version 5.0 – July 2017 Serious eye damage/eye irritation Work out the actual levels of the 'base powder' ingredients in Mixture B and carry out the ingredients: summation method (CLP Annex I, Table 3.3.3) using the relevant 1 (18% substance X Mixture B contains 40.6 % ingredients classified as Eye Dam. , 11% - ionic surfactant), thus Mixture B is also silicates, 8 % anionic surfactant and 3.6 % non Eye Dam.1 . classified as Respiratory sensitisation s 0.7% of the ingredient 'enzymes' classified for respiratory sensitisation Mixture B contain category 1 . However this is below the concentration triggering classification (CLP Annex I, Table 5 ) thus Mixture B is not classified as a respiratory sensitiser. However ingre dient 'enzymes' 3.4. trigger additional labelling information (CLP Annex II, 2.8). EUH208 STOT Mixture B does not contain any ingredients classified as STOT RE or STOT SE 1 or 2, but it contains 11% of an ingredient classified as STOT SE 3 (respiratory tract ir ritation). The generic concentration limit is 20 % for extrapolating the classification as STOT SE 3 from an ingredient to the mixture (CLP Annex I, 3.8.3.4.5.), thus Mixture B does not trigger classification as STOT SE 3 (respiratory tract irritation). summary, mixture B is classified as Skin Irrit. 2, Eye Dam. 1 and additional labelling In — ‘Contains (name of sensitising substance). May information is needed on the label. EUH208 produce an allergic reaction’. Table 1 . 4 Ingredients in Mixture B (rat) Ingredient % w/w Oral LD Classification 50 Base powder Eye Dam.1 20.00 not tested (see list of ingredients below) Skin Irrit. 2 Ox. Sol. 1 Acute Tox. 4 (oral) 770 mg/kg 18.00 Substance X Eye Dam. 1 Eye Dam. 1 3400 mg/kg Silicates Skin Irrit. 2 11.00 respiratory tract irritation STOT SE 3 ( ) 7.00 Carbonate 4090 mg/kg Eye Irrit. 2 Inorganic processing aid 11.30 > 5000 mg/kg Not classified 16.00 > 5000 mg/kg Not classified Builder Acute Tox. 4 (oral) 8.00 1800 mg/kg Anionic surfactant Eye Dam. 1 Skin Irrit. 2 Substance Y 5.00 > 5000 mg/kg Not classified Enzymes 0.70 > 2000 mg/kg Resp. Sens. 1

83 Guidance on the Application of the CLP Criteria 5.0 – July 2017 83 Version Oral LD % w/w (rat) Classification Ingredient 50 3.00 > 5000 mg/kg Not classified Polycarboxylate 100.00 Total: ‘ . Ingredient 1 5 base powder’ Table s % w/w % in Mixture B Ingredient Oral LD (rat) Classification 50 Acute Tox. 4 (oral) Eye Dam. 1 3.6 500 mg/kg - 18.00 ionic surfactant Non Aquatic Acute 1 Skin Irrit. 2 Anionic surfactant > 2000 mg/kg 20.00 4.0 Eye Irrit. 2 50.00 Builder > 5000 mg/kg Not classified 10.0 Carbonate 8.00 1.6 4090 mg/kg Eye Irrit. 2 Inorganic processing aid 0.8 > 5000 mg/kg Not classified 4.00 20.00 Total: 100.00 ANNEX VII TO CLP 1.7. Article 61(5) Where a substance or mixture has been classified in accordance with Directive 67/548/EEC or 1999/45/EC before 1 December 2010 or 1 June 2015 respectively, manufacturers, importers and downstream users may amend the classification of the re using the conversion table in Annex VII to this Regulation. substance or mixtu NOTE: Article 61 uses the term ‘conversion table’ and Annex VII uses the term to ‘translation table’. These terms have the same meaning i.e. the tables in Annex VII that relate CLP classifications according to DSD or DPD to a classification according to CLP. The tables contained in Annex VII to CLP show how classifications in accordance with the DSD were converted into the correspon ding classification under CLP and included in Table 3.1 of 46 to CLP in classifications . The tables also aimed to support translation of existing self Annex VI - CLP . into classifications in accordance with DSD accordance with Although conceptually similar, the coverage of CLP and the DSD or DPD is different. In some cases, the relationship between the category of danger and corresponding R phrases and the - hazard categories and corresponding hazard statements is but in other cases, , clear it is less well defined. Additionally, CLP introduced new hazard classes reflecting hazards that were not covered or were only partly covered by DSD and DPD. 46 Note that the 8th ATP has corrected the Annex VII to CLP. The current Annex VII suggests R34 = Skin in Corr. 1B. Corr. 1 whereas the original translation was to Sk

84 Guidance on the Application of the CLP Criteria 84 Version 5.0 – July 2017 While the tables explicitly point out where no translation wa s possible or where minimum classification where CLP hazard classes or be applied, they do not identify would situations D, are required under CLP. In the particular case of SD and DP categories, not covered by the D any indication for would not provide the a reasonable ‘no classification’ under ble DPD, the ta translation to a CLP classification. translation tables d As mentioned, the Annex VII not always give a direct translation. (to CLP) id a translation eated exposure, For certain hazard classes, including acute toxicity and STOT rep to CLP , from DSD to CLP ed in a recommended minimum according to Annex VII result This minimum classification is also classification. indicated as such in Table 3.1 in Annex VI, and information is available (see also CLP Annex VI, should only be used if no additional hazard 1.2.1). be noted that w hen ever data for a should It substance or mixture is available for a hazard class, the CLP criteria and the Annex the substance or mixture must be classified in accordance with (to CLP) tables must no VII be used. longer Table . 6 identifies where no direct translation was possible according to 1 (to CLP) the Annex VII translation t ables for substances and mixtures requiring classification under DSD or DP D . In addition to the differences indicated in Table 1 . 6 , i t should be note d that for some hazards , the generic concentration limits to be applied for mixtures, were lowered under CLP as . Lower generic concentration limits were set for skin corrosion (R34 and R35), compared to DPD severe eye damage and eye irritation (R41 and R36) , skin irritancy (R38) and reproductive toxicity (R60, R61, R62 and R63). . 6 Hazard classes where the translation tables in Annex VII to CLP indicate that no Table 1 direct translation was possible from DSD t o CLP Potential translation Classifications Comments under DSD or DPD outcomes Change of classification criteria and method; - 1) Explosive. case E, R2 case considerations - by 2) Organic peroxide E, R3 for additional to this Guidance Annex VII See 3) Flammable solid information on transport classifications 4) Oxidising solid - reactive 5) Self 6) No classification O, R8 (liquid) All liquid substances or mixtures classified O,R8 are Oxidising liquid classified as oxidising liquids under CLP. See VII to this Guidance for additional Annex information on transport classifications O, R8 (solid) The test methods for oxidising solids in 67/548/EEC Oxidising solid were different. Most solids classified O, R8 and CLP are also classified as oxidising solids under CLP. See Annex VII to this Guidance for additional information on transport classifications F, R11 (solid) 1) Flammable solid Solid substances or mixtures classified F, R11 may be classified as flammable solids or self reactives heating 1a) Possibly self - under CLP. If classified as flammable solids, they in addition may additionally be classified as self - heating.

85 Guidance on the Application of the CLP Criteria 5.0 July 2017 85 Version – Potential translation Comments Classifications under DSD or DPD outcomes 2) Self - reactive See Annex VII to this Guidance for addit ional information transport classifications on F, R15 Substance or mixture or additional f to this Guidance See Annex VII on which, in contact with information transport classifications water, emit(s) flammable gas(es)

86 ication of the CLP Criteria Guidance on the Appl 86 Version 5.0 – July 2017 2. 2: PHYSICAL HAZARDS PART INTRODUCTION 2.0. General remarks about the prerequisites for 2.0.1 classification and testing The purpose of this chapter is to give some general guidance with respect to the classification of physical hazards, the generation of test data and the ir interpretation. The intention of CLP is to – identify hazards of chemical substances and mixtures and to provide a systematic approach - to communicate them based on harmonized criteria. The classification using classification ps: process involves three ste 1. gathering of relevant information regarding the hazards of a substance or mixture – 8 of CLP); (Articles 5 2. evaluation of hazard information to ascertain the hazards associated with the substance or mixture (Article 9 of CLP); and 3. the substance or mixture will be classified as hazardous and the a decision on whether degree of hazard, where appropriate, by comparison of the data with agreed hazard classification criteria (Article 13 of CLP). Generally n Annex I of CLP must be , for bothsubstances and mixtures, the tests required i performed unless there is adequate and reliable information already available. Testing is required to determine physical hazards including the physico - chemical properties necessary for ve methods are specifically permitted. Before the respective classification unless alternati undertaking testing of a substance or mixture, enquiries should be made to ascertain the availability of data, e.g. flash points, on the substance or mixture. Safety 2.0.2 n data derived from testing. Special care is required In most cases, the classification is based o when new or unknown substances or mixtures are tested. If possible, preliminary tests should be carried out before larger quantities are handled. Appendix 6 of the UN Recommendations on the transport of dangerous goods Manual of Tests and Criteria (UN - MTC) 'Screening procedures' chemical properties based on small - - allows gathering valuable information about physico scale tests. Further aspects of safety are given in the general introduction, Section 1.4 o f the UN MTC - or within the individual test procedures. General conditions for testing 2.0.3 Samples offered for testing must in all aspects be representative of the substance or mixture to be classified. Therefore, it is helpful to characterise or specify the sa mple for the purposes of documentation (i.e. batch number, production code, impurities etc.). Further characterisation (i.e. analysis) is highly recommended in cases where the presence of diluents, activators, stabilisers or moisture may influence the outc ome of the test. In some cases, additional parameters like (e.g.) physical condition, particle size and shape, specific surface area, density, crystal structure, may influence the test result. Therefore, these properties should be mentioned in the test rep ort. The tests must be performed on the substance or mixture in the appropriate physical form where changes in that form may influence the outcome of the test (see also Articles 5 and 6 of CLP).

87 Guidance on the Application of the CLP Criteria 5.0 – July 2017 87 Version 2.0.4 Physical state The physical state determines which hazard cl asses should be considered for testing. As the CLP 47 , hazard classification is based on intrinsic properties of the substance or mixture which states are determined not only by its physical state but also its form. of this guidance As mentioned in Chapter the same solid substance or mixture may have 1.2 , different forms such as flakes, prills, or powder. Furthermore, e.g. a powder may contain particles of different size, and particles of the same size may have different shapes, crystallinity r allotropy etc. These differences may result in different intrinsic properties, and consequently, o different physical hazards of the powder. Particle size is crucial for several classes such as explosives, flammable solids, self reactive substances, pyroph oric solids, self - heating - substances, solid organic peroxides and substances which, in contact with water, emit flammable gases. Therefore not only the physical appearance, but also other parameters should the y may trigger different classifications of the be considered when identifying the form, since same substance or mixture. An example of different classification due to different intrinsic properties of forms is red phosphorus (flammable solid) and white phosphorus (pyrophoric solid) (different allotrope s). It is therefore important to evaluate case by case whether available information on the physical properties of the substance and mixture placed on the market, is applicable to the examined form, and whether additional testing should be performed. The form of a substance or mixture as placed on the market might be such that it is not possible to test it in this form, e.g. if it is in the form of tablets or pellets. In such circumstances, the physical hazards of the substance or mixture must be considere d for classification especially if they are friable and produce secondary effects due to abrasion or crushing during supply and use. If phase separation does occur, the hazardous properties of the mmunicated. most hazardous phase of the substance or mixture must be co If further testing is required, the choice of the test method should be done after thorough evaluation of its suitability for the substance or mixture, as the properties of the form (e.g. for powders especially size and shape of the particle) m ay have a significant effect on the test results. The definitions for gases, liquids and solids are given in Annex I, Part 1 of CLP: 1.0. Definitions Annex I: Part 1, Gas means a substance which: bsolute); or (i) at 50 °C has a vapour pressure greater than 300 kPa (a (ii) is completely gaseous at 20 °C at a standard pressure of 101 . 3 kPa; Liquid means a substance or mixture which: (i) at 50 °C has a vapour pressure of not more than 300 kPa (3 bar); is not completely gaseous at 20 °C and at a standard pressure of 101,3 kPa; and (ii) (iii) which has a melting point or initial melting point of 20 °C or less at a standard pressure of 101,3 kPa; Solid means a substance or mixture which does not meet the definitions of liquid or gas. In some cases (i.e. viscous substances or mixtures), a specific melting point cannot be determined. Such a substance or mixture must be regarded as a liquid if either the result of the 47 CLP Article 5(1), 6(1) and 8(6).

88 ication of the CLP Criteria Guidance on the Appl 88 Version 5.0 – July 2017 ASTM D 4359 standard test metho d for determining whether a material is - 90 test as amended ( indicates ‘liquid’ or the result of the test for determining fluidity a liquid or a solid) (penetrometer test) prescribed in Section 2.3.4 of Annex A of ADR indicates ‘not pasty’. 2.0.5 Quality st be based on the methods named in Annex I, Part 2 of CLP. For The determination of data mu - most hazard classes in Annex I, Part 2 of CLP there is reference made to the UN MTC which gives very detailed descriptions of the test methods. For the classification of flammable gases, sing gases and for the determination of the flash point there are references to international oxidi standards in Annex I, Part 2 of CLP. Whenever possible, the laboratory should validate the performance of the methods used e.g. by participating in inter laborato ry testing or by using - reference materials. Any deviation from the test procedure or standard should be documented and, if necessary, justified. The reliability of all test results used for the classification of hazardous substances and mixtures is importa nt and therefore their transparency and comparability must be ensured. For these purposes, CLP requires in Article 8 the following: Article 8 (5) [...] Where new tests for physical hazards are carried out for the purposes of this Regulation, they shall be carried out, at the latest from 1 January 2014, in compliance with a relevant recognised quality system or by laboratories complying with a relevant recognised standard. [...] In general, the following alternative strategies can be pursued: 1. the principles of good laboratory practice (GLP) (as formerly required by compliance with the DSD); application of EN ISO/IEC 17025 General requirements for the competence of testing 2. and calibration laboratories as amended as a relevant recognised standard; other internat ionally recognised standards of comparable scope. 3. Any laboratory that carries out physical hazard tests for classification purposes can therefore choose how to fulfil the quality requirements of CLP. 2.1. EXPLOSIVES 2.1.1. Introduction The requirements in Chapter 2.1 ‘Explosives’ of Annex I of CLP are identical to those in Chapter 2.1 of GHS. The classification of explosives according to the GHS is almost entirely adopted based on the UN Recommendations on the Transport of Dangerous Goods – Model Regulations (UN RTDG M odel Regulations), which are appropriate for transport and also storage of packaged explosives. The classification of substances, mixtures and articles in the class of explosives and further allocation to a division is a very complex procedure. References to Part I of the UN - MTC and related expertise are necessary.

89 Guidance on the Application of the CLP Criteria 5.0 – July 2017 89 Version 2.1.2. Definitions and general considerations for the classification of explosives The following definition is given in CLP for the class of explosives. 2.1.1.1. The class of explosives Annex I: comprises (a) explosive substances and mixtures; (b) explosive articles, except devices containing explosive substances or mixtures in such quantity or of such a character that their inadvertent or accidental ignition or initiation effect external to the device either by projection, fire, smoke, heat shall not cause any or loud noise; and (c) substances, mixtures and articles not mentioned in points (a) and (b) which are manufactured with a view to producing a practical, explosive or pyrotechnic effect. Additional remark related to the applicability of 2.1.1.1 (a) (see also UN RTDG Model Regulations, 2.1.1.1 (a)):  a substance or mixture which is not itself an explosive but which can form an explosive atmosphere of gas, vapour or dust is not included in this class;  explosive behaviour related to the thermal decomposition of organic peroxides and of self - reactive substances and mixtures is covered by those specific hazard classes and therefore not included in the hazard class explosives. llowing definitions apply for explosives: In addition the fo 2.1.1.2. Annex I: [...] An explosive substance or mixture is a solid or liquid substance or mixture of substances which is in itself capable by chemical reaction of producing gas at such a temperature and pressure and at such a speed as to cause damage to the surroundings. Pyrotechnic substances are included even when they do not evolve gases. A pyrotechnic substance or mixture is a substance or mixture of substances designed to und, gas or smoke or a combination of these as the result produce an effect by heat, light, so sustaining exothermic chemical reactions. - detonative self - of non An unstable explosive is an explosive which is thermally unstable and/or too sensitive for normal handling, transport and use. An exp losive article is an article containing one or more explosive substances or mixtures. A pyrotechnic article is an article containing one or more pyrotechnic substances or mixtures. ctured with a view An intentional explosive is a substance, mixture or article which is manufa to produce a practical explosive or pyrotechnic effect. Certain physical hazards (due to explosive properties) are altered by dilution, as is the case for desensitized explosives, by inclusion in a mixture or article, packaging or other factors. Explosive substances and mixtures wetted with water or alcohols, or diluted with other substances to suppress their explosive properties, may be treated differently to their non - y apply, depending on the wetted or non - diluted counterparts i.e. different hazard classes ma physical properties of the wetted/diluted substance or mixture.

90 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 90 2.1.3. Relation to other physical hazards For safety reasons, substances, mixtures or articles which have already been classified as RTDG Model Regulations) should not be considered for Explosives (Class 1 according to the UN classification in any other physical hazard classes. Since the explosion hazard is more severe than other physical hazards there is no need to further perform classification tests for other sical hazards. potential phy When considering substances and mixtures for classification within the hazard class explosives, the following checks should be performed with respect to other hazard classes: Substances, mixtures and articles that have been manufactured with a view to producing a practical explosive or pyrotechnic effect, are classified as explosives by definition according to 2.1.1.1(c) of Annex I of the CLP. It should be checked whether such a substance or mixture is an unstable explosive. Thermally unstabl e substances or mixtures that are not classified as explosives should be - considered for classification as self reactive substances and mixtures. Mixtures of oxidising substances and mixtures with combustible material that are not classified hould be considered for classification as self as explosives s reactive substances and mixtures, - oxidising liquids or oxidising solids. Due to the complexity of these issues, expert advice should always be sought when dealing with classification of substances and mixtures with potentially explosive properties. 2.1.4. Classification of substances, mixtures or articles as explosives 2.1.4.1. Identification of hazard information Information on the following types of hazards is relevant for the evaluation of substances, mixtures and articles for the class of explosives:  sensitivity to shock;  effects of heating and ignition under confinement; thermal stability;  sensitiveness to impact and friction;  mass explosion hazard;  projection hazard;  fire and radiant heat hazard.  Screening procedures and waiving of testing 2.1.4.2. The screening procedure is described in: CLP, Annex I, Part 2, paragraphs 2.1.4.2 and 2.1.4.3; Appendix 6 of the UN - MTC. The screening procedure may be used for new substances or mixtures which are suspected of es. It should not be used for substances and mixtures manufactured having explosive properti with the intention of producing a practical explosive or pyrotechnic effect. Explosive properties are associated with the presence of certain chemical groups in a molecule which can react t o produce very rapid increases in temperature and/or pressure. The screening procedure is aimed at identifying the presence of such reactive groups and the potential for rapid energy release. Examples of groups which may indicate explosive properties in o rganic materials are:  - C unsaturation (e.g. acetylenes, acetylides, 1, 2 - dienes); C  C - Metal, N - Metal (e.g. Grignard reagents, organo - lithium compounds);

91 Guidance on the Application of the CLP Criteria – Version 91 5.0 July 2017 Contiguous nitrogen atoms (e.g. azides, aliphatic azo compounds, diazonium salts,  hydrazines, sulphonylh ydrazides); Contiguous oxygen atoms (e.g. peroxides, ozonides);  N - O (e.g. hydroxyl amines, nitrates, nitro compounds, nitroso compounds, N - oxides,  - 1,2 oxazoles); - halogen (e.g. chloramines, fluoroamines);  N O - halogen (e.g. chlorates, perchlorates, iodosyl compounds).  A substance or mixture is not classified as explosive: when there are no chemical groups associated with explosive properties present in the a. molecule; or b. when the substance or mixture contains chemical groups associated with explosive - 200; properties which include oxygen and the calculated oxygen balance is less than The oxygen balance is calculated for the chemical reaction: y     z y       x   x        CO O H   O O C H 2 2 y 2 x z 2 2 4         Using the formula:       z 2 y 2 x Oxygen balance = 1600   molecular weight or c. when the organic substance or a homogenous mixture of organic substances contains chemical groups associated with explosive properties but the exothermic decomposition energy is less than 500 J/g and the onset of exothermic decompositio n is below 500 ºC. temperature (The limit is to prevent the procedure being applied to a large number of organic materials which are not explosive but which will decompose slowly above 500 ºC to release more than 500 J/g.) The exothermic decomposition energy may be determined using a suitable calorimetric technique; or d. for mixtures of inorganic oxidising substances with organic material(s), the concentration of the inorganic oxidising substance is: less than 15 % by mass, i f the oxidising substance is assigned to Categories 1 or 2;   less than 30 % by mass, if the oxidising substance is assigned to Category 3. If the screening procedure identifies the substance or mixture to be a potential explosive or if it ining any known explosives, the classification (acceptance) procedure for the is a mixture conta class of explosives (see Section 2.1.4.5.1 ) has to be applied. If the exothermic decomposition energy of organic materials is less than 800 J/g, a UN gap test is not required, neither according to Series 1 Type (a) nor according to Series 2 Type (a). 2.1.4.3. Classification criteria The criteria for the classification of explosives are given in the following tables. Annex I: 2.1.2.1. Substances, mixtures and articles of this class are classified as an unstable explosive on the basis of the flowchart in Figure 2.1.2. The test methods are described in Part I of the UN RTDG, Manual of Tests and Criteria.

92 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 92 2.1.2.2 . Substances, mixtures and articles of this class, which are not classified as an unstable explosive, shall be assigned to one of the following six divisions depending on the type of hazard they present: Substances, mixtures and articles which have a mass explosion hazard (a) Division 1.1 (a mass explosion is one which affects almost the entire quantity present virtually instantaneously); Substances, mixtures and articles which have a projection hazard but (b) Division 1.2 not a mass explosion hazard; (c) Division 1.3 Substances, mixtures and articles which have a fire hazard and either a minor blast hazard or a minor projection hazard or both, but not a mass explosion hazard: (i) ble radiant heat; or combustion of which gives rise to considera which burn one after another, producing minor blast or projection effects or (ii) both; (d) Division 1.4 Substances, mixtures and articles which present no significant hazard:  substances, mixtures and articles which present only a smal l hazard in the event of ignition or initiation. The effects are largely confined to the package and no projection of fragments of appreciable size or range is to be expected. An external fire shall not cause virtually instantaneous explosion of almost the entire contents of the package; (e) Division 1.5 Very insensitive substances or mixtures which have a mass explosion hazard:  substances and mixtures which have a mass explosion hazard but are so insensitive that there is very little probability of initiat ion or of transition from burning to detonation under normal conditions; Extremely insensitive articles which do not have a mass explosion (f) Division 1.6 hazard:  articles which contain only extremely insensitive substances or mixtures and which te a negligible probability of accidental initiation or propagation. demonstra 2.1.2.3 . Explosives, which are not classified as an unstable explosive, shall be classified in I of one of the six divisions referred to in section 2.1.2.2 based on Test Series 2 to 8 in Part the UN RTDG, Manual of Tests and Criteria according to the results of the tests laid down in 2.1.1: Table Table 2.1.1 Criteria for explosives Criteria Category For explosives of Divisions 1.1 to 1.6, the following are the core set of tests that need to be performed: Explosibility: according to UN Test Series 2 (section 12 of the UN RTDG, Manual of Tests and Criteria). Intentional explosives (¹) Unstable explosives or 2. shall not be subject to UN Test Series explosives of Divisions 1.1 to 1.6 Sensiti 3 (section 13 of the veness: according to UN Test Series UN RTDG, Manual of Tests and Criteria). Thermal stability: according to UN Test 3(c) (sub - section 13.6.1 of the UN RTDG, Manual of Tests and Criteria).

93 Guidance on the Application of the CLP Criteria 5.0 – July 2017 93 Version rect Division. Further tests are necessary to allocate the cor (¹) This comprises substances, mixtures and articles which are manufactured with a view to producing a practical, explosive or pyrotechnic effect. Where the test is conducted in the package form and the packaging is changed, a further test must be conducted where it is considered that the change in packaging will affect the outcome of the test. Classification tests must be performed on the substance or mixture as supplied. If the same chemical is to be presented in a physical form different from that which was tested and which is considered likely to materially alter its performance in a classification test, the substance or mixture must also be tested in the new form. 2.1.4.4. Testing and evaluation of hazard information le, these must be evaluated against the set criteria for classification. Where test data are availab When the screening procedure indicates that a substance or mixture may possess explosive properties, a cautious approach when performing the tests is necessary to ensure safe handling . 2.1.4.5 where the individual For information on the test procedures see the following Section test series are described in context with the respec tive decision logic. The test procedures for the classification of explosives are described in detail in the Part I of the UN - MTC. 2.1.4.5. Classification procedure and decision logics Any substance, mixture or article having , or suspected of having , explosives cha racteristics must be considered for classification in the hazard class of explosives. Substances, mixtures and articles classified in this hazard class must be assigned to the appropriate division or must be classified as unstable explosive. process tion The classifica is divided into two stages, the acceptance procedure and the assignment procedure. In the acceptance procedure, intrinsic explosive properties of a substance, mixture or article are xplosion effects. If the substance, determined through tests of its sensitivity, stability and e mixture or article is not characterised as unstable explosive and is provisionally accepted into the class of explosives, it is then necessary to ascertain the correct division by applying the assignment procedure. urther subdivision into compatibility groups A to S is described in The f detail in the UN RTDG Model Regulations , Section 2.1.2. The compatibility groups and their recommended combination identify types of explosives which are deemed to be compatible, e.g. for combined storage or transportation and can therefore be used to distinguish technical requirements (especially) in these sectors. However, assignment of compatibility groups is not part of the classification system according to CLP. The tests for acceptanc e and the further tests to determine the correct division are grouped into eight test series. Classification procedures, test methods and criteria are described in detail in Part I of the UN - MTC. NOTE: The person responsible for the classification of explosives should be experienced in this field and be familiar with the criteria for classification.

94 Guidance on the Appl ication of the CLP Criteria Version 5.0 – 94 July 2017 2.1.4.5.1. Acceptance procedure The acceptance procedure is used to determine whether or not a substance, mixture or article is a candidate for the class of explosives or is an unstable explosive. The test methods used for deciding on provisional acceptance into the class of explosives are grouped into four series, numbered 1 to 4 (see CLP Annex I, Figure 2.1.2 reported below).

95 Guidance on the Application of the CLP Criteria 5.0 – Version 95 July 2017 Annex I: Figure 2.1.2 Procedure for provisional acceptance of a substance, mixture or article in the class of explosives (Class 1 for transport)

96 ication of the CLP Criteria Guidance on the Appl 96 Version 5.0 – July 2017 The numbering of Test Series 1 to 4 relates to the sequence of assessing the results rather than the order in which the tests should be conducted. It may be important for the safety of test amounts of material, be conducted first personnel that certain tests, using small before proceeding to experiment with larger quantities. Starting the testing procedure with Test Series 3 is highly recommended, because these tests involve relatively small sample sizes, which reduces the risk to t est personnel. Test Series 1 Within Test Series 1 the question ‘Is it an explosive substance / mixture?’ is answered on the basis of the results of three types of tests to assess possible explosive effects. The question is answered ‘Yes’ if a ‘+’ is obtain ed in any of the three types of tests. If the answer is ‘No’, the substance / mixture is rejected from this class; it is not an explosive. Under certain - conditions the test Type 1 (a) can be replaced by certain tests of Test Series F, see UN MTC, 1.3.5. Section 1 The three types of test used are (recommended test is indicated within brackets): a shock test with defined booster and confinement to determine the ability Type 1 (a): , zero gap ); of the substance to propagate a detonation (UN Gap test a test to determine the effect of heating under confinement (Koenen test); Type 1 (b): and Type 1 (c): a test to determine the effect of ignition under confinement (time/pressure test). Test Series 2 mixture too insensitive for Series 2 tests are used to answer the question ‘Is the substance / acceptance into this Class?’. In general, the basic apparatus and method used is the same as that for Test Series 1 but with less stringent criteria, e.g. in the case of gap tests, the gap is answered ‘No’ if a ‘+’ is obtained in any of the used is greater than zero. The question three types of test. If the answer is ‘Yes’, the substance / mixture is rejected from this class; it is not an explosive. Under certain conditions test Type 2 (a) can be replaced by certain Series F, see UN tests of Test MTC, Section 12.3.4. - The following three types of test are used (recommended test is indicated within brackets): a shock test with defined initiation system and confinement to determine Type 2 (a): (with a defined gap e.g. 50 mm) ; sensitivity to shock (UN gap test) Type 2 (b): a test to determine the effect of heating under confinement (Koenen test); and Type 2 (c): a test to determine the effect of ignition under confinement (Time/pressure test). If the substance or mixture is manufacture d with a view to produce a practical explosive or pyrotechnic effect, it is unnecessary to conduct Test Series 1 and 2 for purposes of classification. Test Series 3 As stated above it is recommended to carry out Test Series 3 before Test Series 1 and 2 for safety reasons due to the small sample amount needed. It is also recommended to carry out Test Series 3 even if negative results have been obtained in Test Series 1 and/or 2 because only Test Series 3 gives information about the thermal stability and the sensitivity to mechanical stimuli (impact and friction). Test Series 3 is used to answer the questions ‘Is the substance / mixture thermally in the form in which it stable?’ and ‘Is the substance / mixture too dangerous for transport

97 Guidance on the Application of the CLP Criteria 5.0 – July 2017 97 Version lves tests for determining the sensitiveness of the substance or mixture was tested?’ This invo to mechanical stimuli (impact and friction), and to heat and flame. The following four types of tests are used (recommended test is indicated within brackets): a falling weight test to determine sensitiveness to impact (BAM Type 3 (a): Fallhammer); Type 3 (b): a friction; or impacted friction test to determine sensitiveness to friction (BAM friction apparatus); an elevated temperature test to determine thermal stability (t Type 3 (c): hermal stability test at 75 °C); and Type 3 (d): an ignition test to determine the response of a substance or mixture to fire (small scale burning test). The first question is answered ‘No’ if a ‘+’ is obtained in Test type 3(c). Then the substance / mixtu re is considered as thermally unstable and either classified as an unstable explosive or - reactive substance or mixture. as a self The second question is answered ‘Yes’ if a ‘+’ is obtained in any of the Test types 3(a), 3(b) or 3(d). If a ‘+’ is obtained, the substance / mixture may be encapsulated or packaged to reduce its sensitiveness to external stimuli or is classified as an unstable explosive. Furthermore, the explosive may be desensitized in order to suppress/reduce its explosive properties in which case the classification procedure has to be restarted. Test Series 4 Series 4 tests are intended to answer the question ‘Is the article , packaged article or packaged substance or mixture too dangerous ?’. Conditions which may to be transported g supply and use include high /low temperature and high relative humidity, occur durin vibration, bumping and dropping. The two types of test to be carried out are: Type 4 (a): a test of thermal stability for articles; and Type 4 (b): rom dropping. a test to determine the hazard f The question is answered ‘Yes’ if a ‘+’ is obtained in either Test type 4 (a) or 4 (b) and the substance or mixture or article is classified as an unstable explosive. It is important to note that a substance / mixture which fails Test Series 2 (i.e. it is sensitive enough for acceptance into the class of explosives) may still, if properly packaged, leave the class of explosives provided that it is not designed to have an explosive effect and does not exhibit any explosive hazard in Test Series 6 of the assignment procedure (see example for musk xylene). Such an exclusion from the class of explosives is restricted to the specific type and size of package tested. est Series 1 Especially for substances / mixtures, which have explosive properties according to T and/or 2 but can leave the class of explosives after Test Series 6 due to proper packaging, it is necessary to communicate these properties in the Safety Data Sheet (SDS). Furthermore, the mented in the SDS when they meet the results from Test types 3 (a) and 3 (b) should be docu criteria of EU test method A.14 in Regulation (EC) No 440/2008 (these are substances with a sensitiveness to impact, determined by UN Test Series 3 (a) (ii) of 40 J or less and/or a sensitiveness to friction, determined by Test Series 3 (b) (i) of 360 N or less). 2.1.4.5.2. Assignment procedure to a division The assignment procedure to one of six divisions, depending on the type of hazard they present, applies to all substances, mixtures and/or articles that are candidates for the class of explosives. A substance, mixture or article must be assigned to the division which corresponds

98 ication of the CLP Criteria Guidance on the Appl July 2017 Version 5.0 – 98 to the results of the tests to which the substance, mixture or article, as offered for supply and use, has been subjected. Other test results, and data gathered from accidents which have occurred, may also be taken into account. numbered 5 – The test methods used for assignment to a division are grouped into three series designed to provide the information necessary to answer the questions in Figu – re 2.1.3 in to 7 CLP. NOTE: The person responsible for the classification of explosives should be experienced in this field and be familiar with the criteria for classification.

99 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 99 Annex I: Figure 2.1.3 Procedure for assignment to a division in the class of explosives (Class 1 for transport)

100 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 100 Test Series 5 Test Series 5 is only carried out for explosive substances/mixtures which are very insensitive and therefore candidates for division 1.5. Typical substances/mixtures are blasting agents such as ANFO, slurries, and emulsion explosives. The results from three types of series 5 tests are used to answer the question ‘Is it a very insensitive explosive substance/mixture with a mass explosion hazard?’. The test types are (recommended test is indicated withi n brackets): Type 5 (a): a shock test to determine the sensitivity to intense mechanical stimulus (cap sensitivity test); Type 5 (b): thermal tests to determine the tendency of transition from deflagration to detonation (French or USA DDT test); and Type 5 (c): a test to determine if a substance, when in large quantities, explodes when subjected to a large fire. The question is answered ‘No’ if a ‘+’ is obtained in any of the three test types. A candidate for Division 1.5 should pass one test of each type. Test Series 6 The results from four types of series 6 tests are used to determine which division, amongst Divisions 1.1, 1.2, 1.3 and 1.4, corresponds most closely to the behaviour of the substance, in a fire resulting from internal or mixture or article to be classified if a load is involved external sources or an explosion from internal sources. The results are also necessary to assess whether a substance, mixture or article can be assigned to Compatibility Group S of ould be excluded from this class. Test Series 6 should be Division 1.4 and whether or not it sh applied to packages of substances, mixtures or articles in the condition and form in which they are offered for supply and use. The four test types are (recommended test is indicated within brackets ): a test on a single package to determine if there is mass explosion of the contents Type 6 (a): (single package test); Type 6 (b): a test on packages of an explosive substance, mixture or explosive articles, or non - packaged explosive articles, to determi ne whether an explosion is propagated from one packaged article to another (stack test); and - package to another or from a non Type 6 (c): a test on packages of an explosive substance, mixture or explosive articles, or non - packaged explosive articles, to de termine whether there is a mass explosion or a hazard from dangerous projections, radiant heat and/or violent burning or any other dangerous effect when involved in a fire (bonfire test); Type 6 (d): a test on an unconfined package of explosive articles to which special provision 347 of Chapter 3.3 of the UN RTDG Model Regulations applies, to determine if there are hazardous effects outside the package arising from accidental ignition or initiation of the contents. Test types 6 (a), 6 (b), 6 (c) and 6 (d) are performed in alphabetical order. However, it is not always necessary to conduct tests of all types. Test type 6 (a) may be waived if explosive articles are carried without packaging or when the package contains only one article. Test type 6 (b) may be waived if in each type 6 (a) test:

101 Guidance on the Application of the CLP Criteria 5.0 – July 2017 101 Version  the exterior of the package is undamaged by internal detonation and/or ignition; or the contents of the package fail to explode, or explode as feebly as would exclude  o another in test type 6(b). propagation of the explosive effect from one package t Test type 6(c) may be waived if, in a type 6(b) test, there is practically instantaneous explosion of virtually the total contents of the stack. In such cases the product is assigned to Division 1.1. sed to determine whether a 1.4S classification is appropriate and is Test type 6 (d) is a test u only used if Special Provision 347 of Chapter 3.3 of the UN RTDG Model Regulations applies. The results of test series 6 (c) and 6 (d) indicate if 1.4S is appropriate, otherwise the ification is 1.4 other than S. class mixture gives a ‘ — ‘ result (no propagation of detonation) in the Series 1 If a substance or type (a) test, the 6(a) test with a detonator may be waived. If a substance gives a ‘ — ‘ result (no or slow deflagration) in a Series 2 type (c) test, the 6 (a) test with an igniter may be waived. Test Series 7 Test Series 7 aims at military explosives (Extremely Insensitive Substance: EIS or article containing an EIS) and is generally not relevant for explosives for civil use. Therefore the - MTC, Part I, individual tests are not described here. If needed, they can be found in the UN Section 17. Test Series 8 The question whether a candidate for ammonium nitrate emulsion or suspension or gel, intermediate for blasting explosives (ANE) is insensitive enough for classification as oxidising is answered by series 8 tests. The three test types are (recommended test is indicated within brackets): Type 8 (a): a test to determine the thermal stability (Thermal Stability Test for ANE); Typ e 8 (b): a shock test to determine sensitivity to intense shock (ANE gap test); and Type 8 (c): a test to determine the effect of heating under confinement (Koenen test). Test Series 8 is used to establish whether an ammonium nitrate emulsion or suspension or gel, intermediate for blasting explosives (ANE) may leave the class of explosives or not. Substances or mixtures failing any of the tests must be classified as explosives (Division 1.1. or 1.5) or as an unstable explosive in accordance with CLP Annex I , Figure 2.1.4. If they pass all three tests they are classified as an oxidising liquid or solid.

102 Guidance on the Appl ication of the CLP Criteria 102 Version 5.0 – July 2017 Annex I: Figure 2.1.4 Procedure for the classification of ammonium nitrate emulsion, suspension or gel (ANE)

103 Guidance on the Application of the CLP Criteria – July 2017 103 Version 5.0 Hazard communication for explosives 2.1.5. Pictograms, signal words, hazard statements and precautionary 2.1.5.1. 48 statements Annex I: Table 2.1.2 Label elements for explosives Unstable Division Division Division Division Division Division Classificati Explosive on 1.3 1.4 1.1 1.5 1.6 1.2 GHS Pictogram s No signal Danger Danger Danger Danger Warning Danger Signal word Word H204: Fire H200: H202: H203: No hazard H201: Hazard H205: May Explosive; mass Explosive; Explosive; Statement statement or Unstable fire, blast explode in Explosive projection severe mass or projection hazard fire explosion projection hazard hazard hazard - P210 P210 P210 - P201 No pre P210 P210 Pre cautionary cautionary P230 P250 P230 P230 P230 P234 Statement statement P240 P280 P234 P234 P234 P234 Prevention P250 P240 P240 P240 P240 P280 P250 P250 P250 P250 P280 P280 P280 P280 - P370 + P370 + Pre P370 + - No pre P370 + P370 + P370 + cautionary P372 + P372 + cautionary P372 + P372 + P372 + P372 + P380 + statement P380 + P380 + Statement P380 + P380 + P380+P3 P373 73 P373 P373 P373 P373 Response P370 + P380 + P375 Pre - - No pre P401 P401 P401 P401 P401 P401 cautionary cautionary Statement statement Storage 48 The combination statement P370+P372+P380+P373 applies to division 1.4 except for compatibility group S in transport packaging, whereas the combination statement P370+P380+P375 applies to division 1.4 compat ibility group S in transport packaging.

104 Guidance on the Appl ication of the CLP Criteria Version 5.0 – 104 July 2017 Pre - - No pre P501 P501 P501 P501 P501 P501 cautionary cautionary Statement statement Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. The intrinsic explosive properties of substances and mixtures regarding their stability and sensitivity are only investigated within Test Series 1, 2 and 3 during the acceptance procedure. Subseque nt tests for the assignment to the Divisions 1.1, 1.2, 1.3 and 1.4 (Test Series 6) are carried out with the packaged substances, mixtures or articles. The type of packaging may significantly influence the test outcome. Consequently, there are some deficien cies in the hazard communication of the GHS for unpacked or repacked explosive substances and mixtures, especially for substances and mixtures, which are provisionally accepted in the class of explosives but are later rejected from this class due to their packaging in the assignment procedure (see CLP Annex I, Figure 2.1.1 and Figure 2.1.3 and Section 2.1.4.5.1 of this guidance). These substances and mixtures have explosive properties but there might be no hazard communication about these properties due to the subsequent classification in a hazard class other than the class of explosives. Musk xylene is ). The results of Test Series 6 for an example which illustrates this issue (see Sec tion 2.1.7.2 musk xylene in the specified packaging lead to the exclusion of this substance from the hazard class of explosives. But musk xylene on its own (unpacked) sh ows explosive properties due to heating under confinement (Koenen test). Also repacking of the substance in a packaging other than the tested one can result in a completely different outcome of Test Series 6. GHS, but should be kept in mind by everyone This issue is not sufficiently clarified under applying the CLP criteria. Additional labelling provisions 2.1.5.2. Packaging dependance 2.1.5.2.1. Explosives are normally classified in their transport packaging. The packaging itself may be crucial for the classification. This is clear from the Figure 2.1.3 in Section especially 2.1.4.5.2 when it comes to Test Series 6. The assignment of an explosive substance or mixture to a particular Division within the hazard class of explosives is thus only valid for the substance and ng in which it was tested, which is usually the transport packaging. mixture in the packagi - Because of the package dependence of the classification, paragraph 2.1.2.4 of the Annex I to the CLP prescribes: Annex I: 2.1.2.4. If explosives are unpackaged or repacked in packaging ot her than the original or similar packaging, they shall be retested. Further, according to NOTE 1 to Table 2.1.2 in Section 2.1.3 of Annex I to CLP, unpackaged explosives or explosives repacked in packaging other than the original or similar packaging must have the following label elements: Annex I: 2.1.3. Hazard communication [...] NOTE 1: Unpackaged explosives or explosives repackaged in packaging other than the original or similar packaging shall include the following label elements: (a) the pictogram : exploding bomb; (b) the signal word: “Danger”; and

105 Guidance on the Application of the CLP Criteria 5.0 – July 2017 105 Version (c) the hazard statement: 'explosive; mass explosion hazard' U nless the hazard is shown to correspond to one of the hazard categories in Table 2.1.2, in which case the corresponding symbol, signal word and/or the hazard statement shall be assigned. Normally, if explosives are unpackaged or repacked in packaging other than the original or similar packaging the classification procedure needs to be performed again in order to determine which Division the explosive belongs to in the new packaging. The label elements prescribed in NOTE 1 to Table 2.1.2, as quoted above, are the same as those of Division 1.1 and in practice this Division constitutes the most severe classification of a repackaged explosive. (P lease note that Table 2.1.2 foresees also the hazard category ‘Unstable explosive’, which is assigned on the basis of the intrinsic properties of a substance or mixture via Test Series 3 and it is not package dependent). Therefore, the CLP allows labelling of a repackaged explosive with labelling corresponding to Division 1.1 instead of retesting. This, however, overestimates the hazardous properties unless the explosive in fact belongs to Division 1.1. Many explosives are supplied in inner packages which a re placed together in an outer package and where the entity as a whole, i.e. the combination of inner and outer packages, constitutes the transport packaging. According to the UN RTDG Model Regulations and the modal transport IMDG Code, ICAO TI) the classification tests are performed in regulations (ADR, RID, ADN and the transport packaging. Under Article 33(1) of CLP where the hazard pictograms(s) required by CLP relate to the same hazard as in the rules for the transport of dangerous goods, the respective CLP hazard pictogram(s) do not need to appear on the outer packaging. The classification in accordance with rules on the transport of dangerous goods is almost entirely identical to the corresponding classification procedure used in CLP and hence the CLP classification will automatically be known for the transport packaging. However, the CLP classification for the inner package alone strictly speaking is not known to the manufacturer, importer or downstream user as this will not have been derived from the classification of the transport packaging. On the other hand, it is normally not practicable to perform the required tests on the inner packages. Therefore, normally the same classification as for the transport ges. The labelling requirements for the inner packaging may be assumed for the inner packa foreseen in Table 2.1.2 of Annex I to the CLP. However, the following packages are th ose exceptions apply:  Transport packages in which the packaging is designed such that mass explosion is prevented by the pack aging, e.g. by arranging the individual inner packages crosswise (so that they are not neighbouring each other) and by separating them with specified material. This is especially the case when packing instruction P101 according to section 4.1.5 of the ADR applies. In this case the inner package should be labelled in accordance with Note 1 to Table 2.1.2 of Annex I to the CLP (i.e. as Division 1.1 unless tested otherwise).  Packages in which explosives of different divisions are contained (for such cases see especially the mixed packing provisions MP 20 to MP 24 in section 4.1.10 of the ADR).  they do not apply if the packaging is changed, as stated in Note 1 to Table Furthermore, 2.1.2 of Annex I to the CLP. 2.1.5.2.2. Supplemental hazard information Some R - phrases under DSD a re not covered by hazard classes in the current GHS. They are included as supplemental hazard statements in Part 1 of Annex II to CLP. The following EU hazard statements are important in connection with explosive properties:

106 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 106 Annex II: E UH001 – ‘Explosive when dry’ 1.1.1. For explosive substances and mixtures as referred to in chapter 2.1 of part 2 of Annex I, placed on the market wetted with water or alcohols or diluted with other substances to suppress their explosives properties. EUH001 must be assigned to explosives which are wetted, diluted, dissolved or suspended with a phlegmatizer in order to reduce or suppress their explosive properties (desensitized explosives in the sense of the foreseen new hazard class for desensitized explosives) a nd which do not meet the criteria of the hazard class of explosives. 1.1.6. EUH044 ‘Risk of explosion if heated under confinement’ Annex II: – For substances and mixtures not in themselves classified as explosive in accordance with 2 of Annex I, but which may nevertheless display explosive properties in section 2.1 of part practice if heated under sufficient confinement. In particular , substances which decompose explosively if heated in a steel drum do not show this effect if heated in less - strong containers. Some substances and mixtures which may react explosively if heated under confinement are not covered adequately by the classification system. This may e.g. be the case for:  substances or mixtures which are exempted from the class of explosives based on their packaging and according to results of the Test Series 6; substances or mixtures with a SADT of more than 75 °C for a 50 kg package which  therefore cannot be classified as self - reactive. of , in order to make the user aware EUH044 must be assigned to such substances or mixtures these properties. Further communication requirements 2.1.5.3. According to Note 2 to Table 2.1.2, explosive properties of certain substances and mixtures to the user via the which are exempted from classification as explosives must be communicated SDS (when one is required). Hazard communication Annex I: 2.1.3. [...] Substances and mixtures, as supplied, with a positive result in Test Series 2 in Part NOTE 2: I, Section 12, of the UN RTDG, Manual of Tests and Criteria, which a re exempted from classification as explosives (based on a negative result in Test Series 6 in Part I, Section 16 of the UN RTDG, Manual of Tests and Criteria,) still have explosive properties. The user shall be informed of these intrinsic explosive propert ies because they have to be considered for – especially if the substance or mixture is removed from its packaging or is handling – repackaged and for storage. For this reason, the explosive properties of the substance or mixture shall be communicated in Se ction 2 (Hazards identification) and Section 9 (Physical and chemical properties) of the Safety Data Sheet and other sections of the Safety Data Sheet, as appropriate 2.1.6. Relation to transport classification Division 1.1 – 1.6 within Class 1 of the UN RTDG Model Regulations covers explosive substances, mixtures and articles. Normally, the transport classification in accordance with the UN RTDG Model Regulations and the modal transport regulations (ADR, RID, AD N and IMDG one when deriving the CLP classification for explosives, Code, ICAO TI) can be used one - to -

107 Guidance on the Application of the CLP Criteria 5.0 – July 2017 107 Version which are packaged in authorised transport packaging. See Annex VII of this guidance for additional information on transport classification in relation to CLP classification. For the use of other packaging or for unpacked substances and mixtures the additional labelling ) have to be observed or re - testing is necessary. provisions (see Section 2.1.5.2 Examples of classification for explosives 2.1.7. Examples are given below for the classification of substances. Equivalent information would be needed for mixtures. 2.1.7.1. Example of substances and mixtures fulfilling the classification criteria a. RESULTS FROM APPLICATION OF THE ACCEPTANCE PROCEDURE Step Test Conclusion Rationale 0. General data: 0.1 Name of the substance / Hexanitrostilbene mixture: 1. Is the substance / mixture a No candidate for ammonium nitrate emulsion, suspension or gel, intermediate for blasting explosive (ANE)? 2. Is the substance / mixutre Yes manufactured with the view to producing a practical explosive or pyrotechnic effect? Test Series 3 3. 75 °C / 48 Result: ‘ — ‘, 3.1 Thermal stability: hour thermally stable test (test 3(c)) ‘ ‘, not too — BAM Result: Limiting 3.2 Impact sensitivity: impact energy 5 J Fallhammer test dangerous (test 3(a)(ii)) in form tested 3.3 Friction sensitivity: BAM friction ‘, not too — ‘ Result: Limiting load > 240 N dangerous test (test 3(b)(i)) in form tested Is the substance / mixture 4. Yes thermally stable? 5. Is the substance / mixture too No dangerous in the form in which it was tested? 6. Conclusion: PROVISIONALLY ACCEPT INTO THIS CLASS

108 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 108 Step Test Conclusion Rationale Apply the 10.1 Exit: assignment procedure b. RESULTS FROM APPLICATION OF THE ASSIGNMENT PROCEDURE Test Step Conclusion Rationale 1. Is the substance a candidate No for Division 1.5? Result: Package the substance Test Series 6 2. initiation in the Test 6(a) with detonator Result: detonation, 2.1 Effect of crater package: 2.2 Effect of propagation: Type 6(b) with detonator Result: detonation of the whole stack of packages, crater 2.4 Effect of fire engulfment: Test 6(c) may be waived of the result of because 6(b) test. the 3. Is the result a mass explosion? Yes 4. Conclusion: Assignment to Division 1.1 Example of substances and mixtures not fulfilling the classification 2.1.7.2. criteria - MTC, Part I, Section 10.5.2, Figure 10.5. This example is taken from the UN RESULTS FROM APPLICATION OF THE ACCEPTANCE PROCEDURE c. Test Conclusion Rationale Step 0. General data: 0.1 Name of the substance / - - 2,4,6 - 5 butyl mixture: tert - xylene (musk - trinitro - m xylene) 1. Is the substance / mixutre a No candidate for ammonium nitrate emulsion, suspension or gel, intermediate for blasting explosive ANE? 2. Is the substance / mixture No manufactured with the view to

109 Guidance on the Application of the CLP Criteria 5.0 Version July 2017 109 – Conclusion Rationale Step Test producing a practical explosive or pyrotechnic effect? 3. Test Series 1 UN gap test Result:’+’, propagation 3.1 Propagation of Detonation: of detonation (test 1(a)) Fragmentation Result: Limiting Koenen test 3.2 Effect of heating under (test 1(b)) diameter 12.0 mm confinement: type ‘F’ ‘+’, shows some explosive effects heating on under confinement — Time/pressure ’, no effect on 3.3 Effect of ignition under Result: ‘ test (test ignition under confinement: confinement 1(c)(i)) Is it an explosive substance / 4. Yes mixture? Test Series 2 5. 5.1 Sensitivity to shock: UN gap test ’, not Result: ‘ — sensitive to shock (test 2(a)) Result: Limiting Koenen test Fragmentation 5.2 Effect of heating under diameter 12.0 mm (test 2(b)) type ‘F’ ‘+’, confinement: violent effect on heating under ment. confine — 5.3 Effect of ignition under Time/pressure ’, no effect on Result: ‘ ignition under test (test confinement: 2(c)(i)) confinement 6. Is the substance / mixture too No insensitive for acceptance into this class? Conclusion: Substance to be considered for this class Test Series 3 7. 75 °C/48 hour 7.1 Thermal stability: Result: ‘ — ’, thermally test (test 3(c)) stable Result: Limiting impact 7.2 Impact sensitivity: BAM Fallhammer test energy 25 J", not too (test 3(a)(ii)) dangerous in form tested.

110 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 110 Step Test Conclusion Rationale 7.3 Friction sensitivity: BAM friction ’, not too — ‘ Result: Limiting load > 360 N test (test dangerous in form tested 3(b)(i)) 8. Is the substance / mixture Yes thermally stable? 9. Is the substance / mixture too No dangerous in the form in which it was tested? 10. Conclusion: PROVISIONALLY ACCEPT INTO THIS CLASS 10.1 Exit Apply the assignment procedure The explosive properties shall be communicated in the safety data sheet in accordance with section 2.1.5.3 above.

111 Guidance on the Application of the CLP Criteria – Version 111 5.0 July 2017 RESULTS FROM APPLICATION OF THE ASSIGNMENT PROCEDURE d. Test Conclusion Step Rationale 1. No Is the substance a candidate for Division 1.5? Result: Package the substance Test Series 6 2. No significant Result: Only localised 2.1 Effect of initiation in the Test 6(a) with package: decomposition around detonator reaction detonator 2.2 Effect of ignition in the Test 6(a) with Result: Only localised No significant package: decomposition around igniter reaction igniter Type 6(b) 2.3 Effect of propagation: test not required as no effect outside package between packages in 6(a) test No effects which Result: Only slow 2.4 Effect of fire engulfment: Test 6 would hinder fire burning with black smoke occurred. fighting Is the result a mass explosion? 3. No 4. Is the major hazard that from No dangerous projections? 5. Is the major hazard radiant heat No and/or violent burning but with no dangerous blast or projection hazard? Is there nevertheless a small 6. No hazard in the event of ignition or initiation? 7. Is the substance manufactured No with the view to producing a practical explosive or pyrotechnic effect? 8. Conclusion: NOT AN EXPLOSIVE 8.1 Exit Consider for another class (e.g. flammable solid)

112 ication of the CLP Criteria Guidance on the Appl 112 Version 5.0 – July 2017 2.2. LUDING CHEMICALLY UNSTABLE GASES) FLAMMABLE GASES (INC Introduction 2.2.1. The criteria for ‘Flammable gases (including chemically unstable gases)’ are found in Annex I, Section 2.2 of CLP and are identical to those in Chapter 2.2 of GHS. 2.2.2. Definitions and general considerations for the classification of flammable gases (including chemically unstable gases) : 2.2.1. Definitions Annex I 2.2.1.1 Flammable gas means a gas or gas mixture having a flammable range with air at 20 °C and a standard pressure of 101.3 kPa. 2.2.1.2. A chemically unstable gas means a flammable gas that is able to explode even in the absence of air or oxygen. The flammable range of a flammable gas is defined between the ‘lower flammability limit’ (LFL) in air and the ‘upper flammability limit’ (UFL) in air. In technical literature, the terms ‘lower explosion limit’ (LEL) and ‘upper explosion limit’ (UEL) are often used instead of the LFL and UFL, respectively. The hazard class of flammable gases also covers chemically unstable gases as defined above. Relation to other physical hazards 2.2.3. Annex I: 2.2.2. Classification criteria [...] sified as flammable gases; see S Note: Aerosols shall not be clas ection 2.3. For flammable gases that are packaged in aerosol dispensers see 2.3 Aerosols. If classified as ey do not have to be classified as flammable gases in addition. aerosols, th 2.2.4. Classification of substances and mixtures as flammable gases (including chemically unstable gases) Identification of hazard information 2.2.4.1. VI of CLP and more gases are classified Many gases are classified as flammable gases in Annex as flammable gases in the UN RTDG Model Regulations. For gases that are not classified as flammable gases in Annex VI of CLP nor in the UN RTDG Model Regulations, there is ample scientific literature giving the flam mability range for most Part 20 - 20 - 1, Explosive atmospheres – - gases (e.g. IEC 60079 1: Material characteristics for gas and vapour classification – Test methods and data as amended). recognised international In the case a gas or gas mixture needs to be tested for flammability, a standard must be used such as the EN 1839, Determination of explosion limits of gases and vapours as amended or ISO 10156, Gases and gas mixtures – Determination of fire potential and oxidising ability for the selection of cylinde r valves outlets as amended. Information on a number of chemically unstable gases can be found in the UN - MTC, Section 35. Tables 35.1 and 35.2 within UN - MTC, Section 35.3.2.1 contain information on a number of chemically unstable gases together with their classification and Category.

113 Guidance on the Application of the CLP Criteria 5.0 – July 2017 113 Version If information on other gases than the ones mentioned in the above tables is needed a test - method for determination of chemical instability of gases and gas mixtures is described in UN MTC, Section 35. However, it should be no ted that this test method is not applicable to liquefied gas mixtures. In case the gaseous phase above a liquefied gas mixture may become chemically unstable after withdrawal, this should be communicated via the SDS. 2.2.4.2. ting for gas mixtures Screening procedures and waiving of tes There are thousands of gas mixtures on the market and there are a limited number of test reports for the flammability of gas mixtures in the scientific literature. Tests to determine the flammability range are time consuming and expen sive for gas mixtures which are often prepared on demand. In most of the cases, the formulator of the gas mixture will use a as described in ISO 10156 as amended (see Section 2.2.4.4 calculation method ) to determine if the mixture is flammable or not. If the calculations in accordance with ISO 10156 as amended show that a gas mixture is not flammable it is also not classified as chemically unstable and therefore it is not necessary to carry out the tests for determining chemical instability for classification purposes. Expert judgement should be applied to decide whether a flammable gas or gas mixture is a candidate for classification as chemically u nstable in order to avoid unnecessary testing of gases where there is no doubt that they are stable. Functional groups indicating chemical instability in gases are triple bonds, adjacent or conjugated double bonds, halogenated double - bonds and - gs. strained rin Gas mixtures containing only one chemically unstable gas are not considered as chemically unstable and therefore do not have to be tested for classification purposes if the concentration of the chemically unstable gas is below the higher of the followi ng generic concentration limits: a. the lower explosion limit (LEL) of the chemically unstable gas; or b. 3 mole%. Furthermore, for some gases there are also specific concentration limits available and these are indicated in the tables 35.1 and 35.2 within UN - MT C, Section 35.3.2.1. 2.2.4.3. Classification criteria The criteria for the classification of flammable gases (including chemically unstable gases) are given in the following tables:

114 ication of the CLP Criteria Guidance on the Appl 114 Version 5.0 – July 2017 Annex I: Table 2.2.1 2.2.2. Criteria for flammable gases Category Criteria kPa: Gases, which at 20 °C and a standard pressure of 101.3 (a) are ignitable when in a mixture of 13 % or less by volume in air; or 1 (b) have a flammable range with air of at least 12 percentage points regardless of the lower flammable limit. Gases, other than those of Category 1, which, at 20 °C and a standard pressure of 2 101.3 kPa, have a flammable range while mixed in air. Table 2.2.2 2.2.2 Annex I: Criteria for chemically unstable gases Category Criteria Flammable gases which are chemically unstable at 20 °C and a pressure of 101.3 A kPa. Flammable gases which are chemically unstable at a temperature greater than 20 B °C and/or a pressure greater than 101.3 kPa. Testing and evaluation of hazard information 2.2.4.4. test method and a calculation method for the classification ISO 10156 as amended describes a of flammable gases. The test method may be used in all cases, but must be used when the calculation method cannot be applied. the T for all The calculation method applies to gas mixtures and can be applied when Ci for all inert components are available. These are listed for a flammable components and the K k number for gases in ISO 10156 as amended. In the absence of T value for a flammable gas, Ci the value of the LFL can be used and ISO 10156 propos es the value of 1.5 where no K value is k listed. The calculation method described in ISO 10156 as amended uses the criterion that a gas mixture is considered non - flammable in air if: n ' A i  1  2.2.4.4 Equation 1 . T 1  i c i where: A i A '  i p n 2 Equation 2.2.4.4 .  B K A   k k i 1 1   k i and where: :th flammable gas in the mixture, in % is the equivalent content of the i A ' i

115 Guidance on the Application of the CLP Criteria 5.0 – July 2017 115 Version T ci is the maximum content of flammable gas i which, when mixed with nitrogen, is not flammable in air, in % is the molar fraction of the i :th flammable gas in the mixture, in % A i :th inert gas in the mixture, in % is the molar fraction of the k B k is the coefficient of equivalency of the inert gas k relative to nitrogen K k is the number of flammable gases in the mixture n is the number of inert gas es in the mixture p The principle of the calculation method is the following: Where a gas mixture contains an inert diluent other than nitrogen, the volume of this diluent is adjusted to the equivalent volume of nitrogen using the equivalency coefficient fo r the inert gas , which 2 . From this the equivalent contents . 2.2.4.4 Equation are then derived through ' A K k i should be viewed as the corresponding concentration of the flammable gases if nitrogen was the 2.2.4.4 1 the equivalent contents are then only inert gas present in the mixture. In . Equation compared to the constants , which have been experimentally found using nitrogen as the T ci (only) inert gas. r most gases It should be noted that ISO 10156 uses molar fractions in some of its equations. Fo - under normal (i.e. non extreme) conditions, however, the volume fraction can be assumed to be equal to the molar fraction, which is the same as assuming ideal gas behaviour for all gases in the mixture. Furthermore, although normally a fractio n is a number ranging from 0 to 1, in this case it is easier to express it as percentage, i.e. the fraction multiplied by 100. The calculation method described in ISO 10156 as amended determines only if the mixture is flammable or not. It does not determin e a flammability range and therefore the calculation method cannot determine if the mixture is flammable Category 1 or Category 2. Therefore, to be on the safe side, mixtures determined to be flammable according the calculation method are classified Flamma ble gas; Category 1. If, however, there is a need to distinguish between Category 1 and Category 2, the lower and the upper explosion limits have to be determined by using a suitable test method (e.g. EN 1839 or ISO 10156 as amended). For mixtures containi ng both flammable and oxidising components, special calculation methods are described in ISO 10156 as amended. Gases or compressed gas mixtures that are classified as flammable have to be considered for classification as chemically unstable in addition. If the screening procedures described in Section 2.2.4.2 The test method is are not conclusive, the gas or gas mixture has to be tested. described in UN - MTC, Section 35. It uses the same equipment as the test method for oxidising gases according to ISO 10156 as amended and therefore could be applied by laboratories that also carry out the tests for oxidising gases. 2.2.4.5. Decision logic Classification of flam mable gases is laid down in the following flow - charts which are applicable according to CLP. NOTE: The person responsible for the classification of flammable gases (including chemically unstable gases) should be experienced in this field and be familiar with the criteria for classification.

116 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 116 2.2.4.5.1. Decision logic for flammable gases Annex I: Figure 2.2.1 Flammable gases Gaseous substance or mixture of gases Does it have a flammable range with air at Not classified NO °C and a standard pressure of 101.3 kPa? 20 YES At 20 °C and a standard pressure of 101.3 kPa, does it: Category 1 a. ignite when in a mixture of 13 % or less by vo lume in air?; or YES have a flammable range with air of at b. least 12 percentage points regardless Danger of the lower flammable limit? NO Category 2 No pictogram Warning

117 Guidance on the Application of the CLP Criteria – July 2017 117 Version 5.0 2.2.4.5.2. chemically unstable gases Decision logic for Annex I: Figure 2.2.2 Chemically unstable gases Flammable gas or gas mixture Category A (chemically unstable gas) Is it chemically unstable at 20 °C and a No additional YES standard pressure of 101.3 kPa? pictogram No additional signal word NO Category B Is it chemically unstable at a temperature greater than 20 °C and/or a pressure (chemically YES greater than 101.3 kPa? unstable gas) No additional pictogram NO No additional signal word classified as Not chemically unstable

118 Guidance on the Appl ication of the CLP Criteria 118 5.0 – July 2017 Version 2.2.5. Hazard communication for flammable gases (including chemically unstable gases) Pictograms, signal words, hazard statements and precautionary 2.2.5.1. statements Annex I: Table 2.2.3 2.2.3. ncluding chemically unstable gases) Label elements for flammable gases (i Flammable gas Chemically unstable gas Classification 1 Category 2 Category A Category B Category No additional No additional No pictogram GHS Pictogram pictogram pictogram No additional No additional Signal Word Danger Warning signal word signal word Additional hazard statement H231: Additional hazard statement H230: May react H221: Flammable May react explosively even H220: Extremely Hazard absence of flammable gas explosively even in the Statement gas air at elevated in the absence of air pressure and/or temperature Precautionary Statement P202 P210 P210 P202 Prevention Precautionary P377 P377 Statement P381 P381 Response Precautionary Statement P403 P403 Storage Precautionary Statement Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2.

119 Guidance on the Application of the CLP Criteria – 119 5.0 Version July 2017 Relation to transport classification 2.2.6. The criteria for flammable gases Category 1 correspond to the criteria that are in use for ations. Consequently all gases listed as classifying flammable gases in the UN RTDG Model Regul flammable in the UN RTDG Model Regulations and in the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI) must be classified as Flam.Gas 1; H220. See Annex VII for additional information on transport classification in relation to CLP classification. 2.2.7. Example of classification for flammable gases EXAMPLE MIXTURE: 2 % (H ) + 6 % (CH ) + 27 % (AR) + 65 % (HE) 4 2 Calculation steps: Step 1 : Assign the gases and state their molar fractions, assuming the molar fractions are equal to the volume fractions (ideal gas behaviour for all gases). is flammable gas 1, H 2 A yielding = 2 mole % 1 CH is flammable gas 2, 4 A yielding = 6 mole % 2 Ar is inert gas 1, B yielding = 27 mole % 1 He is inert gas 2, B = 65 mole % yielding 2 =2 since there are two flammable gases in the mixture n =2 since there are two inert gases in the mixture p Step 2: and in ISO 10156 as amended. Look up the values of T K ci k = 5.5 mole % T c 1 = 8.7 mole % T 2 c 0.55 K = 1 0.9 K = 2 Calculate the equivalent gas contents for the flammable gases according to Step 3: A ' i Equation 2.2.4.4 . 2 2 = 2.46 mole % '  A 1     6 2  65 9    .  0 27 55 . 0 6 = 7.38 mole % ' A  2     6 2 65 9 .      0 27 55 . 0 Step 4: Calculate the flammability of the gas mixture according to Equation 2.2.4.4 . 1

120 ication of the CLP Criteria Guidance on the Appl – Version 5.0 120 July 2017 2 A ' A ' A ' 38 7 . 2 . 46 i 2 1     = 1.29  5 8 . T . T 5 T 7 1 i  ci c 1 2 c Step 5 : Compare the outcome to the criterion in Equation 2.2.4.4 . 1 Since 1.29 > 1, this particular gas mixture is considered to be flammable.

121 Guidance on the Application of the CLP Criteria 5.0 – July 2017 121 Version AEROSOLS 2.3. Introduction 2.3.1. Identical criteria related to the flammability of aerosols are found in Annex I, Section 2.3 of CLP, Chapter 2.3 of GHS as well as in the Aerosol Dispensers Directive (ADD) 75/324/EEC. 2.3.2. Definitions and general considerations for the classification of ols aeros Annex I: Aerosols, this means aerosol dispensers, are any non - refillable receptacles 2.3.1. made of metal, glass or plastics and containing a gas compressed, liquefied or dissolved under pressure, with or without a liquid, paste or powder, and fitted w ith a release device allowing the contents to be ejected as solid or liquid particles in suspension in a gas, as a foam, paste or powder or in a liquid state or in a gaseous state. 2.3.3. Relation to other physical hazards physical hazards. There is no direct relation to other Annex I, 2.3.2.1. 1. [...] Note 2: Aerosols do not fall additionally within the scope of Sections (flammable gases), 2.5 2.2 (gases under pressure), 2.6 (flammable liquids) and 2.7 (flammable solids). Depending on their contents, aerosols may however fall within the scope of other hazard classes, including their labelling elements.

122 ication of the CLP Criteria Guidance on the Appl 122 Version 5.0 – July 2017 2.3.4. Classification of aerosols Classification criteria 2.3.4.1. Aerosols shall be classified in one of the three categories of this hazard Annex I: 2.3.2.1. class, depending on their flammable properties and their heat of combustion. They shall be 2 if they contain more than 1% components (by considered for classification in Category 1 or mass) which are classified as flammable according to the following criteria set out in this Part: Flammable gases (see S ection 2.2 ); – C, which includes Flammable Liquids according to section ° – Liquids with a flash point ≤ 93 2.6; Flammable solids (see S ection 2.7 – ); or their heat of combustion is at least 20kJ/g. Note 1: Flammable components do not cover pyrophoric, self - - reactive substances heating or water r used as aerosol contents. and mixtures because such components are neve [...] 2.3.2.2. An aerosol shall be classified in one of the three categories for this Class on the basis of its components, of its chemical heat of combustion and, if applicable, of the results of the and of the ignition distance test and enclosed space test (for foam test (for foam aerosols) spray aerosols) in accordance with Figures 2.3.1(a) to 2.3.1(c) of this Annex and sub - sections 31.4, 31.5 and 31.6 of Part III of the UN RTDG, Manual of Tests and Criteria. Aerosols which ot meet the criteria for inclusion in Category 1 or Category 2 shall be classified in do n Category 3. Note: Aerosols containing more than 1% flammable components or with a heat of combustion of at sification procedures in this least 20 kJ/g, which are not submitted to the flammability clas section shall be classified as aerosols, Category 1. - MTC, Section 31, flammability classification for aerosols refers to Under the ADD and also in UN - corresponds to the flammable’. This respectively ‘extremely flammable’, ‘flammable’ and ‘non terms ‘Aerosol, Category 1’, ‘Aerosol, Category 2’ and ‘Aerosol, Category 3’ which are used in . CLP The following identical criteria can be found in both CLP and ADD: The aerosol is classified as ‘Aerosol, Category 3’ if it contains 1 % or less flammable 49 components and the chemical heat of combustion is less than 20 kJ/g. The aerosol is classified as ‘Aerosol, Category 1’ if it contains 85 % or more flammable and the chemical heat of combustion is 30 components kJ/g or more. All other ae rosols should be submitted to the appropriate flammability classification procedures in order to select the appropriate Category 1, 2 or 3. However, if these are not submitted to the 49 Depending on their flash point value, also certain liquids not classified under CLP as Flam. Liq., Cat. 1, 2 or 3, will be considered as flammable components in an aerosol. The CLP hazard class of Flammable liquids covers liquids of flash point ≤ 60 °C while a liquid component in an aerosol is considered flammable when its flash point is ≤ 93 °C.

123 Guidance on the Application of the CLP Criteria 5.0 July 2017 123 Version – flammability classification procedures they must be automatically classif ied as ‘Aerosol, Category 1’. The chemical heat of combustion is determined in accordance with CLP Annex I, 2.3.4.1 which is identical to point 1.10 of the Annex to ADD. 2.3.4.2. Testing and evaluation of hazard information Results from the ignition distance test, the enclosed space test and the foam flammability test may be used for classification related to the flammability of aerosols. These test methods are described under point 6.3 of the Annex to ADD and are therefore available in all EU languages. They are al - MTC Section 31. so described in the UN After evaluation according to the appropriate criteria (see previous sections) the aerosol is classified in one of the three categories. 2.3.4.3. Decision logic The classification procedure is also laid down in the following - charts which are applicable flow according to CLP. NOTE: The person responsible for the classification of aerosols should be experienced in this field and be familiar with the criteria for classification.

124 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 124 2.3.4.3.1. Decision logic for aerosols Figure 2.3.1 (a) Annex I: for aerosols AEROSOL Category 3 YES ≤ 1% flammable Does it contain No pictogram does it have a and components (by mass) heat of combustion < 20 kJ/g? Warning NO Category 1 Does it contain ≥ 85% flammable and does it have a components (by mass) YES heat of combustion ≥ 30 kJ/g? Danger NO ' For spray aerosols, go to decision logic 2.3.1(b) For foam aerosols, go to decision logic 2.3.1(c)'

125 Guidance on the Application of the CLP Criteria 5.0 – July 2017 125 Version Decision logic for spray aerosols 2.3.4.3.2. Annex I: (b) Figure 2.3.1 Spray aerosols SPRAY AEROSOL Category 1 In the ignition distance test, does ignition occur YES 75 cm? at a distance ≥ Danger NO Category 2 Does it have a heat of combustion < 20 kJ/g? NO Warning YES In the ignition distance test, does ignition occur Category 2 at a distance ≥ 15 cm? YES NO Warning In the enclosed space ignition test; is: Category 2 ( a) the time equivalent ≤ 300 s/m³or YES b) the deflagration density ≤ 300 g/m³? ( Warning NO Category 3 No pictogram Warning

126 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 126 2.3.4.3.3. Decision logic for foam aerosols Annex I: Figure 2.3.1 (c) Foam aerosols FOAM AEROSOL Category 1 In the foam test, is: YES ( a) the flame height ≥ 20 cm and the flame duration ≥ 2 s; or duration ≥ 7 s ? ( b) the flame height ≥ 4 cm and the flame Danger NO Category 2 In the foam test; is the flame height ≥ 4 cm and the flame YES ? duration ≥ 2 s Warning NO Category 3 No pictogram Warning

127 Guidance on the Application of the CLP Criteria July 2017 127 Version 5.0 – Hazard communication for aerosols 2.3.5. Pictograms, signal 2.3.5.1. words, hazard statements and precautionary statements Table 2.3.1 Annex I: Label elements for aerosols Category 3 Category 2 Classification Category 1 No pictogram GHS Pictograms Danger Warning Signal Word Warning H222: Extremely H223: Flammable aerosol flammable aerosol Hazard Statement H229: Pressurised H229: Pressurised H229: Pressurised container: May burst container: May burst container: May burst if heated. if heated. if heated. P210 P210 P210 Precautionary P211 P211 P251 Statement Prevention P251 P251 Precautionary Statement Response Precautionary P410 + P412 P410 + P412 P410 + P412 Statement Storage Precautionary Statement Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. 2.3.5.2. Additional labelling provisions The ADD imposes additional labelling requirements on all aerosols, flammable or not. For example: Where an aerosol dispenser contains flammable components but is not classified as flammable (i.e. ‘Aerosol, Category 3’), the quantity of flammable material contained in the aerosol dispenser must be stated clearly on the label, in the form of the following legible and indelible wording: ‘X % by mass of the contents are flammable’.

128 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 128 2.3.6. Relation to transport classification Aerosol di spensers (UN 1950) belong to Class 2 in the UN RTDG Model Regulations and in the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI). Flammability , classification criteria are harmonised between CLP and in the modal transport regulations (ADR RID, ADN and IMDG Code, ICAO TI). Aerosols, Category 1 and 2 fall under Division 2.1 (sometimes referred to as Class 2.1 or Group F, FC, TF or TFC depending on their contents with hazardous properties). Aerosols, Category 3 imes referred to as Class 2.2 or Group A, O, T, C, CO, TC or TOC fall under Division 2.2 (somet depending on their contents with hazardous properties). See Annex VII for additional information on transport classification in relation to CLP classification. 2.3.7. Examples of classification for aerosols For reasons of simplification the active materials chosen in the examples have been considered as non - combustible materials (  H = 0 kJ/g). However this is not the case in practice. c 2.3.7.1. Examples of aerosols fulfilling the classification criteria orant: Deod Composition: Hc = 43.5 kJ/g)  70 % (flammable components, Butane/propane: Hc = 24.7 kJ/g)  25 % (flammable components, Ethanol: Hc = 0 kJ/g) 5 % (non - flammable components,  Others: This spray aerosol contains 95 % of flammable components, and its chemical heat of combustion kJ/g (= 0.70 * 43.5 + 0.25 * 24.7). equals 36.6 This aerosol is classified as Aerosol, Category 1. Air freshener (wet): Composition: 30 % (flammable components,  Hc = 43.5 kJ/g) Butane/propane: flammable components, 70 % (non -  Hc = 0 kJ/g) Others: This spray aerosol contains 30 % of flammable components and its chemical heat of combustion equals 13.1 kJ/g. In the ignition distance test, the ignition occurs at less than 75 cm but more than 15 cm. This aerosol is classified as Aerosol, Category 2. Shaving foam: Composition: 4 % (flammable components,  Hc = 43.5 kJ/g) Butane/propane: Hc = 0 kJ/g)  flammable components, - 96 % (non Others:

129 Guidance on the Application of the CLP Criteria 5.0 July 2017 129 Version – This foam aerosol contains 4 % of flammable components and its chemical heat of combustion equals 1.7 kJ/g. In the foam test, the flame height is less than 4 cm and the flame duration less than 2 s. This aerosol is classified as . Aerosol, Category 3 However, according to the requirements of ADD, the quantity of flammable components must be stated clearly on the label: ‘4% by mass of the contents are flammable’. 2.3.7.2. Examples of aerosols not fulfilling the classification criteria By definition, all aerosol dispensers fall under one of the three categories for this hazard class.

130 ication of the CLP Criteria Guidance on the Appl 130 Version 5.0 – July 2017 2.4. OXIDISING GASES Introduction 2.4.1. The requirements in Chapter 2.4 ‘Oxidising gases’ of Annex I of CLP are identical to those in chapter 2.4 of the GHS. 2.4.2. Definitions and general considerations for the classification of oxidising gases 2.4.1. ture which may, generally by Annex I: Oxidising gas means any gas or gas mix providing oxygen, cause or contribute to the combustion of other material more than air does. Relation to other physical hazards 2.4.3. Oxidising gases do not need to be classified in any other hazard class apart from ‘Gases under essure’ where appropriate. pr Classification of substances and mixtures as oxidising gases 2.4.4. 2.4.4.1. Identification of hazard information There are not many pure gases that are oxidising. Most oxidising gases are identified as such in Determination of the UN RTDG Model Regulations a nd in ISO 10156 Gases and gas mixtures : fire potential and oxidizing ability for the selection of cylinder valve outlets as amended . Screening procedures and waiving of testing 2.4.4.2. There are thousands of gas mixtures containing oxidising gases on the market and there are very few test reports on oxidising potential of gas mixtures in the scientific literature. Tests according to ISO 10156 as amended in order to determine the oxidising potential are time which are often prepared on demand. In most of the consuming and expensive for gas mixtures cases, the formulator of the gas mixture will use a calculation method as described in ISO 10156 as amended. 2.4.4.3. Classification criteria 2.4.2 . Table 2.4.1 Annex I: Criteria for oxidising gases Category Criteria Any gas which may, generally by providing oxygen, cause or contribute to the 1 combustion of other material more than air does. Note: ‘Gases which cause or contribute to the combustion of other material more than air does’ means pure gases or gas mixtures with an oxidising power greater than 23.5 % as determined by a method specified in ISO 10156 as amended. Please note that ISO 10156 - 2:2005 has been integrated into the revised version ISO 10156:2010. ISO 10156:2010 supersedes EN 720 - 2:1996 an d ISO 10156 - 2:2005.

131 Guidance on the Application of the CLP Criteria July 2017 131 5.0 Version – Testing and evaluation of hazard information 2.4.4.4. ISO 10156 as amended describes a test method and a calculation method for the classification of oxidising gases. The test method may be used in all cases, but must be used when the calculation method cannot be applied. for all The calculation method applies to gas mixtures and can be applied only when the C i oxidising components and the K for all inert components are available. These are listed for a k number of gases in ISO 10156 as amended. For gas mixtures the calculation method described criterion that a gas mixture should be considered as more in ISO 10156 as amended uses the oxidising than air if the ‘Oxidising Power’ (OP) of the gas mixture is higher than 0.235 (23.5 %). The OP is calculated as follows: n x C  i i  i 1 . 1 Equation 2.4.4.4  OP p n x B K    k i k 1 1   i k Where: x is the molar fraction of the i :th oxidising gas in the mixture, in % i C is the coefficient of oxygen equivalency of th e i :th oxidising gas in the mixture i is the coefficient of equivalency of the inert gas k relative to nitrogen K k B is the molar fraction of the k :th inert gas in the mixture, in % k is the number of oxidising gases in the mixture n is the number of inert gases in the mixture p For mixtures containing both flammable and oxidising components, special calculation methods are described in ISO 10156 as amended. 2.4.4.5. Decision logic Classification of oxidising gases is done according to decision logic 2.4.4.1 as included in the GHS. NOTE: The person responsible for the classification of oxidising gases should be experienced in this field and be familiar with the criteria for classification. 2 . 1 Decision logic for oxidising gases (Decision logic 2.4 of GHS) Figure Gaseous substance or mixture of gases Category 1 YES Does the gas contribute to the combustion of other material more than air does? Danger NO Not classified

132 ication of the CLP Criteria Guidance on the Appl 132 Version 5.0 – July 2017 2.4.5. Hazard communication for oxidising gases Pictograms, signal words, hazard statements and precautionary 2.4.5.1. statements Annex I: Table 2.4.2 Label elements for oxidising gases Category 1 Classification GHS Pictogram Signal word Danger H270: May cause or intensify fire; oxidiser Hazard statement P220 Precautionary Statement Prevention P244 Precautionary Statement Response P370 + P376 Precautionary Statement Storage P403 Precautionary Statement Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. 2.4.6. Relation to transport classification Most oxidising gases are classified as such with subsidiary risk 5.1 in the UN RTDG Model Model Regulations and in Regulations. Consequently all gases listed as oxidising in the UN RTDG the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI) must be classified as Ox. Gas 1. See Annex VII for additional information on transport classification in relation to CLP classification. Example of classificat ion for oxidising gases 2.4.7. 2.4.7.1. Example of substances and mixtures not fulfilling the classification criteria EXAMPLE OF A CLASSIFICATION USING THE CALCULATION METHOD OF ISO 10156 AS AMENDED Example Mixture: 9 % (O ) + 16 % (N O) + 75 % (N ) 2 2 2 Calculation steps Step 1 : Ascertain the coefficient of oxygen equivalency ( C ) for the oxidising gases in the mixture and i the nitrogen equivalency factors (K oxidising gases. ) for the non - flammable, non - k

133 Guidance on the Application of the CLP Criteria 5.0 Version July 2017 133 – C (N O) = 0.6 (nitrous oxide) i 2 (O) = 1 (oxygen) C i K (N ) = 1 (nitrogen) 2 k : Calculate the Oxidising Power (OP) of the gas mixture according to Equation 2.4.4.4 . 1 Step 2 n x C  i i    0 . 09 1 6 . 0 16 . 0  1 i  OP  . 0  186 p n 0 09 0 0   . 1  . 16 75 .  B K x   k i k   k i 1 1 0.186 < 0.235 (18.6 % < 23.5 %), therefore the mixture is not considered as an oxidising gas.

134 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 134 2.5. GASES UNDER PRESSURE 2.5.1. Introduction The requirements in Chapter 2.5 ‘Gases under pressure’ of Annex I of CLP are identical to those in Chapter 2.5 of GHS. The hazard class ‘Gases under pressure’ corresponds to Class 2 ‘Gases’ in the UN RTDG Model Regulations. 2.5.2. Definitions and general consid erations for the classification of gases under pressure Definition of ‘gas’ 2.5.2.1. Annex I: Gas means a substance which (i) at 50 °C has a vapour pressure greater 1.0. than 300 kPa (absolute); or (ii) is completely gaseous at 20 °C at a standard pressure of 101.3 kPa; This definition means that substances and mixtures are considered as gases when their boiling point or initial boiling point (BP) is not higher than 20 °C. Substances and mixtures with a boiling point or initial boiling point higher than 20 °C are l iquids except those few that develop a vapour pressure higher than 300 kPa at 50 °C; these substances and mixtures are considered as gases because of the pressure hazard when packaged. Hydrogen fluoride (HF) with a BP of 19.4 °C is a borderline line case t hat has always been classified as a liquid. 2.5.2.2. Definition of gases under pressure Gases under pressure are gases or gas mixtures which are contained in a Annex I: 2.5.1.1. kPa (gauge) or more at 20 °C , or which are liquefied or receptacle at a pressure of 200 liquefied and refrigerated. They comprise compressed gases, liquefied gases, dissolved gases and refrigerated liquefied gases. This definition means in practice that compressed gases or dissolved gases that are packaged at a pressure less than 200 kPa are not classified for this hazard. Dissolved gases packaged at a pressure less than 200 kPa (gauge) are liquids and should be classified as such if they have other hazardous properties, e.g. flammable liquids. Also, liquids packaged under a layer of inert gas (e.g. nitrogen or helium) remain to be classified as liquids and not as gases under pressure. 2.5.3. Relation to other physical hazards Gases under pressure may also need to be classified for the hazard classes flammable gases and oxidising gases where relev ant. 2.5.4. Classification of substances and mixtures as gases under pressure 2.5.4.1. Identification of hazard information Many gases are identified as such in the UN RTDG Model Regulations and many flammable gases and some oxidising gases are identified as gases in An nex VI of CLP. The UN RTDG Model Regulations identifies further if the gas can be packaged as a ‘compressed gas’, a ‘liquefied gas’, a ‘refrigerated liquefied gas’ and a ‘dissolved gas’. To determine whether a substance is a gas in

135 Guidance on the Application of the CLP Criteria 5.0 – July 2017 135 Version case it is not listed in the UN RTDG Model Regulations and in case of doubt, the following physical characteristics are necessary:  the boiling point; the vapour pressure at 50 °C.  , Section R.7.1.3 (Boiling point), IR & CSA, Chapter R.7a: Endpoint specific guidance See also R.7. 1.5 (Vapour pressure). For those substances that meet the definition of a gas (see Section ), the critical 2.5.2 temperature is also necessary. For t he classification of gas mixtures based on the pseudo - 2.5.4.3 critical temperature see Section . The references according to Section provide good quality data on boiling points, vapour 2.6.8 pressure and the critical temperature of substances. Classification criteria 2.5.4.2. Annex I: Table 2.5.1 Criteria for gases under pressure Criteria Group Compressed A gas which when packaged under pressure is entirely gaseous at  - 50 °C; including all gases with a critical temperature - 50 °C. gas is partially liquid at A gas which, when packaged under pressure, - 50 °C. A distinction is made between: temperatures above i) high pressure liquefied gas: a gas with a critical temperature between Liquefied gas - 50 °C and + 65 °C; and ii) low pressure liquefied gas: a gas with a critical temperature above + 5 °C. 6 A gas which when packaged is made partially liquid because of its low Refrigerated liquefied gas temperature. A gas which when packaged under pressure is dissolved in a liquid phase Dissolved gas solvent. Note: Aerosols shall not be classifie d as gases under pressure. See S ection 2.3 . 2.5.4.3. Testing and evaluation of hazard information The critical temperature of pure gases is well defined and can be found in technical literature, e.g. EN 13096 — Conditions for filling gases into receptacles — Transportable gas cylinders as amended. Single component gases For gas mixtures, the classification is based on the ‘pseudo - critical temperature’ which can be estimated as the mole weighted average of the c omponents’ critical temperatures. n T x  i Pseudo critical temperature - =  Crit i i  1

136 ication of the CLP Criteria Guidance on the Appl 5.0 – 136 July 2017 Version where x is the molar concentration of component i and T is the critical temperature (in °C i Crit i . i or in K) of the component Decision logic 2.5.4.4. Classification of gases under pressure is done according to decision logic 2.5.4.1 as included in the GHS. NOTE: The person responsible for the classification of gases under pressure should be experienced in this field and be familiar with the criteria f or classification.

137 Guidance on the Application of the CLP Criteria – July 2017 5.0 137 Version 2 . 2 Decision logic for gases under pressure (Decision logic 2.5 of GHS) Figure The substance or mixture is a gas pressure of 200 kPa Is the gas contained in a receptacle at a No Not classified as a (gauge) or more at 20 °C, or is the gas liquefied or liquefied and gas under refrigerated? pressure Yes Dissolved gas Yes Is the gas dissolved in a liquid phase solvent? Warning No Refrigerated liquefied gas Yes Is the gas partially liquid because of its low temperature? Warning No No Is the gas partially liquid at temperatures above – 50 °C? (Low pressure) No Liquefied gas Yes Is its critical temperature above + 65 °C? Warning No (High pressure) Liquefied gas Yes Is its critical temperature between – 50 °C + 65°C? Warning Compressed gas Yes entirely in gaseous state at – 50 °C? Is the gas Warning

138 ication of the CLP Criteria Guidance on the Appl 138 Version 5.0 – July 2017 2.5.5. Hazard communication for gases under pressure Pictograms, signal words, hazard statements and precautionary 2.5.5.1. statements Annex I: Table 2.5.2 Label elements for gases under pressure Refrigerated Compressed gas Liquefied gas Classification Dissolved gas liquefied gas GHS Pictogram Warning Signal Word Warning Warning Warning H280: H281: Contains H280: Contains Contains gas H280: Contains refrigerated gas under under gas under Hazard gas; may cause Statement pressure; may pressure; may pressure; cryogenic burns explode if heated may explode explode if heated or injury if heate d Precautionary Statements P282 Prevention Precautionary Statements P336 + P315 Response Precautionary Statements P410 + P403 P410 + P403 P410 + P403 P403 Storage Precautionary Statements Disposal Note: Pictogram GHS04 is not required for gases under pressure where pictogram GHS02 or pictogram GHS06 appears. The wording of the Precautionary Statements is found in CLP Annex IV, Part 2.

139 Guidance on the Application of the CLP Criteria 5.0 – July 2017 139 Version Relation to transport classification 2.5.6. Gases are listed in UN RTDG Model Regulations and in the transport regulations (ADR, RID, 50 with an indication of the physical state in their name for compressed gases (e.g. Argon, ADN) compressed), for refrigerated liquefied gas (e.g. Oxygen, refrigerated liquid) and for dissolved gas (e.g. Acetylene, dissolved). These indications of the physical s tate can be used to identify the group of gases under pressure according to CLP. The gas names without an indication of the physical state are ‘liquefied gases’ by default. See Annex VII elation to CLP classification. for additional information on transport classification in r 50 The classification code s according to the ADR, Sections 2.2.2.1.2 and 2.2.2.1.3 are: 1. Compressed gas; 2. Liquefied gas; 3. Refrigerated liquefied gas; 4. Dissolved gas. A asphyxiant; O oxidizing; F flammable; T toxic; TF toxic, flammable; TC toxic, corrosive; TO toxic, oxidizin g; TFC toxic, flammable, corrosive; TOC toxic, oxidizing, corrosive.

140 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 140 Examples of classification for gases under pressure 2.5.7. Examples of substances and mixtures fulfilling the classification 2.5.7.1. criteria 2.5.7.1.1. Example mixture: 9 % (O ) + 16 % (N ) O) + 75 % (N 2 2 2 ) + 16 % (N EXAMPLE MIXTURE: 9 % (O (N ) O) + 75 % 2 2 2 Calculation steps: Step 1 : Ascertain the critical temperatures in Kelvin for the gases in the mixture: Oxygen (O ): 118.4 °C (= 154.75 K) T = - 2 Crit Nitrous Oxide (N T = +36.4 °C (= 309.55 K) O): Crit 2 T ): Nitrogen (N = - 147 °C (= 126.15 K) Crit 2 Step 2 : Calculate the pseudo - critical temperature: 0.09  154.75 K + 0.16  309.55 K + 0.75  126.15 K= 158.7 Kelvin = - 115.08 °C The pseudo - critical temperature is lower than - 50 °C, therefore the mixture is a ‘compressed gas’ .

141 Guidance on the Application of the CLP Criteria 5.0 – July 2017 141 Version 2.6. FLAMMABLE LIQUIDS 2.6.1. Introduction The criteria for ‘Flammable liquids’ are found in Annex I, Section 2.6 of CLP and are not identical to those of GHS as the respective GHS Chapter 2.6 contains additional classification - Category 4 for flammable liquids. criteria 2.6.2. Definitions and general considerations for the classification of flammable liquids Annex I: Flammable liquid means a liquid having a flash point of not more than 60 °C. 2.6.1. pressure of The flash point is the lowest temperature of the liquid, corrected to a barometric kPa, at which application of a test flame causes the vapour of the liquid to ignite 101.3 momentarily and a flame to propagate across the surface of the liquid under the specified conditions of test. This means, the lower explosion limit is exc eeded at the flash point. Relation to other physical hazards 2.6.3. For flammable liquids that are packaged in aerosol dispensers, see Section on Aerosols. If 2.3 classified as flammable aerosols, they must not be classified as flammable liquids in addition Section 2.3 ). (see 2.6.4. Classification of substances and mixtures as flammable liquids 2.6.4.1. Identification of hazard information r mixture is a liquid see Section . For the decision if a substance o 2.0.4 For the classification of a substance or mixture as a flammable liquid, data on the flash point the boiling point (or the initial boiling point) are needed. For experimental determination and on of the flash point information on the viscosity of the liquid is needed, in order to select a suitable method. Furthermore, in order to make use of the derogation for classification in 2.6.4.3 ), information on Category 3 according to Annex I Section 2.6.4.5 of CLP (see Section sustained combustibility is necessar y. Experimentally determined data or data taken from reliable data sources are to be preferred over calculated ones. See also IR & CSA, Chapter R.7a: Endpoint specific guidance , Section R.7.1.3 (Boiling point), R.7.1.9 (Flash point). The references in Section 2.6.8 provide good quality data on boiling points (all three references) and flash point (first reference) of substances. Special care is required when viscous substances or mixtures are tested or when halogenated compounds are present (see Section 2.6.4.4.1 ). 2.6.4.2. Screeni ng procedures and waiving of testing 2.6.4.2.1. Boiling point Normally calculation methods based on increments give satisfying results for substances and mixtures. With respect to the criterion for distinguishing between Category 1 and 2 (boiling point of 35 °C) only that method with a mean absolute error lower than 5 °C could be recommended for screening.

142 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 142 2.6.4.2.2. Flash point Calculation should work for pure liquids, neglecting impurities, if the vapour pressure curve and res, calculation of the flash point is lower explosion limit are accurately known. For mixtu of a sometimes not reliable and at this time, it is not possible to predict what the accuracy calculated value is. Calculation can be used as a screening test for mixtures, and a flash point erimentally if the calculated value using the method cited in CLP need not be determined exp Annex I, 2.6.4.3 is 5 °C greater than the relevant classification criterion (23 °C and 60 °C, respectively). However, the restrictions outlined in the CLP Annex I, 2.6.4.2 must be taken acco unt of. Calculation based on structural similarity or properties is often only applicable to a narrowly defined set of substances. For mixtures they are not yet applicable. Therefore for both flash point and boiling point experimental determination is recommended. 2.6.4.3. Classification criteria A flammable liquid has to be classified in one of the 3 categories of this class. Annex I: Table 2.6.1 Label elements for flammable liquids Category Criteria 1 Flash point < 23 °C and initial boiling point ≤ 35 °C 2 Flash point < 23 °C and initial boiling point > 35 °C 1 Flash point ≥ 23 °C and ≤ 60 °C 3 (1) For the purpose of this Regulation gas oils, diesel and light heating oils having a flash 3. point between > 55 °C and ≤ 75 °C may be regarded as Category Note: Aerosols shall not be classified as flammable liquids; see section 2.3. 2.6.4.5. Liquids with a flash point of more than 35 °C and not more than 60 °C Annex I: need not be classified in Category 3 if negative results have been obtained in the sustained c ombustibility test L.2, Part III, section 32 of the UN RTDG, Manual of Tests and Criteria. Gas oils, diesel and light heating oils in the flash point range of 55 °C to 75 °C may be regarded as a whole. The reason is that these hydrocarbon mixtures have varying flash points in that Automotive fuels – Die sel - Requirements and Test range due to seasonal requirements (EN 590 Methods as amended). If they are regarded as a whole for CLP they have to be regarded as 3. This states however no preliminary decision with respect to downstream Category Regulations and legislation. 2.6.4.4. Testing and evaluation of hazard information The assignment to the respective hazard category will determine the technical means to be taken to avoid dangerous events. In combination with other safety characteristics like explosion to clear restrictions in the conditions of use. limits or auto ignition temperature this can lead The relevant data are to be communicated via the CSR and SDS (see IR&CSA Part F: Chemical Safety Report , Part G: Extending the SDS and Guidance on compilation of safety data sheets respectively).

143 Guidance on the Application of the CLP Criteria 5.0 – July 2017 143 Version 2.6.4.4.1. Testing Su itable methods are listed in CLP Annex I, Table 2.6.3. In case of substances with a high decomposition potential, a method using small amounts of as Rapid equilibrium closed cup method liquid (e.g. EN ISO 3679 Determination of flash point - ommended to reduce the amount of substance under test. amended) is rec The method to be used has to be chosen taking into account the properties of the liquid (viscosity, halogenated compounds present) and the scope of the standard. mended to use the mean of at least two test runs. One of For classification purposes it is recom these runs may be automated. In case of a deviation between manual and automated determination above the tolerance limits of the method, the lower value should be taken or the repeated with manual observation. If the experimentally determined determination should be flash point is found to be within ± 2 °C a threshold limit when using a non equilibrium method, - it is recommended to repeat the determination with an equilibrium method. If no flash point is found up to 60 °C and (partly) halogenated compounds are present or if there is the possibility of loss of volatile flammable or non flammable components (i.e. the - e liquid is a candidate for the assignment of EUH018, EUH209 or EUH209A) or if in doubt, th explosion limits should be determined in order to decide whether labelling with EUH018, EUH209 or EUH209A is appropriate. Determination of explosion limits should be carried out according to EN 1839 Determination of explosion limits of gases and vapours as amended or ISO 10156 Gases and gas mixtures – Determination of fire potential and oxidising ability for the selection of cylinder valves outlets Determination of explosion points as amended or EN 15794 of flammable liquids as amended. Substances For non - halogenated substances, the flash point is usually found 80 °C to 130 °C below the boiling point. Special care has to be taken when a sample contains impurities with a lower boiling point than the main compound. Even if their concentration is below 0.5 %, especially if their boiling point is substantially lower, they may have a strong effect on the test result. Impurities with a higher boiling point will normally have no effect on the flash point. Within the respective scope, every standard is applicable. Mixtures The flash point may be lower than the lowest flash point of the components and non - volatile components may influence the flash point. Equilibrium methods are advised if the boiling points of the components of the mixture cover a eratures or their concentrations are very different. They are also advised in wide range of temp case of viscous mixtures (alternatively: test methods with low heating rates (1 °C per min) using a stirrer). In case of viscous mixtures or if an inerting substance is present a t low concentrations and this is a highly volatile compound, the ignitability of the mixture may depend on the temperature at which the tests are started. When an inerting substance is present temperature ranges may exist where the vapour phase is inerted and other temperature ranges where it is not. Halogenated compounds The difference between boiling point and flash point may be lower than with non - halogenated compounds. It is highly recommended to run the tests under careful control with manual observation. Test results may be very difficult to reproduce. In such cases, classification should be based on the lowest value found (flash or burning inside or outside the cup) or on the value obtained

144 Guidance on the Appl ication of the CLP Criteria Version 5.0 – 144 July 2017 during the screening run if in the main trial performed in accordance with the standard, no flash could be found. 2.6.4.4.2. Evaluation of hazard information Flash points determined by testing or from the mentioned internationally recognised qualified literature are to be preferred over those derived by calculation because of the error of most of the QSAR methods and their limited application range. If in literature different flash points are found for the same substance the one found as evalu ated or recommended has to be preferred. If in literature different flash points are found for the same substance where none is found as evaluated/recommended the lower one has to be preferred because of safety reasons or an ld be carried out. experimental determination shou According to the criteria either Category 1, Category 2 or Category 3, including the relevant 2.6.5 ). In case the criteria hazard statement and signal word, have to be assigned (see Section for EUH018, EUH209 or EUH209A are met, the liquid has to be labelled with the respective supplemental hazard statement as well. In the majority of cases EUH018 covers E UH209 and EUH209A. 2.6.4.5. Decision logic Compared to the decision logic 2.6 for flammable liquids contained in the GHS chapter 2.6.4.1, this decision logic below is amended to include derogations for gas oil, diesel, light heating, sustained combustibility and fo r phrases EUH018, EUH209 and EUH209A. NOTE: The person responsible for the classification of flammable liquids should be experienced in this field and be familiar with the criteria for classification.

145 Guidance on the Application of the CLP Criteria – 145 5.0 Version July 2017 Figure 2 Amended GHS decision logic for flammable liquids to include derogations for gas oil, . 3 diesel, light heating, sustained combustibility and for phrases EUH018, EUH209 and EUH209A The substance or mixture is a liquid Halogenated substance, mixture containing Gas oil, diesel, light No No No halogenated, volatile heating oil with flash Flash point ≤ 60 °C or non volatile point up to 75 °C flammable Yes substances Yes Yes Yes No Explosive vapour/air Yes Flash point < 23 °C mixture possible (EN 1839, EN 15794) No No Category 3 No Not subject of Flash point > 35 °C hazard class ‘flammable liquid Warning Yes Yes EUH209, Sustained combustibility EUH209A, EUH018 No Category 2 No Boiling point ≤ 35 °C No need to be classified as Danger ‘flammable liquid’ Yes Category 1 Danger

146 ication of the CLP Criteria Guidance on the Appl 146 Version 5.0 – July 2017 2.6.5. Hazard communication for flammable liquids Pictograms, signal words, hazard statements and precautionary 2.6.5.1. statements 2.6.3. Annex I: Table 2.6.2 Label elements for flammable liquids Category 1 Category 2 Category 3 Classification GHS Pictograms Danger Signal Word Warning Danger H224: Extremely Flammable liquid Hazard H226: H225: Highly flammable liquid and flammable liquid and and vapour Statement vapour vapour P210 Precautionary P210 P210 P233 P233 Statement P233 P240 P240 Prevention P240 P241 P241 P241 P242 P242 P242 P243 P243 P243 P280 P280 P280 P303 + P361 + P353 P303 + P361 + P353 Precautionary P303 + P361 + P353 P370 + P378 P370 + P378 P370 + P378 Statement Response Precautionary P403 + P235 P403 + P235 P403 + P235 Statement Storage Precautionary P501 P501 P501 Statement Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. Additional labelling provisions for flammable liquids 2.6.5.2. - 1.1.4. EUH018 – 'In use, may form flammable/explosive vapour air Annex II: mixture' For substances and mixtures not classified as flammable themselves, which may form flammable/explosive vapour - air mixtures. F or substances this might be the case for halogenated hydrocarbons and for mixtures this might be the case due to a volatile flammable component or due to the loss of a volatile non - flammable component.

147 Guidance on the Application of the CLP Criteria 5.0 – July 2017 147 Version point but do have an explosion range or may Substances or mixtures which do not show a flash become flammable in use have to be labelled with EUH018. Annex II: 2.9. Liquid mixtures containing halogenated hydrocarbons For liquid mixtures which show no flashpoint or a flashpoint higher than 60 ̊C but not m ore than 93 ̊C and contain a halogenated hydrocarbon and more than 5 % highly flammable or flammable substances, the label on the packaging shall bear one of the following statements, le or flammable: depending on whether the substances referred to above are highly flammab EUH209 — ‘Can become highly flammable in use’ or — EUH209A ‘Can become flammable in use’ Note: EUH209 and EUH209A are limited to special types of mixtures whereas EUH018 covers a wider range of mixtures. In the majority of cases EUH018 covers EUH209 and EUH209A. Information about testing can be found in Section paragraph 5. 2.6.4.4.1 Re - classification of substances and mixtures classified as flammable 2.6.6. liquids according to DSD and DPD or already classified for transport 2.6.6.1. Relation to transport classification Class 3 of the UN RTDG Model Regulations and the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI) cover flammable liquids based on the same criteria as the CLP hazard class flammable liquid. In general there is a correspondence between transport packing groups and CLP hazard categories. However, in many cases specific exceptions apply. Further, the UN RTDG Model Regulations cover substances and mixtures transported above their flash point and desensitized explosive s. In practice the information on flash point and boiling point needed for classification is available and it is recommended to classify based on the data rather than use direct translation. See Annex VII for additional information on transport classificat ion in relation to CLP classification. 2.6.7. Examples of classification for flammable liquids 2.6.7.1. Examples of substances and mixtures fulfilling the classification criteria 2.6.7.1.1. Example 1 - BUTYLACETATE + P MIXTURE OF: N XYLENE + 1,3,5 - TRIMETHYLBENZENE - (7.9 MOL % + 60.3 MOL % + 31.7 MOL %) Initial boiling point (calculated): 140 °C 26 °C Flash point (calculated): calculated flash point is within 5 °C to the limiting value of 23 °C  flash point has to be measured. Dyn. Viscosity at 20 °C (DIN 53019): 8 mPas Flash point (EN ISO 3679): 30.0 °C  According to boiling point and measured flash point result: Flam.Liq. Category 3

148 ication of the CLP Criteria Guidance on the Appl 148 Version 5.0 – July 2017 2.6.7.1.2. Example 2 YDROCARBONS AND DICHLOROMETHANE (70 VOL % + 30 VOL %) H 52 °C Initial Boiling point (calculated): no flash point according to a standard Flash point: Because the hydrocarbon part of the mixture has a flash point by itself ( - 12 °C) the question ‘Is  an explosive vapour/air mixture possible’ (EN 1839 as amended, EN 15794 as amended) or ‘Can it swered. become highly flammable / flammable during use?’ has to be an Answer: Yes an explosion range exists; yes it can become highly flammable during use. According to the answer, the mixture has to be labelled with EUH018 or EUH209  Note 1: In that case EUH018 covers EUH209 Note 2: The EUH018 must only be assigne d if the substance or mixture is classified as hazardous (Article 25 (1) of CLP) Cannot be classified as flammable liquid because the mixture has no flash point. 2.6.7.2. Examples of substances and mixtures not fulfilling the classification criteria 2.6.7.2.1. Example 3 AQUEOUS FORMULATION OF ALIPHATIC POLYURETHANE RESIN Boiling point (EC 440/2008, EU test method A.2): 92 °C Dyn. Viscosity at 20 °C (DIN 53019 as amended): 1938 mPas Sample is highly viscous, use low heating rate for flash point determination (1 °C /min). Flash point (EN ISO 13736 as amended): 42.5 °C - combustion not sustained Sustained combustibility test (UN MTC L.2) at 60.5 °C: MTC L.2)at 75 °C: combustion not sustained Sustained combustibility test (UN -  According to the flash point result: Category 3 However, does not necessarily have to be classified as flammable liquid Category 3 because it did not sustain combustion. 2.6.8. References Brandes, E. and Möller, W.: Safety Characteristic Data , Volume 1, Flammab le gases and liquids, nw - Verlag, 2008 et al. William M. Haynes CRC Handbook of Chemistry and Physics 93rd Edition . CRC (2012) Press, Taylor and Francis, Boca Raton, FL O'Neil, Maryadele J. et al. © (2016, 2012) The Merck Index - An Encyclopedia of Chemicals, Drugs, and Biologicals (14th Edition – Version 14.9). Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc.

149 Guidance on the Application of the CLP Criteria – Version 149 July 2017 5.0 FLAMMABLE SOLIDS 2.7. Introduction 2.7.1. The criteria for ‘Flammable solids’ are found in Annex I, Section 2.7 of CLP and are identical to those in Chapter 2.7 of GHS. 2.7.2. Definitions and general considerations for the classification of flammable solids Annex I: 2.7.1.1. A flammable solid means a solid which is readily combustible, or may cause or contribute to fire through friction. Readily combustible solids are powdered, granular, or pasty substances or mixtures which are dangerous if they can be easily ignited by brief contact with an ignition source, such as a burning match, and if the flame spreads rapidly. Special consideration on part icle size 2.7.2.3. Annex I: [...] Note 1: The test shall be performed on the substance or mixture in its physical form as presented. If for example, for the purposes of supply or transport, the same chemical is to be presented in from that which was tested and which is considered likely to a physical form different materially alter its performance in a classification test, the substance shall also be tested in the new form. [...] ea exposed to air The finer the particle size of a solid substance or mixture, the greater the ar will be, and since flammability is a reaction with the oxygen in air, the particle size will greatly influence the ability to ignite. Hence it is very important that flammable properties for solids are how it can investigated on the substance or mix ture as it is actually presented (including reasonably be expected to be used , see Article 8 (6) of CLP). This is indicated by the Note cited in CLP Annex I, 2.7.2.3.For further information please see Section 1.2 within this Guidance. 2.7.3. Relation to other physical hazards reactive substances and mixtures as well as pyrophoric or - Explosives, organic peroxides, self be considered for classification as flammable solids since flammability oxidising solids should not is an intrinsic hazard in these classes. However, flammable solids can present other physical hazards at the same time, i.e. they might - heating or corrosive or emit flammable gases in contact with water. be self , Aerosols. If 2.3 For flammable solids that are packaged in aerosol dispensers, see Section osols, they must not be classified as flammable solids in addition classified as flammable aer 2.7 ). (see Section

150 ication of the CLP Criteria Guidance on the Appl 1 Version 5.0 – July 2017 50 2.7.4. Classification of substances and mixtures as flammable solids 2.7.4.1. Identification of hazard information ng For the classification of a substance or mixture as a flammable solid data on the followi properties are needed:  melting point;  information on water reactivity;  information on flash point for solids containing flammable liquids. IR & CSA, Chapter R.7a: Endpoint specific guidance , Section R.7.1.2 (Melting/freezing See also (Flash point). point), R.7.1.9 Many organic solid substances or mixtures fulfil the criteria to be classified as flammable solids. For inorganic solids, the classification as flammable is rather rare. 2.7.4.2. Screening procedures and waiving of testing In general, a possible cla ssification as a flammable solid should be considered for any solid organic substance or mixture containing such material. For inorganic material, testing may be waived in cases where the substance is commonly known to be not flammable (i.e. stable salts o r metal oxides) or where a flammability hazard can be excluded by any other scientific reasoning. In many cases, a simple screening test (see Section 2.7.4.4 ) can be used to determine whether a solid should be classified as flammable. Solid substances and mixtures are classified as flammable according to their burning behaviour. The test method as described in Part III, Sub - - MTC should be section 33.2.1.4.3.1 in th e UN applied for screening purposes. Alternatively, the burning index (referred to as ‘class number’ in VDI 2263) as obtained from the Burning Behaviour test (VDI 2263, part 1) may be used. If a nce or mixture should not be classified as a burning index of 3 or less is found, the substa flammable solid and no further testing is required. However, if smouldering or a flame is observed, the full test must be carried out. Classification criteria 2.7.4.3. with the GHS system. The classification criteria are fully in accordance 2.7.2.1. Powdered, granular or pasty substances or mixtures (except powders of Annex I: – see metals or metal alloys 2.7.2.2) shall be classified as readily combustible solids when the time of burning of one or more of the test runs, performed in accordance with the test - section method described in Part III, sub 33.2.1, of the UN RTDG, Manual of Tests and Criteria, is less than 45 seconds or the rate of burning is more than 2,2 mm/s. 2.7.2.2. Powders of metals or metal alloys shall be classified as flammable solids when they can be ignited and the reaction spreads over the whole length of the sample in 10 minutes or less. 2.7.2.3. A flammable solid shall be classified in one of the two categories for this class using Method N.1 as descr ibed in 33.2.1 of the UN RTDG, Manual of Tests and Criteria in accordance with Table 2.7.1;

151 Guidance on the Application of the CLP Criteria 5.0 – July 2017 151 Version Table 2.7.1 Criteria for flammable solids Category Criteria Burning rate test Substances and mixtures other than metal powders: (a) 1 wetted zone does not stop fire and (b) seconds or burning rate > 2,2 mm/s burning time < 45 Metal powders:  5 minutes burning time Burning rate test Substances and mixtures other than metal powders: (a) 2 wetted zone stops the fire for at least 4 minutes and (b) burning time < 45 seconds or burning rate > 2,2 mm/s Metal powders: minutes and  10 minutes burning time > 5 [...] Note 2: Aerosols shall not be classified as flammable solids; see section 2.3. Testing and evaluation of hazard information 2.7.4.4. - reactive properties first and to For safety reasons, it is advisable to test for explosive and self rule out pyrophoric behaviour before performing this test. The classification test is described in Part III, Sub - section 33.2.1.4.3.2 of the UN - MTC. The sam ple should be tested in its commercially relevant form. Special care has to be taken that the sample forms an unbroken strip or powder train in the test mould. Large pieces that do not fit into the mould should be gently crushed. For pasty or sticking subs tances it may be helpful to line the mould with a thin plastic foil which is withdrawn after having formed the train. Classification is based upon the fastest burning rate / shortest burning time obtained in six test runs, unless a positive result is obser ved earlier. For substances and mixtures other than metal powders, the category is assigned depending on whether the wetted zone is able to stop the flame. 2.7.4.5. Decision logic Classification of flammable solids is done according to decision logic 2.7.4 as incl uded in the GHS. NOTE: The person responsible for the classification of flammable solids should be experienced in this field and be familiar with the criteria for classification.

152 Guidance on the Appl ication of the CLP Criteria Version 5.0 – 152 July 2017 Figure 2 . 4 Decision logic for flammable solids (Decision logic 2.7 of GHS) The substance/mixture is a solid Negative Not classified Screening test Positive Burning rate test: No a. For substances or mixtures other than metal powders: Not classified Burning time < 45 s or burning rate > 2.2 mm/s?  10 min? b. Metal powders: Burning time Yes Category 1 For substances or mixtures other than metal powders: a. No Does the wetted zone stop propagation of the flame? b. Metal powders: Burning time > 5 min? Danger Yes Category 2 Warning

153 Guidance on the Application of the CLP Criteria – Version 153 July 2017 5.0 Hazard communication for flammable solids 2.7.5. Pictograms, signal words, hazard statements and precautionary 2.7.5.1. statements Annex I : 2.7.3. Table 2.7.2 Label elements for flammable solids Category 1 Category 2 Classification Pictograms GHS Signal Word Warning Danger H228: Flammable Solid H228: Flammable Solid Hazard Statement P210 P210 P240 P240 Precautionary Statement Prevention P241 P241 P280 P280 Precautionary Statement Response P370 + P378 P370 + P378 Statement Storage Precautionary Precautionary Statement Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. Relation to transport classification 2.7.6. Division 4.1 within Class 4 of the UN RTDG Model Regulations covers flammable substances, solid desensitized explosives and self - reactive liquids or solids. If a transport classification according to the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI) is available it should be kept in mind that transport classific ation is based on prioritisation of hazards (see UN RTDG Model Regulations, Section 2.0.3) and that flammable solids have a relatively low rank in the precedence of hazards. Therefore, the translation from transport e if a transport classification for a flammable solid is classification to CLP should be only don explicitly available. The conclusion that a substance or mixture not classified as a flammable solid for transport should not be classified as a flammable solid according to CLP is, in general, not c See Annex VII for additional information on transport classification in relation to orrect. CLP classification. 2.7.7. Examples of classification for flammable solids 2.7.7.1. Example of substances and mixtures fulfilling the classification criteria The following example shows a classification based on test data:

154 Guidance on the Appl ication of the CLP Criteria Version 5.0 – 154 July 2017 TEST SUBSTANCE: ‘FLAMMALENE’ (ORGANIC MATERIAL, SOLID) Screening test (VDI 2263, part 1): burning index: 5 (burning with an open flame or emission of sparks) Conclusion: Substance is candidate for classification as a flammable solid, further testing required. UN Test N.1 (Test method for readily Burning times for a distance of 100 mm (6 runs): 44 combustible solids): s; 40 s; 49 s; 45 s; 37 s; 41 s. substance is Shortest burning time is less than 45 s; a flammable solid. Wetted zone stops the fire, no reignition. Conclusion: Classify as flammable solid, Category 2. 2.7.7.2. Examples of substances and mixtures not fulfilling the classification criteria Many inorganic salts and oxides are not flammable such as NaCl, NaBr, KI, FeO, MnO etc. Urea or phthalic acid anhydride are examples of organic substances that would not be classified as flammable solids. 2.7.8. References VDI guideline 2263, part 1, 1990, Test methods for the Determination of the Safe ty Characteristics of Dusts

155 Guidance on the Application of the CLP Criteria 5.0 – July 2017 155 Version REACTIVE SUBSTANCES AND MIXTURES SELF - 2.8. Introduction 2.8.1. reactive substances and mixtures’ are found in Annex I, Section 2.8 of CLP The criteria for ‘Self - and are identical to those in Chapter 2.8 of GHS. - reactive substances and mixtures can In general, substances or mixtures classified as self decompose strongly exothermically when 50 kg are exposed to temperatures of 75 °C or lower - Accelerating Decomposition Temperature (SADT) of the substance or depending on the Self mixture. Sel f - reactive substances and mixtures display a very wide range of properties. The most hazardous type is - reactive substances and mixtures that are too dangerous to TYPE A of self transport commercially though they can be stored safely with appropriate precau tions. At the other end of the scale this classification includes substances and mixtures that only decompose slowly at temperatures well above the normal storage and transport temperatures (e.g. 75 °C). reactive substances and mi The decomposition of self xtures can be initiated by heat, contact - with catalytic impurities (e.g. acids, heavy - metal compounds, and bases), friction or impact. The rate of decomposition increases with temperature and varies with the substance or mixture. y if no ignition occurs, may result in the evolution of toxic gases or Decomposition, particularl vapours. For certain self - reactive substances and mixtures, the temperature must be controlled during storage and handling. Some self reactive substances and mixtures may decompose - expl osively, particularly if confined. This characteristic may be modified by the addition of - reactive substances and mixtures diluents or by the use of appropriate packaging. Some self - burn vigorously. Self of the types listed reactive substances are, for example, some compounds below: c. Aliphatic azo compounds ( - C - N=N - C - ); d. - C - N Organic azides ( ); 3 + - Z e. - Diazonium salts ( CN ); 2 N - nitroso compounds ( - f. - N=O); and N g. Aromatic sulfohydrazides ( - SO - - NH ). NH 2 2 This list is not exhaustive and substances with other reactive gro ups, combination of groups and some mixtures of substances may have similar properties. Additional guidance on substances, which may have self reactive properties, is given in Appendix 6, Section 5.1 of the UN - MTC. - g in subsidiary labelling, are indicated in the list of Additional hazardous properties, resultin - reactive substances and mixtures included in the UN RTDG Model already classified self Regulations, Section 2.4.2.3.2.3. Commercial self - reactive substances and mixtures are commonly formulated by dil ution with solid and liquid substances with which they are compatible. 2.8.2. Definitions and general considerations for the classification of self - reactives In CLP the following definition is given for self - reactive substances and mixtures: Annex I: 2.8.1.1. Self - reactive substances or mixtures are thermally unstable liquid or solid substances or mixtures liable to undergo a strongly exothermic decomposition even without participation of oxygen (air). This definition excludes substances and mixtures classifie d according to this Part as explosives, organic peroxides or as oxidising.

156 ication of the CLP Criteria Guidance on the Appl 156 Version 5.0 – July 2017 2.8.1.2. - reactive substance or mixture is regarded as possessing explosive properties A self or to when in laboratory testing the formulation is liable to detonate, to deflagrate rapidly show a violent effect when heated under confinement. General considerations 2.8.3. Hazard communication Annex I: Type G has no hazard communication elements assigned but shall be considered for properties belonging to other hazard classes. Relation to other physical hazards 2.8.3. Neither the burning properties nor the sensitivity to impact and friction form part of the classification procedure for self reactive substances and mixtures in CLP. These properties may - of self reactive substances and mixtures (see additional tests be of importance in safe handling - in Section 2.8.4.3.2 ). In addition, the following should be noted: properties Explosive The explosive properties do not have to be determined according to the CLP Annex I, Chapter - reactive 2.1, because explosive properties are incorporated in the decision logic for self substances and mixtures. Note that substances and mixtures may have explosive properties when handled under higher confinement. 2.8.4. Classification of substances and mixtures as self - reactive 2.8.4.1. Identification of hazard information reactive substance or mixture in one of the seven categories ‘ types A - The classification of a self to G’ is dependent on its detonation, deflagration and thermal explosion properties, its response to heating under confinement, its explosive power and the concentration and the type of diluent added to desensitize the substance or mixture. Specifi cations of acceptable diluents that can be used safely are given in the UN RTDG Model Regulations, Section 2.4.2.3.5. - reactive substance or mixture as type A, B or C is also dependent on The classification of a self ance or mixture is tested as it affects the degree of the type of packaging in which the subst confinement to which the substance or mixture is subjected. This has to be considered when handling the substance or mixture; stronger packaging may result in more violent reactions when the substance o r mixture decomposes. This is why it is important that storage and transport is done in packaging, allowed for the type of self - reactive substance and mixture, that conforms the requirements of the UN - packaging or IBC instruction (P520/IBC520) or tank inst ruction (T23). The traditional aspects of explosive properties, such as detonation, deflagration and thermal explosion, are incorporated in the decision logic Figure 2.8.1 of CLP (see Section 2.8.4.4 ). Consequently, the determination of explosive properties as prescribed in the hazard class explosives needs not to be conducted for self - reactive substances and mixtures. 2.8.4.2. Classification criteria Accordi ng to CLP, substances and mixtures must be considered for classification in this hazard class as a self - reactive substance or mixture unless: Annex I: 2.8.2.1. [...]

157 Guidance on the Application of the CLP Criteria 5.0 – July 2017 157 Version (a) they are explosives, according to the criteria given in 2.1; ids or solids, according to the criteria given in 2.13 or 2.14, except (b) they are oxidising liqu that mixtures of oxidising substances, which contain 5 % or more of combustible organic reactive substances according to the procedure defined - substances shall be classified as self 2.8.2.2; in (c) they are organic peroxides, according to the criteria given in 2.15; (d) their heat of decomposition is less than 300 J/g; or (e) their self accelerating decomposition temperature (SADT) is greater than 75 °C for a 50 - Manual of Test and Criteria, sub sections 28.1, 28.2, 28.3 and kg package (See UN RTDG, - 28.3.) Table 2.8.2.2 . Mixtures of oxidising substances, meeting the criteria for classification as oxidising substances, which contain 5 % or more of combustible organic substances and which do not meet the criteria mentioned in (a), (c), (d) or (e) in 2.8.2.1, shall be subjected to the self - reactive substances classification procedure; Such a mixture showing the properties of a self - reactive substance type B to F (see 2.8.2.3) shall be classif ied as a self reactive substance. - [...] In addition to the above, substances and mixtures must be considered for classification in this hazard class unless: 2.8.4.2. Annex I: [...] (a) There are no chemical groups present in the molecule associated with explosive or self - reactive properties; examples of such groups are given in Tables A6.1 and A6.2 in Appendix 6 of the UN RTDG, Manual of Tests and Criteria. [...] In the CLP decision logic (see Section ), classification of self - reactive substances or 2.8.4.4 mixtures is based on performance based testing in both small scale tests and, where necessary, some larger scale tests with the substance or mi xture in its packaging. The concept of ‘intrinsic properties’ is, therefore, not necessarily, applicable to this hazard class. Self - reactive substances or mixtures are classified in one of the seven categories of ‘types A to G’ according to the classificat ion criteria given in Section 2.8.2.3 of Annex I, CLP. The classification principles are given in the decision logic in Figure 2.8.1 of CLP (see Section 2.8.4.4 ) and the Test Series A to H, as described in the Part II of the UN - MTC, should be performed.

158 ication of the CLP Criteria Guidance on the Appl 158 Version 5.0 – July 2017 Annex I: Self - reactive substances and mixtures shall be classified in one of the 2.8.2.3. to the following principles: seven categories of ‘types A to G’ for this class, according (a) any self - reactive substance or mixture which can detonate or deflagrate rapidly, as reactive substance TYPE packaged, shall be defined as self - A; - reactive substance or mixture possessing explosive properties a nd which, as (b) any self packaged, neither detonates nor deflagrates rapidly, but is liable to undergo a thermal - reactive substance TYPE B; explosion in that package shall be defined as self - en the substance (c) any self reactive substance or mixture possessing explosive properties wh or mixture as packaged cannot detonate or deflagrate rapidly or undergo a thermal explosion shall be defined as self C; - reactive substance TYPE - (d) any self reactive substance or mixture which in laboratory testing: (i) detonates partially, does not deflagrate rapidly and shows no violent effect when heated under confinement; or (ii) does not detonate at all, deflagrates slowly and shows no violent effect when heated under confinement; or l and shows a medium effect when heated (iii) does not detonate or deflagrate at al under confinement; - reactive substance TYPE D; shall be defined as self - reactive substance or mixture which, in laboratory testing, neither detonates nor (e) any self deflagrates at all and shows low or no effect when heated under confinement shall be defined as self - reactive substance TYPE E; (f) any self - reactive substance or mixture which, in laboratory testing, neither detonates in the cavitated state nor deflagrates at all and shows only a low or no effect when he ated under confinement as well as low or no explosive power shall be defined as self reactive - substance TYPE F; reactive substance or mixture which, in laboratory testing, neither detonates in (g) any self - ws no effect when heated under the cavitated state nor deflagrates at all and sho o confinement nor any explosive power, provided that it is thermally stable (SADT is 60 C o to 75 C for a 50 kg package), and, for liquid mixtures, a diluent having a boiling point not o reactive substance C is used for desensitisati on shall be defined as self - less than 150 G. If the mixture is not thermally stable or a diluent having a boiling point less than TYPE o 150 C is used for desensitisation, the mixture shall be defined as self - reactive substance TYPE F. s conducted in the package form and the packaging is changed, a further test Where the test i shall be conducted where it is considered that the change in packaging will affect the outcome of the test. A list of currently classified self reactive substances and mixtures i s included in the UN RTDG - Model Regulations, Section 2.4.2.3.2.3. 2.8.4.3. Testing and evaluation of hazard information 2.8.4.3.1. Thermal stability tests and temperature control In addition to the classification tests given in decision logic Figure 2.8.1 of CLP, the thermal stability of the self - reactive substances and mixtures has to be assessed in order to determine the SADT. The SADT is defined as the lowest temperature at which self - accelerating decomposition of a n transport, handling and storage. substance or mixture may occur in the packaging as used i

159 Guidance on the Application of the CLP Criteria 5.0 – July 2017 159 Version The SADT is a measure of the combined effect of the ambient temperature, decomposition kinetics, package size and the heat transfer properties of the substance or mixture and its packaging. There is no relation between the SADT of a self - reactive substance and mixture and its classification in one of the seven categories ‘types A to G’. The SADT is used to derive safe handling, storage and transport temperatures (control temperature) and alar m temperature (emergency temperature). reactive substance and mixture needs temperature control and Depending on its SADT a self - the rules as given in CLP Annex I, 2.8.2.4, consist of the following two elements: Criteria for temperature control: 1. Self - 2. ctive substances and mixtures need to be subjected to temperature control when rea the SADT is ≤ 55 ° C. Derivation of control and emergency temperatures: 3. SADT* Emergency temperature Type of receptacle Control temperature 20 °C or Single packagings less 10 °C below SADT 20 °C below SADT and IBC’s 15 °C below SADT over 20 °C to 35 °C 10 °C below SADT over 35 °C 10 °C below SADT 5 °C below SADT Tanks 10 °C below SADT 5 °C below SADT < 50 °C handling and storage. *i.e. the SADT of the substance/mixture as packaged for transport, reactive It should be emphasized that the SADT is dependent on the nature of the self - substance or mixture itself, together with the volume and heat loss characteristics of the - packaging or vessel in which the substance or mixture is handled. The temperature at which - accelerating decomposition occurs falls: self  as the size of the packaging or vessel increases; and with increasing efficiency of the insulation on the package or vessel.  The SADT is only valid for the substance or mixtu re as tested and when handled properly. Mixing the self - reactive substances and mixtures with other chemicals, or contact with incompatible materials (including incompatible packaging or vessel material) may reduce the thermal stability due to catalytic de composition, and lower the SADT. This may increase the risk of decomposition and has to be avoided. Additional considerations and testing 2.8.4.3.2. Explosive properties The sensitivity of self reactive substances and mixtures to impact (solids and liquids) and - frict ion (solids only) may be of importance for the safe handling of the substances and mixtures, in the event that these substances and mixtures have pronounced explosive properties (e.g. rapid deflagration and/or violent heating under confinement). Test metho ds to determine these properties are described in Test Series 3 (a) (ii) and 3 (b) (i) of the UN - MTC. This information should be documented in the SDS. Burning properties Although there are currently no dedicated storage guidelines for self - reactive subst ances and mixtures (although in some countries under development), often the regulations for organic peroxides are referred to. For storage classification the burning rate is commonly used, see Section 2.15 on organic peroxides.

160 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 160 Flash point - reactive substances or mixtures is only relevant in the The flash point for liquid self the self - temperature range where the product is thermally stable. Above the SADT of reactive substance or mixture, flash point determination is not relevant because decomposition products are evolved. NOTE: In case a flash point determination seems reasonable (expected flash point below the SADT) a test method using small amount of sample is recommended. In case the self - reactive substance or mixture is diluted or dissolved, the diluent may determine the flash point. - ignition temperature Auto self - reactive substances The determination of the auto ignition temperature is not relevant for and mixtures, because the vapours decompose during the execution of the test. Available test methods are for non - decomposing vapour phases. Auto ignition of self - reactive substance and mixtures vapours when they decompose, can neve r be excluded. This information should be documented in the SDS. Self - ignition temperature Also self - ignition temperature determination (test applicable for solids) is not relevant. The tively given by the SADT thermal stability of self - reactive substances and mixtures is quantita test. Control and Emergency temperatures The Control and Emergency temperatures are based on the SADT as determined by UN Test H.4. The Dewar vessel used in the UN Test H.4 is supposed to be representative for the substance or mix ture handled in packages. For handling of the substance or mixture in larger quantities (IBCs/tanks/vessels etc.) and/or in better (thermally) insulated containers under more thermal insulated conditions, the SADT has to be determined for that quantity wit h the given degree of insulation. From that SADT the Control and Emergency temperatures can be 2.15.4.3 ) derived (see also Section Additional classification considerations 2.8.4.3.3. lf - reactives Currently, the following properties are not incorporated in the classification of se under the CLP:  mechanical sensitivity i.e. impact and friction sensitivity (for handling purposes);  burning properties (for storage purposes);  flash point for liquids; and  burning rate for solids. In addition to the GHS criteria CLP mentions that: Annex I: 2.8.2.2 [...] Where the test is conducted in the package form and the packaging is changed, a further test shall be conducted where it is considered that the change in packaging will affect the outcome of the test. Please note that polymerising substances do not fulfil the criteria for classification as self - reactives. However, there are on - going discussions at the UNSCEGHS on this subject.

161 Guidance on the Application of the CLP Criteria 5.0 July 2017 Version – 161 2.8.4.4. Decision logic Classification of self - reactive substances and mixtures is done according to de cision logic 2.8 as included in the GHS. NOTE: The person responsible for the classification of self - reactive substances and mixtures should be experienced in this field and be familiar with the criteria for classification.

162 ication of the CLP Criteria Guidance on the Appl Version 5.0 – July 2017 162 2 . 5 Figure Decision logic 2.8 for self - reactive substances and mixtures

163 Guidance on the Application of the CLP Criteria July 2017 163 Version 5.0 – Hazard communication for self - 2.8.5. reactives Pictograms, signal words, hazard statements and precautionary 2.8.5.1. statements must be used for substances and mixtures According to CLP the following label elements meeting the criteria for this hazard class: Table 2.8.1 Annex I: - Label elements for self reactive substances and mixtures 2 Type C & D Classification Type E & F Type B Type G Type A GHS pictograms Danger Signal Word Danger Danger Warning H241: H242: Hazard Statement H242: H240: Heating Heating may Heating may Heating may may cause a cause a fire or cause a fire cause an fire There are explosion explosion no label elements P210 P210 Precautionary P210 P210 allocated statement to this P234 P234 P234 P234 hazard Prevention P235 P235 P235 P235 category P240 P240 P240 P240 P280 P280 P280 P280 + P378 P370 P370 + P372 P370 + P378 P370 + P380 Precautionary + P380 + statement + P375 P373 1 [+P378] Response Precautionary P403 P403 P403 P403 statement P411 P411 P411 P411 Storage P420 P420 P420 P420 Precautionary P501 P501 P501 P501 statement Disposal

164 ication of the CLP Criteria Guidance on the Appl 164 Version 5.0 – July 2017 1 See the introduction to Annex IV for details on the use of square brackets. 2 Type G has no hazard communication elements assigned but should be considered for properties belonging to other hazard classes. The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. 2.8.6. DSD and DPD or Relation to transport classificationaccording to already classified for transport Division 4.1 within Class 4 of the UN RTDG Model Regulations covers flammable substances, - - solid desensitized explosives and self reactive liquids or solids. A list of already classified self ces is included in UN RTDG Model Regulations, Section 2.4.2.3.2.3. This table reactive substan includes self reactive substances of various types from type B to type F. See Annex VII for - additional information on transport classification in relation to CLP classification. Examples of classification for self - reactives 2.8.7. 2.8.7.1. Examples of substances and mixtures fulfilling the classification criteria Substance to be classified: NP Molecular formula: n.a. According to CLP Annex I, Section 2.8.2.1, the substance has: an energy content of 1452 kJ/kg; and  a SADT of 45 °C (in 50 kg package);  and consequently it has to be considered for classification in the hazard class self - reactive substances and mixtures. - rea ctive substances Test results and classification according to CLP decision logic 2.8.1 for self - and mixtures and the UN MTC, Part II, is as follows: CLASSIFICATION TEST RESULTS reactive substance or mixture: 1. Name of the self - NP 2. General data NP, technically pure 2.1. Composition Molecular formula n.a. 2.2. Physical form solid, fine powder 2.3. Colour brown 2.4. 3 Density (apparent) 460 kg/m 2.5. 3. Detonation (test series A) Box 1 of the decision logic Does the substance propagate a detonation? 3.1. Method UN Test A.1: BAM 50/60 steel tube test 3.2. Sample conditions technically pure substance 3.3. Observations fragmented part of the tube: 12, 18cm

165 Guidance on the Application of the CLP Criteria 5.0 – July 2017 165 Version CLASSIFICATION TEST RESULTS No 3.4. Result 3.5. 1.3 Exit 4. Deflagration (test series C) Does the substance propagate a Box 5 of the decision logic deflagration? Time/pressure test (test C.1) 4.1. Method 1 Sample conditions ambient temperature 4.1.1. 498, 966, 3395 ms 4.1.2. Observations Result 4.1.3. Yes, slowly 4.2. Method 2 Deflagration test (test C.2) Sample temperature: 20 °C 4.2.1. conditions deflagration rate: 0.90, 0.87 mm/s 4.2.2. Observations Result Yes, slowly 4.2.3. 4.3. Yes, slowly Final result 4.4. 5.2 Exit 5. Heating under confinement (test series E) Box 8 of the decision logic: What is the effect of heating it under defined confinement? 5.1. Method 1 Koenen test (test E.1) 5.1.1. Sample conditions Observations Limiting diameter: < 1.0 mm 5.1.2. fragmentation type ‘A’ Low 5.1.3. Result Method 2 Dutch pressure vessel test 5.2. (test E.2) 5.2.1. Sample conditions Observations 5.2.2. Limiting diameter: <1.0 mm (with 10 g), 1.0 mm (50 g) Result 5.2.3. low 5.3. Final result low 5.4. 8.3 Exit 6. Thermal stability (outside of the decision logic) Method Heat accumulation storage test (test H.4) 6.1. mass 232.5 g. Half life time of cooling of Dewar 6.2. : Sample conditions vessel with 400 ml water: 10.0 hrs.(representing substance in package) self 6.3. Observations - accelerating decomposition at 45 °C no self - accelerating decomposition at 40 °C 6.4. Result SADT 45 °C (in 50 kg package)

166 Guidance on the Appl ication of the CLP Criteria Version 5.0 – 166 July 2017 CLASSIFICATION TEST RESULTS 7. General remarks The decision logic is given in Figure 2 . 6 8. Final classification reactive substance, Type D, solid, temperature Hazard / hazard class: Self - controlled Flame (GHS02) Label Danger Signal word may cause a fire H242: Heating Hazard statement Temperature control Needed based on SADT (45 °C, in package) Control temperature* 35 °C (in package) Emergency temperature* 40 °C (in package) *See UN - MTC, table 28.2.

167 Guidance on the Application of the CLP Criteria 5.0 – Version 167 July 2017 Figure 2 . 6 Decision logic for self - reactive substance example: NP, technically pure

168 ication of the CLP Criteria Guidance on the Appl 168 Version 5.0 – July 2017 2.9. PYROPHORIC LIQUIDS Introduction 2.9.1. The criteria for ‘Pyrophoric liquids’ are found in Annex I, Section 2.9 of CLP and are identical to those in Chapter 2.9 of GHS. Pyrophoricity, i.e. the ability to spontaneously ignite in air, is the result of a reaction of a substance or mixture with the oxygen in the air. The reaction is exothermic and has the flame, heat or particularity that it starts spontaneously, i.e. without the aid of a supplied spark, other energy source. Another way of saying this is that the auto - ignition temperature for a pyrophoric substance or mixture is lower than room (ambient) temperature. - s well as their - metalloids may be suspected of being pyrophores, a Organo metals and organo - derivatives. Also organo phosphines and their derivatives, hydrides and their derivatives and haloacetylene derivatives may show pyrophoricity (Urben, 2007). There are also pyrophoric substances or mixtures that do not belong to the above mentioned groups of chemicals, i.e. the list above is not exhaustive. Since pyrophoric substances or ignite spontaneously mixtures in air, pyrophoricity is a very dangerous property. In case of doubt it should therefore be thoroughly investigated whether a given substance or mixture is More information on pyrophoric substances can e.g. be found in pyrophoric. Bretherick’s Handbook of Reactive Chemical Hazards (Urben, 2007). 2.9.2. Definitions and general considerations for the classification pyrophoric liquids The definition in CLP for pyrophoric liquids is as follows: Annex I: 2.9.1. Definition Pyrophoric liquid means a liquid substance or mixture which, even in small quantities, is liable to ignite within five minutes after coming into contact with air. Rel ation to other physical hazards 2.9.3. Pyrophoric substances and mixtures will react spontaneously with air already in small amounts and more or less instantaneously (within minutes). This differentiates them from self heating - react spontaneously with air but only when in larger substances and mixtures, which also amounts and after an extended period of time (hours or days). While liquids in themselves generally do not exhibit self - heating properties due to the limited contact with air (which can occur only at th e surface), liquids that are adsorbed onto solid particles should, in general, be considered for classification in the hazard class self - heating substances and mixtures, see Chapter 2.11 of this guidance. Pyrophoricity may be expected for certain reactive metals and some of their compounds (e.g. hydrides and other organo metal compounds). Many of these substances and mixtures will also - react vigorously with water under the production of flammable gases. Suc h substances and mixtures may thus be classified in the hazard class substances and mixtures which in contact with water emit flammable gases in addition, see Chapter 2.12 of this guidance. It should be noted in this context that water reactive substances and mixtures may also to some extent - react with the humidity in air, although such a reaction is seldom vigorous. A substance or mixture that spont aneously ignites in air in accordance with the test procedures is to be considered pyrophoric, regardless of the reaction mechanism. Liquids not classified as pyrophoric but that can burn may belong to the hazard class flammable liquids depending on their flash point and ability to sustain combustion, see Section 2.6 of this guidance.

169 Guidance on the Application of the CLP Criteria 5.0 – July 2017 169 Version 2.9.4. Classification of substances and mixtures as pyrophoric liquids Identification of hazard information 2.9.4.1. Since the tests to determine pyrophoricity are simple and require no special equipment, see 2.9.4.4 below, there is in general no reason to go to data sources instead of performing Section tests. Furthermore, the possibilities of waiving tests are ample both for known pyrophores and 2.9.4.2 for substances and mixtures known not to be pyrophoric, see Section below. If information anyway is taken from literature or other data sources, it is of utmost importance 2.0.4 . Naturally, all data sources that the correct physical form is considered, see Section should be carefully evaluated with regard to reliability and scientific validity. 2.9.4.2. Screening procedures and waiving of testing In case a liquid is known from practical handling to be pyrophoric no testing is necessary. Such liquids are classified as pyrophoric liquids without testing. This would also be the case if the form the tests for liquid spontaneously ignites upon opening of the receptacle when trying to per classification. According to the additional classification considerations in CLP Annex I, 2.9.4, the classification procedure for pyrophoric liquids need not be applied when experience in manufacture or s not ignite spontaneously on coming into contact with air at handling shows that the liquid doe normal temperatures (i.e. the liquid is known to be stable at room temperature for prolonged periods of time (days)). 2.9.4.3. Classification criteria Section 2.9.2.1 of Annex I of CLP specifies the cla ssification criteria: Annex I: Table 2.9.1 Criteria for pyrophoric liquids Category Criteria The liquid ignites within 5 1 min when added to an inert carrier and exposed to air, or it ignites or chars a filter paper on contact with air within 5 min. Testing and evaluation of hazard information 2.9.4.4. In Section 2.9.2.1 of Annex I of CLP reference to the test - methods are made: Annex I: 2.9.2.1. A pyrophoric liquid shall be classified in a single category for this class using test N.3 in part III, sub - section 33.3.1.5 of the UN RTDG, Manual of Tests and Criteria according to Table 2.9.1: The UN Test N.3 for pyrophoricity is quite simple and sufficiently described in Part III, Section 33 of the UN - MTC. No special equipment is needed. Essentially the substance or mixture is exposed to air to see if it ignites. For liquids which do not spontaneously ignite when poured, the surface in contact with air is increased using a filter paper. Ignition or charring of the filter paper is regarded as a positive response in the test, i.e. such a liquid is considered to be pyrophoric. It is important that samples for testing of pyrophoric properties are carefully packed and sealed. Furthermore, the material offered for testing should be freshly prepared, since the reactive pro perties may diminish due to aging or agglomeration. Whenever experiments are to be done

170 ication of the CLP Criteria Guidance on the Appl 170 Version 5.0 – July 2017 one should be careful a pyrophoric substance or mixture may well ignite already upon opening – the receptacle! general, very complex, and that the It should be noted that the mechanism of oxidation is, in humidity of air might influence the rate of reaction. Therefore a false negative may result when performing the tests in an extremely dry environment, and this condition must be avoided when ssification for pyrophoricity. The filter paper test of UN Test N.3 for performing the tests for cla pyrophoric liquids should be carried out at 25 ± 2 °C and a relative humidity of 50 ± 5 % (see - MTC, Section 33.3.1.5). UN 2.9.4.5. Decision logic ccording to decision logic 2.9.4.1 as included in the Classification of pyrophoric liquids is done a GHS. NOTE: The person responsible for the classification of pyrophoric liquids should be experienced in this field and be familiar with the criteria for classification. 2.9.4.5.1. Decision logic for pyrophoric liquids . Figure Decision logic for pyrophoric liquids (Decision logic 2.9 of GHS) 2 7 The substance/mixture is a liquid Category 1 Yes Does it ignite within 5 min when poured into a porcelain cup filled with diatomaceous earth or silica gel? Danger No Category 1 Yes Does it ignite or char a filter paper within 5 min? Danger No Not classified

171 Guidance on the Application of the CLP Criteria 5.0 – July 2017 171 Version Hazard communication for pyrophoric liquids 2.9.5. statements and precautionary 2.9.5.1. Pictograms, signal words, hazard statements 2.9.3 Annex I: Table 2.9.2 Label elements for pyrophoric liquids 1 Classification Category GHS Pictogram Signal Word Danger Hazard Statement H250: Catches fire spontaneously if exposed to air Precautionary Statement Prevention P210 P222 P231 + P232 P233 P280 Precautionary Statement Response P302 + P334 P370 + P378 Precautionary Statement Storage Precautionary Statement Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. 2.9.6. Relation to transport classification Division 4.2 within Class 4 of the UN RTDG Model Regulations covers pyrophoric solids, liquids - heating substances and mixtures. UN Test N.3 that is used for classification for and self pyrophoricity for liquids accor ding to CLP is also used for classification in the subdivision pyrophoric substances and mixtures in Division 4.2: Substances liable to spontaneous combustion according to the UN RTDG Model Regulations. The criteria for Category 1 according to CLP (which i s the only category for pyrophoric liquids) and for packing group I in Division 4.2 according to the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI) are also ned to packing exactly the same. Furthermore, all pyrophoric substances and mixtures are assig group I within Division 4.2, which is used exclusively for pyrophoric substances and mixtures. Therefore, any liquid assigned to Division 4.2, packing group I according to the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI ) will be classified in Category 1 of the hazard class pyrophoric liquids according to CLP. See Annex VII for additional information on transport classification in relation to CLP classification.

172 ication of the CLP Criteria Guidance on the Appl 172 Version 5.0 – July 2017 2.9.7. Examples of classification for pyrophoric liquids e that the substance and mixture names in this chapter are fictitious Please not . 2.9.7.1. Examples of substances and mixtures fulfilling the classification criteria 2.9.7.1.1. Example 1 Name: Pyrpherdine Physical state: Liquid Pyrophoric properties: Unknown, therefore the UN Test N.3 of the UN - MTC was applied. However, when opening the receptacle in order to perform the test, Pyrpherdine self - ignited. Classification: Pyrophoric liquid, Category 1

173 Guidance on the Application of the CLP Criteria 5.0 – July 2017 173 Version 2.9.7.1.2. Example 2 Qulipyr Name: Physical state: Liquid Pyrophoric properties: Unknown, therefore the UN Test N.3 of the UN - MTC was applied. Test result: When poured according to the test procedure, nothing happened. The procedure was repeated six times, each time giving a negative result (i.e. no ignition). Therefore Qulipyr was s upplied to a filter paper in accordance with the test method. In the second trial the filter paper was charred within five minutes. Classification: Pyrophoric liquid, Category 1 2.9.7.2. Examples of substances and mixtures not fulfilling the classification criteria 2.9.7.2.1. Example 3 Name: Notpyratal Liquid Physical state: Pyrophoric properties: Unknown, therefore UN Test N.3 of the UN - MTC was applied. Test result: When poured according to the test procedure nothing happened in either of six trials. Therefore Notpyratal was supplied to a filter paper in accordance with the test method, whereupon no ignition or charring occurred in either of three trials. Classification: Not a pyrophoric liquid 2.9.8. References Urben, Peter G. (2007). Bretherick's Handbook of Reactive Chemical Hazards , Volumes 1 - 2 (7th Edition). Elsevier.

174 ication of the CLP Criteria Guidance on the Appl 174 Version 5.0 – July 2017 2.10. PYROPHORIC SOLIDS Introduction 2.10.1. The criteria for ‘Pyrophoric solids’ are found in Annex I, Section 2.10 of CLP and are identical to those in Chapter 2.10 of GHS. Pyrophoricity, i.e. the ability to spontaneously ignite in air, is the result of a reaction of a substance or mixture with the oxygen in the air. The reaction is exothermic and has the particularity that it starts spontaneously, i.e. without the aid of a supplied spark, flame, he at or - ignition temperature for a other energy source. Another way of saying this is that the self pyrophoric substance or mixture is lower than room (ambient) temperature. - metals and organo - metalloids may be suspected of being pyrophores, as well as their Organo - derivatives. Also organo phosphines and their derivatives, hydrides and their derivatives, haloacetylene derivatives, and complex acetylides may show pyrophoricity (Urben, 2007). However, although many Furthermore, powders or fine particles of metals could be pyrophoric. solid metallic substances, like e.g. aluminium, would be suspected of being pyrophoric when considering their general reactivity, they form a protective oxide coat upon reaction with air. - This thin coat of metal oxide prevents the me tal from reacting further, and hence such substances may not show pyrophoric behaviour in reality. There are also pyrophoric solids that do not belong to the above mentioned groups of chemicals, i.e. the list above is not exhaustive. Since pyrophoric soli in ds ignite spontaneously air, pyrophoricity is a very dangerous property. In case of doubt it should therefore be thoroughly investigated whether a given solid is pyrophoric. More information on pyrophoric k of Reactive Chemical Hazards (Urben, 2007). solids can e.g. be found in Bretherick’s Handboo 2.10.2. Definitions and general considerations for the classification pyrophoric solids The definition in CLP for pyrophoric solids is as follows: 2.10.1. Definition Annex I: Pyrophoric solid means a solid substan ce or mixture which, even in small quantities, is liable to ignite within five minutes after coming into contact with air. Special consideration on particle size 2.10.2.1. Annex I: [...] physical form as Note: The test shall be performed on the substance or mixture in its presented. If for example, for the purposes of supply or transport, the same chemical is to be presented in a physical form different from that which was tested and which is considered likely to materially alter its performance in a classi fication test, the substance shall also be tested in the new form. The finer the particle size of a solid, the greater the area exposed to air will be, and since the pyrophoricity is a reaction with the oxygen in air, the particle size will greatly influence ability to spontaneously ignite. Hence it is very important that pyrophoric properties for solids how it can are investigated on the substance or mixture as it is actually presented (including ). This is indicated by the Note cited , see Article 8 (6) of CLP reasonably be expected to be used in CLP Annex I, 2.10.2.1.

175 Guidance on the Application of the CLP Criteria 5.0 – July 2017 175 Version 2.10.3. Relation to other physical hazards Pyrophoric solids will react spontaneously with air already in small amounts and more or less heating substances and - instantaneously (within minutes). This differentiates them from self ger amounts and after an mixtures, which also react spontaneously with air but only when in lar extended period of time (hours or days). A solid which is not classified as a pyrophoric solid may thus belong to the hazard class self - heating substances and mixtures, and should be ss, see Chapter 2.11 of this guidance. considered for classification in that hazard cla Pyrophoricity may be expected for certain reactive metals and some of their compounds (e.g. hydrides and othe - metal compounds). Many of these substances will also react r organo vigorously with water under the production of flammable gases. Such substances may thus be classified in the hazard class substances and mixtures which in contact with water emit gases in addition see Chapter 2.12 of this guidance. It should be noted in this flammable - reactive substances or mixtures may also to some extent react with the context that water humidity in air, although such a reaction is seldom vigo rous. A substance that spontaneously ignites in air in accordance with the test procedures is to be considered pyrophoric, regardless of the reaction mechanism. Solids not classified as pyrophoric may still be able to burn rapidly if subjected to enough in itiating energy, such as the flame from a gas burner, to start the reaction. Therefore they may be subject to classification in the hazard class flammable solids, see Chapter 2.7 of this guidance, i.e. they may be 'readily combustible solids'. 2.10.4. Classification of substances and mixtures as pyrophoric solids 2.10.4.1. Identification of hazard information Since the tests to determine pyrophoricity are simple and r equire no special equipment, see Section 2.10.4.4 below, there is in general no reason to go to data sources instead of re, the possibilities of waiving tests are ample both for known performing tests. Furthermo pyrophores and for substances and mixtures known not to be pyrophoric, see Section 2.10.4.2 below. If information is taken from literature or other data sources anyway, it is of utmost importance that the correct physical form is considered, see Section 2.0.4 . Naturally, all data sources should be carefully evaluated with regard to reliability and scientific validity. 2.10.4.2. Screening procedures and waiving of testing In case a solid is known from practical handling to be pyrophoric no tes ting is necessary. Such solids are classified as pyrophoric solids without testing. This would also be the case if the solid spontaneously ignites upon opening of the receptacle when trying to perform the tests for classification. According to the addition al classification considerations in CLP Annex I, 2.10.4, the classification procedure for pyrophoric solids need not be applied when experience in manufacture or handling shows that the substance or mixture does not ignite spontaneously on coming into cont act with air at normal temperatures (i.e. the substance or mixture is known to be stable at room temperature for prolonged periods of time (days)). 2.10.4.3. Classification criteria Section 2.10.2.1 of Annex I of CLP specifies the classification criteria: Annex I: Table 2.10.1 Criteria for pyrophoric solids Category Criteria

176 ication of the CLP Criteria Guidance on the Appl 176 Version 5.0 – July 2017 1 The solid ignites within 5 minutes of coming into contact with air. Testing and evaluation of hazard information 2.10.4.4. - methods are made: In Section 2.10.2.1 of Annex I of CLP reference to the test 2.10.2.1. A pyrophoric solid shall be classified in a single category for this class Annex I: III, sub - section 33.3.1.4 of the UN RTDG, Manual of Tests and Criteria using test N.2 in part in accordance with Table 2.10.1: y is quite simple and sufficiently described in Part III, Section 33 of UN Test N.2 for pyrophoricit - MTC. No special equipment is needed. Essentially the solid is exposed to air to see if it the UN ignites. It is important that samples for testing of pyrophoric properties are careful ly packed and sealed. Furthermore, the material offered for testing should be freshly prepared, since the reactive properties may diminish due to aging or agglomeration. Whenever experiments are to be done l ignite already upon opening the receptacle! one should be careful – a pyrophoric solid may wel It should be noted that the mechanism of oxidation is, in general, very complex, and that the humidity of air might influence the rate of reaction. It is known that certain metals will not react reas in the presence of moisture the reaction is almost instantaneous (often even in dry air, whe trace amounts of moisture are sufficient). Therefore a false negative may result when performing the tests in an extremely dry environment, and this condition must be avoided when performing the tests for classification for pyrophoricity. Decision logic 2.10.4.5. Classification of pyrophoric solids is done according to decision logic 2.10.4.1 as included in the GHS. NOTE: The person responsible for the classification of pyrophoric s olids should be experienced in this field and be familiar with the criteria for classification. Decision logic for pyrophoric solids 2.10.4.5.1. GHS) Figure 8 Decision logic for pyrophoric solids (Decision logic 2.10 of . 2 The substance/mixture is a solid Category 1 Yes Does it ignite within 5 min after exposure to air? Danger No Not classified

177 Guidance on the Application of the CLP Criteria 5.0 – July 2017 177 Version 2.10.5. Hazard communication for pyrophoric solids Pictograms, signal words, hazard statements and precautionary 2.10.5.1. statements Annex I: 2.10.3 Table 2.10.2 Label elements for pyrophoric solids Classification Category 1 GHS Pictogram Danger Signal Word Hazard Statement H250: Catches fire spontaneously if exposed to air Precautionary Statement Prevention P210 P222 P231 + P232 P233 P280 Precautionary Statement Response P302 + P335 + P334 P370 + P378 Precautionary Statement Storage Precautionary Statement Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. 2.10.6. Relation to transport classification Division 4.2 within Class 4 of the UN RTDG Model Regulations covers pyrophoric solids, liquids and self heating substances and mixtures. The UN Tests N.2 that is used for classification for - pyrophoricity for solids according to CLP is also used for classification in the subdivision pyrophoric substances and mixtures in Division 4.2: Substances liable to spontaneous combustion according to the UN RTDG Model Regulations. The criteria for Category 1 according to CLP (which is the only category for pyrophoric solids) and for packing group I in Division 4.2 according to the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI) are also exactly the same. Furthermore, all pyrophoric substances and mixtures are assigned to packi ng group I within Division 4.2, which is used exclusively for pyrophoric substances and mixtures. Therefore, any solid substance or mixture assigned to Division 4.2, packing group I according to the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI) will be classified in Category 1 of the hazard class pyrophoric solids according to CLP. See Annex VII for additional information on transport classification in relation to CLP classification.

178 ication of the CLP Criteria Guidance on the Appl 178 Version 5.0 – July 2017 2.10.7. Examples of classification for pyrophoric solids fictitious . se note that the substance and mixture names in this chapter are Plea 2.10.7.1. Examples of substances and mixtures fulfilling the classification criteria Example 1 2.10.7.1.1. Name: Pyroferil Physical state: Solid Pyroferil is known to self - ignite upon contact with air at ambient conditions. Pyrophoric properties: Classification: Pyrophoric solid, Category 1 2.10.7.1.2. Example 2 Name: Zorapyrole Physical state: Solid Pyrophoric properties: Unknown, therefore the UN Test N.2 of the UN - MTC was applied. When poured from one meter height according to the test procedure, Test result: - ignited after two minutes already in the first trial. Zorapyrole self Classification: Pyrophoric solid, Category 1 Examples of substances and mixtures not fulfilling the classification 2.10.7.2. criteria 2.10.7.2.1. Example 3 Name: Nonopyr Physical state: Solid Pyrophoric properties: Nonopyr has been handled extensively in air and has never self - ignited. From the chemical structure no pyrophoricity is expected. Classification: Not a pyrophoric solid

179 Guidance on the Application of the CLP Criteria 5.0 July 2017 179 Version – 2.10.7.2.2. Example 4 Name: Pyronot Physical state: Solid Pyrophoric properties: Unknown, therefore UN Test N.2 of the UN - MTC was applied. Test result: When poured from one meter height according to the test procedure no procedure was repeated six times ignition occurred within five minutes. The and each time the result was negative. Classification: Not a pyrophoric solid 2.10.8. References Urben, Peter G. (2007). Bretherick's Handbook of Reactive Chemical Hazards , Volumes 1 - 2 (7th Edition). Elsevier.

180 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 180 2.11. SELF - HEATING SUBSTANCES AND MIXTU RES 2.11.1. Introduction The criteria for ‘Self - heating substances and mixtures’ are found in Annex I, Section 2.11 of CLP and are identical to those in Chapter 2.11 of GHS. Self - heating is the result of an exothermic reaction of a substance o r mixture with the oxygen in the air. Initially, the reaction rate may be very low. However, when the heat produced cannot be removed rapidly enough (i.e. heat accumulation), the substance or mixture will self heat, - - nition. The phenomenon can occur only where a large with the possible consequence of self ig surface of substance or mixture is in contact with air or oxygen (for example, piles of powders, crystals, splinters, any other rough surface etc.). The initiation occurs usually at or near the the substance or mixture pile with the available air in the interspace between the centre of particles. Since the surface area of a solid substance or mixture exposed to air increases with decreasing particle size, it follows that particle size and shape will greatl y influence the propensity of a substance or mixture to self heat. Therefore it is very important that self - heating properties for - solids, and especially powders, are determined for the substance or mixture in the form it is supplied and expected to be use d. 2.11.2. Definitions and general considerations for the classification of self - heating substances and mixtures The definitions in CLP for self - heating substances and mixtures are as follows: Annex I: 2.11.1.1. - heating substance or mixture is a liquid or s olid substance or A self mixture, other than a pyrophoric liquid or solid, which, by reaction with air and without energy - supply, is liable to self heat; this substance or mixture differs from a pyrophoric liquid or solid in that it will ignite only when in large amounts (kilograms) and after long periods of time (hours or days). Self - heating of a substance or a mixture is a process where the gradual reaction of 2.11.1.2. that substance or mixture with oxygen (in the air) generates heat. If the rate of heat produc tion exceeds the rate of heat loss, then the temperature of the substance or mixture - ignition and combustion. will rise which, after an induction time, may lead to self Relation to other physical hazards 2.11.3. Pyrophoric solids and liquids should not be consid ered for classification as self - heating substances and mixtures. Classification of self - heating substances and mixtures 2.11.4. 2.11.4.1. Identification of hazard information Self - heating is a very complex phenomenon which is influenced by many parameters (some of them being volume, temperature, particle shape and size, heat conductivity and bulk density). - heating behaviour cannot be predicted from any theoretic al model. In some Therefore, self cases, properties might even differ between producers of seemingly very similar substances or mixtures. Differences in self - heating behaviour are especially to be anticipated where surface treatment occurs in the production process. Hence , all data sources should be carefully evaluated with regard to reliability and scientific validity. It is of utmost importance that in compliance with Articles 5 and 6 of CLP authentic and used for testing. In many representative material in the correct form and physical state be

181 Guidance on the Application of the CLP Criteria 5.0 – July 2017 181 Version 2.11.4.2 ) can be used to determine whether self cases, a simple screening test (see Section - heating occurs or not. 2.11.4.2. ning procedures and waiving of testing Scree The classification procedure for self - Annex I: 2.11.4.2. heating substances or mixtures need not be applied if the results of a screening test can be adequately correlated with the classification test and an appropria te safety margin is applied. Examples of screening tests are: (a) The Grewer Oven test (VDI guideline 2263, part 1, 1990, Test methods for the De - termination of the Safety Characteristics of Dusts) with an onset temperature 80 K above the reference tem perature for a volume of 1 l; (b) The Bulk Powder Screening Test (Gibson, N. Harper, D.J. Rogers, R. Evaluation of the fire and explosion risks in drying powders, Plant Operations Progress, 4 (3), 181 189, 1985) - with an onset temperature 60 K above the reference temperature for a volume of 1 l. - heating EU test method A.16 as described in Regulation (EC) No 440/2008 checks for self properties. However, the method used is generally inappropriate for a sound assessment, and lassification. Therefore, special care must be taken if results from the findings do not lead to a c EU test method A.16 are interpreted towards a CLP classification for self heating substances and - mixtures. - heating applies only to solids. The surface o f liquids is not In general, the phenomenon of self large enough for reaction with air and the test method is not applicable to liquids. Therefore liquids are not classified as self heating. However, if liquids are adsorbed on a large surface - - d should be considered. (e.g. on powder particles), a self heating hazar Substances or mixtures with a low melting point (< 160 °C) should not be considered for - air classification in this class since the melting process is endothermic and the substance surface is drastically reduced. However, this crite rion is only applicable if the substance or mixture is completely molten up to this temperature. 2.11.4.3. Classification criteria heating substance or mixture must be classified in one of the two categories for this class A self - III, Sub - nce with UN Test N.4 in Part 33.3.1.6 of the if, in a test performed in accorda section MTC, the result meets the criteria according to following table: UN - Table 2.11.1 Annex I: heating substances and mixtures - Criteria for self Criteria Category A positive result is obtained in a test using a 25 mm sample cube at 140 °C 1 (a) a positive result is obtained in a test using a 100 mm sample cube at 140 °C and a negative result is obtained in a test using a 25 mm cube sample at 140 °C and volume the substance or mixture is to be packed in packages with a 3 of more than 3 m ; or 2 (b) a positive result is obtained in a test using a 100 mm sample cube at 140 °C and a negative result is obtained in a test using a 25 mm cube sample at 140 °C, a po sitive result is obtained in a test using a 100 mm cube sample at 120 °C and the substance or mixture is to be packed in packages with a volume of more than 450 litres; or

182 ication of the CLP Criteria Guidance on the Appl 182 Version 5.0 – July 2017 (c) mm sample cube at 140 °C a positive result is obtained in a test using a 100 mm cube sample at 140 and a negative result is obtained in a test using a 25 and a positive result is obtained in a test using a 100 mm cube sample at °C 100 °C. Note The test sh all be performed on the substance or mixture in its physical form as presented. If, for example, for the purposes of supply or transport, the same chemical is to be presented ikely to in a physical form different from that which was tested and which is considered l materially alter its performance in a classification test, the substance shall also be tested in the new form. Substances and mixtures with a temperature of spontaneous combustion higher than 2.11.2.3. sified as a self - heating substance or mixture. 50 °C for a volume of 27 m³ shall not be clas 2.11.2.4. Substances and mixtures with a spontaneous ignition temperature higher than 50 °C litres shall not be assigned to Category for a volume of 450 1 of this class. 2.11.4.4. Testing and evaluation of hazard information A self - heating substance or mixture must be classified in one of the two categories for this class 33.3.1.6 of the UN using UN Test N.4 in Part III, Sub - section - MTC. 2.11.4.4.1. General remarks - test, further testing becomes If self heating behaviour cannot be ruled out by a screening necessary. UN Test N.4 as described in the latest version of the UN - MTC should be used. Explosive substances and mixtures should not be tested according to this method. For safety and self - reactive properties and to rule out reasons, it is advisable to test for explosive pyrophoric behaviour before performing this test. The oven should be equipped with an appropriate pressure release device in case an energetic decomposition is triggered by a - ining flammable solvents explosion protection measures temperature rise. For samples conta have to be taken. The tests may be performed in any order. It is suggested to start with the 25 mm sample cube at 140 °C. If a positive result is obtained, the substance or mixture must be classified a s a self - heating substance or mixture, Category 1, and no further testing is necessary. The test procedure need not be applied if the substance or mixture is completely molten at 160 °C. 2.11.4.4.2. Sample preparation The sample (powder or granular) in its commercial form should be used and should not be milled or ground. It should be filled to the brim of the sample container and the container tapped several times. If the sample settles, more is added. If the sample is heaped it should be levelled to the brim. The sam ple container is placed in the oven as described in the UN - MTC. 2.11.4.4.3. Criteria and evaluation A positive result is obtained if spontaneous ignition occurs or if the temperature of the sample time count starts exceeds the oven temperature by 60 K. The testing time is 24 hours. The when the temperature in the centre of the sample has reached a value of 2 K below the oven temperature. This is especially important when the sample contains solvents which evaporate under the test conditions or when larger test volumes are used for extrapolation purposes (see below). Before starting UN Test N.4, the decomposition behaviour of the sample should be known. In general, it is sufficient to perform a screening with Differential Scanning Calorimetry. Special o the interpretation of the test data is necessary when exothermic care with respect t

183 Guidance on the Application of the CLP Criteria 5.0 July 2017 Version – 183 decomposition may occur at the test temperatures. In such cases, a test under an inert atmosphere (i.e. nitrogen) should be run to determine the temperature rise due to decomposition. Caref ul flushing with the chosen inert gas is essential in such cases since otherwise much air may be retained between the crystals of the sample in the container. 2.11.4.5. Decision logic The following decision logic for self - heating substances and mixtures is applicab le according to CLP. NOTE: The person responsible for the classification of self - heating substances and mixtures should be experienced in this field and be familiar with the criteria for classification.

184 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 184 Annex I: Figure 2.11.1. Self - heating substances and mixtures SUBSTANCE/MIXTURE heating when - Does it undergo dangerous self NOT CLASSIFIED tested in a 100 mm sample cube at 140 °C? NO Category 1 YES Does it undergo dangerous self heating when - YES sample cube at 140 °C? tested in a 25 mm Danger NO Category 2 3 ? Is it packaged in more than 3 m YES NO Warning heating when - Does it undergo dangerous self tested in a 100 mm sample cube at 120 °C? NOT CLASSIFIED NO YES Category 2 Is it packaged in more than 450 litres volume? YES NO Warning heating when Does it undergo dangerous self - Category 2 tested in a 100 mm sample cube at 100 °C? YES NO Warning NOT CLASSIFIED Exemption 2.11.4.6. 2.11.4.3 The following exemptions apply (see Section ):  Substances and mixtures with a temperature of spontaneous combustion higher than 50 heating substance or mixture. °C for a - volume of 27 m³ must not be classified as a self

185 Guidance on the Application of the CLP Criteria 5.0 – July 2017 185 Version  Substances and mixtures with a spontaneous ignition temperature higher than 50 °C for a volume of 450 litres must not be assigned to Category 1 of this class. MT C does not provide any guidance on how these values should be However, the UN - determined. The UN test regime is based on the assumption of a cubic sample shape. For the extrapolation to larger volumes, an improved model has to be used. According to Grewer (Grewer, 1994), plotting the logarithm of the volume to surface ratio (log (V/A)) versus the reciprocal temperature gives good results without knowledge of the Frank - Kamenetzskii (Frank - Kamenetzskii, 1969) shape factor. m³ can be found by extrapolation of the The critical temperature for a volume of 450 l or 27 Figure critical temperature in a log (V/A) vs. 1/T plot (see . 9 ): 2 2 9 Extrapolation towards large volumes Figure . Temperature in °C 220 60 70 240 50 200 180 160 140 120 100 90 80 27 m3 -0,4 10 m3 -0,6 Extrapolation to 27 m³ 1 m3 -0,8 -1,0 100 L -1,2 Note: sample gave positive -1,4 result at 140 °C / 1 liter -1,6 1,6 L -1,8 400 mL log(V/A) in m positive result -2,0 negative result 110 mL Sample volume -2,2 16 mL -2,4 -2,6 Linear regression lines -2,8 -3,0 0,0030 0,0028 0,0027 0,0026 0,0025 0,0024 0,0023 0,0022 0,0029 0,0021 0,0020 0,0019 0,0031 reciprocal abs. Temperature in 1/K The test setup is essentially the same as in UN Test N.4 of the UN - MTC but now the sample size 2.11.4.3 ection ically varied. The criteria of S and possibly the shape are systemat apply as well. The critical temperature must be determined over a range of at least four different volumes and with a volume not smaller than 16 ml. If possible, larger volumes should be also tested. The 1 borderline temperature should be determined as precisely as possible. For small volumes (< heating may be considerably less than 60 K; in this case litre), the temperature rise due to self - a noticeable temperature rise is interpreted as a po sitive result. A conservative approach is required for the evaluation. The uncertainty of measurement must be taken into account. The extrapolation must be based on a linear regression of the negative and positive borderline data sets in the log (V/A) vs. 1/T diagram. The maximum permissible difference between a positive and a negative result should be 5 K. An exemption may be claimed if the more conservative endpoint for the particular volume is well beyond 50 °C (i.e. 55 °C or higher).

186 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 186 2.11.5. Hazard communicatio n for self - heating substances and mixtures 2.11.5.1. Pictograms, signal words, hazard statements and precautionary statements : Annex I Table 2.11.2 Label elements for self - heating substances and mixtures Classification Category 1 Category 2 GHS Pictograms Signal Word Danger Warning heating in large - H251: Self - heating; may H252: Self Hazard Statement quantities; may catch fire catch fire Precautionary Statement P235 P235 Prevention P280 P280 Precautionary Statement Response P407 P407 Storage Precautionary Statement P413 P413 P420 P420 Precautionary Statement Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. Relation to transport classification 2.11.6. – substances and mixtures liable to spontaneous combustion – Division 4.2 within Class 4 of the UN RTDG Model Regulations comprises the following entries: a. pyrophoric substances and mixtures ; heating substances and mixtures. - b. self , Whereas pyrophoric substances and mixtures in the modal transport regulations (ADR, RID ADN and IMDG Code, ICAO TI) are assigned to packing group I, self - heating substances and mixtures are assigned to packing groups II and III. In cases where a substance or mixture is classified in Division 4.2, packing group II or III, the translation int o the CLP system is straightforward. It should be kept in mind that transport classification is based on prioritisation of hazards (see UN RTDG Model Regulations, Section 2.0.3) and that self - heating substances and mixtures have a relatively low rank in th e precedence of hazards. Therefore, the translation from the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI) to CLP should be only done if a transport classification as self - heating is explicitly available. The conclusion that a substanc e or heating mixture not classified as self - heating for transport should not be classified as a self -

187 Guidance on the Application of the CLP Criteria 5.0 – July 2017 187 Version substance or mixture according to CLP is, in general, not correct. See Annex VII for additional information on transport classification in relation to CLP classification. Examples of classification for self - 2.11.7. heating substances and mixtures Examples of substances and mixtures fulfilling the classification 2.11.7.1. criteria  many organometallic compounds, especially substances or mixtures containing transition metals; m any organic substances or mixtures; the tendency to self - heat increases with  decreasing particle size; many metals, especially catalysts.  2.11.7.2. Examples of substances and mixtures not fulfilling the classification criteria In general, liquids show no self - heati ng behaviour unless adsorbed on a large surface. Scientific background - A basic model for the thermal explosion of solids was first developed by Frank Kamenetzskii (Frank - Kamenetzskii, 1969). It is based on the assumption that only the heat loss by thermal conduction is relevant for the phenomenon. In this case, the critical criterion for a thermal runaway reaction can be described as a linear relationship between the reciprocal absolute temperature and the logarithm of volume. he UN for self - heating substances and mixtures is based on The classification scheme of t charcoal as a reference system. The critical temperature for a 1 litre cube of charcoal is 140 °C and for a cube of 27 m³ 50 °C. When a parallel line is drawn in the 1/T vs. logarithm of volume dia gram from the reference points 1 litre / 120 °C and 1 litre / 100 °C, the corresponding volumes for a critical temperature of 50 °C are found to be 3 m³ and 450 l, respectively (see Figure 2 . 10 ). The black dotted line in Figure 2 . 10 separates Category 1 from Category 2. For examples of results followi UN - MTC. ng the Test N.2 see Section 33.3.1.4.5 of However, the slope of the line in the 1/T vs. volume diagram depends on the individual activation energy of the substance or mixture, and therefore it may vary within certain limits. It must be born in mind that this test regime has been developed to facilitate classification and that it may not suffice to solve safety issues in storage.

188 Guidance on the Appl ication of the CLP Criteria Version 5.0 – 188 July 2017 Figure 2 . 10 Volume dependency of the critical temperature for charcoal 0,002 220 Grewer test 200 0,0022 180 Charcoal line reference 160 (UN Manual T + C) 0,0024 140 recipr. Temp. 120 self heating - Not 0,0026 Cat . 1 Exemption Exemption 100 90 Temperature (°C) < 450 0,0028 < 3 m 80 ltr ³ symbolize dots Yellow 70 points UN Manual test 0,003 60 Cat 50 . 2 0,0032 1 10 100 1000 10000 100000 0,001 0,01 0,1 Volume (liter) References 2.11.8. Grewer, T. (1994). Thermal hazards of chemical reactions , Elsevier. Frank Kamenetzskii, D.A. (1969). - , 2nd edition, Diffusion and heat transfer in chemical kinetics Plenum Press, New York, London.

189 Guidance on the Application of the CLP Criteria – Version 189 5.0 July 2017 RES WHICH, IN CONTAC T WITH WATER, 2.12. SUBSTANCES AND MIXTU EMIT FLAMMABLE GASES Introduction 2.12.1. The criteria for ‘Substances and mixtures which, in contact with water, emit flammable gases’ are found in Annex I, Section 2.12 of CLP and are identical to those in Chapter 2.12 of GHS. cal structure and/or the physical state (e.g. particle size) substances or Depending on the chemi mixtures may be able to react with water (even damp / air humidity) under normal ambient temperature conditions. Sometimes this reaction can be violent and/or with significant ation of heat. Especially if gases are evolved this reaction may become very dangerous gener during use. In addition, it is important to know whether a substance or mixture emits flammable gases after contact with water because special precautions are necessary especially with regard to explosion protection. Examples are demonstrated in the following table. . Examples of hazards, depending on the property of the emitted gas, when 2 1 Table substances and mixtures are i n contact with water Type of Example of the hazard CLP Reference emitted gas Annex II, 1.1.3:  Heating up of the substance Gas (in general) Supplemental hazard  Splashing of the substance and thus e.g. information: contact with skin etc. or additional risk during fire fighting EUH014* Pressure rise and bursting  of e.g. the packaging, tank Ignition  Annex I, 2.12: Flammable gas H260/H261 Flash of fire   Toxic gas Annex II, 1.2.1: Damage to health: intoxication (acute) Supplemental hazard information: EUH029 * For s upplemental hazard information: see Section 2.4.2 2.1 Definitions and general considerations for the classification of 2.12.2. substances and mixtures which, in contact with water, emit flammable gases The following definition is given in CLP for substa nces and mixtures which, in contact with water, emit flammable gases (CLP Annex I, 2.12). Annex I: 2.12.1. Substances or mixtures which, in contact with water, emit flammable gases means solid or liquid substances or mixtures which, by interaction with water, are liable to become spontaneously flammable or to give off flammable gases in dangerous quantities. 2.12.3. Relation to other physical hazards If the chemical identity of the emitted gas is unknown, the gas must be tested for flammability (unless it ignit es spontaneously). Other than under DSD/DPD, pyrophporic liquids and

190 ication of the CLP Criteria Guidance on the Appl 190 Version 5.0 – July 2017 pyrophoric solids have to be considered for classification in this hazard class as well and data about pyrophoric properties are needed prior to testing for this hazard class. Classification of substances and mixtures which, in contact with 2.12.4. water, emit flammable gases Identification of hazard information 2.12.4.1. For the classification of substances and mixtures which, in contact with water, emit flammable gases the following data are n eeded, if applicable: chemical structure;   water solubility; chemical identity and flammability of the emitted gas;   pyrophoric properties of the tested substance or mixture; particle size in case of solids;   friability in case of solids; hazard properties i  n general;  information concerning the experience in production or handling. IR & CSA, Chapter R.7a: Endpoint specific guidance See also , Section R.7.1.7 (Water solubility), R.7.1.14 (Granulometry). whether the substance or mixture Information about the chemical structure is used to check . contains metals and/or metalloids The water solubility is used to decide whether the substance or mixture is soluble in water to n form a stable mixture. This may also be decided based on information concerning experience i handling or use, e.g. the substance or mixture is manufactured with water or washed with water 2.12.4.4.1 ). (see Section the emitted gas is used to decide whether the evolved gas is flammable The chemical identity of or not. If the chemical identity of the emitted gas is unknown, the gas must be tested for 2.2 ). flammability (see Section of the UN In case of pyrophoric substances and mixtures the UN Test N.5 MTC, Part III, Section - 33.4.1.3.1 must be executed under nitrogen atmosphere. Therefore, data about pyrophoric properties are needed prior to testing. The melting point, boiling point and information about viscosity are necessary to identify the physical state of the substance or mixture. See also IR & CSA, Chapter R.7a: Endpoint specific guidance , Section R.7.1.2 (Melting point/freezing point), R.7.1.3 (Boiling point), R.7.1.18 (Viscosity). Even though the UN Test N.5 can be applied to both, soli ds and liquids, these data are necessary to decide whether information concerning the friability (for solids) in accordance with the test method is necessary. The particle size and the friability of a solid substance or mixture are crucial parameters for the classification of substances and mixtures which, in contact with water, emit flammable gases. These parameters have a significant effect on the test result. Thus specific requirements regarding the particle size and the friability are prescribed in the UN Test N.5. For further details regarding the test procedure see Section 2.12.4.4.1 . The references in Section 2.12.8 provide good quality data on physical hazards.

191 Guidance on the Application of the CLP Criteria 5.0 – July 2017 191 Version 2.12.4.2. Screening procedures and waiving of testing For the majority of substances and mixtures, flam mability as a result of contact with water is not a typical property and testing can be waived based on a consideration of the structure and experiences in handling and use. 2.12.4.1. The classification procedure for this class need not be applied if: Annex I: a) the chemical structure of the substance or mixture does not contain metals or metalloids; or b) experience in handling and use shows that the substance or mixture does not react with water, e.g. the substance is manufactured with water or washed with water; or c) the substance or mixture is known to be soluble in water to form a stable mixture. Classification criteria 2.12.4.3. Annex I: Table 2.12.1 Criteria for substances or mixtures which in contact with water emit flammable gas Category Criteria Any substance or mixture which reacts vigorously with water at ambient temperatures and demonstrates generally a tendency for the gas produced to ignite spontaneously, or which reacts readily with water at ambient temperatures 1 on of flammable gas is equal to or greater than 10 such that the rate of evoluti litres per kilogram of substance over any one minute. Any substance or mixture which reacts readily with water at ambient temperatures such that the maximum rate of evolution of flammable gas is equal to or greater 2 than 20 litres per kilogram of substance per hour, and which does not meet the criteria for Category 1. Any substance or mixture which reacts slowly with water at ambient temperatures such that the maximum rate of evolution of flammable ga s is equal to or greater 3 than 1 litre per kilogram of substance per hour, and which does not meet the criteria for Categories 1 and 2. Note: The test shall be performed on the substance or mixture in its physical form as presented. If for example, for the purposes of supply or transport, the same chemical is to be presented in a physical form different from that which was tested and which is considered likely to materially alter its performance in a classification test, the substance must also be tested in the new form. 2.12.2.2. A substance or mixture shall be classified as a substance or mixture which in contact with water emits flammable gases if spontaneous ignition takes place in any step of the test procedure. 2.12.4.4. Testing and evaluation of hazard information 2.12.4.4.1. Testing procedure Care must be taken during testing as the emitted gas might be toxic or corrosive.

192 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 192 The testing procedure for substances and mixtures which in contact with water emit flammable ors and therefore must be carried out by gases is sensitive to a number of influencing fact experienced personnel. Some of these factors are described in the following: Apparatus / measuring technique 2. In UN Test N.5 no special laboratory apparatus / measuring technique to determine the rate of n is required and no reference material is prescribed. As demonstrated in the past gas evolutio by a round robin test (Kunath, K. et al . 2011), the gas evolution rate measured by different apparatuses may vary widely. Therefore in order to avoid measuring and classific ation errors adequate quality control measures are necessary to validate the results and should be noted in the test report. 3. Particle size and/or friability The particle size of a solid has a significant effect on the test result. Therefore, if for solids the percentage of powder with a particle size of less than 500 μm constitutes more than 1 % of the total mass, or if the substance or mixture is friable, then the complete sample must be ground in particle size during handling to a powder before testing to account for a possible reduction and transport. In certain cases, grinding may not be applicable and/or the sample cannot be ground completely to a particle size of less than 500 μm (e.g. metal granules). ctive procedures, the particle size and the Information on these pre - treatments and the respe friability has to be provided in the test report. Atmospheric parameters 4. Variations of the atmospheric parameters (mainly air pressure and temperature) during the test have a considerable influence on the test r esult. Therefore the substance or mixture must be tested at 20 °C, i.e. make sure that the test apparatus is acclimatised to 20 °C. On the other hand it is difficult to regulate and stabilise the air pressure during the testing. To characterise this influe ncing factor and to avoid false positive results, an additional ‘blank test’ is highly recommended. The results of the blank test should be noted in the test report. Test with demineralised (distilled) water 5. The UN Test N.5 is performed with demineralised (distilled) water. In practice, contact with water can be to water in the liquid state (fresh water, sea water) or humid air, respectively. Note that the reactivity and thus the gas evolution rate observed in practice may differ from the gas evolution rat e value measured using demineralised water. This should be taken into account when handling substances and mixtures which in contact with water emit flammable gases. 6. Stirring procedures during the test t may have a considerable effect on the Stirring of the sample or water mixture during the tes test result (e.g. significant increase or decrease of the gas evolution rate). Therefore, the sample or water mixture should not be stirred continuously during the test, e.g. by an automatic magnetic stirrer, even if the test sample has hydrophobic properties and moistening of the sample becomes impossible (see Kunath K. ., 2011). et al 7. Spontaneous ignition Spontaneous ignition of the evolved gas without contact with an additional ignition source, i.e. without the flam e of the gas burner results in classification as Category 1. This does not necessarily mean that the evolved gas is pyrophoric but often the heat of reaction is sufficient to ignite the evolved gas (e.g. the hydrogen evolved when sodium reacts with water).

193 Guidance on the Application of the CLP Criteria 5.0 – July 2017 193 Version Evaluation of hazard information 2.12.4.4.2. In order to accurately interpret the test results the evaluating person must have sufficient experience in the application of the test methods and in the disturbing / influencing factors as described above. of data comprises two steps: The evaluation evaluation of all available data; and  identification of the study or studies giving rise to the highest concern (key studies).  The criteria for assignment to Category 2 or 3 are gas evolution rates of 20 and 1 litre per kilog ram of substance or mixture per hour, respectively, but for Category 1 the relevant criterion is 10 litres per kilogram of substance or mixture over any one minute period (if the gas nd for correct does not ignite spontaneously). This has to be considered while testing a evaluation of the test results. The assignment to the respective hazard class/category will further determine the technical means to be taken to avoid dangerous events which, in combination with other safety characteristics such as i) explos ion limits, ii) flash points (applicable only for liquids) or iii) self - ignition temperature, can lead to clear restrictions in the conditions of use. 2.12.4.5. Decision logic Classification of substances and mixtures which, in contact with water, emit flammable g ases is done according to decision logic 2.12.4.1 as included in the GHS. NOTE: The person responsible for the classification of substances and mixtures which, in contact with water, emit flammable gases should be experienced in this field and be familiar with the criteria for classification.

194 ication of the CLP Criteria Guidance on the Appl 194 Version 5.0 – July 2017 Figure . 11 2 Decision logic for substances and mixtures which, in contact with water, emit flammable gases (Decision logic 2.12 of GHS) Substance/mixture No water, does it react slowly at ambient temperatures such In contact with 1 litre per kg that the maximum rate of evolution of flammable gas is ≥ classified Not of substance per hour? Yes Category 1 In contact with water, does the substance react vigorously with water at Yes ambient temperatures and demo nstrate generally a tendency for the gas produced to ignite spontaneously, or does it react readily with water at ambient temperatures such that the rate of evolution of flammable gas is Danger ≥ 10 litres per kg of substance over any one minute? No 2 Category Yes In contact with water, does it react readily with water at ambient temperatures such that the maximum rate of evolution of flammable 20 gas is ≥ litres per kg of substance per hour? Danger No 3 Category Warning

195 Guidance on the Application of the CLP Criteria – Version 195 5.0 July 2017 Hazard communication for substances and mixtures which, in contact 2.12.5. with water, emit flammable gases Pictograms, signal words, hazard statements and precautionary 2.12.5.1. statements for substances and mixtures Table 2.12.2 : Annex I Label elements for substances or mixtures which in contact with water emit flammable gases Category 1 Category 2 Category 3 Classification GHS Pictograms Danger Signal Word Warning Danger H260: H261: Hazard Statement H261: In contact with with water In contact In contact with releases flammable water releases water releases flammable gases gases which may flammable gases ignite spontaneously P223 P223 Precautionary Statement P231 + P232 P231 + P232 P231 + P232 Prevention P280 P280 P280 P302 + P335 + P334 P302 + P335 + Precautionary Statement Response P334 P370 + P378 P370 + P378 P370 + P378 P402 + P404 Precautionary P402 + P404 P402 + P404 Statement Storage Precautionary P501 P501 P501 Statement Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. 2.12.5.2. Additional labelling provisions Annex II of CLP provides the following additional labelling provisions for water - reactive substances and mixtures. These statements must be assigned in accor dance with CLP, Article 25 (1), to substances and mixtures classified for physical, health or environmental hazards. There are no criteria or test methods provided for these EUH statements.

196 ication of the CLP Criteria Guidance on the Appl 196 Version 5.0 – July 2017 – ' Reacts violently with water ' 1.1.3. EUH014 Annex II: substances and mixtures which react violently with water, such as acetyl chloride, alkali For metals, titanium tetrachloride. 1.2.1. EUH029 - ' Contact with water liberates toxic gas ' Annex II: damp air, evolve gases classified For substances and mixtures which in contact with water or for acute toxicity in category 1, 2 or 3 in potentially dangerous amounts, such as aluminium phosphide, phosphorus pentasulphide. Relation to transport classification 2.12.6. ations covers substances and mixtures Division 4.3 within Class 4 of the UN RTDG Model Regul which in contact with water emit flammable gasses. Substances and mixtures which are classified and/or labelled in Division 4.3 in the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI) are classified a s substances and mixtures which, in contact with water, emit flammable gases under CLP. See Annex VII for additional information on transport classification in relation to CLP classification. 2.12.7. Examples of classification for substances and mixtures which, in contact with water, emit flammable gases 2.12.7.1. Example of a substance fulfilling the classification criteria Many different types of chemicals may belong to the hazard class of substances and mixtures which, in contact with water, emit flammable gases, for exa mple, alkali metals, alkyl aluminium derivatives, alkyl metals, metal hydrides, metal phosphides, certain metal powders. A Bretherick’s Handbook of Reactive Chemical Hazards comprehensive list can be found in (Urben, 2007). Example 1 2.12.7.1.1. PYROPHORIC SUBSTANCE FULFILLING THE CRITERIA FOR CLP CLASSIFICATION Magnesium alkyls (Index No. 012 - 003 - 00 - 4) Substance: Mg R Chemical structure: 2 Hydrogen Flammable gas: not applicable Gas evolution rate: Spontaneous ignition: not possible due to the nitrogen atmosphere during the UN Test N.5 DSD classification: F; R14 - 17 Transport classification: - Reference: Former Annex I to DSD and Annex VI to CLP  CLP Classification: Water - react. 1; H260 Pyr. Sol. 1; H250 Supplemental Hazard Information: EUH014

197 Guidance on the Application of the CLP Criteria 5.0 – July 2017 197 Version 2.12.7.2. Example of a substance not fulfilling the classification criteria Example 2 2.12.7.2.1. MANGANESE ETHYLENE BIS (DITHIOCARBAMATE) COMPLEX WITH ZINC SALT 88 % (MANCOZEB) Gas evolution rate: 0 litres per kilogram of substance per hour. Spontaneous ignition: not applicable Transport classification: not Class 4.3 - MTC Table 33.4.1.4.5 Reference: UN Test N.5, UN CLP Classification:  Not classified as substance which, in contact with water, emit flammable gases 2.12.8. References rd et al. (2012) CRC Handbook of Chemistry and Physics 93 . CRC Edition William M. Haynes Press, Taylor and Francis, Boca Raton, FL - : GESTIS database on hazardous substances http://www.dguv. de/bgia/en/gestis/stoffdb/index.jsp et al. (2016, 2012) The Merck Index O'Neil, Maryadele J. An Encyclopaedia of Chemicals, - Drugs, and Biologicals (14th Edition – Version 14.9). Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc. Urben, Peter G. (2007). Bretherick's Handbook of Reactive Chemical Hazards , Volumes 1 - 2 (7th Edition). Elsevier. Kunath, K., Lüth, P., Uhlig, S. (2011). Interlaboratory test on the method UN Test N.5 / EC A.12 “Substances which, in contact with water, emit flammable gases” 2007 .Short report. BAM Bundesanstalt für Materialforschung und – prüfung. Berlin. ISBN 978 - 3 - 9814634 - 1 - 5. http://www.bam.de/de/serv ice/publikationen/publikationen_medien/short__report_rv_un_n_5. pdf

198 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 198 2.13. OXIDISING LIQUIDS Introduction 2.13.1. The criteria for ‘Oxidising liquids’ are found in Annex I, Section 2.13 of CLP and are identical to those in Chapter 2.13 of GHS. The hazard class oxidis ing liquids comprises liquid substances and mixtures whose hazard is characterised by the fact that, in contact with other materials, they are able to cause or contribute to the combustion of those materials. The other materials do not necessarily have to belong to a certain hazard class in order to be able to be affected by the presence of oxidising substances or mixtures. This is for example the case when a solid material (e.g. wood) is soaked with an oxidising liquid. Certain combinations of combustible materials and oxidising substances or mixtures may even result in spontaneous combustion, thermal instability or form an explosive mixture, this means that they may have explosive properties or may be regarded as self - reactive substances or mixtures. Alth ough widely known as oxidising materials, their hazard and behaviour might be better understood by considering them to be fire enhancing substances or mixtures. The hazards communication of oxidising liquids intends to communicate the property that it may cause fire or explosion or that it may intensify fire. Apart from the combustion hazard, the production of toxic and/or irritating fumes may cause an additional hazard. For example, when nitrates are involved in a fire, nitrous fumes may be formed. The te sting procedure and criteria for oxidising substances or mixtures do not work properly for ammonium nitrate compounds or solutions, ammonium nitrate based fertilizers and ammonium nitrate emulsions, suspensions or gels. Therefore for classification and lab elling of substances or mixtures containing ammonium nitrate, known experience should be used and expert judgement should be sought. For the classification procedures for ammonium nitrate emulsions, – intermediate for blasting explosive s, see Chapter 2.1 of this guidance. suspensions or gels 2.13.4.3 Annex I: In the event of divergence between test results and known experience in the handling and use of substances or mixtures which shows them to be oxidising, judgments based on known experience shall take precedence over test results. 2.13.2. Definitions and general considerations for the classification of oxidising liquids The CLP text comprises the followi ng definition for oxidising liquids. Annex I: 2.13.1. Definition Oxidising liquid means a liquid substance or mixture which, while in itself not necessarily combustible, may, generally by yielding oxygen, cause, or contribute to, the combustion of othe r material. 2.13.3. Relation to other physical hazards Oxidising liquids that are mixed with combustible materials or reducing agents may have explosive properties and should be considered for classification in the hazard class Explosives (including the applicable screening procedures), see Chapter 2.1 of this guidance.

199 Guidance on the Application of the CLP Criteria 5.0 – July 2017 199 Version - In rare cases, mixtures with oxidising liquids may exhibit self reactive behaviour, see Chapter of this guidance. Expert judgement should be sought in case of doubt. 2.8 The classification procedure and criteria for oxidising substances or mixtures is not applicable ganic peroxides. Under DSD organic peroxides were considered to be oxidising substances for or O – or mixtures because of the presence of the – bond. The majority of the organic peroxides – O do not possess oxidising properties; their main hazards are reactivity and flammability. Under CLP organic peroxides are comprised in a separate hazard class (CLP Annex I, 2.15) and they must not be considered according to the procedures described for oxidising liquids. Organic peroxides were classified as oxidising (O; R7) accor ding to the DSD, which was not appropriate since the vast majority of them do not exhibit oxidising properties. Inorganic oxidising liquids are not flammable and therefore do not have to be subjected to the classification procedures for the hazard classes flammable liquids or pyrophoric liquids. Also other liquids that are classified as oxidising liquids are normally not flammable, although a few exemptions may exist. Expert judgement should be sought in case of doubt. 2.13.4. Classification of substances and mixt ures as oxidising liquids 2.13.4.1. Identification of hazard information Oxidising liquids may cause, or contribute to, the combustion of other material. Although the definition states that they generally do this by yielding oxygen, halogens can behave in a simila r way. Therefore, any substance or mixture containing oxygen and/or halogen atoms should in principle be considered for inclusion into the hazard class oxidising liquids. This does not necessarily mean that every substance or mixture containing oxygen and/ or halogen atoms should be subjected to the full testing procedure. Screening procedures and waiving of testing 2.13.4.1.1. Liquids that are classified as explosives should not be subjected to the testing procedures for oxidising liquids. Organic peroxides should be considered for classification within the hazard class organic 2.15 peroxides, see Chapter of this guidance. Experience in the handling and use of s ubstances or mixtures which shows them to be oxidising is an important additional factor in considering classification as oxidising liquids. In the event of divergence between test results and known experience, judgement based on known experience should ta ke precedence over test results. Before submitting a substance or a mixture to the full test procedure, an evaluation of its chemical structure may be very useful as it may prevent unnecessary testing. The person applying this procedure should have suffici ent experience in testing and in theoretical evaluation of hazardous substances and mixtures. The following text provides a guideline for the theoretical evaluation of potential oxidising properties on basis of its composition and chemical se of doubt, the full test must be performed. structure. In ca For organic substances or mixtures the classification procedure for this hazard class need not to be applied if: a. the substance or mixture does not contain oxygen, fluorine or chlorine; or b. the substance or mixtu re contains oxygen, fluorine or chlorine and these elements are chemically bonded only to carbon or hydrogen. For inorganic substances or mixtures, the classification procedure for this hazard class need not be applied if they do not contain oxygen or halogen. On basis of this theoretical evaluation only a distinction can be made between ‘potentially oxidising’ (i.e. further testing required) and ‘non - oxidising’ (i.e. no further testing for this

200 ication of the CLP Criteria Guidance on the Appl 200 Version 5.0 – July 2017 hazard class required). It is not possible to assign a haz ard category on basis of a theoretical evaluation. Any substance or mixture that complies with the above waiving criteria can be safely regarded to have no oxidising properties and, hence, needs not to be tested and needs not to be liquid. However, such a substance or mixture may still possess other regarded as an oxidising hazardous properties that require classification into another hazard class. In case a mixture of an oxidising substance and a non - hazardous inert substance is offered for , the following should be taken into account: classification  An inert material by definition does not contribute to the oxidising capability of the oxidising substance. Hence, the mixture can never be classified into a more severe hazard category. If an oxidising substan ce is mixed with an inert material, the oxidising capability of the  mixture does not linearly decrease with decreasing content of oxidising substance. The relationship is more or less logarithmic and depends on the characteristics of the e. For instance, a mixture containing 50 % of a strong oxidiser and 50 oxidising substanc % of an inert material may retain 90 % of the oxidising capability of the original oxidising component. Non - testing classification of mixtures based solely on test data for the original oxidising substance should therefore be done with extreme care and only, if sufficient experience in testing exists. The determination of the oxidising properties of an aqueous solution of solid oxidising  substances and the classification as an oxidising mixture is not necessary provided that the total concentration of all solid oxidisers in the aqueous solution is less than or equal to 20 % (w/w). Classification criteria 2.13.4.2. sing liquid to The testing procedures for oxidising liquids are based on the capability of an oxidi enhance the combustion of a combustible material. Therefore, substances and mixtures that are submitted for classification testing are mixed with a combustible material. In principle, dried . The mixture of the potentially oxidising fibrous cellulose is used as a combustible material liquid and cellulose is then ignited and its behaviour is observed and compared to the behaviour of reference materials. For liquids the mixture with cellulose is ignited under confinement in an autoclave and the pressure rise rate that is caused by the ignition and the subsequent reaction is recorded. The pressure rise rate is compared to that of three reference material mixtures. The higher the pressure rise rate, the stronger the oxidising capability of the liqu id tested. Annex I: 2.13.2.1. An oxidising liquid shall be classified in one of the three categories for this class using test - section 34.4.2 of the UN RTDG, Manual of Tests and Criteria in O.2 in Part III, sub accordance with Table 2.13.1: Table 2.13.1 Criteria for oxidising liquids Category Criteria Any substance or mixture which, in the 1:1 mixture, by mass, of substance (or mixture) and cellulose tested, spontaneously ignites; or the mean pressure 1 rise time of a 1:1 mixture, by mass, of substance (or mixture) and cellulose is less than that of a 1:1 mixture, by mass, of 50 % perchloric acid and cellulose.

201 Guidance on the Application of the CLP Criteria 5.0 July 2017 201 Version – Any substance or mixture which, in the 1:1 mixture, by mass, of substance (or time less than or mixture) and cellulose tested, exhibits a mean pressure rise 2 equal to the mean pressure rise time of a 1:1 mixture, by mass, of 40 % aqueous sodium chlorate solution and cellulose; and the criteria for Category 1 are not met. Any substance or mixture which, in the 1:1 mixture, by mass, of subst ance (or mixture) and cellulose tested, exhibits a mean pressure rise time less than or 3 equal to the mean pressure rise time of a 1:1 mixture, by mass, of 65 % aqueous nitric acid and cellulose; and the criteria for Category 1 and 2 are not met. For addit ional information regarding the use of non - testing data see Section 2.13.4.3 below and Urben, 2007 (see Section 2.13.7 ). Testing and evaluation of hazard information 2.13.4.3. The test methods for oxidising liquids are designed to give a final decision regarding their classification. Apart from testing, a lso experience in the handling and use of substances or mixtures which shows them to be oxidising is an important additional factor in considering classification in this hazard class. In the event of divergence between test results and known experience, ju dgement based on known experience should take precedence over test results. However, a substance or mixture must not be classified into a less severe Category based on experience only. 2.13.4.4. Decision logic Classification of oxidising liquids is done according t o decision logic 2.13 as included in the GHS. NOTE: The person responsible for the classification of oxidising liquids should be experienced in this field and be familiar with the criteria for classification.

202 ication of the CLP Criteria Guidance on the Appl 5.0 – 202 July 2017 Version 2 . 12 Decision logic for oxidising liquids (Decision logic 2.13 of GHS) Figure The substance/mixtures is a liquid No Not Does it, in the 1:1 mixture, by mass, of substance (or mixture) classified and cellulose tested, exhibits a pressure rise ≥ 20 70 kPa (gauge)? Yes Does it, in the 1:1 mixture, by mass, of substance (or mixture) No and cellulose tested, exhibit a mean pressure rise time less than or Not equal to the mean pressure rise time of a 1:1 mixture, by mass, of classified ? 65 % aqueous nitric acid and cellulose Yes Category 3 Does it, in the 1:1 mixture, by mass, of substance (or mixture) No and cellulose tested, exhibit a mean pressure rise time less than or equal to the mean pressure rise time of a 1:1 mixture, by mass, of 40 % aqueous sodium chlorate and cellulose? Warning Yes gory 2 Cate Does it, in the 1:1 mixture, by mass, of substance (or mixture) No and cellulose tested, spontaneously ignite or exhibit a mean pressure rise time less than that of a 1:1 mixture, by mass, of 50 % perchloric acid and cellulose? Warning Yes Category 1 Danger

203 Guidance on the Application of the CLP Criteria 5.0 – July 2017 203 Version 2.13.4.5. Hazard communication for oxidising liquids 2.13.4.5.1. Pictograms, signal words, hazard statements and precautionary statements The pictograms and hazard statements are designed to indicate that oxidising substances and mixtures may cause or contribute to fire or explosion and therefore in principle should be separated from combustible materials. : Annex I Table 2.13.2 Label elements for oxidising liquids Category 2 Category 3 Category 1 GHS Pictograms Danger Danger Warning Signal Word Hazard Statement H272: May intensify H271: May cause fire or H272: May intensify explosion; strong oxidiser fire; oxidiser fire; oxidiser Precautionary P210 P210 P210 P220 P220 Statement P220 P280 Prevention P280 P280 P283 P306 + P360 Precautionary P370 + P378 P370 + P378 Statement P371 + P380 + P375 P370 + P378 Response Precautionary Statement P420 Storage Precautionary Statement P501 P501 P501 Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. 2.13.5. Relation to transport classification Division 5.1 within Class 5 of the UN RTDG Model Regulations covers oxidising liquids and e same tests and criteria as the CLP. Therefore, a liquid substance or oxidising solids, using th mixture classified as Division 5.1 (sometimes referred to as Class 5.1) according to any of the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI) is normally also class ified as an oxidising liquid according to the CLP. Packing Groups I, II and III of the transport regulations correspond directly to Categories 1, 2 and 3 of the CLP, respectively. See Annex VII for additional information on transport classification in rela tion to CLP classification.

204 ication of the CLP Criteria Guidance on the Appl 204 Version 5.0 – July 2017 2.13.6. Examples of classification for oxidising liquids 2.13.6.1. Examples of substances and mixtures fulfilling the classification criteria The list of substances and mixtures fulfilling the criteria for classification is only presented for information purposes. This list is not exhaustive. For examples of results see Section 34.4.2.5 of - MTC. UN  Ferric nitrate, saturated aqueous solution Lithium perchlorate, saturated aqueous solution   Magnesium perchlorate, saturated aqueous solution  Perchlor ic acid, 55 %  Sodium nitrate, 45 % aqueous solution 2.13.6.2. Examples of substances and mixtures not fulfilling the classification criteria Nickel nitrate, saturated aqueous solution  Potassium nitrate, 30 % aqueous solution   Silver nitrate, saturated aqueous soluti on 2.13.7. Reference Urben, Peter G. (2007). Bretherick's Handbook of Reactive Chemical Hazards, Volumes 1 - 2 (7th Edition). Elsevier.

205 Guidance on the Application of the CLP Criteria 5.0 – July 2017 205 Version 2.14. OXIDISING SOLIDS Introduction 2.14.1. o The criteria for ‘Oxidising solids’ are found in Annex I, Section 2.14 of CLP and are identical t those in Chapter 2.14 of GHS. The hazard class oxidising solids comprises substances and mixtures whose hazard is characterised by the fact that, in contact with other materials, they are able to cause or contribute to the combustion of those materials. The other materials do not necessarily have to belong to a certain hazard class in order to be affected by the presence of an oxidising solid. This is for example the case when a liquid fuel (e.g. gas oil) mixes with an oxidising solid. Certain combinatio ns of combustible materials and oxidising substances or mixtures may even result in spontaneous combustion, thermal instability or form an explosive mixture, this means that they may have explosive properties or may be regarded as self - reactive substances or mixtures. Although widely known as ‘oxidising materials’, their hazard and behaviour might be better understood by considering them to be ‘fire enhancing substances’. The hazards communication of oxidising solids intends to communicate the property tha t it may cause fire or explosion or that it may intensify fire. Apart from the combustion hazard, the production of toxic and/or irritating fumes may cause an additional hazard. For example, when nitrates are involved in a fire, nitrous fumes may be formed . The testing procedure and criteria for oxidising substances or mixtures do not work properly for ammonium nitrate, ammonium nitrate compounds, ammonium nitrate based fertilizers and bstances and mixtures ammonium nitrate gels. Therefore, for classification and labelling of su containing ammonium nitrate, known experience should be used and expert judgement should be sought. For the classification procedures for ammonium nitrate gels – intermediate for 2.1 of this guidance. blasting explosives, see Section 2.14.4.3 Annex I: In the event of divergence between test results and known experience in the handling and use of substances or mixtures which shows them to be oxidising, judgments based on known experience shall take precedence over test results. 2.14.2. Definitions and general considerations for the classification of oxidising solids The CLP text comprises the following definition for oxidising solids . Annex I: 2.14.1. Definition Oxidising solid means a solid substance or mixture which, while in itself is not necessarily combustible, may, generally by yielding oxygen, cause, or contribute to, the combustion of other material. Special consideration on particle size The oxidising properties of a solid depend on its particle size. Smaller particles enable a more intimate contact between the solid oxidiser and a combustible solid. The smaller the particle size, the higher the oxidising capability of th e solid. As a consequence, it may happen that large particles of a certain solid are considered to be non - hazardous, while small particles of the same solid need to be classified into the hazard class of oxidising solids.

206 ication of the CLP Criteria Guidance on the Appl 206 Version 5.0 – July 2017 Hence it is very important that ox idising properties for solids are investigated on the substance how it can reasonably be expected to be used or mixture as it is actually presented (including , see Article 8 (6) of CLP). This is indicated by the Note 2 cited in CLP Annex I, 2.14.2.1. I: Annex 2.14.2.1. [...] Note 2: The test shall be performed on the substance or mixture in its physical form as presented. If for example, for the purposes of supply or transport, the same chemical is to was tested and which is considered be presented in a physical form different from that which likely to materially alter its performance in a classification test, the substance shall also be tested in the new form. 2.14.3. Relation to other physical hazards Oxidising solids that are mixed with combustible materials or r educing agents may have explosive properties and should be considered for classification in the hazard class Explosives 2.1 of this guidance. (including the applicable screening procedures), see Chapter In rare cases, mixtures with oxidising solids may exhibit self - reactive behaviour, see Chapter 2.8 of this guidance. Expert judgement should be sought in case of doubt. The classification procedure and criteria for oxidising substances and mixtures is not applicable for organic peroxides. Under DSD organic peroxides were considered to be oxidising substances rganic peroxides do not because of the presence of the – O – O – bond. The majority of the o possess oxidising properties; their main hazards are reactivity and flammability. Under CLP organic peroxides comprise a separate hazard class (CLP Annex I, 2.15) and they must not be for oxidising solids. Organic peroxides were considered according to the procedures described classified as oxidising (O; R7) according to the DSD, which was not appropriate since the vast majority of them do not exhibit oxidising properties. o not need to be subject to the Inorganic oxidising solids are not flammable and therefore d classification procedures for the hazard classes flammable solids or pyrophoric solids. Also other solids that are classified as oxidising solids are normally not flammable, although a few nt should be sought in case of doubt. exeptions may exist. Expert judgeme Classification of substances and mixtures as oxidising solids 2.14.4. 2.14.4.1. Identification of hazard information Oxidising solids may cause, or contribute to, the combustion of other material. Although the in Annex I: 2 definition .14.1, quoted above, states that they generally do this by yielding oxygen, halogens can behave in a similar way. Therefore, any substance or mixture containing oxygen and/or halogen atoms should in principle be considered for inclusion into the hazard cat egories oxidising solids. This does not necessarily mean that every substance or mixture containing oxygen and/or halogen atoms should be subjected to the full testing procedure. 2.14.4.1.1. Screening procedures and waiving of testing Solids that are classified as explosives should not be subjected to the testing procedures for oxidising solids. Organic peroxides should be considered for classification within the hazard class organic peroxides, see Chapter 2.15 of this guidance. Experience in the handling and use of substances or mixtures which shows them to be oxidising event of is an important additional factor in considering classification as oxidising solids. In the

207 Guidance on the Application of the CLP Criteria 5.0 – July 2017 207 Version divergence between test results and known experience, judgement based on known experience should take precedence over test results. Before submitting a substance or a mixture to the full test procedure, an evaluation of its be very useful as it may prevent unnecessary testing. The person chemical structure may applying this procedure should have sufficient experience in testing and in theoretical evaluation of hazardous substances and mixtures. The following text provides a guideline for the etical evaluation of potential oxidising properties on the basis of its composition and theor chemical structure. In case of doubt, the full test must be performed. For organic substances or mixtures the classification procedure for this hazard class need not e applied if: b a. the substance or mixture does not contain oxygen, fluorine or chlorine; or b. the substance or mixture contains oxygen, fluorine or chlorine and these elements are chemically bonded only to carbon or hydrogen. For inorganic substances or mixture s, the classification procedure for this hazard class need not be applied if they do not contain oxygen or halogen. the On only be made between ‘potentially basis of this theoretical evaluation a distinction can oxidising’ (i.e. further testing required) a nd ‘non - oxidising’ (i.e. no further testing for this hazard class required). It is not possible to assign a hazard category on the basis of a theoretical evaluation. Any substance or mixture that complies with the above waiving criteria can be safely regar ded to have no oxidising properties and, hence, need not be tested and need not be regarded as an oxidising solid. However, such a substance or mixture may still possess other hazardous properties that require classification into another hazard class. In case a mixture of an oxidising substance and a non - hazardous inert substance is offered for classification, the following should be taken into account: An inert material by definition does not contribute to the oxidising capability of the  ce. Hence, the mixture can never be classified into a more severe oxidising substan hazard category.  If an oxidising substance is mixed with an inert material, the oxidising capability of the mixture does not linearly decrease with decreasing content of oxidising substance. The relationship is more or less logarithmic and depends on the characteristics of the oxidising substance. For instance, a mixture containing 50 % of a strong oxidiser and 50 % of an inert material may retain 90 % of the oxidising capability of the origi nal - testing classification of mixtures based solely on test data for oxidising component. Non the original oxidising substance should therefore be done with extreme care and only if sufficient experience in testing exists. 2.14.4.2. Classification criteria The testi ng procedures for oxidising solids are based on the capability of an oxidising solid to enhance the combustion of a combustible material. Therefore, solids that are submitted to ied fibrous cellulose classification testing are mixed with a combustible material. In principle, dr is used as a combustible material. The mixture of the potentially oxidising solid and cellulose is then ignited and its behaviour is observed and compared to the behaviour of reference material mixtures. For solids the mixture with ce llulose is ignited at atmospheric conditions and the time necessary for the combustion reaction to consume the mixture is recorded. The faster the combustion rate, the stronger the oxidising capability of the solid tested.

208 Guidance on the Appl ication of the CLP Criteria 208 5.0 – July 2017 Version An oxidising s olid shall be classified in one of the three categories for this Annex I: 2.14.2.1. - section 34.4.1 or test O.3 in Part III, sub - class using test O.1 in Part III, sub section 34.4 3 of the UN RTDG, Manual of Tests and Criteria, in accordance with Table 2.14.1: Table 2.14.1 Cr iteria for oxidising solids Criteria using test O.1 Criteria using test O.3 Category 1 Any substance or mixture which, in the Any substance or mixture which, in the to - cellulose ratio (by - cellulose ratio (by - to 4:1 or 1:1 sample 4:1 or 1:1 sample - mass) tested, exhibits a mean burning mass) tested, exhibits a mean burning time less than the mean burning time of rate greater than the mean burning rate mixture, (by mass), of potassium a 3:2 of a 3:1 mixture (by mass) of calcium pero bromate and cellulose. xide and cellulose. Any substance or mixture which, in the 2 Any substance or mixture which, in the 4:1 or 1:1 sample - to - cellulose ratio (by - to cellulose ratio (by 4:1 or 1:1 sample - mass) tested, exhibits a mean burning mass) tested, exhibits a mean burning time equal to or less than the mean rate equal to or greater than the mean burning rate of a 1:1 mixture (by mass) burning time of a 2:3 mixture (by mass) of calcium peroxide and cellulo se and of potassium bromate and the criteria the criteria for Category 1 are not met. for Category 1 are not met. Any substance or mixture which, in the 3 Any substance or mixture which, in the - to - cellulose ratio (by cellulose ratio (by - to - 4:1 or 1:1 sample 4:1 or 1:1 sample mass) tested, exhibits a mean burning mass) tested, exhibits a mean burning time equal to or less than the mean rate equal to or greater than the mean mixture (by mass) burning time of a 3:7 mixture (by mass) burning rate of a 1:2 of calcium peroxide and cellulose and m bromate and cellulose and of potassiu the criteria for Categories 1 and 2 are the criteria for Categories 1 and 2 are not met. not met. Note 1 Some oxidising solids also present explosion hazards under certain conditions (when stored in e may give rise to an explosion hazard large quantities). Some types of ammonium nitrat under extreme conditions and the 'Resistance to detonation test' (IMSBC Code (International Maritime Solid Bulk Cargoes Code, IMO), Appendix 2, Section 5) can be used to assess this hazard. Appropriate information sha ll be made in the SDS. Note 1 may also apply to other oxidising ammonium salts. Experience indicates that the conditions required for ammonium nitrate to present an explosion hazard involve a combination of factors, such as storage in large volumes (multiple tonnes) and either contamination (e.g. with metals, acids, organics) or excessive heat (e.g. under conditions of fire). The resistance to detonation (RTD) test is extensively described in Regulation (EC) No 2003/2003 for ammonium nitrate. For add itional information regarding the use of non - testing data see Section 2.14.4.3 below and Urben, 2007 (see Section 2.14.7 ).

209 Guidance on the Application of the CLP Criteria 5.0 July 2017 209 Version – Testing and evaluation of hazard information 2.14.4.3. 51 for oxidising solids are designed to give a final decision regarding their The test methods classification. ed that experience in the handling and use of substances or It should be recall mixtures, besides testing, is an important additional factor in considering classification in this hazard class. 2.14.4.4. Decision logic gic 2.14 as included in the GHS. Classification of oxidising solids is done according to decision lo NOTE: The person responsible for the classification of oxidising solids should be experienced in this field and be familiar with the criteria for classification. 51 As from December 2012 an alternative test method for oxidising solids, Test O.3, has been included in the UN MTC (see document ST/SG/AC.10/40/Add.2). Test O.3 is an improved version of Test O.1 using a different reference substance and gravimetric measurements of the burning rate. Reference to Test O.3 has been included in the 5th revised edition of the GHS.

210 ication of the CLP Criteria Guidance on the Appl 5.0 – 210 July 2017 Version 2 . 13 Decision logic for oxidising solids (Decision logic 2.14 of GHS) Figure The substance/mixture is a solid No Not to cellulose, by mass, test ignite - Does it, in the 4:1 or 1:1 sample - classified or burn? Yes - cellulose, by mass, test exhibit Does it, in the 4:1 or 1:1 sample - to a mean burning time ≤ the mean burning time of a 3:7 mixture, by mass, by potassium bromate and cellulose? No Not Or classified a mean burning rate greater than or equal to the mean burning rate of a 1:2 mixture, by mass, of a calcium peroxide and cellulose? Yes Category 3 - n the 4:1 or 1:1 sample cellulose, by mass, test exhibit - to Does it, i a mean burning time ≤ the mean burning time of a 2:3 mixture, by mass, by potassium bromate and cellulose? No Or a mean burning rate greater than or equal to the mean burning Warning y mass, of a calcium peroxide and rate of a 1:1 mixture, b cellulose? Yes Category 2 to cellulose, by mass, test exhibit - Does it, in the 4:1 or 1:1 sample - a mean burning time < the mean burning time of a 3:2 mixture, by No mass, by potassium bromate and cellulose? Or Dangerd a mean burning rate greater than the mean burning rate of a 3:1 mixture, by mass, of a calcium peroxide and cellulose? Category 1 Yes Danger

211 Guidance on the Application of the CLP Criteria 5.0 – July 2017 211 Version 2.14.4.5. Hazard communication for oxidising solids 2.14.4.5.1. Pictograms, signal words, hazard statements and precautionary statements The pictograms and hazard statements are designed to indicate that oxidising substances and mixtures may cause or contribute to fire or explosion and therefore in principle should be separated from combustible materials. Annex I : Table 2.14.2 Label elements for oxidising solids Category 1 Category 2 Category 3 GHS Pictograms Danger Danger Warning Signal Word Hazard Statement H272: May intensify H271: May cause fire or H272: May intensify explosion; strong oxidiser fire; oxidiser fire; oxidiser Precautionary P210 P210 P210 P220 P220 Statement Prevention P220 P280 P280 P280 P283 Precautionary P306 + P360 P370 + P378 P370 + P378 P371 + P380 + P375 Statement Response P370 + P378 Precautionary P420 Statement Storage Precautionary P501 P501 P501 Statement Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. 2.14.5. Relation to transport classification Division 5.1 within Class 5 of the UN RTDG Model Regulations covers oxidising liquids and oxidising solids, using the same tests and criteria as the CLP. Therefore, a solid substance or mixture classified as Division 5.1 (sometimes referred to as Class 5.1) according to any of the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI) is normally also classified as an oxidising solid according to CLP. Packing Groups I, II and III of the transport regulations correspond directly to Categories 1, 2 and 3 of CLP, respectively. See Annex VII for additional information on transport classification in relation to CLP classification.

212 ication of the CLP Criteria Guidance on the Appl 212 Version 5.0 – July 2017 2.14.6. Examples of classification for oxidising solids Examples of substances and mixtures fulfilling the 2.14.6.1. classification criteria The list of substances and mixtures fulfilling the criteria for classification is only presented for For examples of results see section 34.4.1.5 of information purposes. This list is not exhaustive. - UN MTC. Calcium nitrate, anhydr ous  Chromium trioxide  Potassium nitrite   Potassium perchlorate  Potassium permanganate  Sodium chlorate  Sodium nitrite Sodium nitrate  Strontium nitrate, anhydrous  2.14.6.2. Examples of substances and mixtures not fulfilling the classification criteria  Calcium nitrate, tetrahydrate  Cobalt nitrate, hexahydrate 2.14.7. Reference Urben, Peter G. (2007). Bretherick's Handbook of Reactive Chemical Hazards, Volumes 1 - 2 (7th Edition). Elsevier.

213 Guidance on the Application of the CLP Criteria – Version 213 5.0 July 2017 2.15. ORGANIC PEROXIDES 2.15.1. Introduction The criteria for ‘Organic peroxides’ are found in Annex I, Section 2.15 of CLP and are identical to those in Chapter 2.15 of GHS. The hazard class organic peroxides is unique in the respect that it is the only category to which not be chemicals are assigned on the basis of their chemical structure. Organic peroxides can seen as an ‘intrinsic property’; it is a family of chemical substances and mixtures which may have various properties. However, the type of peroxide is determined by testing. 2.15.2. Definitions and general considerations for the classification of organic eroxides p In CLP, the following definition is given for organic peroxides. 2.15.1. Definition Annex I: O - O - Organic peroxides means liquid or solid organic substances which contain the bivalent - structure and may be considered derivatives of hydrogen peroxide, where one or both of the hydrogen atoms have been replaced by organic radicals. The term organic peroxide includes organic peroxide mixtures (formulations) containing at least one organic peroxide. Organic ces or mixtures, which can undergo exothermic self - peroxides are thermally unstable substan accelerating decomposition. In addition, they can have one or more of the following properties: (i) be liable to explosive decomposition; (ii) burn rapidly; (iii) be sensitive to impact or friction; (iv) r eact dangerously with other substances. 2.15.1.2. An organic peroxide is regarded as possessing explosive properties when in laboratory testing the mixture (formulation) is liable to detonate, to deflagrate rapidly or to nder confinement. show a violent effect when heated u Relation to other physical hazards 2.15.3. In addition to the definition (CLP Annex I, 2.15.1), organic peroxides may: a. be flammable; b. emit flammable gas when heated. properties. In general, organic peroxides do not have or have only weak oxidising The additional (subsidiary) labelling, as indicated in the list of classified organic peroxides included in the UN RTDG Model Regulations, Section 2.5.3.2.4, represents the additional hazardous properties. Neither the burning properties nor th e sensitivity to impact and friction form part of the classification procedure for organic peroxides in CLP. However, these properties may be of importance for the safe handling of organic peroxides (see Section 2.15.4.3.2 , additional testing). In addition, the following should be noted:

214 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 214 Explosive properties The explosive properties do not have to be determined according to the CLP Annex I, Chapter 2.1, because explosive properties are incorporated in the decision logic for organic peroxides. Note that organic peroxides may have explosive properties when handl ed under higher confinement. Flammable properties The hazard statement for flammable properties for liquid organic peroxides should be based on the appropriate category for flammable liquids, as long as the flash point is relevant, (see Section ). The translation table in Annex VII to CLP can be used for this. 2.15.4.3.2 2.15.4. Classification of substances and mixtures as organic peroxides 2.15.4.1. Identification of hazard information The classification of an organic peroxide in one of the seven categories ‘Types A to G’ is dependent on its detonation, deflagration and thermal explosion properties, its response to heating under confinement, its explosive power and the concentration and the type of diluent added to desensitize the organic peroxide. Specifications of acceptable diluents that can be used safely are given in the UN RTDG Model Regulations, 2.5.3.5. The classification of an organic peroxide as Type A, B or C is dependent on the type of packaging in which the organic peroxide is tested as it affects the degree of confinement to which the organic peroxide is subjected. This lt in has to be considered when handling the organic peroxide; stronger packaging may resu more violent reactions when the organic peroxide decomposes. This is why it is important that storage and transport is done in packaging, allowed for the type of organic peroxide, that - (P520/IBC520) or tank conforms the requirements of the UN packaging or IBC instruction instruction (T23). The traditional aspects of explosive properties, such as detonation, deflagration and thermal explosion, are incorporated in the decision logic of CLP Figure 2.15.1. Consequently, explosive prescribed for the hazard class ‘explosives’ needs not to be property determination as conducted for organic peroxides. A list of currently classified organic peroxides is included in the UN RTDG Model regulations, Section 2.5.3.2.4. 2.15.4.2. Classification criteria In CLP, organic peroxid es are not classified as oxidisers but they are a distinct hazard class. Annex I: 2.15.2.1. Any organic peroxide shall be considered for classification in this class, unless it contains: en containing not a) not more than 1,0 % available oxygen from the organic peroxides wh more than 1,0 % hydrogen peroxide; or b) not more than 0,5% available oxygen from the organic peroxides when containing more than 1,0 % but not more than 7,0 % hydrogen peroxide. [...] In CLP decision logic Annex I, Figure 2.15.1, classification of organic peroxides is based on performance based testing both small scale tests and, where necessary, some larger scale test with the organic peroxide in its packaging. The concept of ‘intrinsic properties’ is, therefore, not applicable to this hazard class. Organic peroxides are classified into one of the seven categories of ‘Types A to G’ according to the classification criteria of CLP. The classification principles are given in decision logic Figure

215 Guidance on the Application of the CLP Criteria 5.0 – July 2017 215 Version - MTC, should be 2.15.1 of CLP and the Test Series A to H, as described in the Part II of the UN performed. Organic peroxides shall be classified in one of the seven categories of 2.15.2.2. Annex I: A to ‘Types G’ for this class, according to the following principles: hich, as packaged, can detonate or deflagrate rapidly shall be (a) any organic peroxide w defined as organic peroxide TYPE A; (b) any organic peroxide possessing explosive properties and which, as packaged, neither l explosion in that detonates nor deflagrates rapidly, but is liable to undergo a therma B; package shall be defined as organic peroxide TYPE (c) any organic peroxide possessing explosive properties when the substance or mixture as packaged cannot detonate or deflagrate rapidly or undergo a thermal explosion shall be d as organic peroxide TYPE C; define (d) any organic peroxide which in laboratory testing: (i) detonates partially, does not deflagrate rapidly and shows no violent effect when heated under confinement; or (ii) does not detonate at all, deflagrates slowly and shows no violent effect when heated under confinement; or (iii) does not detonate or deflagrate at all and shows a medium effect when heated under confinement; shall be defined as organic peroxide TYPE D; (e) any organic peroxide which, in laboratory testing, neither detonates nor deflagrates at all and shows low or no effect when heated under confinement shall be defined as organic peroxide TYPE E; (f) any organic peroxide which, in laboratory testing, neither detonates in the cavitated state nor defl agrates at all and shows only a low or no effect when heated under confinement as F; well as low or no explosive power shall be defined as organic peroxide TYPE (g) any organic peroxide which, in laboratory testing, neither detonates in the cavitated state nor deflagrates at all and shows no effect when heated under confinement nor any o explosive power, provided that it is thermally stable, i.e. the SADT is 60 C or higher for a (1) kg package 50 , and, for liquid mixtures, a diluent having a boiling point of not less than o 150 C is used for desensitisation, shall be defined as organic peroxide TYPE G. If the o organic peroxide is not thermally stable or a diluent having a boiling point less than 150 C efined as organic peroxide is used for desensitisation, the organic peroxide shall be d TYPE F. Where the test is conducted in the package form and the packaging is changed, a further test shall be conducted where it is considered that the change in packaging will affect the outcome of the test. (1) See UN RTDG, Man ual of Test and Criteria, sub - sections 28.1, 28.2, 28.3 and Table 28.3. A list of currently classified organic peroxides is included in the UN RTDG Model Regulations, Section 2.5.3.2.4.

216 ication of the CLP Criteria Guidance on the Appl 216 Version 5.0 – July 2017 2.15.4.3. Testing and evaluation of hazard information Thermal stability tests and temperature control 2.15.4.3.1. In addition to the classification tests given in decision logic Figure 2.15.1 of CLP, the thermal stability of the organic peroxide has to be assessed in order to determine the SADT. For the - MTC, Part II, Section 28, may be used. determination of the SADT, the testing me thod in UN - accelerating decomposition of an The SADT is defined as the lowest temperature at which self organic peroxide may occur in the packaging as used in transport, handling and storage. The SADT is a measure of the combined effect of the ambient temperature, decomposition kinetics, organic peroxide and its packaging. package size and the heat transfer properties of the There is no relation between the SADT of an organic peroxide and its classification in one of the seven categories ‘Types A to G’. The SADT is used to derive safe handling, storage and rgency transport temperatures (control temperature) and alarm temperature (eme temperature). Depending on its SADT an organic peroxide needs temperature control and the rules as given in CLP Annex I, 2.15.2.3, consist of the following two elements: 4. Criteria for temperature control: The following organic peroxides need to be subjected to temperature control: a. Organic peroxide types B and C with a SADT ≤ 50 ° C; b. Organic peroxide type D showing a medium effect when heated under confinement with a SADT ≤ 50 ° C or showing a low or no effect when heated under confinement with a SAD T ≤ 45 ° C; and c. Organic peroxide types E and F with a SADT ≤ 45 ° C. 5. Derivation of control and emergency temperatures: Type of receptacle Control temperature Emergency SADT * temperature 20 °C below SADT 20 °C or less Single packagings and 10 °C below SADT IBC’s 15 °C below SADT 10 °C below SADT over 20 °C to 35 °C 35 °C over 10 °C below SADT 5 °C below SADT < 50 °C Tanks 10 °C below SADT 5 °C below SADT * i.e. the SADT of the organic peroxide as packaged for transport, handling and storage It should be emphasized that the SADT is dependent on the nature of the organic peroxide itself, together with the volume and heat - loss characteristics of the packaging or vessel in which the organic peroxide is handled. The temperature at which self accelerating deco mposition - occurs falls:  as the size of the packaging or vessel increases; and  with increasing efficiency of the insulation on the package or vessel. The SADT is only valid for the organic peroxide as tested and when handled properly. Mixing the organic p eroxide with other chemicals, or contact with incompatible materials (including incompatible packaging or vessel material) may reduce the thermal stability due to catalytic decomposition, and lower the SADT. This may increase the risk of decomposition and has to be avoided.

217 Guidance on the Application of the CLP Criteria 5.0 – July 2017 217 Version 2.15.4.3.2. Additional considerations and testing Explosive properties The sensitivity of organic peroxides to impact (solids and liquids) and friction (solids only) may nounced be of importance for the safe handling of the organic peroxide if they have pro explosive properties (e.g. they are liable to detonate, to deflagrate rapidly or show a violent effect when heated under confinement). Test methods to determine these properties are described in Test Series 3 of the UN - MTC (see Test 3 (a) (ii) and 3 (b) (i)). This information on the mechanical sensitivity should be included in the SDS. Burning properties In some national storage guidelines the burning rate is commonly used for classification for the irements. Test methods are incorporated in purposes of storage and consequential storage requ these national storage regulations. Flash point The flash point for liquid organic peroxides is only relevant in the temperature range where the peroxide determination of organic peroxide is thermally stable. Above the SADT of the organic the flash point is not relevant because decomposition products are evolved. NOTE: In case a flash point determination seems reasonable (expected flash point below the SADT) a test method using small amount of sample is recommended. In case the organic peroxide is diluted or dissolved, the diluent may determine the flash point. Auto - ignition temperature The determination of the auto ignition temperature is not relevant for organic peroxides. Available test methods are fo r non - decomposing vapour phases but the vapours of organic peroxides decompose during execution of the test and auto ignition of these organic peroxide vapours can never be excluded. This information should be included in the SDS. - e Self ignition temperatur Also the determination of the self - ignition temperature (applicable for solids) is not relevant. The thermal stability of organic peroxides is quantitatively given by the SADT. Control and Emergency temperatures e based on the SADT as in most cases determined The Control and Emergency temperatures ar by UN Test H.4. The Dewar vessel used in the UN Test H.4 is supposed to be representative for the organic peroxide handled in packages. For handling the organic peroxide in larger quantities (IBCs/tanks/vesse ls etc.) and/or in (thermally) insulated containers, the SADT has to be determined for that quantity with that degree of insulation. From that SADT the Control and Emergency temperatures can be derived (see also Section 2.15.4.3.1 ). 2.15.4.3.3. Additional classification considerations Currently the following properties are not incorporated in the classification of organic peroxides under the CLP:  mechanical sensi tivity i.e. impact and friction sensitivity (for handling purposes);  burning properties (for storage purposes);  flash point for liquids; and  burning rate for solids. Furthermore:

218 ication of the CLP Criteria Guidance on the Appl Version 5.0 – July 2017 218 Annex I: 2.15.4.2. Mixtures of already classified organic peroxides may be classified as the same type of organic peroxide as that of the most dangerous component. However, as two stable components can form a thermally less stable mixture, the SADT of the mixture shall be d etermined. Note: The sum of the individual parts can be more hazardous than the individual components. Formulated commercial organic peroxides are classified according to their SADT. 2.15.4.4. Decision logic The decision logic for organic peroxides is applicable a ccording to CLP. NOTE: The person responsible for the classification of organic peroxides should be experienced in this field and be familiar with the criteria for classification.

219 Guidance on the Application of the CLP Criteria Version 5.0 – July 2017 219 Figure 2 . 14 Decision logic 2.15 for organic peroxides

220 Guidance on the Appl ication of the CLP Criteria 5.0 – July 2017 220 Version 2.15.5. Hazard communication for organic peroxides Pictograms, signal words, hazard statements and precautionary 2.15.5.1. statements organic peroxide meeting the According to CLP the following label elements must be used for criteria for this hazard class: Table 2.15.1 Annex I: Label elements for organic peroxides Type A Type B Type C & D Type E & F Type G Classification GHS pictograms Warning Danger Signal Word Danger Danger H240: H241: H242: Hazard H242: Heating may Heating may Statement Heating may Heating may cause a fire cause an cause a fire cause a fire or explosion explosion There are no label elements P210 P210 Precautionary P210 P210 allocated to statement this hazard P234 P234 P234 P234 category Prevention P235 P235 P235 P235 P240 P240 P240 P240 P280 P280 P280 P280 P370 + P380 P370 + P378 Precautionary P370 + P372 P370 + P378 + + statement P375[+ Response P380 + P373 1 P378] Precautionary P403 P403 P403 P403 statement P410 P410 P410 P410 Storage P411 P411 P411 P411 P420 P420 P420 P420 Precautionary P501 P501 P501 P501 statement Disposal

221 Guidance on the Application of the CLP Criteria 5.0 – July 2017 221 Version 1 See introduction to Annex I for details on the use of square brackets. The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. Additional labelling provisions for organic peroxides 2.15.5.2. Additional hazardous properties, resulting in additional (subsidiary) labelling, are indicated in the list of classified organic peroxides included in the UN RTDG Model Regulations, section 2.5.3.2.4 . 2.15.6. Relation to transport classification Division 5.2 within Class 5 of the UN RTDG Model Regulations covers organic peroxides. A list of currently classified organic peroxides is included in the UN RTDG Model Regulations, Section 2.5.3.2.4. This table incl - Type F (and some formulations Type udes organic peroxides Type B - G, so called exempted organic peroxides). An exceptional case in this respect is a peroxyacetic acid formulation, as currently classified in the UN RTDG Model Regulations under UN 3149, wi th the following description: HYDROGEN PEROXIDE AND PEROXYACETIC ACID MIXTURE with acid(s), water and not more than 5 % peroxyacetic acid, STABILISED. In the classification procedure for organic peroxides, see 2.15.4.4 , this formulation will be assigned to organic peroxide Type G, decision logic in Section In view of the above, this and consequently no label elements are allocated. formulation can be classified, also in accordance with CLP, as an Oxidising liquid, Category 2. See Annex VII for additional information on transport classification in relation to CLP classification. 2.15.7. Examples of classification for organic peroxides 2.15.7.1. Exampl es of substances and mixtures fulfilling the classification criteria Substance to be classified: Example Peroxide Molecular formula: n.a. According to CLP Annex I, Section 2.15.2.1, the substance has an active oxygen content of 7.40 onsidered for classification in the hazard class organic peroxides. % and thus has to be c Test results and classification according to CLP decision logic 2.15.1 for organic peroxides and the UN - MTC, Part II, is as follows: CLASSIFICATION TEST RESULTS 1. Name of the organic peroxide: Example Peroxide 2. General data 2.1. Composition: Example Peroxide, technically pure (97 %) 2.2. Molecular formula: n.a. 2.3. Active oxygen content: 7.18 % 2.4. Physical form: liquid 2.5. Colour: colourless 3 Density (apparent): 900 kg/m 2.6.

222 ication of the CLP Criteria Guidance on the Appl 222 Version 5.0 – July 2017 CLASSIFICATION TEST RESULTS 3. Detonation (test series A) Does the peroxide propagate a detonation? Box 1 of the decision logic: UN Test A.1: BAM 50/60 steel tube test Method: 3.1. Sample conditions: peroxide assay 97 % 3.2. Observations: fragmented part of the tube: 18 cm 3.3. Result: No 3.4. 3.6. 1.3 Exit: 4. Deflagration (test series C) Box 5 of the decision logic: Can the peroxide propagate a deflagration? Method 1: 4.1. Time/pressure test (test C.1) Sample conditions: 4.1.1. ambient temperature 4.1.2. Observations: 4000 ms Result: Yes, slowly 4.1.3. 4.2. Method 2: Deflagration test (test C.2) 4.2.1. temperature: 25 °C Sample conditions: Observations: deflagration rate: 0.74 mm/s 4.2.2. Result: Yes, slowly 4.2.3. Final result: Yes, slowly 4.3. 5.2 Exit: 4.4. 5. Heating under confinement (test series E) What is the effect of heating it under confinement? Box 8 of the decision logic: Method 1: 5.1. Koenen test (test E.1) 5.1.1. Sample conditions: - 5.1.2. limiting diameter: 2.0 mm Observations: fragmentation type ‘F’ 5.1.3. Result: Violent 5.2. Method 2: Dutch pressure vessel test (test E.2) 5.2.1. Sample conditions: - 5.2.2. Observations: limiting diameter: 6.0 mm (with 10 g)

223 Guidance on the Application of the CLP Criteria 5.0 – July 2017 223 Version CLASSIFICATION TEST RESULTS Result: Medium 5.2.3. Final result: Violent 5.3. Exit: 8.1 5.4. 6. Explosion test in package (test series G) Can it explode as packaged? Box 10 of the decision logic: Method: 6.1. Thermal explosion test in package (test G.1) 6.2. Sample conditions: 30 litre packaging, no fragmentation (N.F.) 6.3. Observations: Result: No 6.4. Exit: 10.2 6.5. 7. Thermal stability (outside of the decision logic) 7.1. Method: Heat accumulation storage test (test H.4) 7.2. Sample conditions: mass 380 g. Half life time of cooling of Dewar vessel with400 ml DMP: 10.0 hrs. (representing substance in package) Observations: self accelerating decomposition at 35 °C 7.3. no self accelerating decomposition at 30 °C Result: 7.4. SADT 35 °C 52 The decision logic is given in Figure x 8. General remarks: 9. Final classification Hazard class: Organic peroxide, Type C, liquid, temperature controlled Label: Flame (GHS02) Signal word: Danger Hazard statement: H242: Heating may cause a fire Temperature control: Needed based on SADT (35 °C, in package) Control temperature*: 20 °C (in package) Emergency temperature*: 25 °C (in package) *see UN - MTC, table 28.2. 52 Not attached to this example.

224 ication of the CLP Criteria Guidance on the Appl 224 Version 5.0 – July 2017 2.15.7.2. Additional remarks Explosive properties 2.15.7.1 a substance and a mixture may have explosive properties when As shown in Section ed in UN Test handled under greater confinement and where the packaging in which it was test G.1 (see point 6 of classification test results above) is changed. Such information should be given in the SDS. The example in Section 2.15.7.1 shows a violent effect when heated under confinement (see point 5.3 of the above results). Consequently, also the impact sensitivity according to UN Test to series 3, test 3 (a) (ii), BAM Fallhammer should be determined. For this example it amounts 20 J. Such information should be given in the SDS. Burning properties 2.15.7.1 the burning properties as determined by the test method For the example in Section described in the storage guidelines, currently in place in France, Germany, Netherlands and Sweden, is 7.0 kg/min/m². Based on this fig ure and the classification as organic peroxide type C, the storage classification can be assigned in those countries. Flash point The example substance thermally decomposes before the temperature at which the vapour can be ignited is reached (see Section 2.15.4.3.2 ) and consequently a flash point cannot be determined.

225 Guidance on the Application of the CLP Criteria 5.0 – July 2017 225 Version 2.16. CORROSIVE TO METALS 2.16.1. Introduction The criteria for ‘Corrosive to metals’ are found in Annex I, Section 2.16 of CLP and are identical to those in Chapter 2.16 of GHS. - The hazard class corrosive to metals is a physico chemical property that is new in the EU classification scheme and appears for the first time in CLP. So far, only the hea lth hazard corrosivity to skin was considered in the classification scheme. To some extent, both properties relate to each other and, in the context of transport of dangerous goods, have been considered for classification in class 8, despite the different nature of the hazard (material damage versus living tissue damage). A substance or a mixture that is corrosive to metal under normal conditions is a substance or a mixture liable to undergo an irreversible electrochemical reaction with metals that leads t o significant damage or, in some cases, even to full destruction of the metallic components. The corrosive to metal property is a quite complex property, since it is a substance (or mixture) related as well as a material (metal) related property. This mean s a corrosive substance or mixture leads to corroded material (metal), according to a number of external conditions. From the material side, many types of corrosion processes may occur, according to configurations, ion process, nature of metal, potential passivation liquid or fluid media inducing the corros occurring by oxide formation during corrosion. From the substance or mixture side, many parameters may influence the corrosion properties of a substance or mixture, such as the nature of the chemical or the pH. From an electochemistry point of view, corrosion conditions are often studied using Pourbaix diagrams, which plot the electrochemical potential (in Volt) that develops according to electrical charges transfer versus the pH - value. Such a diagram is shown for the case of iron and applies only for carbon steel corrosion (Jones, 1996). Figure . 15 Potential pH (also called Pourbaix) diagram for iron in water at 25 °C, indicating 2 stable form of the Fe element and implicitly, corrosion domains For the purposes of CLP, corrosion to metal will only be considered, by pure convention, for substances and mixtures that are liable to attack carbon steel or aluminium, two of the most common metals that may com e in contact with chemical substances (containment material, reactor material). The classification scheme applied here must not be considered as a material

226 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 226 (metal) classification method for metals regarding resistance to corrosion. By no means steel or minium specimens that are treated to resist to corrosion, must be selected for testing. alu 2.16.2. Definitions and general considerations for the classification of substances and mixtures corrosive to metals CLP comprises the following definition for substances and m ixtures that are corrosive to metal. Annex I: 2.16.1. Definition A substance or a mixture that is corrosive to metals means a substance or a mixture which by chemical action will materially damage, or even destroy, metals. 2.16.3. Relation to other physical hazards There is no direct relation to other physical hazards. 2.16.4. Classification of substances and mixtures as corrosive to metals 2.16.4.1. Identification of hazard information Importance of the physical state of the test substance or mixture There is no reference in the definition (CLP Annex I, 2.16.1) to the physical state of the substances or mixtures that needs consideration for potential classification in this hazard class. rd According to the test method to be employed for considering classification under this haza class, we may state at least that gases are out of the scope of the corrosive to metal hazard class. Neither the corrosivity of gases nor the formation of corrosive gases is currently covered applicable here. not by CLP classes and are therefore Accordi ng to the classification criteria only substances and mixtures for which the application of the UN Test C.1 (described in part III, Section 37.4.1.1 of the UN - MTC) is relevant and needs to - MTC , Section 37.4 excludes solids, be considered. Application of classification criteria in the UN while ‘liquids and solids that may become liquids (during transport)’, have to be considered for such a classification. The wording ‘solids that may become liquids’ was developed for UN RTDG Model Regulations classification purposes, and needs further explanation. Solids may become liquids by melting (due to increase in temperature). Solids having a melting point lower than 55 °C (which is the test temperature required in UN Test C.1) must then be taken into consideration. Th e other physical way to transform a solid into liquid is by dissolution in water or another solvent. Classification of solid substances that may become liquids by dissolution is subject to further expert judgement, and may need adaptation of the classifica tion criteria or test protocol (see 2.16.4.4.2 ). Interaction with liquids may come from air moisture or unintentional Section contact with water. Ot her solvent traces may result from the extraction process during manufacturing and these may induce corrosion in practice. Substances and mixtures in a liquid state must be tested without any modification before testing. For other cases (solids that may be come liquids), appropriate testing procedures require further work by the Committees of experts in charge of developing and updating the GHS at UN level. It needs to be further specified how such substances or mixtures must be quids) to be able to determine their corrosivity to metals. As an prepared (transformed into li example, it is thought that the quantity of solvent (water or any other solvent) to liquefy the test substance before testing would greatly influence results of the UN Test C.1 test and may not necessarily represent the real life situation of a product during transport, handling or use. Non - testing data Following parameters are helpful to evaluate corrosive properties before testing:

227 Guidance on the Application of the CLP Criteria 5.0 – July 2017 227 Version  melting points for solids; chemical nature of the substanc es and mixtures under evaluation (e.g. strong acids);  pH values (liquids).  See also IR & CSA, Chapter R.7a: Endpoint specific guidance, Section R.7.1.2 (Melting point/freezing point). Literature may also provide information on widely used substances and liquids ‘compatibility tables’, taking account of the corrosiveness of the products that may serve to decide whether testing must be conducted before assigning the corrosive to metals hazard class, on basis of expert judgement. The following substances and mixtures should be considered for classification in this class:  substances and mixtures having acidic or basic functional groups; substances or mixtures containing halogen;   substances able to form complexes with metals and mixtures containing such ces. substan Screening procedures and waiving of testing 2.16.4.2. Experience may have proven the corrosivity of given substances and mixtures. In such case no more testing is needed (see examples in Section 2.16.7 ). Generally extreme pH - values point to a higher likelihood that the substance or mixture is corrosive. However, it cannot lead to immediate classification in the hazard class corrosive to shows that immunity zones (where steel does not 2 . 15 Figure metals. As a pro of of that, s concerned. corrode) still exist on the full spectrum of pH values as far as carbon steel i Corrosivity is so complex that the evaluation of a mixture cannot be extrapolated from similar behaviour of constituents of a mixture. However, if one significant component of a mixture is corrosive to metals the mixture is likely to be corros ive to metals as well. Testing the actual mixture is therefore highly recommended. As already mentioned, solids are currently difficult to test according to the current CLP requirements, as the UN Test C.1 was designed for liquids. Where an initial test on either steel or aluminium indicates the substance or mixture being tested is corrosive, the follow up test on the other metal is not required . 2.16.4.3. Classification criteria Substances and mixtures of hazard class corrosive to metals are classified in a single h azard - MTC, Part III, Section 37, category on the basis of the outcome of the UN Test C.1 (UN paragraph 37.4). Annex I: Table 2.16.1 Criteria for substances and mixtures corrosive to metals Category Criteria 1 Corrosion rate on either steel or aluminium surfaces exceeding 6,25 mm per o year at a test temperature of 55 C when tested on both materials.

228 ication of the CLP Criteria Guidance on the Appl 228 Version 5.0 – July 2017 2.16.4.4. Testing and evaluation of hazard information General considerations 2.16.4.4.1. a of corrosion rate will never be applied in an absolute It is important to point out that the criteri way, but by extrapolating the measured rate of corrosion over the test period to the annual assumed correlating corrosion rate. This exercise has to take account of the fact that the s not necessarily constant over time. Expert judgement may be required to corrosion rate i consolidate the optimum test duration and to ascertain test results. However, the possibility of increasing the testing period from minimum one week to four weeks as well as the use of two UN Test C.1 act as barriers against erroneous classification. different metals in the Whatever the result of the classification may be, the classification as corrosive to metals relates to steel and/or aluminium only and does not provide information with regard to the corrosivity potential to other metals than those tested. Two types of corrosion phenomena need to be distinguished for classification of substances and ttack and mixtures in this hazard class, although not reported in CLP: the uniform corrosion a the localised corrosion (e.g. pitting corrosion, shallow pit corrosion). 2 . 2 (Section 37.4.1.4.1 of the UN - MTC) translates the corres ponding minimum mass loss Table rates leading to classify the test substance or mixture as corrosive to metals for standard metal specimens (2 mm of thickness), according to time of exposure, for reasons of uniform corrosion process. In case of use of metal plat es of a thickness that differs from the specified 2 mm (see comments in Section ), the values in Table 2 . 2 2.4.2 and Table 2 . 3 need adjustments due to the fact that the corrosion process depends on the surface of specimen.

229 Guidance on the Application of the CLP Criteria 5.0 – July 2017 229 Version 2 2 Table Minimum mass loss of specimens after different exposure times (corresponding to . the criterion of 6.25 mm/year) Exposure time Mass loss 7 days 13.5 % 14 days 26.5 % 21 days 39.2 % 51.5 % 28 days Table . 3 (Section 37.4.1.4.2 of the UN - MTC) indicates the criteria leading to classification of 2 the test substance or mixture as corrosive to metals for standard metal specimens, according to time of exposure, for reasons of localised corrosion process. Table 2 3 Minimum intrusion depths after exposure times (corresponding to the criterion of . localized corrosion of 6.25 mm/year) Min. intrusion depth Exposure time 120  m 7 days m 240  14 days 360  m 21 days m 480  28 days It is not mentioned explicitly in the text that localised corrosion as well as uniform corrosion has also be taken into account. However, localised corrosion, that is entirely part of UN Test C.1 protocol, has actually to be taken into account. In addition, although the type of corrosion is not reflected in the classification result, this valuable information should be given in the SDS .

230 ication of the CLP Criteria Guidance on the Appl 230 Version 5.0 – July 2017 2.16.4.4.2. testing Additional notes on best practice for Competence required for testing The overall evaluation of appropriate data for considering the corrosion properties of a substance or a mixture and in particular for testing it according to the mentioned criteria for this hazard class requires cer tain qualifications and experience. Expertise is often needed for this - hazard class, which relates to a complex and multi faceted hazardous phenomenon. Selection of metal specimens rate specifications CLP refers to two types of metals (carbon steel and aluminium) meeting accu (technical characteristics of metal sheets and plate thickness). Thicker metal sheets, such as cast materials, of which the thickness is reduced by any form of mechanical treatment, may l) thickness could induce corrosion enhanced never be used. Mechanical reduction of sheet (meta process due to cross section heterogeneity in metal grain and impurities. It is far better to use slightly different specifications of metal in the correct thickness or slightly different specimen plate thicknes ses. It is recognised that it will not always be easy to obtain metal specimens with the profile as described above. - MTC, Sub - section Regarding the type of aluminium or steel to be used for this test see UN 37.4.1.2. Minimum corrosive media volume to prevent any limitation on the corrosion process due to full consumption of the In order corrosive media before the end of the testing period, a minimum volume of substance or mixture (1.5 L, according to the UN - MTC) has to be used. (Note: volume/surface ratio o f 10 mL/cm² is stated in DIN 50905, similar in ASTM G31 – 72.) Adjustment of the test temperature Corrosion processes are temperature dependent. In the context of CLP, the property corrosive ed temperature of 55 °C 1 to metals is assessed through testing metal specimens at a specifi  °C. In practice, it may be difficult with standard testing equipment to stay within the  1 °C) of the gas phase, all over the test period. In such case, the temperature window (55 °C ightly higher temperature and somewhat lower test can be performed conservatively at a sl  3 °C). accuracy (e.g. 57 °C Selecting the appropriate test duration The evaluation of the criterion of 6.25 mm/year is generally based on a test duration not to stop the test proce dure already after 1 exceeding 1 month. There is, however, the option able 1). For the decision on test duration, the non - week (see T linear behaviour of the corrosion process must be taken due account of. In borderline cases a non - appropriate test duration may positive or false negative results. result in either false Specimen cleaning Attention must be paid to the correct cleaning of the corroded residue before measurement of the corrosion characteristics. In case of adhesive corroded layer, the same cleaning process needs to be car ried out on a non corroded sample to verify if the cleaning procedure is not significantly abrasive. For further information see UN - MTC, Sub - section 37.4.1.3. Testing soluble solids As said in Section 2.15.4.1 , for solids that may become liquids through dissolution in water or in a solvent, the adequate testing procedure is more complex (not explicitly describe in the UN C.1 test protocol). In no cas e will simple dilution of the solid substance or mixture in any quantity of water lead to satisfactory testing of the substance or mixture for corrosion to metals.

231 Guidance on the Application of the CLP Criteria 5.0 July 2017 Version – 231 For the specific case where the corrosion potential is linked to the presence of solvent tr aces (other than water), expert judgement is needed to determine if further testing must be performed (where the solid is put in interaction with the metallic part considered). Example of equipment relevant for the performance UN Test C.1 Figure 2 . 16 Example of testing equipment available on the market to perform UN Test C.1

232 ication of the CLP Criteria Guidance on the Appl 232 Version 5.0 – July 2017 2.16.4.5. Decision logic substances and mixtures corrosive to metals is done according to decision Classification of included in the GHS. logics 2.16.4.1 as NOTE: The person responsible for the classification of substances and mixtures corrosive to metals should be experienced in this field and be familiar with the criteria for classification. Decision logic for substances and mixtures corrosive to metals (Decision logic Figure 2 . 17 2.16 of GHS) Substance/mixture it corrode either steel or aluminium surfaces at a rate Does No exceeding 6.25 mm/year at a test temperature of 55 °C Not classified when tested on both materials? Category 1 Yes Warning

233 Guidance on the Application of the CLP Criteria 5.0 – July 2017 233 Version 2.16.5. to Hazard communication for substances and mixtures corrosive metals 2.16.5.1. Pictograms, signal words, hazard statements and precautionary statements Table 2.16.2 of CLP Annex I provides the label elements for hazard class corrosive to metals. ion under this The hazard statement H290, using the wording ‘may’, reflects that classificat hazard class does not cover all metals (testing only considers carbon steel and aluminium). Thus we may find examples of substances and mixtures that are classified in this hazard class n on other more corrosive resistant metals corrosive to metals but will not induce corrosive actio (e.g. platinum) than those serving as reference materials. Label elements must be used for substances and mixtures meeting the criteria for classification 2.16.2. in this hazard class in accordance with Table Ann 2.16.3. Table 2.16.2 ex I: Label elements for substances and mixtures corrosive to metals Classification Category 1 GHS Pictogram Signal Word Warning Hazard Statement H290: May be corrosive to metals Precautionary Statement, Prevention P234 Precautionary Statement, Response P390 Precautionary Statement, Storage P406 Precautionary Statement, Disposal Note: Where a substance or mixture is classified as corrosive to metals but not corrosive to skin and/or eyes, the labelling pro visions set out in S ection 1.3.6 shall be used. The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. Further, in S ection 1.3.6 of CLP Annex I a derogation from labelling requirements for substances or mixtures classified as corrosive to metals but not corrosive to skin and/or eyes is provided. Annex I: 1.3.6 Substances or mixtures classified as corrosive to metals but not classified as skin corrosion or as serious eye damage (Catgory 1) Substances or mixtures classified as corrosive to me tals but not classified as skin corrosion or as serious eye damage (Catgory 1) which are in the finished state as packaged for consumer use do not require on the label the hazard pictogram GHS05.

234 ication of the CLP Criteria Guidance on the Appl 234 Version 5.0 – July 2017 2.16.6. Relation to transport classification G Model Regulations covers substances and mixtures that are classified Class 8 of the UN RTD for corrosivity to skin, metals or both. Valuable information can be obtained from UN RTDG Model Regulations and the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI). Existing test results obtained in the context of the modal transport regulations (ADR, RID, ADN and IMDG Code, ICAO TI) may be applied since the UN Test C.1 serves as reference for ation on transport testing in both classification systems. See Annex VII for additional inform classification in relation to CLP classification. 2.16.7. Examples of classification for substances and mixtures corrosive to metals The following table lists some examples of substances and mixtures that should be classified or not in Class 2. 16 (according to known UN Test C.1 results) in comparison with predicted results skin corrosion hazard. for 2 4 Examples of classified and non classified substances and mixtures in Class 2.16 Table . ‘Corroded’ means corrosion attack in the sense of UN Test C.1; Note: ‘Not corroded’ means corrosion resistant in the sense of UN Test C.1; ‘Positive’ or ‘Negative’ are results from skin corrosion. Substance or mixture Aluminiu CLP Annex I, 2.16 Skin (for Steel classific ation comparison) m Classified Hydrofluoric acid Not corroded Corroded Positive > 70 % (UN1790) Classified Not corroded Corroded Highly concentrated nitric Positive acid (97 %) (UN2031) Positive red fuming (UN2032) Not corroded HNO Not Not classified 3 corroded Hydrochloric acid (diluted) Corroded Corroded Classified Negative (UN1789) NaOH solutions (UN1824) Not corroded Corroded Classified Positive

235 Guidance on the Application of the CLP Criteria 5.0 – July 2017 235 Version Example of metal specimen plates after exposure to a corrosive 2.16.7.1. mixture Figure . 18 Example of corroded metal plates after testing according to UN Test C.1 for a 2 classified mixture Plate located in the liquid phase Plate located in the interface Plate located in the vapour phase This example shows that the corrosion may develop at different rates according to the accurate position of the specimen related to the corroding mixture (sunk in the liquid, placed in the gas phase above liquid or at the liquid/gas interface). 2.16.8. References 72(2004) . Standard Practice for Laboratory Immersion Corrosion Testing of Metals - ASTM G31 , 2nd edition, 1996, Prentice Hall, Upper Principles and Prevention of Corrosion Jones, D.A., 52. Saddle River, NJ. ISBN 0 - 13 - 359993 - 0 Page 50 - Part 1: General guidance DIN 50905 - 1: 2007, Corr osion of metals - - Corrosion testing - Korrosionsuntersuchungen - (Korrosion der Metalle Teil 1: Grundsätze).

236 ication of the CLP Criteria Guidance on the Appl 236 Version 5.0 – July 2017 3. DS PART 3: HEALTH HAZAR ACUTE TOXICITY 3.1. Definitions and general considerations for acute toxicity 3.1.1. 3.1.1.1. Acute toxicity means those adverse effects occurring following oral or Annex I: dermal administration of a single dose of a substance or a mixture, or multiple doses given within 24 hours, or an inhalation exposure of 4 hours. Acute toxicity relates to effects occurring after a single or relatively brief exposure to a substance or mixture. The definition in CLP reflects the fact that the evidence for acute toxicity is usually obtained from animal testing. In particular, acute toxicity is usually characterised in terms of lethality and exposure times are based around those used in experimental protocols. However, classification for acute toxicity can also be based on human evidence which shows lethality following human exposure. covering effects after single or brief exposure – ‘Acute There are different hazard classes - – Single Exposure)’, skin toxicity’ and ‘STOT SE (Specific Target Organ Toxicity irritation/corrosion and eye damage. These are independent of each other and may all be ixture if the respective criteria are met. However, care should be assigned to a substance or a m taken not to assign each class for the same effect, essentially giving a multiple classification, riate even where the criteria for different classes are fulfilled. In such a case the most approp (the most severe hazard) class should be assigned. Acute toxicity classification is generally assigned on the basis of evident lethality (e.g. an LD /LC value), or, where the potential to cause lethality can be concluded from evident 50 50 toxicity (e.g SE should be considered where there is . from the fixed dose procedure). STOT - clear evidence of toxicity to a specific organ, when it is observed in the absence of a classification for lethality 3.8 of this Guidance ). Mortalities during the first 72 h (see Section after first treatment (in a repeated dose study) may also be considered for the assessment of acute toxicity. For more details see Guidanc e on IR&CSA , Section R.7.4.1.1. 3.1.1.2. The hazard class Acute Toxicity is differentiated into: Annex I: Acute oral toxicity; – – Acute dermal toxicity; Acute inhalation toxicity. – The classification must be considered for each route of exposure, using the appropriate approach as described in Section and Section 3.1.2.3 of this Guidance. If different 3.1.2.2 hazard categories are assigned, the most severe hazard category must be used to select the appropriate pictogram and signal word on the label for acute toxicity. For each relevant route of exposure, the hazard statement will correspond to the classification of this specific route. 3.1.2. Classification of substances for acute toxicity 3.1.2.1. Identification of hazard information 3.1.2.1.1. Iden tification of human data Relevant information with respect to acute toxicity may be available from sources such as case reports, epidemiological studies, medical surveillance and reporting schemes and national or acute toxicity should report severe effects poison centres. Human data to be considered f

237 Guidance on the Application of the CLP Criteria – Version 237 5.0 July 2017 after single exposure or exposure of less than 24h, but data on severe effects after a few exposures over a few days can also be considered on a case by case basis. CSA, For more details see Guidance on . IR & Section R.7.4.3.2 Identification of non - human data 3.1.2.1.2. - Non testing data: Physicochemical data Physico chemical properties, such as pH, physical state, form, solubility, vapour pressure and - particle size, can be important parameters in evaluating toxicity stu dies and in determining the most appropriate classification. This is especially valid with respect to inhalation where physical 3.1.2.3.2 of this form and particle size can have a significant impact on toxicity (see Section Guidance). (Q)SAR models, expert systems and grouping methods - testing data can be provided by the following approaches: a) structure - activity Non tive structure relationships (SARs) and quantita - activity relationships (QSARs), collectively called expert systems incorporating (Q)SARs and/or expert rules; and c) grouping (Q)SARs; b) across and categories. These approaches can be used to assess acute toxicity if methods (read - they prov ide relevant and reliable (adequate) data for the chemical of interest. [...] Compared with some endpoints, there are relatively few (Q)SAR models and expert systems capable of predicting acute toxicity.’ (Guidance on IR&CSA , Section R.7.4.3.1). : Testing data In vitro data in vitro tests that have been officially adopted by the EU or OECD for There are currently no assessment of acute toxicity (see Guidance on IR&CSA , Section R.7.4.3.1, for further ng expert judgement. information). Any available studies should be assessed by usi Animal data A number of different types of studies have been used to investigate acute toxicity. Older standard studies were designed to determine lethality and estimate the LD /LC . In contrast, 50 50 contemporary study protocols, such as the fixed dose procedure, use signs of evident toxicity rather than lethality as indications of acute toxicity. , Section R.7.4.3.1 . The animal studies are listed in the Guidance on IR&CSA Classification criteria 3.1.2.2. Annex I: 3.1.2.1. Substances can be allocated to one of four hazard categories based on acute toxicity by the oral, dermal or inhalation route according to the numeric criteria shown in Table (oral, dermal) o r LC 3.1.1. Acute toxicity values are expressed as (approximate) LD 50 50 (inhalation) values or as acute toxicity estimates (ATE). Explanatory notes are shown following Table 3.1.1. Table 3.1.1 Acute toxicity hazard categories and acute toxicity estimates (ATE) defining the respective categories Exposure Route Category 1 Category 2 Category 3 Category 4 Oral (mg/kg < ATE 300 < ATE ATE ≤ 5 5 < ATE ≤ 50 50 bodyweight) 2000 ≤ 300 ≤ See: Note (a)

238 ication of the CLP Criteria Guidance on the Appl 238 Version 5.0 – July 2017 Note (b) Dermal 1000 < ATE ATE ≤ 50 50 < ATE (mg/kg 200 < ATE 200 2000 ≤ bodyweight) ≤ 1000 ≤ Note (a) See: Note (b) 1 (ppmV ( )) 2500 < ATE 500 < ATE 100 < ATE 100 Gases ATE ≤ ≤ 500 20000 ≤ 2500 ≤ Note (a) see: Note (b) Note (c) 10.0 < ATE 2.0 < ATE ≤ Vapours (mg/l) 0.5 < ATE ≤ 0.5 ATE ≤ 2.0 20.0 ≤ 10.0 see: Note (a) Note (b) Note (c) Note (d) Dusts and mists 1.0 < ATE ≤ 5.0 0.5 < ATE ≤ 0.05 < ATE ATE ≤ 0.05 0.5 (mg/l) ≤ 1.0 Note (a) see: Note (b) Note (c) 1 ) Gas concentrations are expressed in parts per million per volume (ppmV). ( Notes to Table 3.1.1: (a) The acute toxicity estimate (ATE) for the classification of a substance is derived using the LD /LC where available. 50 50 (b) The acute toxicity estimate (ATE) for the classification of a substance in a mixture is derived using: - the LD where available, /LC 50 50 - the appropriate conversion value from Ta ble 3.1.2 that relates to the results of a range test, or - the appropriate conversion value from Table 3.1.2 that relates to a classification category. (c) The ranges of the acute toxicity estimates (ATE) for inhalation toxicity in the table are based on 4 - hour testing exposures. Conversion of existing inhalation toxicity data which have been generated using a 1 - hour exposure can be carried out by dividing by a factor of 2 for gases and vapours and 4 for dusts and mists. For some substances the test atmo sphere will not just be a vapour but will consist of a (d) mixture of liquid and vapour phases. For other substances the test atmosphere may consist of a vapour which is near the gaseous phase. In these latter cases, classification shall be based 1 (100 llows: Category ppmV), Category on ppmV as fo 2 (500 ppmV), Category 3 (2500 ppmV), Category 4 (20 000 ppmV).

239 Guidance on the Application of the CLP Criteria 5.0 – July 2017 239 Version The terms ‘dust’, ‘mist’ and ‘vapour’ are defined as follows: dust: solid particles of a substance or mixture suspended in a gas (usually air), - mist: liquid droplets of a substance or mixture suspended in a gas (usually air), - vapour: the gaseous form of a substance or mixture released from its liquid or solid state. - Dust is generally formed by mechanical processes. Mist is generally formed by condens ation of supersaturated vapours or by physical shearing of liquids. Dusts and mists generally have μm. sizes ranging from less than 1 to about 100 NOTE regarding CLP Annex I, Table 3.1.1, Note (c): toxicity relate to a 4 - The classification criteria for acute inhalation hour experimental values have been obtained in studies using exposure exposure period. Where LC 50 durations shorter or longer than 4 hours these values may be adjusted to a 4 hour - equivalent using Haber’s law (C·t=k) for direct compa rison with the criteria. The n formula may be refined to (C ·t=k) where the value of n, which is specific to individual substances, should be chosen using expert judgement. If an appropriate value of n is not available in the literature then it may sometimes be derived from the available inverse mortality data using probits (i.e. the s cumulative distribution function associated with the standard normal distribution ). Alternatively, some default values are recommended (Guidanc e on IR&CSA , Section R.7.4.4.1) . Particular care should be taken when using Haber’s law to assess inhalation data on substances which are corrosive or locally active. In all cases, Haber’s law should only be used in conjunction with expert judgement. noted that the statements in the Guidance on IR&CSA , Section R.7.4.4.1, It is with respect to Haber’s law are not consistent with those of CLP. However, the CLP approach must be used for classification and labelling. Harmonised ATE values 3.1.2.2.1. ATE values are gradually included in Annex VI. These values must From 2016 harmonised be applied when classifying mixtures containing the substance just as any other harmonised regardless of any other ATE value derived from testing of the substance item . 3.1.2.2.2. Minimum classification For certain entries in Annex VI there is an asterisk indicating that it is the minimum classification. In case the substance has a minimum classification this is the lowest classification possible, however, if there is data indicating that a more stringen t classification is warranted the classification has to be adapted accordingly. This is due to translation from the old DSD legislation. Evaluation of hazard information 3.1.2.3. 3.1.2.3.1. Evaluation of human data The evaluation of human data often becomes difficult due to various limitations frequently found with the types of studies and data highlighted in Section of this Guidance. These 3.1.2.1.1 include uncertainties relating to exposure assessment (i.e. unreliable information on the amount of substance the subjects were exposed to) and uncertain exposure to other substances. As such, human data needs careful expert evalua tion to properly judge the reliability of the findings. It should be acknowledged that human data often do not provide sufficiently robust evidence on their own to support classification. They may, however, contribute to a weight of evidence assessment wit h other available information such as data from animal studies.

240 ication of the CLP Criteria Guidance on the Appl 240 Version 5.0 – July 2017 The classification for acute toxicity is based primarily on the dose/concentration that causes mortality (the Acute Toxicity Estimate, ATE), which is then related to the numerical values in 3.1.2.2 e classification criteria according to CLP Annex I, Table 3.1.1 (see Section th of this ormula in CLP Annex I, 3.1.3.6.1 and Guidance) for substances or for use in the additivity f 3.1.3.3 3.1.3.6.2.3 for mixtures (see Section of this Guidance). The ATE is usually obtained from animal studie s but in principle suitable human data can also be used if available. Where human data are available, they should be used to estimate the ATE which can be used directly for classification as described above. or expected to cause mortality after a The minimum dose or concentration or range shown single human exposure can be used to derive the human ATE directly, without any adjustments 3.1.5.1.1 or uncertainty factors. See Example 1 (methanol) in Section of this Guidance. If there are no exact or quantitative lethal dose data the procedure described in CLP Annex I, 3.1.3.3.5 of this Guidance) would have to be followed using Table 3.1.3.6.2.1(b) (see Section 3.1.3.3 is Guidance) with an assessment of the available information on of th 3.1.2 (see Section - quantitative or qualitative basis. a semi Expert judgement is needed in a total weight of evidence approach taking relevance, reliability, ample 2 (N,N - and adequacy of the information into account. See Ex dimethylaniline) in Section 3.1.5.1.2 of this Guidance. Evaluation of non - human data 3.1.2.3.2. Annex I: Specific considerations for classification of substances as acutely toxic 3.1.2.2. 3.1.2.2.1. evaluation of acute toxicity by the oral and Annex I: The preferred test species for inhalation routes is the rat, while the rat or rabbit are preferred for evaluation of acute dermal toxicity. When experimental data for acute toxicity are available in several animal species, t shall be used in selecting the most appropriate LD value from among scientific judgemen 50 valid, well - performed tests. Evaluation of non - in vitro data: testing and - across may be used instead of testing, and substances Results of (Q)SAR, grouping and read will be classified and labelled on this basis if the method fulfils the criteria described in Annex XI IR&CSA , Section R.7.4.4.1 . data cannot be used as of REACH. See also the Guidance on In vitro a stand alone. However, NRU data can be used as part of a weight of evidence evaluation. Animal data: – establishing: ATE /LC  : An available LD Basis LD /LC is an ATE at first stage. 50 50 50 50  Results from a range test: According to CLP Annex I, Table 3.1.2 results from range tests (i.e. doses/exposure concentrations that cause acute toxicity in the range of numeric criteria values) can be assigned to the four different categories of acute toxicity ure (centre column). Further, CLP Annex I, Table 3.1.2 for each possible route of expos allows allocating a single value, the converted acute toxicity point estimate (cATpE), to each experimentally obtained acute toxicity range estimate or classification category (right column), see Note (b) to Table 3.1.1. This cATpE can be used in the additivity formulae (CLP Annex I, 3.1.3.6.1 and 3.1.3.6.2.3) to calculate the acute toxicity of mixtures. In case of multiple LD /LC values or LD /LC values from several species:  50 50 50 50 values or ATE derived from Where several experimentally determined ATE values (i.e. LD , LC 50 50 studies using signs of non - lethal toxicity) are available, expert judgement needs to be used to choose the most appropriate value for classification purposes. Each study needs to be assessed for its suitability in terms of study quality and reliability, and also for its relevance to the

241 Guidance on the Application of the CLP Criteria 5.0 – July 2017 241 Version substance in question in terms of technical specification and physical form. Studies not considered suitable on reliability or other grounds sh ould not be used for classification. In general, classification is based on the lowest ATE value available i.e. the lowest ATE in the most sensitive appropriate species tested. However, expert judgement may allow another ATE provided this can be supported by a robust justification. If there value to be used in preference, is information available to inform on species relevance, then the studies conducted in the species most relevant for humans should normally be given precedence over the studies in other sp ecies. If there is a wide range of ATE values from the same species, it may be informative to consider the studies collectively, to understand possible reasons for the different results obtained. This would include consideration of factors such as the sex and age of the animals, the animal strains used, the experimental protocols, the purity of the substance and form or phase in which it was tested (e.g. the particle size distribution of any dusts or mists nical factors in inhalation studies. This tested), as well as exposure mode and numerous tech assessment may aid selection of the most appropriate study on which to base the classification. values from tests using different vehicles (e.g. water vs. corn oil or If there are different LD 50 neat substance vs. corn oil), generally the lowest valid value would be the basis for classification. It is not considered appropriate to combine or average the available ATE values. The studies may not be equivalent (in terms of experimental design such as protocol, purity of material tested, species of animal used, etc.) making such a collation or combination unsound. - If there is a study available with a post observation period of less than the 14 days, the time to bserved at the end of the study, the be used according to the OECD guidelines, and effects are o resulting LD might be misleading. Such information should be included in the weight of 50 evidence consideration. If there is available test data from a 28 day study to 1000 mg/kg bw/day and no effects are seen, it can b e concluded that the substance does not fullfill the criteria for acute toxicity (for - 1 to Guidance R.7a, especially S further details see Appendx 7.4 ection 2.4). If a substance is cutely toxic by the not acutely toxic by the oral route it can also be assumed that it is not a dermal route. .1.2.3. Specific considerations for classification of substances as acutely toxic by Annex I: 3 the inhalation route Annex I: Units for inhalation toxicity are a function of the form of the inhaled 3.1.2.3.1. ues for dusts and mists are expressed in mg/l. Values for gases are expressed in material. Val ppmV. Acknowledging the difficulties in testing vapours, some of which consist of mixtures of liquid and vapour phases, the table provides values in units of mg/l. However, fo r those vapours which are near the gaseous phase, classification shall be based on ppmV. Conversions: Differentiation between vapour and mist will be made on the basis of the saturated vapour imated as follows: concentration (SVC) for a volatile substance, which can be est SVC [mg/l] = 0.0412 x MW x vapour pressure (vapour pressure in hPa at 20°C). The conversion from mg/l to ppm assuming an ambient pressure of 1 atm = 101.3 kPa and 25°C is: ppm= 24,450 x mg/l x 1/MW. An LC well below the SVC will be considered for classification according to the criteria for 50 close to or above the SVC will be considered for classification vapours; whereas an LC 50 according to the criteria for mists (see also OECD GD 39). Considerations with respect to physical forms or states or bioavailability: Article 9(5) When evaluating the available information for the purposes of classification, the manufacturers, importers and downstream users shall consider the forms or physical states in

242 ication of the CLP Criteria Guidance on the Appl 242 Version 5.0 – July 2017 which the su bstance or mixture is placed on the market and in which it can reasonably be expected to be used. of this Guidance. and For further details see Sections 1.2 1.3 Special considerations concerning aerosols (dusts and mists): 3.1.2.3.2. Of particular importance in classifying for inhalation toxicity is the use of Annex I: well articulated values in the highest hazard categories for dusts and mists. Inhaled particles between 1 and 4 microns mean mass aerodynamic diameter (MMAD) will deposit in all regions of the rat respiratory tract. This particle size range corresponds to a maximum dose of about 2 mg/l. In order to achieve applicability of animal experiments to human exposure, dusts and mists would ideally be tested in this range in rats. st guidelines for acute inhalation toxicity with aerosols require rodents to be exposed to The te an aerosol containing primarily respirable particles (with a Mass Median Aerodynamic Diameter (MMAD) of 1 4 μm), so that particles can reach all regions of the res piratory tract. The use of – such fine aerosols helps to avoid partial overloading of extra - thoracic airways in obligate nasal breathing species like rats. Results from studies in which substances with particle size with a nerally not be used for classification, but expert MMAD > 4 μm have been tested can ge judgement is needed in cases where there are indications of high toxicity. The use of highly respirable dusts and mists is ideal to fully investigate the potential inhalation hazard of the substance. Howev er, it is acknowledged that these exposures may not necessarily reflect realistic conditions. For instance, solid materials are often micronised to a highly respirable form for testing, but in practice exposures will be to a dust of much lower respirabilit y. Similarly, pastes or highly viscous materials with low vapour pressure need strong measures to be taken to generate airborne particulates of sufficiently high respirability, whereas for other materials this may occur spontaneously. In such situations, s pecific problems may arise with respect to classification and labelling, as these substances are tested in a form (i.e. forms in which these substances specific particle size distribution) that is different from all the h they can reasonably be expected to be used. are placed on the market and in whic A scientific concept has been developed as a basis for relating the conditions of acute inhalation tests to those occurring in real - life, in order to derive an adequate hazard classification. This concept is ap plicable only to substances or mixtures which are proven to cause acute toxicity through local effects and do not cause systemic toxicity (Pauluhn, 2008). Corrosive substances Annex I: 3.1.2.3.3. In addition to classification for inhalation toxicity, if da ta are available that indicates that the mechanism of toxicity was corrosivity, the substance or mixture shall also be labelled as ‘corrosive to the respiratory tract’ (see note 1 in 3.1.4.1). Corrosion of the the respiratory tract tissue after a single, limited respiratory tract is defined by destruction of period of exposure analogous to skin corrosion; this includes destruction of the mucosa. The corrosivity evaluation can be based on expert judgment using such evidence as: human and animal experience, e xisting (in vitro) data, pH values, information from similar substances or any other pertinent data. It is presumed that corrosive substances (and mixtures) will cause toxicity by inhalation exposure. In cases where no acute inhalation test has been performed special consideration should be given to the need to communicate this potential hazard. Corrosive substances (and mixtures) may be acutely toxic after inhalation to a varying degree and by different modes of action. Therefore, it is not possible to estimate the acute inhalation toxicity from the corrosivity data alone.

243 Guidance on the Application of the CLP Criteria 5.0 – July 2017 243 Version There are special provisions for hazard communication of acutely toxic substances by a corrosive effect, see Section of this Guidance. 3.1.4.2 Weight of evidence 3.1.2.3.3. In cases where there is sufficient human evidence that meets the criteria given in Section of this Guidance then this will normally lead to classification for acute toxicity, 3.1.2.2 irrespective of other information available. Please refer also to the Guidance R7a and in particular to especially to Appendix R7.4 - 1. If there are human data indicating no classification but there are also non human data - indicating classification then the classification is based on the non human data unless it is - - a or that the non - shown that the human data cover the exposure range of the non human dat human data are not relevant for humans. If the human and non - human data both indicate no classification then classification is not required. - If there are no human data then the classification is based on the non human data. For the role a nd application of expert judgement and weight of evidence determination, see CLP Annex I, 1.1.1. 3.1.2.4. Decision on classification The classification has to be performed with respect to all routes of exposure (oral, dermal, inhalation) on the basis of all adequa te and reliable available information. 3.1.2.5. Setting of specific concentration limits Specific concentration limits are not applicable for acute toxicity classification. Rather, the relative potency of substances is implicitly taken into account in the additiv ity formula (see Section 3.1.3 .3.3 of this Guidance). For this reason specific concentration limits for acute toxicity will not appear in CLP Annex VI, Table 3.1 or in the classification and labelling inventory (CLP Article 42). 3.1.2.6. Decision logic for classification of substances The decision logic below is provided as additional guidance. It is strongly recommended that the person responsible for classif ication is fully familiar with the criteria for acute toxicity classification before using the decision logic. For a complete classification of a substance, the decision logic must be worked out for each route of exposure for which data and/or information is available. For example, if a certain substance is classified in Category 1 based on an oral LD odyweight (the answer  5 mg/kg b 50 was 'Yes' in box 2 for item (a)), it is still necessary to go back to box 2 in the decision logic and - (e) route of exposure, when complete the classification for the dermal (b) and inhalation (c) data available for one or both of these routes of exposure. In case there are data for all are three routes of exposure, the classification for acute toxicity of the substance will include the three differentiations of the hazard class, which might result in three different categories being as signed to the different routes . The route of exposure will then be specified in the corresponding hazard statement.

244 ication of the CLP Criteria Guidance on the Appl – Version 5.0 244 July 2017 No Are there data and/or information Classification not possible to evaluate - (including WoE, see R.7.4 1) acute toxicity? Yes Category 1 According to the criteria in CLP Annex I, 3.1.2 to 3.1.3.4, does it have an: Oral LD (a)  5 mg/kg bodyweight; or 50 Yes  Dermal LD 50 mg/kg bodyweight; or (b) 50 (c) 100 ppm; or  Inhalation (gas) LC 50 Danger Inhalation (vapour) LC  (d) 0.5 mg/l ; or 50 0.05 mg/l?  Inhalation (dust/mist) LC50 (e) No Category 2 According to the criteria in CLP Annex I, 3.1.2 to 3.1.3.4, does it have an: (a) Oral LD >5 but  50 mg/kg bodyweight; or 50 Yes dyweight; or Dermal LD >50 but  200 mg/kg bo (b) 50 500 ppm; or < (c) Inhalation (gas) LC >100 but 50 Danger 2.0 mg/l; or Inhalation (vapour) LC (d) > 0.5 but < 50 > 0.05 but  0.5 mg/l? (e) Inhalation (dust/mist) LC 50 No Category 3 According to the criteria in CLP Annex I, 3.1.2 to 3.1.3.4, does it have an: (a) 300 mg/kg bodyweight; or >50 but ≤ Oral LD 50 Yes > 200 but ≤ 1000 mg/kg bodyweight; or (b) Dermal LD 50 (c) Inhalation (gas) LC >500 but ≤ 2500 ppm; or 50 Danger >2 but ≤ 10.0 mg/l; or Inhalation (vapour) LC (d) 50 (e) >0.5 but ≤ 1.0 mg/l? Inhalation (dust/mist) LC 50 No Category 4 According to the criteria in CLP Annex I, 3.1.2 to 3.1.3.4, does it have an: Oral LD >300 but ≤ 2000 mg/kg bodyweight; or (a) 50 Yes (b) Dermal LD >1000 but ≤ 2000 mg/kg bodyweight; or 50 >2500 but ≤ 20000 ppm; (c) or Inhalation (gas) LC 50 Warning >10 but ≤ 20 mg/l; (d) Inhalation (vapo ur) LC or 50 (e) Inhalation (dust/mist) LC >1 but ≤ 5 mg/l? 50 No No classification

245 Guidance on the Application of the CLP Criteria 5.0 – July 2017 245 Version acute toxicity 3.1.3. Classification of mixtures for 3.1.3.1. General considerations for classification The criteria for classification of substances for Annex I: acute toxicity as outlined in 3.1.3.1. ection 3.1.2 are based on lethal dose data (tested or deri ved). For mixtures, it is necessary s to obtain or derive information that allows the criteria to be applied to the mixture for the purpose of classification. The approach to classification for acute toxicity is tiered, and is dependent upon the amount of in formation available for the mixture itself and for its ingredients. The procedure for classifying mixtures is a tiered i.e. a stepwise approach based on a hierarchy principle and depending on the type and amount of available data/information. If valid te st data are available for the whole mixture they have precedence. If no such data exist, the so - called bridging principles have to be applied if possible. If the bridging principles are not applicable an assessment on the basis of ingredient information wi ll be applied (see Sections 3.1.3.3.3 , , 3.1.3.3.6 3.1.3.3.5 and 3.1.3.4 of this Guidance). 3.1.3.2. Identification of hazard information Where relevant and reliable toxicological information from human evidence or animal studies is available on a mixture, this should be used to derive the appropriate classification. Where such information on the mixture itself is not available, information on similar tes ted mixtures and, the component substances in the mixture must be used, as described in Section 3.1.3.3 of this Guidance. information on all individual components in the mixture could be Alternatively, the hazard identified as described in Section 3.1.2.2 of this Guidance. Classification criter ia 3.1.3.3. 3.1.3.2. Annex I: For acute toxicity each route of exposure shall be considered for the classification of mixtures, but only one route of exposure is needed as long as this route is followed (estimated or tested) for all components and there is no relevant evidence to sugg est acute toxicity by multiple routes. When there is relevant evidence of toxicity by multiple routes of exposure, classification is to be conducted for all appropriate routes of exposure. All available information shall be considered. The pictogram and si gnal word used shall reflect the most severe hazard category and all relevant hazard statements shall be used. The classification must be considered for each route of exposure. If different hazard categories are assigned, the most severe hazard category will be used to select the appropriate pictogram and signal word on the label for acute toxicity. For each relevant route of exposure, the hazard statement will correspond to the classification of this specific route. te mixture 3.1.3.3.1. When data are available for the comple 3.1.3.4.1. Annex I: mixture itself has been tested to determine its acute toxicity, Where the it shall be classified according to the same criteria as those used for substances, presented in Table 3.1.1. [...] In general, where a mixture has been tested those data should be used to support classification according to the same criteria as used for substances (as described in Section 3.1.2.3 o f this Guidance). However, there should be some consideration of whether the test is appropriate. For instance, if the mixture contains a substance for which the test species is not considered appropriate (for instance a mixture containing methanol tested in rats which are not sensitive to

246 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 246 methanol toxicity), then the appropriateness of these data for classification should be considered using expert judgement. With respect to the classification of mixtures in the form of dust or mist for acute inhalation oxicity, the particle size can affect the toxicity and the resulting classification should take this t 3.1.2.3.2 of this Gui dance). into account (see Section 3.1.3.3.2. When data are not available for the complete mixture: bridging principles Annex I: 3 .1.3.5.1. Where the mixture itself has not been tested to determine its acute toxicity, but there are sufficient data on the individual ingredients and similar tested mixtures to adequately characterise the hazards of the mixture, these data shall be used in accord ance 1.1.3. with the bridging rules set out in section In order to apply bridging principles, there needs to be sufficient data on similar tested mixtures as well as the ingredients of the mixture (see Section 1.6.3 of this Guidance). When the available i dentified information is inappropriate for the application of bridging principles then the mixture should be classified based on its ingredients as in Section 3.1.3.3.3 , 3.1.3.3.5 , 3.1.3.3.6 and 3.1.3.4 of this Guidance. 3.1.3.3.3. When data are available for all ingredients Annex I: 3.1.3.3. (c) If the converted acute toxicity point estimates for all components of a mixture are within the same category, then the mixture should be classified in that category. (d) When only range data (or acute toxicity hazard category information) are available for components in a mixture, they may be converted to point estimates in accordance with Table 3.1.2 when calculating the classification of the new mixture using the formulas in sections 3.1.3.6.1 and 3.1.3.6.2.3. Annex I: Classification of mixtures based on ingredients of the mixture (Additivity 3.1.3.6. formula) 3.1.3.6.1. Data available for all ingredients Annex I: In order to ensure that classification of the mixture is accurate, and that the calculation need only be performed o nce for all systems, sectors, and categories, the acute toxicity estimate (ATE) of ingredients shall be considered as follows: (a) include ingredients with a known acute toxicity, which fall into any of the acute hazard categories shown in Table 3.1.1; (b) ignore ingredients that are presumed not acutely toxic (e.g., water, sugar); (c) ignore components if the data available are from a limit dose test (at the upper threshold for Category 4 for the appropriate route of exposure as provided in Table 3.1.1) and do not sho w acute toxicity. Components that fall within the scope of this section are considered to be components with a known acute toxicity estimate (ATE). See note (b) to Table 3.1.1 and section 3.1.3.3 for appropriate application of available data to the equatio n below, and section 3.1.3.6.2.3. The ATE of the mixture is determined by calculation from the ATE values for all relevant ingredients according to the following formula below for Oral, Dermal or Inhalation Toxicity:

247 Guidance on the Application of the CLP Criteria 5.0 – July 2017 Version 247 100 C i   ATE ATE n i mix where: ingredient i (% w/w or % v/v) C = concentration of i i = the individual ingredient from 1 to n n the number of ingredients = = Acute Toxicity Estimate of ingredient i. ATE i In case an ingredient has a harmonised ATE this value must be used in the formula above. If no harmonise d ATE is available, then the ATE should be derived as stated in 3.1.2.3. The cATpE (mentioned in 3.1.2.3.2) is used when ATE values are not known. If there is a harmonised not support classification in that classification and the only known ATE value hazard does category, then the cATpE should be considered. 3.1.3.3.4. Special case for acute inhalation toxicity substance ( s ) tested for inhalation toxicity as vapours and others For mixtures containing some directly as the ATE ranges are as dust/mist or gas, the additivity formula cannot be used to be used separately for different. Therefore for acute inhalation toxicity additivity has initially each relevant physical form (i.e. gas, vapour and/or dust/mist), using the appropriate category limit in CLP Annex I, Table 3.1.1. As a first step, the fraction of toxicity is calculated for each form/state: /100 fraction = ∑ (limit / ATE) x concentration s Where limit = the upper border of the range of ATE values of a hazard category (Table 3.1.1 of = the concentration (%) of components CLP) for the state/form in question and concentration s tested for this state/form. The most severe category where the sum of fractions for the three states/forms are ≥ 1 would 13 in section 3.1.5. 5 ). apply (see example redient(s) with unknown acute toxicity, the value is corrected as 1 In case of > 10% of ing minus concentration of unknowns/100. In case no ATE values but only classification of the ingredients is known, the converted Acute Toxicity point Estimates (cATpEs) as shown in Table 3.1. 2 of Annex I (see below) should be used. In addtiton to the new example 13, examples 12a and 12b are also provided in section 3.1.5 (see note to the examples). Annex I: Table 3.1.2 Conversion from experimentally obtained acute toxicity range values (or acute toxicity hazard categories) to acute toxicity point estimates for use in the formulas for the classification of mixtures Classification category or experimentally Converted acute toxicity point Exposure estimate (see Note 1) obtained acute toxicity range estimate routes 0 < Category 1  5 Oral 0.5 (mg/kg 2  50 5 < Category 5 bodyweight) 50 100  3 < Category 300 500 2000 < Category  300 4

248 ication of the CLP Criteria Guidance on the Appl – July 2017 Version 248 5.0 Dermal 0 < Category 1  50 5 (mg/kg 2  200 50 50 < Category bodyweight) 1000 200 300 < Category 3  1100  4 Category < 1000 2000  < Category 1 0 100 Gases 10 (ppmV) < Category  500 100 2 100 700 2500 500 3  < Category 4500 4 < Category 20000  2500 0 < Category Vapours  0,5 1 0,05 (mg/l) < Category 2  2 0.5 0,5  10,0 3 2,0 < Category 3 11 Category 4 20,0  10,0 < 0< Category 1  Dust/mist 0,05 0,005 (mg/l) 2  0,05 0,5 < Category 0,05  1,0 0,5 < Category 3 0,5  5,0 1,0 < Category 4 1,5 Note 1: of a These values are designed to be used in the calculation of the ATE for classification mixture based on its components and do not represent test results. Some cATpEs are equal to the upper limit of the next lower category, for example the cATpE of oral Category 2 (5 mg/kg bw) is equal to the upper limit of oral Category 1 (also 5 mg/kg bw). This can lead to a problem when using the cATpE values for calculating the acute toxicity of mixtures. For instance, using the cATpEs for a mixture containing only substances classified in r the mixture. Similarly, a mixture Category 2 actually results in a Category 1 classification fo containing substances classified as Category 3 for dust/mist results in a Category 2 classification. Clearly these outcomes are incorrect and are an unintended side - effect of the approach. In such cases, CLP Annex I, 3.1 .3.3.(c) should be applied. Annex I: 3.1.3.3.(c) If the converted acute toxicity point estimates for all components of a mixture are within the same category, then the mixture should be classified in that category. As a result, the mixtures in the examples highlighted above would be classified in Categories 2 and 3, respectively. Annex I: 3.1.3.3.(b) where a classified mixture is used as an ingredient of another mixture, the actual or derived acute toxicity estimate (ATE) for that mixture may be used, when calculating the classification of the new mixture using the formulas in s ection 3.1.3.6.1 and paragraph 3.1.3.6.2.3. It is important that the downstream user has sufficient information in order to enable him to pe rform a correct classification of mixtures.

249 Guidance on the Application of the CLP Criteria 5.0 – July 2017 249 Version available for all ingredients not When data are 3.1.3.3.5. Where an ATE is not available for an individual ingredient of the 3.1.3.6.2.1. Annex I: mixture, but available information such as that listed below can provide a derived conversion 3.1.2, the formula in paragraph 3.1.3.6.1 shall be value such as those laid out in Table applied. This includes evaluation of: 1 extrapolation between oral, dermal and inhalation acute toxicity estimates ( ). Such an (a) evaluation could r equire appropriate pharmacodynamic and pharmacokinetic data; (b) evidence from human exposure that indicates toxic effects but does not provide lethal dose data; evidence from any other toxicity tests/assays available on the substance that indicates (c) xic acute effects but does not necessarily provide lethal dose data; or to data from closely analogous substances using structure/activity relationships. (d) _______________ 1 ( oute of ) When mixtures contain components that do not have acute toxicity data for each r to exposure, acute toxicity estimates may be extrapolated from the available data and applied ection 3.1.3.2). However, specific legislation may require testing the appropriate routes (see S for a specific route. In those cases, classification shall be performed for that route based upon the legal requirements. Derivation of ATEs from available information: When ingredients have a known acute toxicity (LC values), this value has to be used in or LD 50 50 ances, acute toxicity data will not be available the additivity formula. However, for many subst for all exposure routes. CLP allows for two ways of deriving acute toxicity conversion values. One option is to use the er option, converted acute toxicity point estimates supplied in CLP Annex I, Table 3.1.2. The oth based on expert judgement in substantiated cases is the use of the directly derived ATE values. , a. Route - to - route extrapolation (CLP Annex I, 3.1.3.6.2.1.(a)) Route - to - route extrapolation is defined as the prediction of the total amount of a substance administered by one route that would produce the same systemic toxic response as that obtained by a given amount of a substance administered by another route. Thus , route - to - route extrapolation is only applicable for the evaluation of systemic effects. It is not appropriate to assess direct local effects. This extrapolation is possible if certain conditions are met, which substantiate the assumption that an interna l dose causing a systemic effect at the target is related to an external dose/concentration; preferably the absorption can be quantified. Therefore information on the - chemical and biokinetic properties should be available and assessed in order to a llow physico such a conclusion and performing an extrapolation across routes. In the absence of any information on absorption, 100% absorption has to be presumed as a worst case for the dermal and inhalation route. Extrapolating from the oral route to other routes , the assumption of an absorption of 100% for the oral route is, however, not a worst case. Absorption of less than 100% by the oral route will lead to lower ATEs. Another important factor is the local and systemic metabolic pathways; in particular it must be ensured that no route - specific metabolism/degradation of substance occurs. If extrapolating from oral data, the influence of first - pass metabolism in the stomach/intestines and the liver should be considered, especially if the substance is detoxified. Such first pass metabolism is unlikely to occur to any significant extent by the dermal or inhalation routes, and

250 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 250 so this would lead to an underestimate of toxicity by these routes. Thus if based on kinetic or - pass effect is ex cluded, oral data may be used for extrapolation (Q)SAR data a specific first purposes. For an extrapolation to the dermal route, information on the potential skin penetration may be derived from the chemical structure (polar vs. nonpolar structure elements, Log P , molecular ow ) if kinetic data are not available which would allow a quantitative comparison. When no weight such information is available 100% dermal absorption should be presumed. Further information and guidance on dermal absorption can be found on the OECD and EFSA websit – OECD es ( http://www.oecd.org/chemicalsafety/testingofchemicals/48532204.pdf ) and EFSA ( http://www.efs a.europa.eu/en/efsajournal/doc/2665.pdf ). Similarly for an extrapolation to the inhalation route if there is no quantitative information on absorption then 100% absorption should be presumed. Inhalation volatility is an important the nd may increase the exposure, but on the other hand may reduce factor which on one ha - absorption due to higher exhalation rates. The solubility (in water and non polar solvents) has to be considered, as well as particle size, which plays a particularly important role in inhalati on toxicity. Route - to - route extrapolation is not always appropriate. For example where there is a substantial difference in absorption between oral and inhalation uptake (e.g. poorly soluble - tract), or where the substance particles, substances that decompose within the gastro intestinal causes local effects, the toxicity by different routes may be significantly different, and route - to - route extrapolation may not be appropriate (ECETOC TR 86, 2003). i. Extrapolation oral  inhalation are met an extrapolation from oral data would be performed as If the mentioned conditions follows: Incorporated dose = concentration x respiratory volume x exposure time 1 mg/kg bw = 0.0052 mg/l/4h using a respiratory volume for a 250 g rat of 0.20 l/min and 100 % absorption and pos tulating Table R.7.12 100% deposition and absorption (Guidance on 10). IR&CSA, Chapter R7c, - indicating that the deposition and/or absorption rate for the extrapolated route Valid information 3.1.5.1.9 Example 9 of this is lower would allow a higher equivalent derived ATE (see Section Guidance). Extrapolation oral  dermal ii. cific first pass - If based on kinetic or SAR data a high penetration rate can be assumed and a spe effect is excluded, oral and dermal toxicity might be regarded as equivalent. This is rarely the case. Solids themselves may have a very low absorption rate, but if diluted in an appropriate solvent there may be an appreciable absorption o f the substance. Thus, depending on the kinetic and physico - chemical properties and kind of mixture, varying ATEs will result. For example, butyn - 1,4 - diol causes no mortality in rats when dermally applied as a solid at 5000 mg/kg bw, s solution of butyn - 1,4 - diol is administered, a dermal LD whereas when an aqueou of 659 and 50 1240 mg/kg bw in male and female rats, respectively, and an oral LD of about 200 mg/kg bw 50 in both sexes can be determined. For more details on inter - route extrapolation see the Guidan ce on IR&CSA, Section R.7c. 12.2.4. e xamples 8 and 9 which illustrate this approach. b. Evidence from human exposure Human evidence can be used to derive an appropriate ATE to use in the additivity approach for ). Therefore it is necessary to extrapolate from mixtures (CLP Annex I, 3.1.3.6.1 and 3.1.3.6.2.3

251 Guidance on the Application of the CLP Criteria 5.0 – July 2017 251 Version by into account the potency (i.e. the magnitude of the taking adequate and reliable data and lethal dose reported) of the effects in humans. Thus an equivalent ATE may be derived on the toxicity data (minimum dose/concentration) and used directly in the basis of valid human Example 1 of this Guidance). The alternative to the additivity formulae (see Section 3.1.5.1.1 derivation of an equivalent ATE is the allocation to a category. The category should be justified by semi - quantitative or qualitative data and a subsequent derivation of a converted ATE to CLP Annex I, Table 3.1.2 and subsequent use in the formulae (see Section (cATpE) according Example 2 of this Guidance). See also Section 3.1.2.3.1 3.1.5.1.2 of this Guidance for more details. c. Evidence from other toxicity tests Standard acute toxicity studies should be the primary source of information for acu te toxicity - reliable studies classification. However, when such data are not available or only data from non exist, information from studies conducted for other endpoints can be used for acute toxicity m repeated dose testing can be used. These classification. For example, data on early effects fro studies will not usually provide an exact ATE value that can be used directly for classification, but they may provide enough information to allow an estimate of acute toxicity to be made, which would be sufficien t to support a decision on classification. Furthermore, it can also be concluded that no classification is warranted for instance by a 28 - day repeated dose toxicity o study that is performed with 1000 mg/kg bw/day and no adverse effects are observed (refer t - Appendix 7.4 1 of Guidance R.7a). In addition, a substance not acutely toxic after oral exposure is not considered as acutely toxic via dermal exposure (see Guidance R.7a). Example: dose toxicity daily oral Available information: In a range finding study with respect to repeated doses of 1000 mg/kg bw over 5 days prove to be neither lethal nor cause serious symptoms in rats at the end of the observation period of 14 days. Conclusion: the ATE is > 2000 mg/kg bw since 2 doses following (within roughly) 24 h are not lethal (see Section 3.1.2.2 of this Guidance). Thus this ingredient can be igno red in the additivity procedure. d. Use of (Q)SAR /LC values predicted by a highly reliable model (see Section 3.1.2.3.2 of this Guidance) LD 50 50 =ATE in the /LC may be used according to Note (a) to CLP Annex I, Table 3.1.1 directly as LD 50 50 additivity formula CLP Annex I, 3.1.3.6.1. If the assessment using (Q)SARs gives a more general result a cATpE according to Table 3.1.2 may be derived. It has to be emphasised that these approaches generally require substantial technical information, and expert judgement, to reliably estimate acute toxicity. Further guidance on how to apply this provision is given in Section of this Guidance. 3.1.3.3.6 Annex I: 3.1.3.6.2.3. If the total concentration of the relevant ingredient(s) with unknown acute toxicity is ≤ 10 % then the formula prese nted in section 3.1.3.6.1 shall be used. If the total concentration of the relevant ingredient(s) with unknown toxicity is > %, the formula 10 presented in section 3.1.3.6.1 shall be corrected to adjust for the total percentage of the unknown ingredient(s) as follows: 100 C if   10% C  umknown i   ATE ATE n i mix

252 ication of the CLP Criteria Guidance on the Appl 252 Version 5.0 – July 2017 3.1.3.3.6. Ingredients that should be taken into account for the purpose of classification Annex I: the ‘relevant ingredients’ of a mixture are those which are present in 3.1.3.3.(a) % (w/w for solids, liquids, dusts, mists and vapours and v/v for gases) or concentrations of 1 greater, unless there is a reason to suspect that an ingredient present at a concentration of % is still relevant for classifying the mixture for acute toxicity (see Table 1.1). less than 1 When a mixture contains a ‘relevant’ ingredient (i.e. constituting ≥ 1%; CLP Annex I, 3.1.3.3 (a)) for which there is no adequate acute toxicity data then the mixture must be classified on the label and in the basis of the ingredients with known toxicity, with an additional statement on the SDS to indicate that the mixture consists of ‘x percent’ of component(s) of unknown acute toxicity (CLP Annex I, 3.1.3.6.2.2). The determination of the classification depends on what n toxicity constitute. If these ingredients proportion of the mixture such ingredients of unknow constitute ≤10% of the total mixture, the additivity formula in CLP Annex I, 3.1.3.6.1 must be used. However, in cases where these ingredients constitute over 10%, a modified additivity 1.3.6.2.3 must be used, which adjusts for the presence of a formula in CLP Annex I, 3. significant proportion of ingredients of unknown toxicity. This reflects the greater uncertainty as to the true toxicity of the mixture). Excerpt of Table 1.1 Annex I: Generic cut - off values Generic cut Hazard class off values to be taken into account - Acute Toxicity: - Category 1 - 3 0,1 % - 1 % Category 4 - off values are in weight percentages except for gaseous mixtures for those Note: Generic cut - off values may be best described in volume percentages. hazard classes where the generic cut concentrations As indicated in CLP Annex I, Table 1.1, when components are present in low they do not need to be taken into account when determining the classification of the mixture, according to the approaches detailed in CLP Annex I, 3.1.3.6.1 and 3.1.3.6.2.3 (see Section Example 11 of this Guidance). Accordingly, all components classified in Categories 1 - 3 3.1.5.3.1 at a concentration <0.1% and Category 4 <1% are not taken into account. Similarly unknown ingredients pr esent at <1% are not taken into account. 3.1.3.3.7. Non classified components - For mixtures containing ingredients with ATE values that are more than 2000 mg/kg (i.e. non - classified components), such ingredients need not be considered in the calculation of ATEs with the formula presented in CLP Annex I: 3.1.3.6.1. However, in cases where no acute toxicity data are available for some ingredients or a mixture contains ingredients with unspecified ATE values la of CLP Annex I: 3.1.3.6.2.3 has which could fall within the classifiable limits, then the formu to be used for calculation of ATEs to adjust for the concentrations of ingredients with unknown acute toxicities.

253 Guidance on the Application of the CLP Criteria 5.0 – July 2017 253 Version 3.1.3.4. Generic concentration limits for substances triggering classification of mixtures Generic concentration li mits as such are not applicable for acute toxicity classification; therefore Section 3.1.2.5 specific concentration limits are also not applicable (see of this Guidance ). Nevertheless, according to CLP Annex VI, 1.2.1 the classification for entries with the reference * in the column specific concentration limits is of special concern; the * means that those entries had an SCL in CLP Annex VI, Table 3.2 originating from Annex I to DSD. When assessing a mixture according to the procedure set out in CLP Annex I, a thorough search for the data (animal, human experience or other information) is necessary. The assessment must take all available information into account using a weight of evidence approach and expert judgement with special emphasis on possibly available human experience or information. These validated data will then be used in the additivity formula in CLP Annex I, 3.1.3.6.1 as ATEs or cATpEs (CLP Annex I, Table 3.1.2). 3.1.3.5. Decision on classification The assessment o f classification has to be performed with respect to all the relevant routes of exposure (oral, dermal, inhalation) on the basis of all adequate reliable data. If there is evidence of se toxicity by multiple routes of exposure classification is warranted for all the routes, however the label should include one pictogram and a signal word reflecting the most severe hazard category. If , for example, a mixture fulfils the criteria for oral toxicity Category 4 and for inhalation Category 2, then the mixture will be classified in Category 4 for oral toxicity and Category 2 for inhalation toxicity and assigned the corresponding hazard statements; it will be labelled with the acute toxicity Cat egory 2 pictogram (skull and cross bones) and the signal word ‘Danger’ and both the hazard statements for inhalation Category 2 (H330) and oral Category 4 (H302) (see CLP Annex I Table 3.1.3 in next section 3.1.4.1 of this Guidance). 3.1.3.6. Decision logic for classification of mixtures The decision logic is provided as additional guidance. It is strongly recommended that the person responsible for classification study the criteria for classification before and during use of the decision logic.

254 ication of the CLP Criteria Guidance on the Appl July 2017 5.0 – 254 Version Classify in appropriate Does the mixture as a whole have Yes data/information to evaluate acute category according to CLP toxicity toxicity? Annex I, Table 3.1.1 No Classify in appropriate Yes category principles be applied? Can bridging No Apply the acute toxicity Yes Is acute toxicity data available for all estimate calculation to ingredients of mixture? determine the ATE of the mixture No Is it possible to estimate missing Yes ATE(s) of the ingredient(s), i.e. can where: conversion value(s) be derived? ATE to = concentration of C mix i Decision ingredient i No logic in i = the individual 3.1.2.6 Is the total concentration of the ingredient from 1 to n Yes ingredient(s) with unknown acute n = the number of 10%? toxicity ≤ ingredients Toxicity ATE = Acute i No Estimate of ingredient i. Apply the acute toxicity estimate calculation (i.e. when the total concentration of ingredients with unknown acute to Decision ATE 10%): toxicity is > mix logic in 3.1.2.6

255 Guidance on the Application of the CLP Criteria 5.0 – July 2017 255 Version form of labelling for acute toxicity Hazard communication in 3.1.4. the Pictograms, signal words, hazard statements and precautionary 3.1.4.1. statements Annex I: Table 3.1.3 Acute toxicity label elements Category 1 Classification Category 2 Category 3 Category 4 GHS Pictograms Signal Word Danger Danger Warning Danger Hazard Statement: H301: Toxic H300: Fatal H300: Fatal H302: Harmful if if swallowed if swallowed if swallowed swallowed – Oral Dermal H310: Fatal – H311: Toxic Fatal H310: H312: Harmful in in contact in contact in contact contact with skin with skin with skin with skin – Inhalation H332: Harmful if H331: Toxic H330: Fatal H330: Fatal if inhaled if inhaled if inhaled inhaled (see Note 1) P264 P264 P264 Precautionary Statement P264 Prevention (oral) P270 P270 P270 P270 P301 + P310 P301 + P312 P301 + P310 P301 + P310 Precautionary Statement Response (oral) P321 P321 P321 P330 P330 P330 P330 Precautionary Statement P405 P405 P405 Storage (oral) P501 P501 P501 P501 Precautionary Statement Disposal (oral) Precautionary Statement P262 P280 P280 P262 Prevention (dermal) P264 P264 P270 P270 P280 P280 P302 + P352 P302 + P350 Precautionary Statement P302 + P350 P302 + P352 Response (dermal) P312 P310 P312 P310

256 ication of the CLP Criteria Guidance on the Appl 256 Version 5.0 – July 2017 P322 P322 P322 P322 P361 P361 P363 P361 P363 P363 P363 Precautionary Statement P302 + P352 P302 + P352 P302 + P352 P302 + P352 Response (dermal) P312 P310 P312 P310 P321 P321 P321 P321 P361 + P361 + P361 + P362 +P364 P364 P364 P364 Precautionary Statement P405 P405 P405 Storage (dermal) P501 Precautionary Statement P501 P501 P501 Disposal (dermal) Precautionary Statement P261 P260 P261 P260 Prevention (inhalation) P271 P271 P271 P271 P284 P284 Precautionary Statement P304 + P340 P304 + P340 P304 + P340 P304 + P340 Response (inhalation) P310 P310 P311 P312 P321 P320 P320 Precautionary Statement P403 + P233 P403 + P233 P403 + P233 Storage (inhalation) P405 P405 P405 Precautionary Statement P501 P501 P501 Disposal (inhalation) Note 1 classification for inhalation toxicity, if data are available that indicates that In addition to the mechanism of toxicity is corrosivity, the substance or mixture shall also be labelled as EUH071: ‘corrosive to the respiratory tract’ — see advice at 3.1.2.3.3. In additi on to an appropriate acute toxicity pictogram, a corrosivity pictogram (used for skin and eye corrosivity) may be added together with the statement ‘corrosive to the respiratory tract’. Note 2 In the event that an ingredient without any useable information at all is used in a mixture at a concentration of 1 % or greater, the mixture shall be labelled with the additional statement that ‘x percent of the mixture consists of ingredient(s) of unknown toxicity’ — see advice at 3.1.3.6.2.2. EUH071 can also be ap plied to inhaled corrosive substances not tested for acute inhalation toxicity according to CLP Annex II, Section 1.2.6 If a substance or a mixture fulfils the classification criteria with respect to different routes the pictogram and signal word will be b ased on the most severe one, however the hazard statements for each route must be included on the label.

257 Guidance on the Application of the CLP Criteria 5.0 – July 2017 257 Version Article 26 1 (b) If the hazard pictogram ‘GHS06’ applies, the hazard pictogram ‘GHS07’ shall not appear. Additional labelling provisions 3.1.4.2. In addition to the statement required under CLP Annex I, 3.1.3.6.2.2, it would be appropriate to - by specify the relevant exposure route of toxicity concerned on a case case basis: For example - ‘x percent of the mixture consists of component(s) of unknown acute oral to ’ . In the case xicity of different values being available for the % of ingredients having unknown acute toxicity (as a result of different route of exposure), the % value to be included in the sentence on the label e the % of ingredients having unknown toxicity is should be selected based on the route wher the highest. 3.1.3.6.2.2. Annex I: In the event that a component without any useable information for classification is used in a mixture at a concentration ≥ 1 %, it is concluded that the mixture cannot be attributed a definitive acute toxicity estimate. In this situation the mixture shall be classified based on the known components only, with the additional statement on the label and in the SDS that: “x percent of the mixture consists of component(s) of unknown acute toxicity”, taking into account the provisions set out in section 3.1.4.2. Annex I: 3.1.4.2 The acute toxicity hazard statements differentiate the hazard based on the route of exposure. Communication of acute toxicity classification should also reflect this differentiation. If a substance or mixture is classified for more than one route of exposure then all relevant classifications should be communicated on the safety data sheet as specified in Annex II to Regulation (EC) No 1907/2006 and the relevant hazard communication elements included on the label as prescribed in section 3.1.3.2. If the statement “x % of the mixture consists of ingredient(s) of unknown acute toxicity” is communicated, as prescribed in section 3.1.3.6.2.2, then, in the information provided in the safety data sheet, it can also be differentiated based on the route of exposure. For example, “x % of the mixture consists of ingredient(s) of unknown acute oral toxicity” and “x % of the mixture consists of ingredient(s) of unknown acute dermal toxicity In case section 3.1.3.6.2.2 applies and the statement ‘x % of the mixture consists of ingredient(s) of unknown acute toxicity’ has to be communicated, the same statement can be differentiated on the bas is of the route of exposure in the safety data sheet (SDS) in accordance with CLP Annex I 3.1.4.2. For example on the label and in the SDS the following should appear: ‘x % of the mixture consists of ingredient(s) of unknown acute toxicity’; in the SDS the route of exposure can also be specified, for example ‘x % of the mixture consists of ingredient(s) of unknown acu te oral toxicity’ and ‘x % of the mixture consists of ingredient(s) of unknown acute dermal toxicity’. In case of different values being available for the % of ingredients having unknown toxicity (as a result of a different route of exposure), the % value to be included in the sentence on the label should be selected based on the route where the % of ingredients having unknown toxicity is the highest.

258 ication of the CLP Criteria Guidance on the Appl 258 Version 5.0 – July 2017 Corrosivity: Annex I: 3.1.2.3.3. available that indicates that the In addition to classification for inhalation toxicity, if data are mechanism of toxicity was corrosivity, the substance or mixture shall also be labelled as ‘corrosive to the respiratory tract’ (see note 1 in 3.1.4.1). Corrosion of the respiratory tract is defined by destruction of the re spiratory tract tissue after a single, limited period of exposure analogous to skin corrosion; this includes destruction of the mucosa. The corrosivity evaluation can be based on expert judgment using such evidence as: human and animal experience, existing (in vitro) data, pH values, information from similar substances or any other pertinent data. In addition to the application of the classification for acute inhalation toxicity, the substance or mixture must also be labelled as EUH071 where data are avail able which indicate that the mode of toxic action was corrosivity (see Note 1 to Table 3.1.3). Such information can be derived from data which warrant classification as corrosive according to the hazard skin 3.2 of this Guidance). In this case the substance or mixture corrosion/irritation (see Chapter has to be classified and labelled for skin corrosion with the pictogram for corrosivity, GHS05, haza rd statement H314 and also labelling with EUH071 (for criteria, see CLP Annex II) is 3.2.4.2 of this Guidance). required (see Chapter 1.2.6. EUH071 Annex II: ‘Corrosive to the respiratory tract’ — For substances and mixtures in addition to classification for inhalation toxicity, if data are available that indicate that the mechanism of toxicity is c orrosivity, in accordance with S ection 3.1.2.3.3 and Note 1 of Table 3.1.3 in Annex I. For substances and mixtures in addition to classification for skin corrosivity, if no acute inhalation test data are available and which may be inhaled. inhalation to a varying degree, Corrosive substances and mixtures may be acutely toxic after although this is only occasionally proved by testing. In case no acute inhalation study is available for a corrosive substance or mixture, and such substance or mixture may be inhaled, a hazard of respiratory tract corrosion may exist. As a consequence, substances and mixtures have to be supplementar il y labelled with EUH071, if there is a possibility of exposure via inhalation taking into consideration the saturated vapour concentration and the possibility of exposure to part size as appropriate (see also C hapter 3.8.2.5 of icles or droplets of inhalable this Guidance. It is strongly recommended to apply the precautionary statement P260: Do not breathe dust/fume/gas/mist/vapours/spray. Toxic by eye contact: 1.2.5 — Annex II: EUH070 ‘Toxic by eye contact’ For substances or mixtures where an eye irritation test has resulted in overt signs of systemic toxicity or mortalit y among the animals tested, which is likely to be attributed to absorption of the substance or mixture through the mucous membranes of the eye. The statement shall also be applied if there is evidence in humans for systemic toxicity after eye contact. The statement shall also be applied where a substance or a mixture contains another substance labelled for this effect, if the concentration of this substance is equal to, or greater than 0,1 %, unless otherwise specified in part 3 of Annex VI. In cases wher e a substance or mixture has shown clear signs of severe systemic toxicity or mortality in an eye irritation study a supplemental labelling phrase EUH070 ‘Toxic by eye contact’ is required. This additional labelling, based on relevant data, is independent of any classification in an acute toxicity category.

259 Guidance on the Application of the CLP Criteria 5.0 July 2017 259 Version – Liberation of toxic gases 1.2.1. EUH029 Annex II: ‘Contact with water liberates toxic gas’ — For substances and mixtures which in contact with water or damp air, evolve gases classified for acute toxici ty in category 1, 2 or 3 in potentially dangerous amounts, such as aluminium phosphide, phosphorus pentasulphide. Annex II: 1.2.1 EUH031 — ‘Contact with acids liberates toxic gas’ For substances and mixtures which react with acids to evolve gases classified for acute toxicity in category 3 in dangerous amounts, such as sodium hypochlorite, barium polysulphide. Annex II: 1.2.3. EUH032 — ‘Contact with acids liberates very toxic gas’ For substances and mixtures which react with acids to evolve gas es classified for acute toxicity in category 1 or 2 in dangerous amounts, such as salts of hydrogen cyanide, sodium azide.

260 Guidance on the Appl ication of the CLP Criteria 260 5.0 – July 2017 Version 3.1.5. Examples of classification for acute toxicity toxicity. NOTE: The classification proposals for the examples refer only to acute Examples of substances fulfilling the criteria for classification 3.1.5.1. Example 1: Methanol 3.1.5.1.1. Use of adequate and reliable human data allowing derivation of an Application equivalent ATE according to CLP Annex I, Table 3.1.1. Animal data not appropriat e. Classification Rationale Test Data Available The rat is known to be Animal data: Classification insensitive to the toxicity of information not possible Oral LD rat ≥ 5000 mg/kg 50 methanol and is thus not bw considered to be a good effects model for human (different effect/mode of action) Human experience: The minimum lethal dose Category 3 reported of 300 mg/kg bw is Methanol is known to cause used as equivalent ATE; lethal intoxications in according to CLP Annex I, humans (mostly via Table 3.1.1 the resulting ingestion) in relatively low classification is Category 3 doses: ‘...minimal lethal dose in the absence of medical treatment is between 300 ’ (IPCS, bw and 1000 mg/kg Environmental Health Criteria 196, Methanol, ) WHO, 1997 in rats from mixtures containing methanol should not be used directly Remarks Test data in additivity formula Example 2: N,N - Dimethylaniline 3.1.5.1.2. Application Use of qualitative human data and of SAR information with extrapolation to an ATE (CLP Annex I, 3.1.3.6.2.1(b) and Table 3.1.2). Animal data are not appropriate. Test Data Classification Rationale Available Animal data: Category 4 information Acute dermal toxicity: LD 50 values > 1690 mg/kg bw rabbit. Human experience: The extensive and consistent 3 Category (oral, dermal, human experience is Broad human experience, inhal ation) considered to be sufficiently reported in many case robust by expert judgement to

261 Guidance on the Application of the CLP Criteria 5.0 July 2017 261 Version – reports, demonstrating be used for classification into death from MetHB Category 3. The rabbit LD 50 suggests lower sensitivity to following relatively low oral/dermal/inhalation MetHB formation than humans which is consistent wi exposure to aromatic th what - amines such as N,N is known from other rabbit tests with substances known to - dimethylaniline. For N,N aniline - Dimethyl no induce MetHB in humans. The itself rabbit data are therefore not exact human toxicity considered to be adequate for values are available. acute toxicity classification. Therefore the human data on this and structurally substances ar e used to related give a converted Acute Toxicity point Estimate (cATpE) according to CLP Annex I, Table 3.1.2 for Category 3; e.g. cATpE dermal = 300 s mg/kg bw, which then fall to a higher category than the in rabbit data. Remarks none 3.1.5.1.3. Example 3 Application No exact LD value available. Expert judgement needed. 50 Test Data Classification Rationale Available Category 4 Since at a dose of 200 mg/kg Corrosive volatile liquid (not information classified for skin corrosion). bw no mortality and only slight transient symptoms Animal data: without necropsy findings were observed, and at 500 compliant acute - In a GLP mg/kg bw the high oral toxicity study in rats, amount/concentration of the the following results were substance caused corrosive observed: serious effect only at the site At a test dose of 200 of action and mortality, based mg/kg bw: no mortality, on expert judgement it can be only transient symptoms assumed that the likely LD is 50 and no necropsy findings. > 300 mg/kg bw. Therefore, the Acute Toxicity Estimate At a test dose of 500 (ATE) value for classification mg/kg: 100% mortality, purpose is between 300 and symptoms: poor general 5 00 mg/kg bw, corresponding state; necropsy findings: to Category 4 classification for hyperemia in stomach acute toxicity. l irritation (due to loca /corrosivity), no other organs affected . Remarks Labelling (in addition to the labelling provisions for Acute tox Cat. 4): Corrosive pictogram (pictogram is not mandatory, it may be added) (see Annex I: Note 1 of Table 3 .1.3) Additional Hazard statement: EUH071 Corrosive to the respiratory tract

262 ication of the CLP Criteria Guidance on the Appl 262 Version 5.0 – July 2017 3.1.5.1.4. Example 4 - - guideline test data. Use of non standard Application Classification Rationale Test Data Rationale for classification: Animal data: Category 2 Available Since the dermal LD is information 50 A study to evaluate the above 50 mg/kg bw and less acute dermal than 200 mg/kg bw, (percutaneous) toxicity Category 2 classification is was performed in rabbits. warranted (see CLP Annex I, The following test data Table 3.1.2) results were reported: At the dose level of 50 - mg/kg bw: no mortality was observed At 200 mg/kg bw: 100% - mortality LD the Therefore, was 50 estimated to be between d 200 50 mg/kg bw an mg/kg bw none Remarks 3.1.5.1.5. Example 5 Application Use of CLP Annex I, Table 3.1.1 and experimentally obtained LC value 50 Test Data Classification Rationale Available Rationale for classification: A gas Category 4 information LC = 4500 ppm is 50 Animal data: considered an Acute Toxicity Estimate (ATE) for A GLP compliant test for - classification purposes; acute inhalation toxicity according to the classification (gaseous form) was criteria for acute inhalation performed in accordance toxicity for gases (CLP Annex with OECD TG 403 in rats. I, Table 3.1.1), this value was The following LC 50 corresponds to Category 4. calculated: Therefore Category 4 Acute : 4500 ppm/4h LC 50 Inhalation Toxicity classification is warranted. Remarks none 3.1.5.1.6. Example 6 Application Time extrapolation; Note (c) in CLP Annex I, Table 3.1.1; Haber’s law Test Data Classification Rationale Available Solid substance on criteria for Category 3 The classificati information acute inhalation toxicity in Animal data: CLP Annex I, Table 3.1.1 refer to a 4h exposure time;

263 Guidance on the Application of the CLP Criteria 5.0 July 2017 263 Version – The acute inhalation therefore to classify a toxicity was studied in rats substance, existing inhalation compliant study in a GLP toxicity data generated from - performed in principle hour exposure should be - 1 according to OECD TG 403 converted accordingly: LC 50 in rats, but with respect values with 1h have to be converted by dividing by 4 h - for transport only with 1 - (1 h) exposure. The LC (Haber’s rule/law, dusts and 50 mists) of 3 mg/l was calculated. 4) (4 - h) = (LC (1 - h) LC : 50 50 : 4) = 0.75 mg/l, = (3 mg/l thus Category 3 classification is warranted according to CLP Annex I, Table 3.1.1. Remarks none 3.1.5.1.7. Example 7: 2,3 - Dichloropropene Application Discrimination from STOT SE - Test Data Classification Rationale Classification according to Category 3 oral Animal data: Available and Category 3 information criteria for acute inhalation - , rat 250 - 320 Oral LD 50 and oral toxicity in CLP inhalation mg/kg bw (assumption: Annex I, Table 3.1.1. results from different tests; lowest LD is valid) 50 rat 2.3 Inhalation LC - 50 mg/l/4h (vapour) Observations: extensive liver and kidney damage following oral and inhalation exposure to lethal doses (insufficient information) Remarks The substance is classified for acute toxicity and not for STOT - SE, since the observed organ toxicity is clearly the cause of the lethality

264 Guidance on the Appl ication of the CLP Criteria 5.0 – July 2017 264 Version Example 8 3.1.5.1.8. Route - to - route extrapolation: oral to inhalation (Section 3.1.3.3.5 of this Application Guidance). Expert judgement. Test Data Extrapolated Rationale inhalation ATE/CATpE Available Animal data: information oral rat: 250 mg/kg LD 50 bw (Category 3) 100 % oral absorption 0.5 mg/l/4h a) Using the extrapolation (cATpE) formula 1 mg/kg bw = assumed 0.0052 mg/l/4h: a) No specific kinetic 2.6 mg/l/4h 250 x 0.0052 mg/l/4h = 1.3 (ATE) information  Category 2 mg/l/4h according to CLP Annex I, b) Robust kinetic information Table 3.1.2 allows the conclusion that only 50% is absorbed due to b)Based on the 50% . an exhalation rate of 50 % inhalation absorption rate the equivalent ATE would be 2.6 Category 3 (2 x 1.3)  ccording to CLP Annex I, a Table 3.1.2 Remarks Robust kinetic and other information would allow the use of directly derived ATEs in the additivity formulae by expert judgement Example 9 3.1.5.1.9. Route - to - route extrapolation: oral to dermal (Section 3.1.3.3.5 of this Application . Guidance). Expert judgement Extrapolated Test Data Rationale dermal ATE/cATpE Animal data: Available information rat oral: 270 mg/kg LD 50 bw; 100 % oral absorption assumed 300 mg/kg bw a) Based on the assumption of sorption the 100% dermal ab a) Assumed dermal converted dermal ATE will be absorption rate: 100% LD dermal 50 derived by using Table 3.1.2 b) Dermal absorption rate 1080 mg/kg for Category 3  300 mg/kg based on robust kinetic/SAR bw bw as cATpE. information: 25% b) Since dermal absorption is only 25%, the dermal ATE has to be accordingly increased  4x270 mg/kg bw = 1080 mg/kg bw. This is regarded as an e quivalent ATE which can be directly used in the additivity formulae.

265 Guidance on the Application of the CLP Criteria – Version 265 5.0 July 2017 Robust kinetic and other information would allow the use of directly derived ATEs Remarks in the additivity formulae by expert judgement Examples of substances not fulfilling the criteria for classification 3.1.5.2. Example 10 3.1.5.2.1. Available data are of different quality. Expert judgement. WoE . Application Rationale Test Data Classification With 3 different available No classification Available A liquid information values a validity check Animal data: proved that the study with LC = 19 mg/l is not fully Three studies for acute 50 valid in contrast to the two inhalation toxicity (vapour) others; thus in a weight of in rats are described. Two evidence approach it is studies were performed in = concluded that the LC 50 accordance with test The ATE > 20 mg/l/4h. guideline 403 and were criteria for Category 4 are compliant. One study - GLP not fulfilled. has deficiencies with respect to study methodology and description of study performance and documentation of the test results; no GLP - pliance. The LC were com 50 as follows: LC50: 19 mg/l/4h (no GLP) – – LC50: 23 mg/l/4h (TG 403, GLP) – LC50: 28 mg/l/4h (TG 403, GLP) Remarks none

266 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 266 3.1.5.3. criteria for classification Example of mixtures fulfilling the Example 11 3.1.5.3.1. Application of the ‘Relevant ingredient’ (CLP Annex I, 3.1.3.3 (a)) and ‘Generic Application cut - off values to be taken into account’ concepts (CLP Annex I, Table 1.1) for CLP Annex I, 3.1.3.6.2.3. mixtures with data gaps using the equation in For dermal and inhalation routes, there is no acute toxicity data available for ingredients 2 and 4. For ingredients 1, 3 and 5 the data indicates no classification for acute toxicity. Test Data Classification Rationale (ingredient) Animal data (oral Available information rat): Ingredient 1 Apply the equation in CLP Annex I, 3.1.3.6.2.3: LD : 125 Oral Category 3 50 (4%) mg/kg bw C   ( 10 if 100 %) C  unknown i   - Ingredient 2 ATE ATE n mix i No data available (92%)  0 . 3 2 4 92 100 : LD Ingredient 3 Oral Category 4 50     (3%) 1500 mg/kg bw 1500 10 125 ATE mix No data available - Ingredient 4 0 002 . 0 032 . 0    054 . 0 02 . = (0.9%) mg/kg bw ATEmix = 148 Oral Category 2 Ingredient 5 LD : 10 50  Category 3 (0.2%) mg/kg bw Rationale for classification of the mixture in Category 3: Remarks 1. Classification via application of substance criteria is not possible since acute toxicity test data was not available for the complete mixture (CLP Annex I, 3.1.3.4). 2. Classification via the application of bridging principles is not possible since data on a similar mixture was not available (CLP Annex I, 3.1.3.5.1). 3. Classification based on ingredient data for the mixture can be considered (CLP Annex I, 3.1.3.6). 4. Applying the ‘relevant ingredients’ concept from CLP Annex I, 3.1.3.3 (a) means that Ingredient 4 is excluded from the ATE calculation since its concentration is < 1%. The mix same reasoning cannot apply to Ingredient 5, though its concentration is below the ’ threshold of 1% but it is higher than the cut ‘relevant ingredients off value of 0.1% for a - Category 2 ingredient in CLP Annex I, Table 1.1. 5. The total concentration of ingredients with unknown acute toxicity (i.e., Ingredient 2) is 92%; therefore, the ATE equation in CL P Annex I, 3.1.3.6.2.3 must be used. This mix corrected calculation adjusts for the total percentage of the ingredient with unknown acute toxicity. 6. Ingredients 1, 3 and 5 are included in the ATE calculation because they have data mix that fall within a CLP a cute toxicity category, CLP Annex I, 3.1.3.6.1 (a). 7. Applying the guidance in Note (b) to CLP Annex I, Table 3.1.1 results in using the data for Ingredients 1, 3 & 5 in the ATE calculation since data is available. actual LD 50 mix Additional Labelling : ‘92 % of the mixture consists of components of unknown acute toxicity.’ (See S ection 3.1.4.2 of this guidance)

267 Guidance on the Application of the CLP Criteria – July 2017 267 Version 5.0 Example 12a 3.1.5.3.2. xamples 12a and 12b assume that it is known that only one physical form (i.e. mist in E Note: example 12a and vapour in example 12b) can occur during any reasonably expected use of the mixture including when the mixture is used to produce a new mixture. This would need to be ed. If toxicity data for more than one form is used, the converted ATE value has to be used justifi even if an ATE value is available, according to these examples. Different phases in inhalation exposure. Extrapolation . Application Rationale Test Data Classification Use/exposure as aerosol Available information (mist) Animal data (rat): LC (mg/L/4 h) 50 Ingredient 1 Category 4 Conv. ATE (mg/L/4 h) = 1.5 mg/L/4 h solid (6%) Ingredient 2 0.6 Category 3 ATE = LC 50 solid (11%) Neglected, since not classified 6 Ingredient 3 (dust) - solid (10%) in any acute category Ingredient 4 Conv. ATE (mg/L/4 h) = 1.5 11 (vapour) Category 4 mg/L/4 h, assuming identical liquid (40%) category for vapour and mist by expert judgement neglected - Ingredient 5 Water; (33%) Remarks Classification: Category 4 No test data available for the whole mixture. Bridging principles not applicable since no test data on similar mixtures available. Classification therefore based on ingredients. Use additivity formula in Annex I, 3.1.3.6.1, as information is available for all ingredients. 100/ATE = (6/1.5) + (11/0.6) + 0 + (40/1.5) + 0 = 49 mix ATE  = 2.04 mg/L/4 h  Category 4 mix NOTE: The mixture of Example 12a has to be classified formally in Category 4 with respect to inhalation toxicity. It is notable that this classification is only derived from the calculation for the aerosol phase, not for the vapour phase. 3.1.5.4. Examples of mixtures not fulfilling the criteria for classification 3.1.5.4.1. Example 12b See Note under example 12a.

268 ication of the CLP Criteria Guidance on the Appl 268 Version 5.0 – July 2017 Application . Different phases in inhalation exposure. Extrapolation Rationale Test Data Classification Available Use/exposure as vapour information Animal data (rat): (mg/L/4 h) LC 50 A solid with no sublimation, Ingredient 1 Category 4 solid (6%) therefore not present in the vapour phase; neglected 0.6 (dust) Category 3 As Ingredient 1 Ingredient 2 solid (11%) Neglected, since not classified 6 (dust) Ingredient 3 - in any acute category solid (10%) 4 Ingredient 4 Category ATE = LC 11 (vapour) 50 liquid (40%) Ingredient 5 Water; not relevant - (33%) Classification: NC Remarks Inhalation is appropriate route since one hazardous ingredient with appreciable vapour pressure. No test data on the whole mixture. Bridging principles not applicable since no test data on similar mixtures available. Classification is therefore based on ingredients. Use additivity formula in CLP Annex I, 3.1.3.6.1 as information is available for all ingredients. There are no contributions from ingredients 1 and 2 in the formula since the diluted solid ingredients do not sublime, and thus are not present in the vapour phase; ingredient 3 is in addition not classified in any acute toxicity category. Ingredient 5 does not sho w acute toxicity. =27.5 mg/L/4 h, which is 100/ATE = 0 + 0 + 0 + 40/11 + 0 = 3.64  ATE mix mix  NC above the upper generic concentration limit for vapour 3.1.5.5. f the application of the additivity method for mixtures for Example o acute inhalation toxicity with ingredient substances in different physical forms (gas, vapour, mist or dust). Example 1 3.1.5.5.1. 3 Application Information on acute inhalation toxicity for all ingredients Test data (LC Classification Reference Tested form 50 acute (ingredient) inhalation)

269 Guidance on the Application of the CLP Criteria 5.0 – July 2017 269 Version mist Category 2 0.19 mg/L RAC 2015 Nicotine (1.9%) 2.25 < LC Diacetyl (6%) < 5.2 BASF. 1993. Study on Category 3 vapour 50 hr] - mg/L [4 the acute inhalation toxicity LC50 of Diacetyl FCC as a vapor in rats 4 hour exposure. Project No. 1310247/927010. BASF Not acutely toxic REACH registration Propylene glycol (65%) REACH registration Not acutely toxic Glycerine (27.1%) No test information on the mixture 1. Rationale No test information on similar mixtures 2. 3. Sufficient information on all ingredients. Therefore the summation method is applicable. As the two ingredients which are acutely toxic have test data for different forms (mist and vapour), it is not clear which ATE range is applicable to the mixture. Therefore, the fraction of the acute toxicity of the mixture is calculated for each ingredient substance and category and added. When the sum of the fractions is one or higher for a category, that category is to the mixture. (See also 3.1.3.3.4) applicable For diacetyl, no LC d was derived but only a range. Therefore, the converte 50 ATE value in accordance with T able 3.1.2 was applied resulting in an ATE of 3 range. mg/L which is inside the observed LC 50 Applied formula: * concentration/100) + ((limit/ATE) x concentration/100) ((limit/ATE) mist vapour limit= the upper border of ATE values for a hazard category (Table 3.1.1., Annex I, CLP) concentration= concentration of a component tested in a state/form Category 1 is not applicable as none of the ingredients are classified as category 1. * 1.9/100 (nicotine) + (2/3) * Category 2: (0.5/0.19) 6/100 (diacetyl) = 0.05 + 0.04 = 0.09 which is below 1 meaning not category 2. Category 3: (1.0/0.19) * 1.9/100 (nicotine) + 10/3 * 6/100 (diacetyl) = 0.10 + 0.20 = 0.30 which is below 1 meaning not category 3. Category 4: (5/0.19) * 1.9/100 (nicotine) + (20/3) * 6/100 (diacetyl) = 0.50 + 0.40 = 0.90 which is below 1 meaning not category 4.

270 ication of the CLP Criteria Guidance on the Appl Version 5.0 – July 2017 270 No classification for acute toxicity by the inhalation route is warranted 3.1.6. References OECD (2009) Series on testing and assessment number 39: Guidance document on acute inhalation toxicity testing ENV/JM/MONO(2009)28 (21 July 2009). ECETOC (2003) TR 86: European Centre for Ecotoxicology and Toxicology of Chemicals, Brussels, Belgium, Technical report N°86. Pauluhn, J. (2008) Inhalation toxicology: methodological and regulatory challenges. Exp Toxicol Pathol. 60 (2 - 3):111 - 24.

271 Guidance on the Application of the CLP Criteria 5.0 – July 2017 271 Version ATION 3.2. SKIN CORROSION/IRRIT Definitions for classification for skin co rrosion/irritation 3.2.1. Skin Corrosion means the production of irreversible damage to the skin; Annex I: 3.2.1.1. namely, visible necrosis through the epidermis and into the dermis, following the application of a test substance for up to 4 hours. Corrosive reac tions are typified by ulcers, bleeding, bloody scabs, and, by the end of observation at 14 days, by discolouration due to blanching of the skin, complete areas of alopecia, and scars. Histopathology shall be considered to evaluate questionable lesions. Ski means the production of reversible damage to the skin following the application n Irritation of a test substance for up to 4 hours. Classification of substances for skin corrosion/irritation 3.2.2. 3.2.2.1. Identification of hazard information Identification of human 3.2.2.1.1. data CLP Article 7(3) specifies that testing on humans is not allowed for the purposes of CLP; however it does acknowledge that existing human data obtained from other sources can be used for classification purposes. Human data may be retrieved from a numb er of sources, e.g. epidemiological studies, clinical studies, well - documented case reports, poison information units and accident databases or occupational experience. In this context the quality and relevance of existing human data for hazard assessment should be critically reviewed. There may be a significant level of uncertainty in human data due to poor reporting and lack of specific information on exposure. Diagnosis confirmed by expert physicians may be missing. Confounding factors may not have been accounted for. Small group sizes may flaw the statistical strength of evidence. Many other factors may compromise the validity of human data. In clinical studies (e.g. for diagnostic purposes) the selection of individuals and the control groups must be car efully considered. A critical review of the value of human studies is IR&CSA Section R.4.3.3 and more specific considerations for skin provided in the Guidance on IR&CSA corrosion/irritation are given in the Guidance on Section R.7.2.4.2. Data indicates th at human skin is, in most cases, less sensitive than the skin of rabbits (ECETOC, 2002). 3.2.2.1.2. Identification of non human data Non human data include physico - chemical properties, results from (Q)SARs and models based on combinations of (Q)SARs and databases (ex pert systems), and results from in vitro and in vivo tests. Available skin corrosion/irritation information on substances may include existing data generated by the test methods in the Test Methods Regulation (Commission Regulation (EC) No 440/2008) or by methods based on internationally recognised scientific principles. Before using the non - testing methods as referred to in the following sections, it should be checked whether the methods are sufficiently validated (or considered valid in case of (Q)SAR an d expert systems) against the criteria for classification according to CLP (and not validated against the old DSD criteria which differed slightly from the CLP criteria). 3.2.2.1.2.1. Consideration of physico - chemical properties Substances with oxidising properties can give rise to highly exothermic reactions in contact with other substances and human tissue. High temperatures thus generated may damage/destroy

272 ication of the CLP Criteria Guidance on the Appl 272 Version 5.0 – July 2017 biological materials. This applies, for example, to organic peroxides, which can be assumed to IR&CSA Section R.7.2.3.1). be skin irritant s, unless evidence suggests otherwise (Guidance on Thus, in the absence of evidence to the contrary, classification as Skin Irritation Category 2 would should be considered for peroxides, whereas the classification for a hydroperoxide normally be Skin Corrosive Category 1. Appropriate evidence must be provided in order to consider no classification of substances with oxidising properties. pH and acid/alkaline reserve 3.2.2.1.2.2. 3.2.2.2.5. Likewise, pH extremes like ≤ 2 and ≥ 11,5 may indicate the potential to Annex I: cause skin effects, especially when associated with significant acid/alkaline reserve (buffering capacity). Generally, such substances are expected to produce significant effects on the skin. ation, a substance is considered as corrosive to skin (Skin In the absence of any other inform Corrosion Category 1) if it has a pH ≤ 2 or a pH ≥ 11,5. However, if consideration of alkali/acid reserve suggests the substance may not be corrosive despite the low or high pH o be confirmed by other data, preferably by data from an appropriate value, this needs t validated in vitro test. Prediction of skin corrosivity based on pH extremes shows a very high specificity ( ˃ 90%) and IR&CSA therefore a low number of false positives (R.7.2.4.1, guidance). The acid/alkaline reserve is a measure of the buffering capacity of chemicals. For details of the methodology, see et al Young , 1988, and Young and How, 1994. The higher the buffer capacity, the higher in general the potential for corrosivity. 3.2.2.1.2.3. Non - testing methods: (Q)SARs and expert systems testing methods such as (Q)SARs and expert systems (a diverse group of models Non - consisting of combinations of SARs, QSARs and databases) may be considered on a case - by - tructures in the substance that are considered to reflect case basis. Structural alerts are subs some kind of chemical or biochemical reactivity that underlies the toxicological effect. The occurrence of a structural alert for a substance suggests the presence of an effect, based on the notion that structural analogues that have exhibited corrosion (or irritation) potential can be used to predict a corrosive or irritant effect for the substance of interest, or to tailor further testing and assessment. The absence of one of the known structural a lerts for irritation and corrosion alone does not prove absence of effect, as knowledge of structural alerts for irritation and corrosion might be incomplete. (Q)SAR systems that also account for skin effects are for example ACD Percepta, Hazard Expert, C - based ASE Ultra, Discovery studio Acellrys (former TOPKAT). Derek Nexus is a knowledge expert system that gives toxicity predictions. These systems go beyond the structural similarity etry and surface considerations encompassing also other parameters such as topology, geom properties. Not all of the models were developed with EU regulatory purposes in mind, so it is important to assess in each case whether the endpoint or effect being predicted corresponds to the regulatory endpoint of interest. 53 ystem BfR - DSS The expert s has been recommended in the Guidance on IR&CSA Section R.7.2.4 since there is no other model that sufficiently describes the absence of effects. currently The BfR rules to predict skin irritation and corrosion have been integrated in the int ernet tool ‘toxtree’, https://eurl - ecvam.jrc.ec.europa.eu/laboratories - research/predictive_toxicology/qsar_tools/toxtree . The BfR alerts (“inclusion rules”) for corrosion and irritation have also been incorporated into the OECD QSAR Toolbox http://www.qsartoolbox.org/ ). ( 53 Decision Support System (DSS) developed by the German Federal Institute for Risk Assessment (BfR) to assess certain hazardous properties of pure chemicals.

273 Guidance on the Application of the CLP Criteria 5.0 – July 2017 273 Version the e ffect can be presence of In the absence of any other existing data, conclusion on an reached if the (Q)SAR or expert system has been shown to adequately predict the presence of the classified effect. In case of negative (Q)SAR data the need for classification cannot be excluded. or in vivo If existing other data (e.g. data) contradicts these conclusions on the in vitro presence or absence of an effect then a weight of evidence approach must be applied. The suitability of the model (reliability, relevance) should be very carefully checked to make sure that the prediction is fit for purpose, and the applicability of the model to the substance should also be justified. Since a formal adoption procedure for the non testing methods (as mentioned above) is not - foreseen and no formal validation process is in place, appropriate document ation is very important. In order to achieve acceptance under REACH the documentation must conform the - called QSAR Model Reporting Format (QMRF). For more details consult the Guidance on so IR&CSA Section R.6.1. Testing methods: methods 3.2.2.1.2.4. in vitro - IR&CSA 2 lists the status of validation and regulatory 2 in the Guidance on Table R.7. test methods for skin corrosion and skin irritation. The information given in vitro acceptance for below is current at the time of publication, however further information on newly adopted OECD Test Guidelines can be found on the OECD website http://www.oecd.org/env/chemicalsafetyandbiosafety/testingofche ( micals/oecdguidelinesforthet estingofchemicals.htm ). Furthermore, up to date information on OECD and EU test guidelines https://www.echa.europa .eu/support/oecd - can be found also on the ECHA website ( - test - eu guidelines ). In vitro methods for skin corrosion The OECD has accepted guidelines for skin corrosion tests as alternatives for the in vitro in vivo in vitro tests for skin corrosiv ity are standard rabbit skin test (OECD TG 404). Accepted found in the EU Test Methods Regulation (EC) No 440/2008 and in OECD Test Guidelines (OECD TG):  The transcutaneous electrical resistance (TER; using rat skin) test (OECD TG 430 / TM B.40) Reconstructed human epidermis (RHE) tests (OECD TG 431 / TM B.40 bis)   The in vitro membrane barrier test method (OECD TG 435) in vitro results on corrosivity do not generally require further testing and can be used Positive in vitro for classification. Negative corrosivity responses must be subject to further evaluation. Whereas the TER test at present does not allow subcategorisation within the corrosive category, the membrane barrier test allows for the differentiation into the three Categories 1A, 1B and 1C. The reconstructed human epidermis (RHE) models included in the OECD TG 431 i.e. ® TM TM TM support the sub , SkinEthic RHE and epiSC SCT, Episkin - categorisation into EpiDerm Category 1A, however they cannot discriminate between Categories 1B and 1C. The applica bility domain of the three tests outlined here (TER - , RHE - and membrane barrier test) with regard to the alkalinity and acidity of the tested substance should be carefully considered to decide which test(s) are most appropriate for the actual substance. Th e TER and the RHE assays have been validated for the classification of skin corrosion. The results of this validation are well founded, because the CLP criteria for skin corrosion are identical with the ones referred to in the past validation study. The me mbrane barrier method has been endorsed as a scientifically validated test for a limited range of substances – mainly acids, bases and their derivatives (ECVAM/ESAC, 2000). In vitro methods for skin irritation

274 ication of the CLP Criteria Guidance on the Appl 274 Version 5.0 – July 2017 The OECD has adopted an skin irritation test guideline i.e. OECD TG 439 (TM B. 46) that in vitro TM TM TM RHE and LabCyte , SkinEthic currently contains four test methods i.e. EpiDerm SIT, EpiSkin – MODEL24 SIT. These test methods can reliably distinguish non - classified from classified EPI s ubstances but cannot distinguish between corrosives and irritants when used alone. Thus, in the case of positive results, the potential corrosive properties should be excluded or confirmed in vitro skin corrosion test. It sho uld be noted that conclusions based on data obtained from an on the applicability domain of the four methods rest mainly on the optimisation and validation data set. All four methods are valid for the classification of substances for skin irritation according to CLP criteria. n on the current developments of in vitro tests and methodology can be found on the Informatio - http://ihcp.jrc.ec.europa.eu/our_labs/eurl ). ECVAM website ( ecvam Other suitable in vitro methods e data from other suitable Positiv methods may be used in a weight of evidence in vitro approach to determine classification as irritant, while negative data are not conclusive for no classification. In this context ‘suitable’ means sufficiently well developed acc ording to - internationally agreed development criteria (see REACH Annex XI, section 1.4). Testing methods: 3.2.2.1.2.5. In vivo data The in vivo test in rabbits according to OECD TG 404 (TM B.4) is the standard in vivo test for r, according to Annex VIII REACH (at or above the hazard assessment under REACH. Howeve of /year ) an in vivo test should only be performed in case the in vitro studies (as 10 tonnes required in Annex VII) are not applicable or the results of these studies are not adequate for classification. l 1987 the OECD standard protocol used occlusive patching for the application of the test Unti substance, which resulted in more rigorous test conditions compared to the semi - occlusive patching used today. Especially in borderline cases of classification the me thod of application should be accounted for in the evaluation of effects. Studies performed according to the USA Federal Hazardous Substances Act (US - FHSA), may be used for classification purposes although they deviate in their study protocol from the OECD TG - - hour test material 404. They do not include a 48 hour observation time and involve a 24 exposure followed by observations at 24 hour and 72 hours. Moreover, the test material is are terminated after patched both on abraded and on intact skin of six rabbits. Studies usually 72 hours. In case of no or minimal responses persisting until the 72 hours time points it is feasible to use such data for classification by calculating the mean values for erythema and oedema on the basis of only the 24 and 72 hours time points. Calculation of mean scores should normally be restricted to the results obtained from intact skin. In case of pronounced responses at the 72 hours time point an expert judgement is needed as to whether the data is appropriate for classificatio n. Data on skin effects on animals may be available from tests that were conducted for other primary purposes than the investigation of skin corrosion / irritation. Such information may be gained from acute or repeated dose dermal toxicity studies on rabbi ts or rats (OECD TG 402; OECD TG 410), guinea pig skin sensitisation studies (OECD TG 406) and from irritation studies in hairless mice.

275 Guidance on the Application of the CLP Criteria 5.0 – July 2017 275 Version Classification criteria 3.2.2.2. Skin corrosion .2.2.1.1. Annex I: 3 Annex I: A substance is corrosive to skin when it produces destruction of skin 3.2.2.1.1.1. tissue, namely, visible necrosis through the epidermis and into the dermis in at least one tested animal after exposure for up to 4 hours. 3.2.2.1.1.2. re data is Annex I: Corrosive substances shall be classified in Category 1 whe categorisation. - not sufficient for sub When data are sufficient substances shall be classified in one of the Annex I: 3.2.2.1.1.3. categories 1A, 1B, or 1C in accordance with the criteria in Table 3.2.1. - three sub Annex I: Three sub - categories are provided within the corrosion category: sub - 3.2.2.1.1.4. category 1A – where corrosive responses are noted following up to 3 minutes exposure and - category 1B – where corrosive responses are described up to 1 hour observation; sub reater than 3 minutes and up to 1 following exposure g hour and observations up to 14 days; and sub category 1C – - where corrosive responses occur after exposures greater than 1 hour and up to 4 hours and observations up to 14 days. Table 3.2.1 Skin corrosion category and subca tegories Category Criteria 1 Category 1 Destruction of skin tissue, namely, visible necrosis through the epidermis and into the dermis, in at least one tested animal after exposure ≤ 4 h Category 1A Sub following exposure ≤ 3 - Corrosive responses in at least one animal min during an observation period ≤ 1 h - Sub Corrosive responses in at least one animal following exposure > 3 Category 1B min and ≤ 1 h ≤ 14 days and observations - Category 1C Sub Corrosive responses in at least one animal after expo sures > 1 h and ≤ 4 h and observations ≤ 14 days 1 Category 1 in paragraph (a) of S ection 3.2.2. See the conditions for the use of Annex I: 3.2.2.1.2. Skin irritation 3.2.2.1.2.1. Annex I: to A substance is irritant to skin when it produces reversible damage the skin following its application for up to 4 hours. The major criterion for the irritation category is that at least 2 of 3 tested animals have a mean score of ≥ 2.3 and ≤ 4.0. Annex I: 3.2.2.1.2.2. A single irritation category (Category 2) is presen ted in Table 3.2.2, using the results of animal testing. Annex I: 3.2.2.1.2.3. Reversibility of skin lesions is also considered in evaluating irritant responses. When inflammation persists to the end of the observation period in 2 or more test animals, tak ing into consideration alopecia (limited area), hyperkeratosis, hyperplasia and scaling, then a material shall be considered to be an irritant. Annex I: 3.2.2.1.2.4. Animal irritant responses within a test can be variable, as they are with corrosion. A sep arate irritant criterion accommodates cases when there is a significant irritant response but less than the mean score criterion for a positive test. For example, a test material might be designated as an irritant if at least 1 of 3 tested animals shows a very

276 ication of the CLP Criteria Guidance on the Appl 276 Version 5.0 – July 2017 elevated mean score throughout the study, including lesions persisting at the end of an observation period of normally 14 days. Other responses could also fulfil this criterion. emical exposure. However, it should be ascertained that the responses are the result of ch Table 3.2.2 a Skin irritation category Category Criteria ≤ 4,0 for erythema/eschar or for oedema in at Irritation (1) Mean score of ≥ 2,3 - least 2 of 3 tested animals from gradings at 24, 48 and 72 hours after (Category patch removal or, if reactions are delayed, from grades on 3 consecutive 2) days after the onset of skin reactions; or Inflammation that persists to the end of the observation period normally (2) 14 days in at least 2 animals, particularly taking into account alopecia ea), hyperkeratosis, hyperplasia, and scaling; or (limited ar In some cases where there is pronounced variability of response among (3) animals, with very definite positive effects related to chemical exposure in a single animal but less than the criteria above. a) Gradi ng criteria are understood as described in Regulation (EC) No 440/2008. 3.2.2.3. Evaluation of hazard information Annex I: 3.2.2.2.1. A tiered approach to the evaluation of initial information shall be elements may be relevant. considered, where applicable, recognising that not all 3.2.2.2.7. Annex I: The tiered approach provides guidance on how to organize existing information on a substance and to make a weight of evidence decision about hazard assessment and hazard classification. Although information might be gained from the evaluation of single parameters within a tier (see S ection 3.2.2.2.1), consideration shall be given to the totality of existing information and making an overall weight of evidence determination. This is especially true when there is conflict in information available on some parameters. The tiered approach for the evalution of the information applied in order to make a decision about the skin corrosion/skin irritation hazard pr operties is illustrated in Figu re 3 . 1 below. The approach in the figure was adopted by the UNSCEGHS in December 2012 (with exception of the added footnotes g) and h)).

277 Guidance on the Application of the CLP Criteria 277 – July 2017 5.0 Version . 1 Tiered evaluation for skin corrosion/skin irritation re Figu 3 Parameter Finding Conclusion Step Existing human or animal 1a: skin corrosion/irritation Classify as skin Skin corrosive a data b corrosive Not corrosive/Insufficient/Inco nclusive/No data 1b: Existing human or animal skin corrosion/irritation Skin irritant Classify as skin a data g irritant Not irritant/Inconclusive Insufficient//No data Existing human or animal 1c: skin corrosion/irritation g Not skin corrosive or Not classified a data skin irritant No/Inconclusive Insufficient/ data Other, existing skin data in 2: Yes; other existing May be deemed to be c b data showing that or skin corrosive animals g substance may cause skin irritant

278 ication of the CLP Criteria Guidance on the Appl 2 Version 5.0 – July 2017 78 Finding Conclusion Step Parameter skin corrosion or skin irritation No/Negative/ Insufficient/Inconclusive data ex vivo/in vitro Classify as skin 3: Positive: Skin Existing d b corrosive corrosivity data corrosive No/Negative/ Insufficient/Inconclusive data Classify as skin Positive: Skin irritant ex vivo/in vitro Existing g irritant irritation data g Negative: not skin Not classified irritant No/ Insufficient/Inconclusive data i 11.5 pH - based assessment (with Classify as skin 4: pH ≤ 2 or ≥ g with high consideration of corrosive acid/alkaline reserve acid/alkaline reserve of the e or no data for chemical) acid/alkaline reserve

279 Guidance on the Application of the CLP Criteria 5.0 – July 2017 279 Version Finding Step Conclusion Parameter Not pH extreme, no pH data or extreme pH with data showing low/no h acid/alkaline reserve 5: Validated Structure Activity Deemed to be skin Skin corrosive b Relationship (SAR) methods corrosive Skin irritant Deemed to be skin irritant No/Inconclusive Insufficient/data 6: Consideration of the total Deemed to be skin Skin corrosive b f corrosive weight of evidence Skin irritant Deemed to be skin irritant Not classified 7: (a) Existing human or animal data could be derived from single or repeated exposure(s), for example in - generated occupational, consumer, transport or emergency response scenarios; or from purposely data from animal studies conducted according to validated and internationally accepted test methods. Although human data from accident or poison centre databases can provide evidence for classification, absence of incidents is not itself evidence for no classification as exposures are generally unknown or uncertain. (b) Classify in the appropriate category/sub - category, as applicable. (c) All existing animal data should be carefully reviewed to determine if sufficient skin corrosion/irritation evidence is available. In evaluating such data, however, the reviewer shoul d bear in mind that the reporting of dermal lesions may be incomplete, testing and observations may be made on a species other than the rabbit, and species may differ in sensitivity in their responses. (d) Evidence from studies using validated protocols wi th isolated human/animal tissues or other, non - tissue - based, though validated, protocols should be assessed.

280 ication of the CLP Criteria Guidance on the Appl 280 Version 5.0 – July 2017 (e) Measurement of pH alone may be adequate, but assessment of acid or alkali reserve (buffering capacity) would be preferable. All information that is available should be considered and an overall determination made on the total (f) weight of evidence. This is especially true when there is conflict in information available on some parameters. Expert judgment should be exercised prior to m aking such a determination. Negative results from applicable validated skin corrosion/irritation in vitro tests are considered in the total weight of evidence evaluation. (g) In case there is a conflict in available data, e.g. negative/irritation human dat a but positive/corrosive in vitro data, a weight of evidence assessment should be performed, see footnote f. (This footnote was not included in the figure in the 5th rev of GHS, but is based on 3.2.1.2. and 3.2.2.2.7, Annex I, CLP ). eds to be confirmed by other data and preferably by data from an appropriate (h) Non corrosivity ne th rev of GHS, but is based validated in vitro test. (This footnote was not included in the figure in the 5 on 3.2.2.2.5, Annex I, CLP). For the case of mixtures with no human or animal data on skin corrosion/irritation but with extreme (i) Figure 3 . 3 in 3.2.3.2.1.1. pH see 3.2.2.3.1. Evaluation of human data The usefulness of human data for classification purposes will depend on the extent to which the effect, and its magnitude, can be reliably attributed to the substance of interest and on the extent and duration of the exposure. Further guidance on evaluation of human data for skin corr osion/irritation can be found in the Guidance on IR&CSA Section R.7.2.4.2. The criteria in CLP Annex I, Tables 3.2.1 and 3.2.2 are not applicable to human data. Evaluation of non human data 3.2.2.3.2. In vitro data 3.2.2.3.2.1. tests the appli cability domain has to be taken into account. in vitro In evaluation of data from in vitro membrane barrier test method is mainly applicable for acids and bases For instance, the and is not applicable for solutions with pH values between 4.5 and 8. Normally, recommendations for classificatio in vitro n according to GHS criteria based on the results of an test are mentioned in the corresponding OECD test guideline. In particular OECD TG 431 concludes that some results fall in the category 1B/1C. Category 1B/1C is not an option in CLP. However, a WoE assessment may lead to a conclusion about the subcategory but if this is not 54 . the case, category 1 should be assigned 3.2.2.3.2.2. In vivo data Tests in albino rabbits (OECD TG 404) Evaluation criteria for local effects on the skin are severity of the damage and reversibility . For the of damage the responses are evaluated according to the Draize score ranking severity from ‘0’ (‘no response’) up to ‘4’ (‘severe response’). Evaluation takes place separately for erythema and oedema. Reversibility of skin lesions is the other decisive factor in evaluating responses in the animal test. The criteria are fulfilled if, for  corrosion 54 Please, note that the issue concerning the subcategorization of skin corrosivity is currently u n der discussion .

281 Guidance on the Application of the CLP Criteria 5.0 – July 2017 281 Version o the full thickness of the skin is destroyed resulting in ulcers, bleeding, bloody scabs discoloration, complete areas of alopecia and scars. In questionable cases a pathologist should be consulted. One animal showing this response at the end of the observation period is sufficient for the classification as corrosive. irritation  a limited degree of alopecia, hyperkeratosis, hyperplasia and scaling occurs. Two o animals showing this response are sufficient for the classification as irritant. o very elevated mean scores throughout the study are revealed, including lesions persisting at the end of an observation period of normally 1 4 days. One animal showing this response throughout and at the end of the observation period is sufficient for the classification as irritant (In cases of suspected corrosives, existing test data may only be available for one animal due to testing restrict ions, see Example 2.). With regard to severity the main criterion for classification of a substance as irritant to skin, is the mean score per animal for either erythema/eschar or oedema. During the observation period following the removal of the patch eac h animal is scored on erythema and oedema. For each of the three test animals the average scores for three consecutive days (usually 24, 48 and 72 hours) are calculated separately for oedema and erythema. If 2/3 animals exceed the the CLP, the classification has to be done accordingly. - values defined in cut - off With regard to reversibility the test report must prove that these effects are transient i.e. the affected sites are repaired within the observation period of the test (see Example 1). - Non tion as corrosive can only be justified if the test was performed with at least three classifica animals and the test results were negative for all three animals. Tests that have been conducted with more than three animals Current guidelines foresee a sequential test ing of rabbits until a response is confirmed. Typically, up to 3 rabbits may be used. The basis for a positive response is the individual rabbit value averaged over days 1, 2, and 3. The mean score for each individual animal is used as a criterion for clas sification. Skin Irritation Category 2 is used if at least 2 animals show a mean score of 2.3 or above. Other test methods, however, have used up to 6 rabbits. This is also the case for the studies performed according to the US - FSHA. with more than three animals, specific guidance needs to be applied For existing test data (adopted by the UNSCEGHS in June 2011): The average score is determi ned per animal (see Example 3, S 3.2.5.1.3 ). ection the case of 6 In the following applies: rabbits a. Classification as skin corrosive – Category 1 if destruction of skin tissue (visible necrosis through the epidermis and into the dermis) occurs in at least one animal after exposure up to 4 hours. b. Classification as skin irritant – Category 2 if at least 4 out of 6 rabbits show a mean score per animal of  2.3 ≤ 4.0 for erythema/eschar or for oedema; the case of In rabbits the following applies: 5 a. Classification as ski n corrosive – Category 1 if destruction of skin tissue (visible necrosis through the epidermis and into the dermis) occurs in at least one animal after exposure up to 4 hours. b. Classification as skin irritant – Category 2 if at least 3 out of 5 rabbits show a mean score per animal of  2.3 ≤ 4.0 for erythema/eschar or for oedema;

282 Guidance on the Appl ication of the CLP Criteria 282 5.0 – July 2017 Version the 4 rabbits the following applies: In case of Classification as skin corrosive Category 1 if destruction of skin tissue (visible a. – dermis) occurs in at least one animal necrosis through the epidermis and into the after exposure up to 4 hours. Classification as skin irritant – Category 2 if at least 3 out of 4 rabbits show a mean b.  2.3 ≤ 4.0 for erythema/eschar or for oedema; score per animal of o Other dermal tests n animals levant data may also be available from animal studies that were conducted for other primary Re purposes than the investigation of skin corrosion/irritation. For example, in line wi ection th S 3.2.2.2.3 of Annex I to CLP, acute dermal toxicity data may be used for classification as skin corrosion/irritation. However, due to the different protocols and the interspecies differences in evaluated on a case - by - case basis. sensitivity, the use of such data in general needs to be These are considered significant if the effects seen are comparable to those described above. If the substance is proven to be either an irritant or a corrosive in an acute dermal toxicity test carried out with rabbits with the undiluted test substance (liquids) or with a suitable suspension (solids), the following applies. In case of signs of skin corrosion, classify as Skin Corrosive estimate (subcategorisation as 1A, 1B or 1C, where possible). In all other cases: calculate or 2 and compare this to the test substance concentration of the amount of test substance per cm 2 80 μl or 80 mg/cm employed in the EU B.4/OECD TG 404 for dermal corrosion/irritation test with rabbits. If in the same range and adequate scoring of skin effects is provided, classify or not as Skin Irritant Category 2. If not in the same range and inadequate scoring of skin effects, use the data in a Weight - Evidence analysis and proceed. of - In case the test was performed in other species, which may be le ss sensitive (e.g. rat), evaluation must be made with caution. Usually, the rat is the preferred species for toxicity studies within the EU. The limit dose level of 2000 mg/kg bw of a solid is normally applied as a g bw onto a skin surface area of about 5x5 cm. 50% suspension in a dose volume of 4 ml/k Assuming a mean body weight of 250 g, a dose of 1 ml of the suspension will be applied to an 2 2 area of 25 cm . In case of an undiluted liquid, 0.5 ml is , i.e 20 mg test substance per cm 2 , i.e. 20 μl/cm2. Considering the fact that (i) the rat skin is less sensitive applied to 25 cm compared to rabbit skin, (ii) much lower exposures are employed and (iii), in general, the scoring of dermal effects is performed less accurately, the results of dermal toxicity testin g in rats will not be adequate for classification with respect to skin irritation. Only in case of evidence of skin corrosivity in the rat dermal toxicity test can the test substance be classified as Skin Corrosive Category 1. All other data should be used in a Weight of Evidence. Regarding data from skin sensitisation studies, the skin of guinea pigs is less sensitive than that of rats which is, in turn, less sensitive than that of rabbits. Only in the case of evidence of skin is corrosivity in the sensit isation test (Maximisation or Buhler) with the neat material or dilutions of solids in water, physiological saline or vegetable oil, should the test substance be classified as Skin Corrosive Category 1. However, care should be exercised when interpreting f indings from guinea pig studies, particularly from maximisation protocols, as intradermal injection with adjuvant readily causes necrosis. All other data should be used for Weight of Evidence only. Information on irritant properties from skin sensitisation tests cannot be used to conclude on a specific classification regarding acute skin irritation but may be used in a Weight of - Evidence - analysis. In general, irritation data from the Local Lymph Node Assay are not usable. The test substance is applied to th e dorsum of the ear by open topical application, and specific vehicles for enhancement of skin penetration are used. 3.2.2.3.3. Weight of evidence According to Article 9(1) CLP, the criteria should be applied to available data. However, rward or simple to apply the criteria and according to Article 9(3) a sometimes it is not straightfo

283 Guidance on the Application of the CLP Criteria 5.0 – July 2017 283 Version weight of evidence and expert judgement should be applied in such cases when the criteria cannot be applied directly. A weight of evidence determination means that all available and scie ntifically justified information bearing on the determination of hazard is considered together, such as physico - chemical parameters (e.g., pH, reserve alkalinity/acidity), information from the application of the category approach (grouping, read across), ( Q)SAR results, the results of suitable in vitro - tests, relevant animal data, skin irritation information/data on other similar mixtures, human experience such as occupational data and data from accident databases, epidemiological and ell clinical studies and w documented case reports and observations. The quality and consistency - of the data should be given appropriate weight. Both positive and negative results should be assembled together in a single weight of evidence determination P and (see 1.1.1.3, Annex I, CL S ection 1.4 in this guidance). Note that non testing methods may normally not enable subcategorsation of corrosive substances. must be performed on a case - by - case basis and with expert judgement. However, Evaluation normally positive results that are adequate for classification should not be overruled by negative findings. Annex I: 1.1.1.4. For the purpose of classification for health hazard s (Part 3) established hazardous effects seen in appropriate animal studies or from human experience that are consistent with the criteria for classification shall normally justify classification. Where d there is a conflict between the evidence is available from both humans and animals an findings, the quality and reliability of the evidence from both sources shall be evaluated in order to resolve the question of classification. Generally, adequate, reliable and representative data on humans (including epid emiological studies, scientifically valid case studies as specified in this Annex or statistically backed experience) shall have precedence over other data. However, even well - designed and conducted epidemiological studies may lack a sufficient number of s ubjects to detect relatively rare but still significant effects, to assess potentially confounding factors. Therefore, positive results from well - conducted animal studies are not necessarily negated by the lack of positive human experience but require an a ssessment of the robustness, quality and statistical power of both the human and animal data. The following Figure 3 . 2 provides an illustration of the assessment of available data, in the case of conflicting results, to decide the weight to be assigned to different types of data (see also Figu re 3 . 1 ). It needs to be noted that the relative weights indicated in the figure assume comparable quality of the data. WoE considerations need to take into account, o n a case - by - case basis, the quality, nature, relevance and applicability domain of the different types of data available. The figure illustrates a decreasing weight of the information from top to bottom. Figure 3 . 2 Simplified illustration of the relative weight of the available information

284 ication of the CLP Criteria Guidance on the Appl 284 Version 5.0 – July 2017 Existing human data Existing animal data In vitro data Other sources (e.g. (Q)SAR) Extreme pH sufficient for Skin Corr classification in absence of other data When contradicting data of comparable quality belongs to different “hierarchical levels”, the following considerations should be made: - e are positive data which belong to a higher level in the hierarchy than the When ther available negative data, more weight should normally be given to the positive data. - When the negative data belong to a level which is higher than the positive data, the full avai lable dataset should be assessed in a WoE approach (as, for example, existing good quality positive animal data could overrule negative human data and negative good data could overrule positive QSAR data). quality in vitro the relevance of the different types of information, as well as More information and guidance on 55 on quality assessment, is provided in OECD guidance no 203 and in the Guidance R.7a. For additional guidance, if both human and animal data are available, see the Guidance on IR&CSA Section R .7.2.3.2. 3.2.2.4. Decision on classification Where the comparison of the information with the criteria leads to a decision that the substance is classified as a skin corrosive but the data used for classification does not allow differentiation between the skin c orrosion subcategories 1A/1B/1C, then the substance should be assigned Skin Corrosion Category 1. 3.2.2.5. Setting of specific concentration limits Article 10(1) Specific concentration limits and generic concentration limits are limits assigned to a substance indicating a threshold at or above which the presence of that substance in another substance or in a mixture as an identified impurity, additive or individua l constituent leads to the classification of the substance or mixture as hazardous. 55 Available at ocuments/publicdisplaydocumentpdf/?cote=env/jm/mono(2014)19&doclangu http://www.oecd.org/officiald . See in particular section B, part 2, module 8. age=en

285 Guidance on the Application of the CLP Criteria 5.0 – July 2017 285 Version Specific concentration limits shall be set by the manufacturer, importer or downstream user where adequate and reliable scientific information shows that the hazard of a su bstance is evident when the substance is present at a level below the concentrations set for any hazard class in Part 2 of Annex I or below the generic concentration limits set for any hazard class in Parts 3, 4 and 5 of Annex I. [..] It is more difficult to prove the absence of a hazardous property; the legal text states that: Article 10(1) [..] In exceptional circumstances specific concentration limits may be set by the manufacturer, conclusive scientific importer or downstream user where he has adequate, reliable and information that a hazard of a substance classified as hazardous is not evident at a level above the concentrations set for the relevant hazard class in Part 2 of Annex I or above the hazard class in Parts 3, 4 and 5 of that Annex. generic concentration limits set for the relevant A specific concentration limit (SCL) set in accordance with the above mentioned provisions shall take precedence over the generic concentration limit (GCL) set out in Tables 3.2.3 and 3.2.4 of Annex I to CLP - specific and should be (Article 10(6)). Furthermore, such an SCL is substance applicable to all mixtures containing the substance instead of any GCL that otherwise would apply to a mixture containing the substance. s for setting a specific concentration limit? What type of information may be the basi - response information is available) Existing human data may in certain cases (especially if dose indicate that the threshold for the irritation hazard in humans for a substance in a mixture, lower than the GCL. A careful evaluation of the usefulness and the validity of would be higher or such human data, as well as their representativeness and predictive value ( IR&CSA , sections R.4.3.3. and R.7.2.4.2), should be performed. As pointed out in 1.1.1.4 (Annex I to C LP), positive results from well - conducted animal studies are not necessarily negated by the lack of positive human experience but require an assessment of robustness, quality and a degree of statistical certainty of both the human and animal data. 56 The aim of the standard test method for ‘Acute Dermal Irritation/Corrosion’ OECD TG 404 is to identify potential skin corrosion or irritation. The test material is generally administered undiluted, thus, no dose response relationship can be obtained from an indiv idual test. - However, if there are adequate, reliable, relevant and conclusive existing data from other already performed animal studies with a sufficient number of animals tested to ensure a high degree of certainty, and with information on dose - response relationships, such data may be considered for setting a lower or, in exceptional cases, a higher SCL on a case - by - case b asis. It should be noted that generating data specifically for the purpose of setting SCLs is not a requirement according to the CLP Regulation. Article 8(1) CLP specifies that new tests may only e or mixture) if all other means of be performed (in order to determine the hazard of a substanc generating information has been exhausted and Article 7(1) specifies that where new tests are carried out, tests on animals must be undertaken only when no other alternatives, which provide adequate reliability and quali ty of data, are possible. The GCLs must be applied for the classification of a mixture on the basis of its ingredient substances classified for skin irritation and corrosivity, if there are no already existing specific data justifying an SCL which is lower or, in exceptional cases, higher than the GCL (see Article 10(1), CLP). Therefore, information will 56 TO NOTE: In OECD TG 404 test substance refers to the test material, test article or test item. The term substance may be used diff erently from the REACH/CLP definition.

286 ication of the CLP Criteria Guidance on the Appl 286 Version 5.0 – July 2017 always be available, for mixtures containing substances already classified for skin mixture by using the GCLs corrosion/irritation, making it possible to identify the hazard for the (Article 9(4), CLP). to use in vitro test methods are being explored as a basis for setting SCLs, but The possibilities an accepted common approach is not yet available. Thus, at the present point in time, it is not possible to provide guidance for the use of in vitro methods for the purpose of setting SCLs. However, this does not exclude that a method to set SCLs based on tests could be in vitro developed in the future, as they provide a promising option for SCL setting An SCL should . apply to any mixture containing the substance instead of the GCL (that otherwise would apply to the mixture containing the substance). Thus, if the SCL is based on data derived from tests to be considered that the derived with dilutions of the substance in a specific solvent, it has concentration should be applicable to all mixtures for which the SCL should apply. Annex VI Part 3 (Table 3.1) to CLP includes e xamples of substances for which a higher or lower SCL was set under Directive 67/548/EEC (old DSD system) and which were transferred to CLP. 3.2.2.6. Decision logic for classification of substances The decision logic, which is based on the one provided in the GHS, is reported as additional guidance here below. It is strongly recommended that the person res ponsible for classification, studies the criteria for classification, as well as the guidance above, before and during use of the decision logic.

287 Guidance on the Application of the CLP Criteria 5.0 – Version 287 July 2017 No Are there data and/or information to Classification not possible skin corrosion/irritation evaluate ? Yes , Category 1 Is the substance corrosive (see criteria in CLP Annex I, 3.2.1.1, Subcategory a 3.2.2.1.1, 3.2.2.2 and Figure 3.1 in this guidance) considering : 1A, 1B or 1C Existing human data showing irreversible damage to (a) skin ; Yes ; Destruction of skin in one or more test animals (b) (c) skin corrosion after Other existing animal data indicating single or repeated exposure; (d) Existing data; ex vivo/in vitro b Danger (e) ; pH extremes of ≤2 or ≥11.5 (f) Information available from validated Structure Activity Relationship methods? No Yes Category 2 Is the substance an irritant (see criteria in CLP, Annex I, 3.2.1.1, 3.2.2.1.2, 3.2.2.2 and Figure 3.1 in this guidance) considering: (a) Existing human data, single or repeated exposure ; (b) Skin irritation data from an animal stud y ; (c) Other existing animal data including single or repeated exposure; Warning ; data (d) in vitro Existing (e) Information available from validated Structure Activity Relationship methods? No No classification a Taking into account consideration of the total weight of evidence if necessary. b Not applicable if consideration of pH and acid/alkaline reserve indicates substances may not be corrosive and test. in vitro opriate validated confirmed by other data, preferably by data from an appr 3.2.3. Classification of mixtures for skin corrosion/irritation 3.2.3.1. Identification of hazard information As for substances, the procedure for evaluating mixtures for classification purposes, is a tiered, i.e. a stepwise, approach bas principle and depending on the type and amount ed on a hierarchy of available data/information starting from evaluating existing human data on the mixture, followed by a thorough examination of the existing in vivo data, in vitro data and finally physico c hemical properties available on the mixture. (The tiered approach to evaluate data for - skin corrosion/irritation as illustrated in , should be taken into account also for re 3 . 1 Figu mixtures in case of relevant and reliable data on the complete mixture).

288 ication of the CLP Criteria Guidance on the Appl 288 Version 5.0 – July 2017 For mixtures that have been on the market for a long time, human data and exper ience may exist that may provide useful information on the skin irritation potential of the respective mixtures. Although human data from accident or poison centre databases can provide evidence for classification, absence of incidents is not itself eviden ce for no classification, as exposures ma y be unknown or uncertain. See S ection 3.2.2.1 of this Guidance for further information on the identificat ion of human data. If valid test data are available for the whole mixture they have precedence. If no such data exist, the so called bridging principles should be applied if possible. If the bridging principles are not applicable, an assessment on the basi s of data for the components of the mixture must be applied. 3.2.3.2. Classification criteria for mixtures Based on available information, the approaches below should be used for classification of a nce (Article 9, CLP and Figure mixture for skin corrosivity and irritation in the following seque . 1 ): 1 a. Classification derived using data on the mixture itself, by applying the substance criteria of Annex I to CLP; b. Classification based on the application of bridging principles, which make use of test data on similar tested mixtures and ingredient substances; c. Classification based on ingredients as described in 3.2.3.3, Annex I, CLP. mplete mixture 3.2.3.2.1. When data are available for the co Annex I: The mixture shall be classified using the criteria for substances, taking 3.2.3.1.1. into account the tiered approach to evaluate data for this hazard class. are encouraged to Annex I: 3.2.3.1.2. When considering testing of the mixture, classifiers use a tiered weight of evidence approach as included in the criteria for classification of substances for skin corrosion and irritation (section 3.2.1.2 and 3.2.2.2), to help ensure an ry animal testing. In the absence of any accurate classification as well as to avoid unnecessa other information, a mixture is considered corrosive to skin (Skin Corrosion Category 1) if it has a pH ≤ 2 or a pH ≥ 11.5. However, if consideration of acid/alkaline reserve suggests the mixture may not be corrosiv e despite the low or high pH value, this needs to be confirmed by other data, preferably by data from an appropriate validated in vitro test. Additional simplified guidelines for the assessment of available data on the mixture when WoE to be applied , is provided in S ection 3.2.2.3.3 (see Figure 3 . 2 ). needs is a range of available in vitro test systems that have been validated for their suitabili There ty in assessing skin corrosion/irritation potential of substances . Some but not all test systems have been validated for mixtures and not all available in vitro test systems work equally well for all fic in vitro types of mixtures. Prior to testing a mixture in a speci assay for classification purposes, it has to be ensured that the respective test has been previously shown to be suitable for the prediction of skin corrosion/irritation properties for the type of mixture to be evaluated. 3.2.3.2.1.1. Mixtures with extreme pH As a general rule, mixtures with a pH of ≤ 2 or ≥ 11.5 should be considered as corrosive. However, assessment of the buffering capacity of the mixture indicated by its acid or alkali reserve should be considered. Low values of acid or alkaline reserv e indicate a low buffer capacity. Mixtures showing a low buffer capacity are less or even not corrosive or irritant. The relation is quantitatively expressed 0.5. If the sums are by: - pH + 1/12 alkaline reserve >= 14.5 or pH - 1/12 acid reserve <= -

289 Guidance on the Application of the CLP Criteria – Version 289 5.0 July 2017 - 0.5 the mixture has to be considered as corrosive (see Decision logic 3.2.3.4, >= 14. 5 or <= step 1a). If the additional consideration of the acid/alkaline reserve according to Young (1987, et al. 1994) suggests that classification for corrosion may not be war ranted, this needs to be confirmed by other data, preferably by data from an appropriate and validated in vitro test, applicable for the mixture. The consideration of acid/alkali reserve should not be used alone to exonerate mixtures from classification. Where it is decided to base the classification of a mixture upon consideration of pH alone, Skin Corrosion Category 1 should be applied. Where the mixture has an extreme pH value but the only corrosive/irritant ingredient present in the mixture is an acid or base with an assigned SCL (either in CLP Annex VI or set by supplier according to Article 10(1)), then the mixture should be classified according to the SCL. In this instance, pH of the mixture should not be considered a second time since it would have already been taken into account when deriving the SCL for the substance. If this is not the case, then the steps to be taken into consideration when classifying a mixture with pH 2 or  11.5 are described in the following decision logic:  . 3 Mixture without human or animal data on skin corrosion/irritation or relevant data 3 Figure  2 or  11.5 from similar tested mixtures, pH is Classify as corrosive, Skin Does the acid alkaline reserve indicate that the mixture may not be cor Corrosion Category 1. rosive? NO  YES  Is the mixture tested in an OECD adopted in vitro Classify as corrosive, Skin skin corrosivity test, considered valid and applicable for the Corrosion Category 1 mixture? NO  YES  If corrosive. as Classify Does the mixture demonstrate corrosive properties in an OECD adopted in vitro skin corrosivity test considered valid discrimination between Skin Corr. sible, Skin and applicable for the mixture? 1A/1B/1C is not pos Corr. 1 must be chosen. YES  NO  Does the mixture demonstrate irritant properties in an OECD Classify as skin irritant, Skin adopted in vitro skin irritation test considered valid and Irritation Category 2 applicable for the mixture? YES  NO  Consideration of the total weight of all available evidence, in particular in case of conflicting data, including the extreme

290 ication of the CLP Criteria Guidance on the Appl 290 Version 5.0 – July 2017 pH, negative/inconclusive results from e.g. validated skin tests, and the results from the corrosion/irritation in vitro ion of the methods based on the ingredients in the applicat - mixture in CLP Annex I, sections 3.2.3.3.2 3.2.3.3.3 (Table Classify: Category 1, 2, no 3.2.3)/3.2.3.3.4.1 - 3.2.3.3.4.3 (Table 3.2.4)  classification The mixture must be classified as Skin corrosion Category 1 should the supplier decide not to carry out the required confirmatory testing. - acid/alkali reserve approach, potentially It is also important to note that the use of the pH leading to a change of the classification from corrosive to irritant, or from irritant to not ue to the effect of the ionic classified, assumes that the potential corrosivity or irritancy is d ionic (non - entities. When this is not the case, especially when the mixture contains non - acid/alkali ionisable) substances themselves classified as corrosive or irritant, then the pH - reserve method cannot be a basis for modify ing the classification but should be considered in the weight of evidence analysis. If a mixture with corrosive constituents also contains surfactants (e.g. tensids or detergent substances), it can be assumed that corrosivity might be amplified (Kartono & Maibach 2006). Even if only one corrosive substance with an assigned SCL is present in such a mixture, the possible synergistic effect has to be taken into account when classifying the mixture. ther corrosive/irritant Where the mixture has an extreme pH value and contains some o ingredients (some of which may have SCLs assigned) in addition to an acid or base with or without an assigned SCL, then the steps described in the above decision logic should be followed. When data are not available for the complete 3.2.3.2.2. mixture: bridging principles Annex I: Where the mixture itself has not been tested to determine its skin 3.2.3.2.1. corrosion/irritation potential, but there are sufficient data on the individual ingredients and similar tested mixtures to adequately chara cterise the hazards of the mixture, these data shall be used in accordance with the bridging rules set out in section 1.1.3. In order to apply bridging principles, there needs to be sufficient data on similar tested mixtures the mixture (see Section 1.6.3.2 of this Guidance). as well as the ingredients of When the available identified information is inappropriate for the application of the bridging principles then the mixture should be classified based on its ingredients as described in Sections and 3.2.3.3 of this Guidance. 3.2.3.2.3 When data are available for all ingredients or only for some ingredients 3.2.3.2.3. 3.2.3.2.3.1. nto account for the purpose of Ingredients that should be taken i classification Annex I: 3 [...] The ‘relevant ingredients’ of a mixture are those which are present .2.3.3.1. in concentrations ≥ 1% (w/w for solids, liquids, dusts, mists and vapours and v/v for gases), unless there is a presumption (e.g., in the case of corrosive ingredients) that an ingredient present at a concentration < 1% can still be relevant for classifying the mixture for skin corrosion/irritation. 3.2.3.2.3.2. The additivity approach is applicable Annex I: 3.2.3.3.2. In general, the approach to classification of mixtures as corrosive or irritant to skin when data are available on the ingredients, but not on the mixture as a whole, is based on the theory of additivity, such that each skin corrosive or skin irritant mixture ingredient contributes to the overall skin corrosive or skin irritant properties of the

291 Guidance on the Application of the CLP Criteria 5.0 – July 2017 291 Version in proportion to its potency and concentration. A weighting factor of 10 is used for skin corrosive ingredients when they are present at a concentration below the generic ration that will concentration limit for classification with Category 1, but are at a concent contribute to the classification of the mixture as skin irritant. The mixture is classified as corrosive or irritant to skin when the sum of the concentrations of such components exceeds a concentration limit. 3.2.3.3.3. Table 3.2.3 provides the generic concentration limits to be used to Annex I: determine if the mixture is considered to be corrosive or irritant to the skin. When the supplier is unable to derive the classification using either data on the mixture itself or bridgi ng principles, he must determine the skin corrosion/irritation properties of the mixture using data on the individual ingredients. Although the general approach is the additivity the supplier must principle, which has been successfully used under the DPD and more recently, ascertain whether the additivity approach is applicable. The first step would then be to identify all the relevant ingredients in the mixture (i.e. their name, chemical type, concentration level, e pH of the mixture. In addition it is important to also hazard classification and any SCLs) and th consider effects that could occur in the mixture, such as surfactant interaction, neutralisation of acids/bases when identifying the properties of the complete mixture (including pH and the acid/alka line reserve) in addition to considering contributions of individual ingredients. Additivity may not apply where the mixture contains substances mentioned in CLP Annex I, - 3.2.3.3.4.3, see Section 3.2.3.2.3.3 3.2.3.3.4.1 of this Guidance. Application of SC Ls when applying the additivity approach The generic concentration limits (GCLs) are specified in CLP Annex I, Table 3.2.3. However, according to CLP Article 10(6), SCLs take precedence over GCLs. Thus, if a given substance has h Article 10(1), CLP, then this limit has to be taken into account an SCL set in accordance wit when applying the summation (additivity) method for skin corrosion/irritation (see Examples 4 and 5). In cases where additivity applies for skin corrosion/irritation to a mixture with two o r more substances some of which may have SCLs assigned, then the following formula should be used: The mixture is classified for skin corrosion/irritation if the: 1 Sum of (ConcA / clA) + (ConcB / clB) + ... + (ConcZ / clZ) is  ConcA = the concentration of substance A in the mixture; Where clA = the concentration limit (either specific or generic) for substance A; ConcB = the concentration of substance B in the mixture; neric) for substance B; etc. clB = the concentration limit (either specific or ge The formula should be used in a stepwise procedure in the following order: 1. Should the mixture be classified in Category 1 A? Only Cat. 1A ingredient substances are added. A and 1B ingredient 2. Should the mixture be classified in Category 1B? Cat. 1 substances are added. Should the mixture be classified in Category 1C? Cat. 1A, 1B and 1C ingredient 3. substances are added. 4. Should the mixture be classified in Category 1? Cat. 1A, 1B, 1C and 1 ingredient substances are added. 3.2.3.2.3.3. The additi vity approach is not applicable Annex I: 3.2.3.3.4.1. Particular care must be taken when classifying certain types of mixtures containing substances such as acids and bases, inorganic salts, aldehydes, phenols,

292 ication of the CLP Criteria Guidance on the Appl 292 Version 5.0 – July 2017 and surfactants. The approach explained in Se ctions 3.2.3.3.1 and 3.2.3.3.2 may not be applicable given that many of such substances are corrosive or irritant to the skin at concentrations < 1%. 3.2.3.3.4.2. For mixtures containing strong acids or bases the pH shall be used as Annex I: a classificati on criterion (see Section 3.2.3.1.2) since pH is a better indicator of skin corrosion than the concentration limits in Table 3.2.3. 3.2.3.3.4.3. A mixture containing ingredients that are corrosive or irritant to the Annex I: skin and that cannot be classif ied on the basis of the additivity approach (Table 3.2.3), due to chemical characteristics that make this approach unworkable, shall be classified as Skin ≥ 1% of an ingredient classified as Skin Corrosion or as Corrosion Category 1 if it contains itation (category 2) when it contains ≥ 3% of a skin irritant ingredient. Classification of Skin Irr mixtures with ingredients for which the approach in Table 3.2.3 does not apply is summarised in Table 3.2.4. 3.2.3.3.5. On occasion, reliable data may show that the skin corrosion/irritation Annex I: hazard of an ingredient will not be evident when present at a level at or above the generic concentration limits mentioned in Tables 3.2.3 and 3.2.4 in Section 3.2.3.3.6. In these cases the mixture shall be classified ac cording to that data (see also Articles 10 and 11). On other occasions, when it is expected that the skin corrosion/irritation hazard of an ingredient is not evident when present at a level at or above the generic concentration limits mentioned in Tables 3 .2.3 and 3.2.4, testing of the mixture shall be considered. In those cases the tiered weight of evidence approach shall be applied, as described in Section 3.2.2.2. If there are data showing that (an) ingredient(s) is/are corrosive or Annex I: 3.2.3.3.6. i rritant to skin at a concentration of < 1 % (skin corrosive) or < 3 % (skin irritant), the mixture shall be classified accordingly. Generic concentration limits for substances triggering classification 3.2.3.3. of mixtures When the additivity approach is applicable 3.2.3.3.1. Table 3.2.3 Annex I: Generic concentration limits of ingredients classified as skin corrosion (Category 1, 1A, 1B or 1C)/skin irritation (Category 2) that trigger classification of the mixture as skin corrosion/skin irritation where the additivity approach applies Sum of ingredients classified as: Concentration triggering classification of a mixture as: Skin Corrosion Skin Irritation Category 2 Category 1 (see note below) Skin corrosion Sub - Category 1A, 1B, 1% but < 5%  5%  1C or Category 1 Skin irritation Category 2  10%

293 Guidance on the Application of the CLP Criteria 5.0 – July 2017 293 Version - Category  (10 x Skin corrosion Sub 10% 1A, 1B, 1C or Category 1) + Skin irritation Category 2 Note Category 1A, 1B or - The sum of all ingredients of a mixture classified as Skin Corrosion Sub respectively, shall each be ≥ 5% respectively in order to classify the mixture as either 1C Skin Corrosion Sub - Category 1A, 1B or 1C. If the sum of the ingredients classified as Skin Corrosion Category 1A is < 5% but the sum of the ingredients classified as S kin Corrosion Category 1A+1B is ≥ 5%, the mixture shall be classified as Skin corrosion Category 1B. Similarly, if the sum of the ingredients classified as Skin Corrosion Category 1A+1B ingredients is < 5% but the sum of the ingredients classified as Sub ategory 1A+1B+1C - C ingredients is ≥ 5% the mixture shall be classified as Skin Corrosion Category 1C. Where at - least one relevant ingredient in a mixture is classified as Category 1 without sub thout sub - categorisation if the categorisation, the mixture shall be classified as Category 1 wi sum of all ingredients corrosive to skin is ≥ 5 %. 3.2.3.3.2. When the additivity approach is not applicable Table 3.2.4 Annex I: Generic concentration limits of ingredients of a mixture that trigger classification of the mixture as skin corrosion/skin irritation, where the additivity approach does not apply Ingredient: Concentration: Mixture classified as: Skin corrosion Category 1 1% Acid with pH 2 ≥ ≤ Skin corrosion Category 1 ≥ 1% Base with pH ≥ 11,5 (Sub - Categories 1A, corrosion Category 1 Other skin corrosive Skin 1% ≥ 1B, 1C or Category 1) ingredients Other skin irritant (Category 2) Skin irritation Category 2 3% ≥ ingredients, including acids and bases 3.2.3.4. Decision logic for classification of mixtures The decision logic, based on the one provided in the GHS, is presented here below as additional guidance. It is strongly recommended that the person responsible for classification, study the criteria for classification, as well as the guidance above, before and during use of the decision logic.

294 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 294 No Does the mixture as a whole or its ingredients have Classification not possible data/information to evaluate skin corrosion/irritation? Yes Yes Does the mixture as a whole have data/information to See decision logic 3.2.2.6 evaluate skin corrosion/irritation? No Classify in Yes appropriate Can bridging principles be applied? category or sub - category No Follow decision logic in Yes 3.2.3.2.1.1 of Section Is pH of the mixture ≤ 2 or ≥ 11.5? this guidance and classify accordingly No Category 1 Yes a 1% of an ingredient which is Does the mixture contain ≥ when the additivity approach may not apply? corrosive Danger No , Category 1 Subcategory one or more corrosive ingredients the Does mixture contain b Yes 1A, 1B or 1C when the additivity approach applies and where the sum of 1 (or any concentrations ingredients classified as Skin Corr. ≥ 5% subcategory) ? Danger No

295 Guidance on the Application of the CLP Criteria – July 2017 5.0 295 Version Category 2 Yes a of an ingredient which is Does the mixture contain ≥ 3% irritant and when the additivity approach may not apply? Warning No Category 2 Does the mixture contain one or more corrosive or irritant ingredients when the additivity approach applies and where the sum of concentrations of ingredients classified Yes as: (a) Skin Corr. Category 1 ≥ 1% but < 5%; or (b) Skin Irrit. Category 2 ≥ 10%; or Warning Cat . 2 ≥ 10%? (c) (10 x Skin Corr.Cat. 1) + Skin Irrit. No Not classified a ection 3.2.3.3.1 of Annex I of CLP. Where relevant < 1%, see S b See note to Table 3.2.3 in Annex I of CLP for details on use of Category 1 subcategories.

296 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 296 3.2.4. Hazard communication in form of labelling for skin corrosion/irritation Pictograms, signal words, hazard statements and precautionary 3.2.4.1. statements 3.2.4.1. Label elements shall be used for substances or mixtures meeting the criteria Annex I: for classification in this hazard class in accordance with Table 3.2.5. Table 3.2.5 Label elements for skin corrosion/irritation Classification Sub - Categories 1A / 1B / Category 2 1C and Category 1 GHS Pictograms Signal Word Danger Warning Hazard Statement H314: Causes severe skin H315: Causes skin burns and eye damage irritation Precautionary Statement P260 P264 Prevention P264 P280 P280 Precautionary Statement P302 + P352 P301 + P330 + P331 Response P321 P303 + P361 + P353 P332 + P313 P363 P304 + P340 P362 + P364 P310 P321 P305 + P351 + P338 Precautionary Statement P405 Storage Precautionary Statement P501 Disposal Article 26 1 (d) If the hazard pictogram ‘GHS05’ applies, the hazard pictogram ‘GHS07’ shall not appear for skin and eye irritation.

297 Guidance on the Application of the CLP Criteria 5.0 – July 2017 297 Version Additional labelling provisions 3.2.4.2. — EUH071 Annex II: 1.2.6. Corrosive to the respiratory tract For substances and mixtures in addition to classification for inhalation toxicity, if data are available that indicate that the mechanism of toxicity is corrosivity, in accordance with section 3.1.2.3.3 and Note 1 of Table 3.1.3 in Annex I. For substances and mixtures in addition to classification for skin corros ivity, if no acute inhalation test data are available and which may be inhaled. Corrosive substances (and mixtures) may be acutely toxic after inhalation to a varying degree, is available for a which is only occasionally proved by testing. In case no acute inhalation study corrosive substance (or mixture) and such substance (or mixture) may be inhaled, a hazard of respiratory tract corrosion may exist. As a consequence, such substances and mixtures have to be supplementary labelled with EUH071, if there is a possibility of exposure via inhalation taking into consideration the saturated vapour concentration and the possibility of exposure to of this particles or droplets of inhalable size as appropriate, (see also Chapter 3.8.2.5 Guidance). Moreover, in such a case it is strongly recommended to apply the precautionary statement P260: ‘Do not breathe dust/fume/gas/mist/vapours/spray.’ EUH066 — Repeated exposure may cause skin dryness or cracking Annex II: 1.2.4. For substances and mixtures which may cause concern as a result of skin dryness, flaking or cracking but which do not meet the criteria for skin irritancy in section 3.2 of Annex I, based her: on eit practical observations; or — — relevant evidence concerning their predicted effects on the skin. Examples of classification for skin corrosion/irritation 3.2.5. Examples of substances fulfilling the criteria for classification 3.2.5.1. 3.2.5.1.1. Example 1: Standard test according to OECD TG 404 with three animals In a guideline test according to OECD TG 404 the test substance was applied for three minutes and 1 hour. No scars or other irreversible effects were found. The scoring results obtained after a 4 hour application time are listed in the following table: - Degree of erythema after Animal Degree of oedema after 24/48/72 h  [observation time] [observation time]  2.3 ? Nr. 1h 24h 48h 72h 7d 14d 1h 24h 48h 72h 7d 14d Erythe - - Oede ma ma 3 3 2 0 1 2 2 2 1 3 Yes No 0  24/48/72 h = 2.7  24/48/72 h = 2.0 =>’positive Responder’ 2 3 3 3 3 0 1 2 2 1 0 Yes No   24/48/72 h = 3 24/48/72 h = 1.7 =>’positive Responder’

298 ication of the CLP Criteria Guidance on the Appl 298 Version 5.0 – July 2017 3 1 1 0 0 1 1 1 1 0 No No 1   24/48/72 h = 24/48/72 h = 1 0.66 Classification: Skin Irritation Category 2 Rationale: The classification is made on the basis of 2/3 ‘positive responder’ exceeding 2.3 mean score for erythema. Example 2: Test carried out with one animal with a test substance which 3.2.5.1.2. is suspected as corrosive Due to the unprecedented structure the biological effects of the substance cannot be h one animal only in anticipated. Therefore, the test according to OECD TG 404 was started wit line with testing restrictions. Exposure times were 3 min and 1h. The following scores/effects were observed: Degree of erythema after Visible Exposure Degree of oedema after ...[observation time] necrosis, time ...[observation time] irreversible skin damage 1h 24h 48h 72h ... 1h 24h 48h 72h ... After 14d 0 0 0 0 0 0 0 3 min No 0 1 2 1h 0 2 2 3 0 Yes 3 Classification: Skin Corrosion Category 1B Rationale: The classification is based on the destruction of the tissue after 1 hour of exposure. Example 3: Test carried out with more than three animals 3.2.5.1.3. substance was tested on acute skin irritation / corrosion according to OECD TG 404. Contact A time was 4 hours. No effects were seen after a contact time of 3 min and one hour. The following scores were obtained after a contact time of 4 hours: Observation time 48h 72h 7d 14d 1h 24h 48h 72h 24h 7d 14d Pos 1h responder Oedema Animal Eryth Erythema Oed - - ma Nr ema e 3 3 2 2 1 0 2 3 2 2 1 0 Yes Yes 1 3 2 2 2 1 0 2 2 2 2 2 1 0 No No 2 3 2 1 1 1 0 2 2 2 2 1 0 No No 4 2 2 1 1 1 0 2 2 2 2 1 0 No No Evaluation is made based on the average score per animal. Only 1/4 of the animals reached the cut - off value of 2.3, i.e. only animal No 1 is a positive responder. No classification is warranted with regard to skin irritation.

299 Guidance on the Application of the CLP Criteria 5.0 – July 2017 299 Version 3.2.5.2. Examples of mixtures fulfilling the criteria for classification Where the mixture is made up of ingredients with no assigned SCLs, the appropriate summation(s) and generic concentration limits from CLP Annex I, Table 3.2.3 should be used. Example 4: Mixture without extreme pH, with ingredients with SCLs 3.2.5.2.1. Skin corrosion / irritation Ingredient Concentration SCL classification (% w/w) Skin Irrit. 2 3.8 Not assigned Substance A Substance B 0.5 Not classified 5.4 C ≥ 10 %: Skin Corr. 1B Base E Skin Corr. 1B 2 5 % ≤ C < 10 %: Skin Irrit. Not classified 4 Substance D Substance F Skin Corr. 1B 2 Not assigned Not classified 84.3 Water pH of the mixture is 10.5 – 11.0, thus extreme pH provisions do not apply. The mixture contains a base but not any surfactant. Additivity is considered to apply. can be disregarded as they are not classified for skin Substance B, substance D and water corrosion/irritation. SCLs are neither assigned to substance F nor substance A, thus GCLs apply for these of this Guidance, Application 3.2.3.2.3.2 ingredients. SCLs are assigned to Base E (see Section of SCLs when applying the additivity approach ). Skin Corr. 1 :  < 1, thus the mix (% substance F/GCL) + (% base E/SCL) = (2/5) + (5.4/10) = 0.94 ture is not classified as Skin Corr. 1 Irrit. 2: Skin (% substance F/GCL) + (% base E/SCL) + (% substance A/GCL) = (2/1) + (5.4/5) + (3.8/10) = 3.46 which is > 1, thus the mixture is classified Skin Irrit. 2 Example 5: Mixture without extreme pH, and non - applicability of the 3.2.5.2.2. additivity approach Ingredient Classification Information Wt% Ingredient 1 4 Skin Corr. 1A pH = 1.8 Ingredient 2 5 Skin Irr. 2 - Ingredient 3 5 Skin Irr. 2 - Ingredient 4 86 - No data available

300 ication of the CLP Criteria Guidance on the Appl 300 Version 5.0 – July 2017 The pH of the mixture is 4.0, thus extreme pH provisions do not apply. There are no test data on the mixture (apart from a pH). Bridging principles do not apply since data on a similar mixture was not available. Classification of the mixture based on ingredient data can be conside red. Ingredient 1 with a pH = 1.8 is an ingredient for which additivity might not apply (see 2 - 3 and Table 3.2.4, Annex I, CLP). Expert judgment would be needed to determine 3.2.3.3.4.1 - ortant. Given the limited whether or not additivity applies. Knowledge of the components is imp information in this example, the classifier of this mixture chose to apply non - additivity as a conservative approach. Without information on the mode of action of Ingredient 1, the mixture could be corrosive regardless of the over all pH. Therefore, the criteria described in paragraph - 2 - 3 were applied (including “ A mixture containing ingredients that are corrosive or 3.2.3.3.4.1 irritant to the skin and that cannot be classified on the basis of the additivity approach (Table 3.2.3), due to chemical characteristics that make this approach unworkable, shall be classified as Skin Corrosive Category 1A, 1B or 1C if it contains ≥ 1% of a an ingredient classified in Category 1A, 1B or 1C respectively or as Category 2 when it contains ≥ 3% of an irritant ingredient .”). Thus, the mixture should be classification as Skin Corrosion Category 1A because the mixture contains an ingredient 1 (Skin Corr. 1A) at a concentration ≥ 1%. Examples of mixtures not fulfilling the criteria for classification 3.2.5.3. 3.2.5.3.1. Example 6: Mixture without extreme pH, with ingredients with SCLs Skin corrosion / Ingredient Concentration SCL irritation classification (% w/w) Surfactant C Skin Irrit. 2 0.4 Not assigned Substance G 3.0 Not assigned Skin Irrit. 2 Irrit. 2 0.7 Not assigned Substance A Skin Substance H 3.0 C ≥ 70 %: Skin Corr. 1A Skin Corr. 1A 50 % ≤ C < 70 %: Skin Corr. 1B 35 % ≤ C < 50 %: Skin Irrit. 2 Substance D Not classified 2 Water 90.9 Not classified – pH of the mixture is: 2.5 3.0, thus extreme pH provisions do not apply. The mixture contains 57 one surfactant. Additivity is considered to apply . Substance D and water can be disregarded as they are not classified for skin corrosion/irritation. Also surfactant C and substance A can be disregarded as bot h are present at below 1%. No SCL is assigned to substance G, thus GCL apply for this ingredient. Skin Corr. 1: 57 Please note that in cases where a mixture with corrosive constituents also contains surfactans, it can be assumed that corrosivity migh be amplified.

301 Guidance on the Application of the CLP Criteria 5.0 – July 2017 301 Version The mixture contains 3% substance H, the only ingredient classified as Skin Corr. 1. As this is below the 50% SCL for substance H, the mixture i s not classified as Skin Corr. 1. 2: Skin Irrit. (% substance H/SCL) + (% substance G/GCL) = (3/35) + (3/10) = 0.39 which is < 1, thus the mixture is not classified Skin Irrit. 2. 3.2.6. References ECETOC (2002), Use of human data in hazard classification for ir ritation and sensitisation, Monograph No 32, Brussels ISSN 0773 6374 - 32 - ECVAM/ESAC (2000) Statement on the application of the CORROSITEX® Assay for skin corrositivity testing. http://ihcp.jrc.ec.europa.eu/our_labs/eurl - ecvam Online: - vitro tests for skin irritation. Online: ECVAM/ESAC (2007) Statement on the validity of in http://ecvam.jrc.it/ ECVAM/ESAC (2008) Statement on the validity of in vitro t ests for skin irritation testing. Online: - http://ecvam.jrc.it/ ECVAM/ESAC (2009) Statement on the performance under UN GHS of three in - vitro assays for skin irritation testing and the adaptation of the reference chemica ls and defined accuracy values http://ecvam.jrc.it/ of the ECVAM skin irritation performance standards. Online: Kartono F & Maibach H. (2006) Irritants in combination with a synergistic or additive effect on the skin re sponse: an overview of tandem irritation studies. Contact Dermatitis 54(6), 303 - 12. Spielmann, H., Hoffmann, S., Liebsch, M., Botham, P., Fentem, J., Eskes, C., Roguet, R., Cotovió, J., Cole, T., Worth, A., Heylings, J., Jones, P., Robles, C., Kandárová, H., Gamer, A., Remmele, M., Curren, R., Raabe, H., Cockshott, A., Gerner, I. and Zuang, V. (2007) The ECVAM International Validation Study on In Vitro Tests for Acute Skin Irritation: Report on the Validity of the EPISKIN and EpiDerm Assays and on the Skin Integrity Function Test. ATLA 35, 559 - 601. Young J.R., How M.J., Walker A.P., Worth W.M.H. (1988): Classification as corrosive or irritant to skin of preparations containing acidic or alkaline substances, without test on animals. Toxicology in Vitro 2 , 19 - 26. Young J.R., How M.J. (1994): Product classification as corrosive or irritant by measuring pH and acid / alkali reserve. In Alternative Methods in Toxicology vol. 10 - In Vitro Skin Toxicology: Irritation, Phototoxicity, Sensitization, eds. A.Rougier, A.M. Goldberg and H.I Maibach, Mary Ann Liebert, Inc. 23 - 27.

302 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 302 3.3. SERIOUS EYE DAMAGE/E YE IRRITATION It should be noted that if a substance or mixture is classified as Skin corrosion Category 1 then serious damage to eyes is implicit as reflected in the hazard statement for skin corrosion (H314: Causes severe skin burns and eye damage). Thus, the corrosive substance or mixture is also a uses serious eye damage) is not classified, but the corresponding hazard statement (H318: C indicated on the label to avoid redu ndancy. 3.3.1. Definitions for classification for serious eye damage/eye irritation Annex I: 3.3.1.1. Serious eye damage means the production of tissue damage in the eye, or serious physical decay of vision, following application of a test substance to the anteri or surface of the eye, which is not fully reversible within 21 days of application. Eye irritation means the production of changes in the eye following the application of test substance to the anterior surface of the eye, which are fully reversible within 21 days of application. 3.3.2. Classification of substances for serious eye damage/eye irritation 3.3.2.1. Identification of hazard information 3.3.2.1.1. Identification of human data Existing data on eye effects in humans may include well - documented epidemiological studies, clinic al studies, case reports, and data from poison information units and accident databases or occupational experience. Their quality and relevance for hazard assessment should be thoroughly reviewed. A critical review of the value of human studies is provided in the Guidance IR&CSA Section R.4.3.3 and more specific considerations for eye damage/irritation are given on in the Guidance on IR&CSA Section R.7.2.9. Identification of non human data 3.3.2.1.2. substances may include existing data Available serious eye damage/eye irritation information on generated by the test methods in the Test Methods Regulation or by methods based on internationally recognised scientific principles. Before using the methods as referred to in the following sections, it should be checked whether the methods are sufficiently validated (or considered valid in case of (Q)SAR and expert systems) against the criteria for classification according to CLP (and not validated against the old DSD criteria which differed slightly from the CLP criteria). 3.3.2.1.3. Consideration of physico - chemical properties Substances with oxidising properties can give rise to highly exothermic reactions in contact with other substances and human tissue. High temperatures thus generated, or direct oxidative amage/destroy biological materials. This applies, for example, to organic impact, may d peroxides, which can be assumed to be eye irritants, unless evidence suggests otherwise IR&CSA Sections R.7.2.8 and R.7.2.4.1). (Guidance on Thus, in the absence of evidence to the c ontrary, a hydro peroxide should be considered to be classified as Eye Damage Category 1, whereas Eye Irritation Category 2 should be considered for peroxides. Appropriate evidence must be provided in order to consider no classification of substances with oxidising properties. 3.3.2.1.4. pH and the acid/alkaline reserve Annex I: 3.3.2.2.4. 11,5 may produce serious eye Likewise, pH extremes like ≤ 2 and ≥ damage, especially when associated with significant acid/alkaline reserve (buffering capacity).

303 Guidance on the Application of the CLP Criteria 5.0 – July 2017 303 Version stances are expected to produce significant effects on the eyes. In the Generally such sub absence of any other information, a substance is considered to cause serious eye damage ≥ 11,5. However, if consideration of acid/alkaline reserve (Category 1) if it has a pH 2 or ≤ ggests the substance may not cause serious eye damage despite the low or high pH value, su this needs to be confirmed by other data, preferably by data from an appropriate validated in vitro test. 2 or  11.5. Where extreme pH is the Substances can be predicted to be corrosive, if the pH is  only basis for classification as serious eye damage, it is important to take into consideration the et al , 1988, and Young and acid/alkaline reserve, a measure of the buffering capacity (Young How, 1994). However, lack o f or low buffering capacity should not be used alone to exonerate from classification as corrosive, which needs to be confirmed by other data, preferably by a validated test (see also S ection 3.2.3.2 of this Guidance). in vitro Further information and/or reasoning is needed to conclude whether the substance caus eye es irritation. Non - testing methods: (Q)SARs and expert systems 3.3.2.1.5. - testing methods such as (Q)SARs and expert systems ( a diverse group of models Non consisting of combinations of SARs, QSARs and databases) - by - may be considered on a case case basis. (Q)SARs are in general not very specific for eye irritancy. In many cases rules are lar manner to those used for skin irritation and corrosion as alert s to indicate an used in a simi effect. (Q)SAR systems that also account for eye effects are for example ACD Percepta, CASE Ultra, Discovery studio Accelrys (former TOPKAT), Derek Nexus. For more detailed guidance, IR&CSA Section R.6 (‘QSAR and grouping of chemicals’). OECD QSAR consult the Guidance on 58 Toolbox and ToxTree contain BfR rules for eye irritation/corrosion. s on the presence or absence of an effect In the absence of any other existing data, conclusion can be made if the (Q)SAR or expert system has been shown to make an adequate prediction (see Figure 3 . 4 ). The suitability of the model (relia bility, relevance) should be very carefully checked to make sure that the prediction is fit for purpose, and the applicability of the model to the substance should also be justified. The predicted endpoint should be adequate for classification and labellin g. In case of negative QSAR data the need for classification cannot be excluded. Since a formal adoption procedure for non - testing methods is not foreseen and no formal validation process is in place, appropriate documentation is crucial. In order to achie ve acceptance under REACH, the documentation must conform to the so - called QSAR Model IR&CSA Section R.6.1. Reporting Format (QMRF). For more details consult the Guidance on Testing methods: in vitro methods 3.3.2.1.5.1. The OECD has at present adopted five in vitro te st guidelines for assessing eye hazard potential. Four in vitro tests methods have been adopted for the identification of substances inducing serious eye damage, i.e. the Isolated Chicken Eye (ICE) test (OECD TG 438; TM B.48), the d Permeability (BCOP) test (OECD TG 437; TM B.47), the Fluorescein Bovine Corneal Opacity an Leakage (FL) test (OECD TG 460), the short time exposure (STE) test (OECD TG 491). In addition, there are three validated test methods without an OECD test guideline i.e. Cytosensor 58 The German Federal Institute for Risk Assessment (BfR) has deve loped a Decision Support System (DSS) to assess certain hazardous properties of pure chemicals.

304 ication of the CLP Criteria Guidance on the Appl 304 Version 5.0 – July 2017 59 Microph test, Isolated Rabbit Eye (IRE) test and the Hen's Egg Test on Chorio - ysiometer (CM) 60 CAM) test . These tests are recommended for use as part of a tiered - - allantoic Membrane (HET 61 Top - testing strategy pproach for regulatory classification and labelling ). A (e.g. Down A substance can be considered as causing serious eye damage (Category 1) based on positive results in the ICE test, the BCOP test, the FL test, the STE test, CM test IRE test or the HET - 62 CAM test identifying substances not causing serious . Four adopted OECD TGs can be used for eye damage/eye irritation which are the ICE test, BCOP test, STE test and Reconstructed human Cornea like Epithelium (RhCE) (OECD TG 492). In addition, the validated CM test - tances not causing serious eye damage or eye irritation. method can be used for identifying subs Negative results from the ICE, BCOP, STE, RhCE and CM test methods can be used for 8 in vitro . For other test methods the negative up approach’ classification purposes, i.e. ‘bottom - ses in these tests must be followed by further testing (see section R.7.2.9.1 in corrosivity respon Guidance on IR&CSA ). the in vitro tests with regulatory acceptance for eye irritation at present. There are no e found on the OECD website: Further information on newly adopted OECD Test Guidelines can b http://www.oecd.org/env/chemicalsafetyandbiosafety/testingofchemicals/oecdguidelinesforthet ( gofchemicals.htm ) estin . Information on the current developments of in vitro tests and methodology can be found on the ecvam ECVAM website ( http://ihcp.jrc.ec.europa.eu/our_labs/eurl - ). 3.3.2.1.5.2. Testing methods: In vivo methods Testing for eye irritation should not be carried out on substances known or predicted to be corrosive to skin and classified as such. Such substances are automatically considered to be severely damaging to the eye and are c lassified but not labelled for serious eye damage in addition to skin corrosion. The in vivo test for in vivo test in rabbits according to OECD TG 405 (TM B.5) is the standard the hazard assessment under REACH. et al 1980) is a modification of the standard OECD TG The Low Volume Eye Test (LVET; Griffith 405 test method. The differences being: the test material is placed directly on the cornea in the LVET test, instead of introducing  it in the conjunctival sac inside the lower lid; a reduction in the volume of test material applied (0.01 ml (or corresponding weight for  solids) in the LVET test, as compared with the standard 0.1 ml). No new tests should be performed according to LVET as stated by ESAC in its conclusion on the use of LVET data for the purpose of classification and labelling in 2009 (ECVAM/ESAC, 2009b). Existing data from the LVET test could be considered for the purpose of classification and labelling, but must be carefully evaluated. The differences mentioned above may result in a classifica tion in a lower category (or no classification) based on LVET data, than if the 59 A draft OECD TG available at http://www.oecd.org/env/ehs/testing/DRAFT%20Cytosensor%20TG%20(V9)%2021%20Dec%2012_clean. . pdf 60 ICCVAM published a report on the HET - CAM in 2010 http://iccvam.niehs.nih.gov/docs/ocutox_docs/InVitro - 2010/Body.pdf . 61 The top - down approach should be used when available information suggests that the substance may cause serious eye damage. The bottom - up approach, on the other hand, should be followed only when available information suggests that the substance may not be irritant to the eye. 62 ICCVAM published a report on the HET - CAM in 2010 http://iccvam.niehs.nih.gov/docs/ocutox_docs/InVitro - 2010/Body.pdf .

305 Guidance on the Application of the CLP Criteria 5.0 – July 2017 305 Version were based on data derived from the standard in vivo test (OECD TG 405 (TM classification B.5)). Thus, positive data from the LVET test could be a trigger for considering class ification in Category 1 on its own, but data from this test indicating Category 2 classification or no classification are not conclusive for a category 2 classification or no classification respectively. - by case bas is to the limited use of LVET data as Consideration should be given on a case - data in a weight of evidence determination in order to assess if the supplementary in vivo criteria for classification are met. A weight of evidence could include, for example, the results of in v appropriate validated tests, relevant and conclusive human and animal data, extreme itro pH. The applicability domain is limited to detergent and cleaning products ECVAM/ESAC, ( 2009b). Classification criteria 3.3.2.2. 3.3.2.1.1 . Serious eye damage (Category 1) Annex I: 3.3.2.1.1.1. A single hazard category (Category 1) is adopted for substances that have potential to seriously damage the eyes. This hazard category includes as criteria the observations listed in Table 3.3.1. These observations include animals with grade 4 cornea lesio ns and other severe reactions (e.g., destruction of cornea) observed at any time during the test, as well as persistent corneal opacity, discoloration of the cornea by a dye substance, adhesion, pannus, and interference with the function of the iris or oth er effects that impair sight. In this context, persistent lesions are considered those which are not fully reversible within an observation period of normally 21 days. Hazard classification as Category 1 also contain substances fulfilling the criteria of c orneal opacity ≥ 3 or iritis > 1,5 observed in at least 2 of 3 tested animals, because severe lesions like these usually do not reverse within a 21 days observation period. [...] Table 3.3.1 a Serious eye damage Criteria Category Category 1 produces: A substance that (a) in at least one animal effects on the cornea, iris or conjunctiva that are not expected to reverse or have not fully reversed within an observation period of normally 21 days; and/or (b) in at least 2 of 3 tested animals, a positive response of: (i) corneal opacity ≥ 3 and/or (ii) iritis > 1,5 calculated as the mean scores following grading at 24, 48 and 72 hours after installation of the test material. a Grading criteria are understood as described in Regulation (EC) No 440/2 008 3.3.2.1.2. Eye irritation (Category 2) Annex I: 3.3.2.1.2.1. Substances that have the potential to induce reversible eye irritation shall be classified in Category 2 (eye irritation).

306 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 306 3.3.2.1.2.2. For those substances where there is pronounced variability among animal responses, this information shall be taken into account in determining the classification [...] Table 3.3 2 a Eye irritation Category Criteria Category 2 least in 2 of 3 tested animals, a positive Substances that produce in at response of: ≥ 1 and/or (a) corneal opacity (b) iritis ≥ 1, and/or (c) conjunctival redness ≥ 2 and/or (d) conjunctival oedema (chemosis) ≥ 2 calculated as the mean scores following grading at 24, 48 and 72 hours after installation of the test material, and which fully reverses within an observation period of 21 days a Grading criteria are understood as described in Regulation (EC) No 440/2008 in vivo test, OECD TG 405, and The classification criteria apply to results of the standard animal are possible to apply to the results of the LVET. However, the differences between the LVET and OECD TG 405 test methods, may result in a classification in a lower category (or no ere based on data derived from classification) based on LVET data, than if the classification w in vivo test (OECD TG 405 (TM B.5)). See also 3.3.2.1.5.2 above. the standard 3.3.2.3. Evaluation of hazard information . 3.3.2.2.1 A Annex I: tiered approach to the evaluation of initial information shall be considered where applicable, recognising that not all elements may be relevant. Annex I: 3.3.2.2.6. The tiered approach provide guidance on how to organize existing information and to make a weight of evidence decision about hazard assessment and hazard classification. Animal testing with corrosive substances shall be avoided whenever possible. Although information might be gained from the evaluation of single parameters within a tier (see 3. 3.2.1.1), consideration should be given to the totality of existing information and making and overall weight of evidence determination. This is especially true when there is conflict in information available in some parameters. The tiered approach for the evaluation of the information applied in order to make a decision about the serious eye damage/eye irritation hazard properties is illustrated by the Figure 3 . 4 below. The figure was adopted by the UNSCEGHS in December 2012 (with exception of the added footnotes g) and h)).

307 Guidance on the Application of the CLP Criteria 5.0 Version 307 – July 2017 63 . Tiered evaluation for serious eye damage/eye 3 irritation 4 Figure also Figure 3.1 ) see ( Parameter Finding Conclusion Step Existing human or animal 1a: serious Classify as causing Serious eye damage serious eye damage/eye eye damage a irritation data f Eye irritant Classify as eye irritant Negative/Insufficient/Inconcl usive/No data Existing human or animal Deemed to cause and classify Skin corrosion 1b: serious eye damage data, skin corrosion as Negative /Insufficient/Inconclusive/No data f 1c: human or animal Existing Not classified Existing data showing that serious eye damage/eye a irritation data substance does not cause serious eye damage or eye irritation No/Insufficient/Inconclusive data Other, existing skin/eye data Yes; other existing 2: May be deemed to cause b data showing that serious eye damage in animals substance may cause serious eye damage Yes; other existing to be an eye May be deemed f data showing that irritant substance may eye irritation cause No/Insufficient/Inconclusive data 63 Adopted by the UNSCEGHS in December 2012.

308 ication of the CLP Criteria Guidance on the Appl 308 Version 5.0 – July 2017 63 4 Tiered evaluation for serious eye damage/eye irritation Figure 3 . ) see ( also Figure 3.1 Parameter Finding Conclusion Step Existing ex vivo/in vitro eye 3: serious Classify as causing Positive: serious eye c eye damage data damage f, h Positive: eye irritant Classify as eye irritant f Negative: not eye Not classified irritant No/Insufficient/Inconclusive data i 4: - based assessment (with serious Classify as causing pH pH ≤ 2 or ≥ 11.5 f with high eye damage consideration of acid/alkaline d acid/alkaline reserve reserve of the chemical) or no data for acid/alkaline reserve Not pH extreme, no pH data or extreme pH with data showing low/no acid/alkaline g reserve Serious eye damage Deemed to cause serious eye damage Validated Structure Activity 5: Eye irritant Deemed to be eye irritant Relationship (SAR) methods Deemed to cause serious eye Skin corrosive damage No/Insufficient/Inconclusive data 6: Consideration of the total serious eye Deemed to cause Serious eye damage e weight of evidence damage Deemed to be eye irritant Eye irritant 7: Not classified (a) Existing human or animal data could be derived from single or repeated exposure(s), for example in occupational, consumer, transport, or emergency response scenarios; or from purposely - generated data from animal studies conducted according to validated and internationally accepted test methods. om accident or poison centre databases can provide evidence for classification, Although human data fr absence of incidents is not itself evidence for no classification as exposures are generally unknown or uncertain; (b) Existing animal data should be carefully reviewed to dete rmine if sufficient serious eye damage/eye irritation evidence is available through other, similar information. It is recognized that not all skin irritants are eye irritants. Expert judgment should be exercised prior to making such a determination;

309 Guidance on the Application of the CLP Criteria 5.0 – July 2017 309 Version - tissue - Ev idence from studies using validated protocols with isolated human/animal tissues or other non (c) based, validated protocols should be assessed. A positive test result from a validated in vitro test on sify as causing serious eye damage; skin corrosion would lead to the conclusion to clas Measurement of pH alone may be adequate, but assessment of acid/alkaline reserve (buffering capacity) (d) would be preferable; All information that is available on a substance should be considered and an overall dete rmination made (e) on the total weight of evidence. This is especially true when there is conflict in information available on some parameters. The weight of evidence including information on skin irritation may lead to classification for eye irritation. Negat ive results from applicable validated in vitro tests are considered in the total weight of evidence evaluation. (f) In case of contradicting data, e.g. negative/irritation human data but positive/serious eye damage in - sment should be performed, see footnote e. (This footnote vitro data, a weight of evidence asses was not included in Figure 3.4 in the 5th rev of GHS, but is based on 3.3.1.2 and 3.3.2.2.6, Annex I, CLP) Non corrosivity needs to be confirmed by other data preferably by data from an appropr (g) iate validated in the 5th rev of GHS, but is based on was not included in Figure 3.4 in vitro test. (This footnote 3.3.2.2.4, Annex I, CLP) (h) Note: currently there are no scientifically valid or internationally accepted in vitro test methods for the ect identification of Cat 2 eye irritants. dir For the cases of mixtures with no human or animal data on serious eye damage/eye irritation but with (i) section 3.3.3.2.1.1 for additional guidance. Figure . 5 in extremeoH, see 3 3.3.2.3.1. Evaluation of human data Quality data on substance - induced eye irritation in humans are likely to be rare. Where human data are available, the usefulness of such data for classification purposes will depend on the extent to which the effect, and its magnitude, can be reliably attributed to the substance of interest. The extent and duration of the exposure needs also to be taken into account as absence of effect may be due to washing off the eyes shortly after e xposure. In such cases the absence of effects may not indicate the absence of hazard. The quality and relevance of such data for hazard assessment should be critically reviewed. decay in vision with If a substance is diagnostically confirmed by a physician to be the cause for the effects not being transient but persistent this should lead to the most serious eye classification, i.e. Eye Damage Category 1. Further information on the evaluation of human data for eye irritation can be found in the IR&CSA Section R7.2.4.2. Guidance on 3.3.2.3.2. - human data Evaluation of non 3.3.2.3.2.1. Ex vivo/ in vitro data A substance can be considered as causing serious eye damage (Category 1) based on positive 64 results in the ICE test, the BCOP test, FL test, STE test, IRE test, CM test or the HET - C AM test . Negative results from the ICE, BCOP, STE, RhCE and CM test methods can be used for classification purposes i.e. ‘bottom - in up approach’, but for other test methods the negative r testing (Guidance on vitro corrosivity responses in these tests must be followed by furthe Section R.7.2.9). IR&CSA Normally, recommendations for classification according to GHS criteria based on the results of an in vitro test are mentioned in the corresponding OECD test guideline. There are currently no validated in vitr o eye irritation test methods available. 64 ICCVAM published a report on the HET - CAM in 2010 http://iccvam.niehs.nih.gov/docs/ocutox_docs/InVitro - 2010/Body.pdf .

310 ication of the CLP Criteria Guidance on the Appl 310 Version 5.0 – July 2017 3.3.2.3.2.2. In vivo data Tests in albino rabbits (OECD TG 405) Evaluation criteria for local effects on the eye are of the damage and reversibility . severity of damage the degree of inflammation is assessed. Responses are graded For the severity according to the grading of ocular lesions in OECD TG 405. Evaluation takes place separately for cornea, iris and conjunctiva (erythema and swelling). If in CLP Annex I, Tables 3.3.1 and 3.3.2, the substances are the scoring meets the criteria classified as Category 1 for serious eye damage or Category 2 for eye irritation, respectively. Reversibility of eye lesions is the other decisive factor in evaluating responses in the animal . If the effects are not transient within the observation time of 21 days but cause persistent test damage, they are considered irreversible and the test substance needs to be classified into Category 1. In the case of studies with a shorter observation period with irreversible effects, classification based on WoE should be considered. If considered as reversible, the test report must prove that these effects are transient, i.e. the affected sites are repaired within the observation peri ection od of the test (see Examp le 1, S ). Evaluation of reversibility or irreversibility of the observed effects does not need to 3.3.5.1.1 exceed 21 days after instillation for the purpose of classification. According to OECD TG 405, in cases of suspected serious eye damage, the test is started with one animal only. If effects in this animal are irreversible until the end of the observation period, available to classify the substance for serious eye damage. For a sufficient information is decision on no classification for serious eye damage and/or irritation or for a decision on classification as irritant, two additional animals have to be tested. nimals the average scores for three consecutive days (usually 24, 48 For each of the three test a and 72 hours) are calculated separately for the cornea, iris and conjunctiva (erythema and swelling). If the mean scores for 2 out of 3 animals exceed the values in CLP Annex I, Tables 3.1 and 3.3.2, classification has to be assigned accordingly. 3. Tests that have been conducted with more than three animals Older test methods us ed up to six rabbits. In such cases, the current UNSCEGHS Guidance needs to be applied (adopted in June 2011) (se e also Example 2, section 3.3.5.1.2): In the case of 6 the following applies: rabbits, Classification for serious eye damage Category 1 if: a. – at least in one animal effects on the cornea, iris or conjunctiva that are not expected i. to reverse or have not ful ly reversed within an observation period of normally 21  3 for days; and/or(ii) at least 4 out of 6 rabbits show a mean score per animal of > 1.5 for iritis corneal opacity and/or Classification for eye irritation – Category 2 if at least 4 out of 6 rabb its show a mean b. score per animal of: i.  1 for corneal opacity and/or ii. 1 for iritis and/or  iii.  2 conjunctival erythema (redness) and/or iv.  2 conjunctival oedema (swelling) (chemosis) and which fully reverses within an observation period of normally 21 days. In the case of rabbits, the following applies: 5 a. Classification for serious eye damage – Category 1 if:

311 Guidance on the Application of the CLP Criteria 5.0 – July 2017 311 Version i. at least in one animal effects on the cornea, iris or conjunctiva that are not expected to reverse or have not fully reversed within an observation period of normally 21 days; and/or at least 3 out of 5 rabbits show a mean score per animal of  b. 3 for corneal opacity and/or > 1.5 for iritis. Classification for eye irritation – i. Category 2 if at least 3 out of 5 rabbits show a mean score per animal of:  1 for corneal opacity and/or ii. iii. 1 for iritis and/or   2 conjunctival erythema (redness) and/or iv.  2 conjunctival oedema (swelling) (chemosis) v. and which fully reverses within an observation period of normally 21 days. 4 In the case of the following applies: rabbits, c. Classification for serious eye damage – Category 1 if: i. at least in one animal effects on the cornea, iris or conjunctiva that are not expected to reverse or have not fully reversed within an observation period of normally 21 days; and/or ii. at least 3 out of 4 rabbits show a mean score per animal of 3 for corneal opacity and/or  > 1.5 for iritis d. Classification for eye irritation Category 2 if at least 3 out of 4 rabbits show a mean – score per animal of:  i. 1 for corneal opacity and/or ii.  1 for iritis and/or  2 conjunctival erythema (redness) and/or iii. iv.  2 conjunctival oedema (swelling) (chemosis) and which fully reverses within an observation period of normally 21 days. ts show a mean score In this case the irritant categories 1 and 2 are used if 4 of 6 rabbi per animal as outlined in the criteria. Likewise, if the test was performed with 4 or 5 animals, for at least 3 individuals the mean score per animal must exceed the values laid down in the classification criteria. A single animal show ing irreversible or otherwise serious effects consistent with corrosion will necessitate classification as serious eye damage Category 1 irrespective of the number of animals used in the test. Other animal tests The LVET uses the same scoring system as for results from the OECD TG 405. However, the differences between the LVET and OECD TG 405 test methods, may result in a classification in a lower category (or no classification) based on LVET data, than if the cla ssification was based on data derived from the standard in vivo test (OECD TG 405 (TM B.5)). See also 3.3.2.1.5.2 above. Note that in case there are test data that originate from non - OECD tests and scoring has not been performed according to the Draize sys tem, the values in CLP Annex I, Tables 3.3.1 and 3.3.2 are not applicable for classification purposes. However these data from non - OECD tests should be considered in a weight of evidence determination.

312 ication of the CLP Criteria Guidance on the Appl 312 Version 5.0 – July 2017 3.3.2.3.3. Weight of evidence criteria should be applied to available information. However, According to Article 9(1) CLP, the sometimes it is not straightforward or simple to apply the criteria and according to Article 9(3) a weight of evidence and expert judgement should be applied in such cases when the criteria cann ot be applied directly. A weight of evidence determination means that all available and scientifically justified information bearing on the determination of hazard is considered together, such as human accident databases, epidemiological and experience (including occupational data and data from clinical studies, and well - documented case reports and observations), relevant animal data, skin - chemical parameters (e.g. pH, reserve alkalinity/acidity), irritation information/data, physico tests, information from the application of the category approach the results of suitable in vitro - across), QSAR results. The quality and consistency of the data shall be given (grouping, read appropriate weight. Both positive and negative results shall be assembled together in a single weight of evidence determination. Evaluation must be performed on a case - by - case basis and with expert judgement. However, normally positive results that are adequate for classification , Annex I, CLP and S 1.4 should not be overruled by negative findings (see also 1.1.1.3 of ection this guidance). Annex I: 1.1.1.4 . For the purpose of classification for health hazards (Part 3) established hazardous effects seen in appropriate animal studies or from human experience that are consistent with the criteria for classification shall normally justify classification. Where evidence is available from both humans and animals and there is a conflict b etween the findings, the quality and reliability of the evidence from both sources shall be evaluated in order to resolve the question of classification. Generally, adequate, reliable and representative ientifically valid case studies as specified data on humans (including epidemiological studies, sc in this Annex or statistically backed experience) shall have precedence over other data. designed and conducted epidemiological studies may lack a sufficient However, even well - number of subjects to detect relat ively rare but still significant effects, to assess potentially confounding factors. Therefore, positive results from well - conducted animal studies are not necessarily negated by the lack of positive human experience but require an assessment of ness, quality and statistical power of both the human animal data. the robust For additional guidance, if both human and animal data are available, see the Guidance on IR&CSA Section R.7.2.3.2. Additional guidelines on the assessment of available information when Wo E need s to be applied is provided in S ection 3.2.2.3.3 (see Figure 3 . 2 ). 3.3.2.4. Decision on classification serious eye damage which is indicated in the A skin corrosive substance is also classified for hazard statement for skin corrosion (H 314: Causes severe skin burns and eye damage). (Skin Corr. 1 and Eye Dam. 1) is required However, although classification for both endpoints and has to be addressed in the safety data sheet, the hazard statement H318 ‘Causes serious eye damage’ is not indicated on the label because of redundancy (CLP Article 27). In other cases, if the comparison of the information related to serious eye damage/eye irritation ia shows that the criteria are met, the substance is classified for serious eye with the criter damage or eye irritation. 3.3.2.5. Setting of specific concentration limits Article 10(1) Specific concentration limits and generic concentration limits are limits nce indicating a threshold at or above which the presence of that assigned to a substa

313 Guidance on the Application of the CLP Criteria 5.0 – July 2017 313 Version substance in another substance or in a mixture as an identified impurity, additive or individual constituent leads to the classification of the substance or mixture as hazardous. Specific co ncentration limits shall be set by the manufacturer, importer or downstream user where adequate and reliable scientific information shows that the hazard of a substance is evident when the substance is present at a level below the concentrations set for an y hazard class in Part 2 of Annex I or below the generic concentration limits set for any hazard class in Parts 3, 4 and 5 of Annex I. [...] It is more difficult to prove the absence of a hazardous property, the legal text states that: Article 10(1) [...] In exceptional circumstances specific concentration limits may be set by the manufacturer, importer or downstream user where he has adequate, reliable and conclusive scientific t a level information that a hazard of a substance classified as hazardous is not evident a above the concentrations set for the relevant hazard class in Part 2 of Annex I or above the generic concentration limits set for the relevant hazard class in Parts 3, 4 and 5 of that Annex. A specific concentration limit (SCL) set in accordance with the above mentioned provisions shall take precedence over the generic concentration limit (GCL) set out in Tables 3.2.3 and 3.2.4 of Annex I to CLP (Article 10(6)). Furthermore, such an SCL is substance - specific and should be applicable to all mixtur es containing the substance instead of any GCL that otherwise would apply to a mixture containing the substance. What type of information may be the basis for setting a specific concentration limit? Existing human data may in certain cases (especially if - response information is available) dose indicate that the threshold for the irritation hazard in humans for a substance in a mixture, would be higher or lower than the GCL. A careful evaluation of the usefulness and the validity of such human data as well as their representativeness and predictive value ( IR&CSA , sections R.4.3.3. and R.7.2.4.2) should be performed. As pointed out in Section 1.1.1.4 of Annex I, CLP, positive results from well conducted animal studies are not necessarily negated by the lack o f - positive human experience but require an assessment of robustness, quality and a degree of statistical certainty of both the human and animal data. 65 The aim of the standard test method for ‘Acute Eye Irritation/Corrosion’ OECD TG 405 is to pote ntial identify . The test material is generally serious eye damage or eye irritation administered undiluted. Thus, no dose - response relationship can be obtained from an individual test. However, if there are adequate, reliable, relevant and conclusive existing data from other already performed animal studies with a sufficient number of animals tested to ensure a high degree of certainty, and with information of dose - response relationships, such data may be igher SCL on a case - by considered for setting a lower or, in exceptional cases, a h case basis. - It should be noted that generating data specifically for the purpose of setting SCLs is not a requirement according to the CLP Regulation. Article 8(1) of CLP specifies that new tests may only be performed (in order to de termine the hazard of a substance or mixture) if all other means of generating information has been exhausted and Article 7(1) specifies that where new tests are carried out, test on animals shall be undertaken only when no other alternatives, 65 TO NOTE: In OECD TG 404 the term test substance refers to the test material, test article or test item. The term substance may be used differently from the REACH/CLP definition .

314 Guidance on the Appl ication of the CLP Criteria Version 5.0 – 314 July 2017 which provid e adequate reliability of data, are possible. The GCLs must be applied for the classification of a mixture on the basis of its ingredient substances classified as causing serious eye damage or as an eye irritant, if there are no already existing specific d ata justifying an SCL which is lower or, in exceptional cases, higher than the GCL (see Article 10(1), CLP). Therefore, always be available, for mixtures containing substances already classified for information will serious eye damage/eye irritation, makin g it possible to identify the hazard for the mixture by using the GCLs (Article 9(4), CLP). The possibilities to use in vitro test methods as a basis for setting SCLs have not yet been explored and therefore, at the present point in time, it is not possib le to provide guidance for methods for the purpose of setting SCLs. However, this does not exclude that the use of in vitro a method to set SCLs based on tests could be developed in the future , and these tests in vitro provide a promising option for SCL setting . An SCL should apply to any mixture containing may the substance instead of the GCL (that otherwise would apply to the mixture containing the substance). Thus, if the SCL is based on data derived from tests with dilutions of the substance in a spec ific solvent, it has to be considered that the derived concentration, should be applicable to all mixtures for which the SCL should apply. Annex VI Part 3 to CLP Regulation includes e xamples of substances for which a higher or lower SCL was set under Dire ctive 67/548/EEC (old Dangerous Substances Directive (DSD) system) which have been included in CLP. 3.3.2.6. Decision logic for classification of substances The decision logic, based on th at provided by the GHS, is reported as additional guidance below. It is strongly recommended that the person responsible for classification study the criteria for classification before and during use of the decision logic.

315 Guidance on the Application of the CLP Criteria 5.0 – Version 315 July 2017 No n to Are there data and/or informatio Classification not possible ? serious eye damage/eye irritation evaluate Yes Does the substance have potential to cause serious eye damage (see criteria in CLP, Annex I, 3.3.1, 3.3.2.1.1, 3.3.2.2 and Figure Category 1 a 3.4 in this guidance) considering : ; Existing human eye data (a) Yes ; versible eye damage in one or more test animals Irre (b) (c) Existing human or animal data indicating skin corrosion; (d) Other existing animal eye data including single or repeated exposure; Danger ex vivo/in vitro Existing (e) eye data; b ( ) pH extremes of ≤2 or ≥11.5 f ; (g) Information available from validated Structure Activity Relationship methods? No Yes Category 2 Is the substance an eye irritant (see criteria in CLP, Annex I, 3.3.1, 3.3.2.1.2, 3.3.2.2 and Figure 3.4 in this guidance) a : considering ; (a) Existing human data, single or repeated exposure (b) Eye irritation data from an animal study ; (c) Other existing animal eye data including single or Warning repeated exposure; data; ex vivo/in vitro Existing (d) able from validated Structure Activity (e) Information avail Relationship methods? No No classification a Taking into account consideration of the total weight of evidence as needed. b Not applicable if consideration of pH and acid/alkaline reserve indicates the substance may not cause serious eye test. damage and confirmed by other data, preferably by data from an appropriate validated in vitro Classification of mixtures for serious ey e damage/eye irritation 3.3.3. 3.3.3.1. Identification of hazard information As for substances, the procedure for classifying mixtures is a tiered i.e. a stepwise approach principle and depending on the type and amount of available based on a hierarchy the mixture, followed by a arting from evaluating existing human data on data/information st ex vivo / in vitro data, - chemical in vivo thorough examination of the existing and finally physico , above). properties, available on the mixture (as illustrated in Figure 3 . 4

316 ication of the CLP Criteria Guidance on the Appl 316 Version 5.0 – July 2017 If valid test data are available for the whole mixture they have precedence. If no such data exist, the so called bridging principles should be applied if poss ible. If the bridging principles are not applicable an assessment on the basis of data for the components of the mixture must be applied. For mixtures that have been on the market for a long time, some human data and experience may exist that could provide useful information on the eye irritation potential of the respective mixtures. However, lack of data on effects in humans may be due to, for example, poor reporting or adequate preventive measures. Therefore, lack of human data cannot be taken as evidence - hazardous. See Section 3.3.2. 1.1 of this Guidance for further of the mixture being non information on the identification of human data. Where it is decided to base the classification of a mixture upon consideration of pH alone, Eye Damage Category 1 should be applied. In this case no further retrieval of information on the mixture itself is needed. Classification criteria for mixtures 3.3.3.2. The information available related to serious eye damage and eye irritation, will determine if the mixture should be classified using the approaches below in the following sequence (CLP Article 9): a. ubstance criteria Classification derived using data on the mixture itself, by applying the s of Annex I to CLP Classification based on the application of bridging principles, which make use of test data b. on similar tested mixtures and ingredient substances c. Classification based on calculation and/or on concentration thresholds, including SCLs and M - factors. 3.3.3.2.1. When data are available for the complete mixture The mixture shall be classified using the criteria for substances, and Annex I: 3.3.3.1.1. taking into a ccount the tiered approach to evaluate data for this hazard class. : 3.3.3.1.2. When considering testing of the mixture classifiers are encouraged to Annex I use a tiered weight of evidence approach as included in the criteria for classification of substanc es for skin corrosion and serious eye damage/eye irritation to help ensure an accurate classification, as well as avoid unnecessary animal testing. In absence of any other information, a mixture is considered to cause serious eye damage (Category 1) if it has a pH ≤ 2,0 or ≥ 11,5. However, if consideration of alkali/acid reserve suggests the mixture may not cause serious eye damage despite the low or high pH value, this needs to be confirmed by other data, preferably data from an appropriate validated in vi tro test. As for substances, where the criteria cannot be applied directly to available identified information, a weight of evidence determination using expert judgement should be used according to CLP Article 9(3) when evaluating the data in order to be able to apply the criteria to the information (according to CLP Article 9(1)) (see 3.3.2.3.3. Weight of evidence above). The integration of all information to come to a final hazard assessment based on weight of evidence in general requires in - depth toxi cological expertise. For guidance on the assessment of the information available for mixtures when WoE needs to be applied, please see Figure 3 . 2 in S e ction 3.2.2.3.3. There are a number of available test systems that have been validated to identify in vitro substances causing serious eye damage (Category 1) and/or no cla ssification (see S ection 3.3.2.1.5.1 ), that are considered to be valid also for mixtures. However, not all available in vitro specific applicability domain, test systems work equally well for all types of mixtures. The

317 Guidance on the Application of the CLP Criteria – Version 317 5.0 July 2017 limitations of the use of the test methods for mixtures should be considered. Thus, including assay for classification purposes, it has to be in vitro prior to testing a mixture in a specific for the prediction of en sured that the respective test has been previously shown to be suitable serious eye damage/eye irritation properties for the type of mixture to be evaluated. There are no in vitro tests with regulatory acceptance for eye irritation at present. A proposal to dentify eye irritants has been presented in a draft combine results of multiple in vitro tests to i OECD Guidance document (ref. OECD 2015). Mixtures with extreme pH 3.3.3.2.1.1. As a general rule, mixtures with a pH of ≤ 2 or ≥ 11.5 should be considered as corrosive. of the mixture indicated by its acid or alkali However, assessment of the buffering capacity reserve should be considered (see 3.2.3.2.1.1.) Where the mixture has an extreme pH value but the only corrosive/irritant ingredient present in nnex VI or set by supplier the mixture is an acid or base with an assigned SCL (either CLP A according to Article 10(1), CLP), then the mixture should be classified according to the SCL. In this instance, pH of the mixture should not be considered a second time since it would have iving the SCL for the substance. already been taken into account when der If this is not the case, then the steps to be taken into consideration when classifying a mixture 2 or  11.5 are described in the following decision logic. with pH  3 . 5 Mixture not classified as Skin Corr. 1 and without animal or human data on serious Figure ta from similar tested mixtures, pH is  2 or  11.5 eye damage/eye irritation or relevant da Does the acid/alkaline reserve indicate that the mixture may not be Classify as corrosive, corrosive? Skin Corr. 1 and serious eye damaging, Eye  NO Dam. 1. YES  Is the mixture tested for serious eye damaging properties in an OECD Classify as serious eye test damaging, Eye Dam. 1. in vitro adopted or internationally accepted scientifically valid considered to be valid and applicable for the mixture? NO  YES  Classify as serious eye Does the mixture demonstrate serious eye damaging properties in an in vitro test OECD adopted or internationally accepted scientifically valid damaging, Eye Dam. 1. considered valid and applicable for the mixture? YES  NO  Consideration of the total weight of available evidence, in particular in case of conflicting data, including extreme pH, negative/inconclusive in vitro tests and results from the results from (e.g.) eye irritation application of the methods based on the i ngredients in the mixture in 3.3.3.3.3 (Table 3.3.3) / 3.3.3.3.4.1 - CLP Annex I, 3.3.3.3.2 -  3.3.3.3.4.3 (Table 3.3.4)

318 ication of the CLP Criteria Guidance on the Appl 318 Version 5.0 – July 2017 Classify: Category 1, no Category 2, classification. Thus, if consideration of extreme pH and acid/alkaline reserve indicates the mi xture may not have the potential to cause serious eye damage, then the supplier should carry out further (CLP Annex I, Section testing to confirm this, preferably an appropriate validated in vitro test Eye damage Category 1 if the supplier 3.3.3.1.2). The mixture must be classified as Serious decides not to carry out the required confirmatory testing. If further testing confirms that the mixture should not be classified for serious eye damage tation either using in vitro eye effects, then the supplier should assess the mixture for eye irri irritation test methods when available and considered appropriately valid and applicable for the mixture or the methods based on ingredients. - acid/alkali reserve method assumes that the potential corrosivity It must be noted that the pH or irritancy is due to the effect of the ionic entities. When this is not the case, especially when the mixture contains non ionic (non - ionisable) substances themselves classified as corrosive or - - acid/alkali rese rve method cannot be a basis for modifying the irritant, then the pH classification but should be considered in the weight of evidence analysis. Where the mixture has an extreme pH value and contains some other corrosive/irritant ingredients (some of which may have SCLs assigne d) in addition to an acid or base with or without an assigned SCL, then the steps described in the above decision logic shall be followed. 3.3.3.2.2. When data are not available for the complete mixture: bridging principles Annex I: 3.3.3.2.1. f has not been tested to determine its skin Where the mixture itsel corrosivity or potential to cause serious eye damage/eye irritation, but there are sufficient data on the individual ingredients and similar tested mixtures to adequately characterise the these data shall be used in accordance with the bridging rules set out hazards of the mixture, in section 1.1.3. In order to apply bridging principles, there needs to be sufficient data on similar tested mixtures on the ingredients of the mixture (see Section 1.6.3 of as well as this Guidance). When the available identified information is inappropriate for the application of the bridging described in principles then the mixture should be classified based on its ingredients as Sections 3.3.3.2 and 3.3.3.3 of this Guidance. 3.3.3.2.3. When data are available for all ingredients or only for some ingredients of the mixture 3.3.3.2.3.1. Ingredients that should be taken into account for the purpose of classification The ‘relevant ingredients’ of a mixture are those which are present [...] Annex I: 3.3.3.3.1. in concentrations ≥ 1% (w/w for solids, liquids, dusts, mists and vapours and v/v for gases), unless there is a presu mption (e.g. in the case of corrosive ingredients) that an ingredient present at a concentration < 1% can still be relevant for classifying the mixture for serious eye damage/eye irritation. 3.3.3.2.3.2. The additivity approach is applicable Annex I: 3.3.3.3.2. In general, the approach to classification of mixtures as seriously damaging to the eye/eye irritant when data are available on the ingredients, but not on the mixture as a whole, is based on the theory of additivity, such that each skin corrosive or eye damaging/eye irritation ingredient contributes to the overall serious eye serious

319 Guidance on the Application of the CLP Criteria 5.0 – July 2017 319 Version damage/eye irritation properties of the mixture in proportion to its potency and concentration. A weighting factor of 10 is used for skin corrosive and serious eye damaging ingre dients when they are present at a concentration below the generic concentration limit for classification with Category 1, but are at a concentration that will contribute to the classification of the mixture as eye irritant. The mixture is classified as ser iously damaging to the eye or eye irritant when the sum of the concentrations of such components exceeds a concentration limit. Annex I: 3.3.3.3.3. Table 3.3.3 provides the generic concentration limits to be used to determine if the mixture shall be classified as seriously damaging to the eye or as eye irritant. When the supplier is unable to derive the classification using either data on the mixture itself or bridging principles, he must determine the serious eye damage/eye irritation properties of his mixture using data on the individual ingredients. Although the general approach is the additivity , the supplier must principle which has been successfully used under the DPD and more recently ascertain whether the additivity approach is applicable wher e all relevant ingredients should be considered. The first step would then be to identify all the relevant ingredients in the mixture (i.e. their name, chemical type, concentration level, hazard classification and any SCLs) and the pH of the mixture. In ad dition, it is important to also consider effects that could occur in the whole mixture, such as surfactant interaction, neutralisation of acids/bases apart from effects of the entire mixture (i.e. pH and the alkaline reserve) and not only consider the cont ribution of individual ingredients. Additivity may not apply where the mixture contains substances mentioned in CLP Annex I, - 3.3.3.3.4.3 which may be corrosive/irritant at concentrations below 1%, see 3.3.3.3.4.1 3.3.3.2.3.3 of this Guidance. Section Application of SCLs when applying the additivity approach The generic concentration limits are specified in Table 3.3.3. However, CLP Article 10(5) ind icates that specific concentration limits (SCLs) take precedence over generic concentration limits. Thus, if a given substance has an SCL set in accordance with Article 10(1), CLP, then this specific concentration limit has to be taken into account when ap plying the summation (additivity) method for serious eye damage/eye irritation (see Examples 4 and 5). In cases where additivity applies for serious eye damage/eye irritation to a mixture with two or more substances some of which may have SCLs assigned, th en the following formula should be used: The mixture is classified for serious eye damage/eye irritation if the Sum of (ConcA / clA) + (ConcB / clB) + ...+ (ConcZ / clZ) is  1 Where ConcA = the concentration of substance A in the mixture; tration limit (either specific or generic) of substance A; clA = the concen ConcB = the concentration of substance B in the mixture; clB = the concentration limit (either specific or generic) of substance B; etc. 3.3.3.2.3.3. The additivity approach is not applicable 3.3.3.3.4.1. Annex I: Particular care must be taken when classifying certain types of mixtures containing substances such as acids and bases, inorganic salts, aldehydes, phenols, and surfactants. The approach explained in paragraphs 3.3.3.3.1 and 3.3.3.3.2 might n ot work given that many of such substances are seriously damaging to the eye/eye irritant at concentrations < 1 %. Annex I: 3.3.3.3.4.2. For mixtures containing strong acids or bases the pH shall be used as classification criteria (see s ection 3.3.3.1.2) since pH will be a better indicator of serious eye

320 ication of the CLP Criteria Guidance on the Appl 320 Version 5.0 – July 2017 damage (subject to consideration of acid/alkali reserve) than the generic concentration limits of Table 3.3.3. 3.3.3.3.4.3. A mixture containing skin corrosive or serious eye damaging/eye Annex I: irritant ingredients that cannot be classified based on the additivity approach (Table 3.3.3), due to chemical characteristics that make this approach unworkable, shall be classified as 1 % of a skin corrosive or se Serious Eye Damage (Category 1) if it contains ≥ rious eye ≥ 3 % of an irritant damaging ingredient and as Eye Irritation (Category 2) when it contains ingredient. Classification of mixtures with ingredients for which the approach in Table 3.3.3 does not apply is summarised in Table 3.3.4. 3.3.3 .3.5 . On occasion, reliable data may show that the effects of serious eye Annex I: damage/eye irritation of an ingredient will not be evident when present at a level at or above d in Tables 3.3.3 and 3.3.4 in s ection the generic concentration limits mentione 3.3.3.3.6. In e these cases the mixture shall be classified according to those data (se also Articles 10 and 11). On other occasions, when it is expected that the skin corrosion/irritation hazards or the effect of serious eye damage/eye irritation an ingred ient will not be evident when present at a at or level testing above the generic concentration limits mentioned in Tables 3.3.3 and 3.3.4, of the mixture shall be considered. In those cases, the tiered weight of evidence strategy shall be applied. Annex I: If there are data showing that (an) ingredient(s) may be corrosive to the 3.3.3.3.6. skin or seriously damaging to the eye/eye irritating at a concentration of < 1 % (corrosive to the skin or seriously damaging the eye) or < 3 % (eye irritant), the mixtur e shall be classified accordingly. 3.3.3.3. Generic concentration limits for substances triggering classification of mixtures 3.3.3.3.1. When the additivity approach is applicable Table 3.3.3 Annex I: n corrosion Generic concentration limits of ingredients of a mixture classified as ski (Category 1, 1A, 1B or 1C) and/or serious eye damage (Category 1) or eye irritation (Category 2) that trigger classification of the mixture as eye damage/eye irritation where additivity approach applies Concentration triggering classification of a mixture as: Sum of ingredients classified as: Eye irritation Serious eye damage Category 1 Category 2 Skin corrosion Sub - Category 1A, 1 % but < 3 %   3 % 1B, 1C or Category 1 + Serious a eye damage ( Category 1)( ) Eye irritation (Category  10 % 2) 10 x (Skin corrosion Sub - 10 %  Category 1A, 1B, 1C or Skin corrosion Category 1 + Serious eye damage (Category 1)) + Eye irritation (Category 2)

321 Guidance on the Application of the CLP Criteria 5.0 – July 2017 321 Version a - ( and Serious Category 1A, 1B, 1C or Category 1 ) If an ingredient is classified as both Skin Corrosion Sub Eye Damage (Category 1), its concentration is considered only once in the calculation. 3.3.3.3.2. When the additivity approach is not applicable Table 3.3.4 Annex I: Generic concentration limits of ingredients of a mixture as serious eye damage the additivity approach does not (Category 1) or eye irritation (Category 2), where apply Concentration Mixture classified as Ingredient Serious eye damage Acid with pH 2 ≥ 1% ≤ (Category 1) Serious eye damage ≥ 1% Base with pH 11,5 ≥ (Category 1) Other ingredient classified as skin corrosion Serious eye damage 1% ≥ (Sub 1A, 1B, 1C or Category 1) Category - (Category 1) 1) or serious eye damage (Category Other ingredient classified as eye irritation Category Eye irritation ( 3% ≥ (Category 2) 2) 3.3.3.4. Decision logic for classification of mixtures The decision logic, based on the one provided in the GHS, is presented here below as additional guidance. It is strongly recommended that the person responsible for classification, study the criteria for classification before and during use of the decision logic.

322 Guidance on the Appl ication of the CLP Criteria 5.0 – July 2017 322 Version No Does the mixture as a whole or its ingredients have Classification not possible data/information to evaluate serious eye damage/eye irritation? Yes Yes data/information to Does the mixture as a whole have See decision evaluate serious eye damage/eye irritation? 3.3.2.6 logic No Yes Classify in appropriate Can bridging principles be applied? category No Yes Follow decision logic in of Section 3.3.3.2.1.1 Is pH of the mixture ≤2 or ≥11.5? this guidance and classify accordingly No Category 1 Yes a the causes ≥1% Does mixture contain of an ingredient which serious eye damage when additivity approach may not apply? Danger No Category 1 the mixture contain one or more ingredients corrosive Does Yes or seriously damaging to the eye when the additivity approach applies and where the sum of concentrations ingredients ? classified as Skin Corr. Cat. 1 + Eye Dam. Cat. 1 ≥3% Danger No

323 Guidance on the Application of the CLP Criteria 5.0 – 323 Version July 2017 Category 2 a Does the mixture contain ≥3% of an ingredient which is Yes an eye irritant and when the additivity approach may not apply? Warning No Does the mixture contain one or more ingredients Category 2 irritant corrosive or seriously damaging to the eye/eye when the additivity approach applies and where the sum of Yes concentrations of ingredients classified as: (a) Eye Dam. Cat. 1 + Skin Corr. Cat. 1 ≥1% but <3%; or (b) Eye Irrit. Cat. 2 ≥10%; or b Warning it. ) + Eye Irr (c) 10 x (Skin Corr. Cat. 1 + Eye Dam. Cat. 1 Cat. 2 ≥10%? No Not classified a Where relevant < 1%, see S ection 3.3.3.3.1 of Annex I of CLP. b If an ingredient is classified as both skin Category 1 and eye Category 1 its concentration is considered only once in the calculation. 3.3.4. Hazard communication in form of labelling for serious eye damage/eye irritation Pictograms, signal words, hazard statem 3.3.4.1. ents and precautionary statements 3.3.4.1 Label elements shall be used for substances or mixtures meeting the criteria Annex I: for classification in this hazard class in accordance with Table 3.3.5. Table 3.3.5 a) Label elements for serious eye damage/eye ir ritation( Category 2 Classification Category 1

324 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 324 GHS Pictograms Danger Warning Signal Word H319: Causes serious eye Hazard Statement H318: Causes serious eye damage irritation P280 Precautionary Statement P264 Prevention P280 Precautionary Statement P305 + P351 + P338 P305 + P351 + P338 Response P310 P337 + P313 Precautionary Statement Storage Precautionary Statement Disposal a ( ) Where a chemical is classified as skin corrosion Sub - Category 1A, 1B, 1C or Category 1, labelling for serious eye damage/eye irritation can be omitted as this information is already included in the hazard statement for skin corrosion Category 1 (H314).' A skin corrosive mixture is considered to also cause serious eye damage which is indicated in the hazard statement for skin corrosion, H314: Causes severe skin burns and eye damage. Thus, in this case a mixture has to be classified for both classifications (Skin Corr. 1 and Eye Dam. 1) but the hazard statement H318 ‘Causes serious eye damage’ is not in dicated on the label because of redundancy (CLP Article 27). 3.3.5. Examples of classification for serious eye damage/eye irritation 3.3.5.1. Examples of substances fulfilling the criteria for classification mals 3.3.5.1.1. Example 1: Standard test according to OECD TG 405 with three ani In a study according to OECD 405 the test substance was applied on the eyes of three rabbits. The scoring results obtained are listed in the following table: Cornea: Evaluation after ... Positive responder? Animal  Score ... No.  1  3 21 days 1 hr 24 hrs 48 hrs 72 hrs 1 0 2 2 2 0  24/48/72 h animal 1 is 2 Yes No 2 2 2 2 2 0

325 Guidance on the Application of the CLP Criteria 5.0 Version 325 – July 2017 24/48/72 h animal 2 is 2  Yes No 3 1 1 0 2 2 24/48/72 h animal 3 is 1.3  Yes No Effects are reversible Iris: Evaluation after ... responder? Positive Animal  Score ... No.  1 72 hrs 21 days 1 hr 48 hrs > 1.5 24 hrs 1 0 1 1 1 0  24/48/72 h animal 1 is 1 Yes No 1 1 1 2 0 1 24/48/72 h animal 2 is 1  Yes No 3 1 1 1 1 0  24/48/72 h animal 3 is 1 Yes No Effects are reversible Erythema: – Conjunctiva Evaluation after ... Positive responder? Animal Score ...  No.  2 72 hrs 21 days 1 hr 24 hrs 48 hrs 0 2 2 2 2 1  24/48/72 h animal 1 is 2 Yes 2 1 1 1 0 1  24/48/72 h animal 2 is 1 No 3 1 1 1 1 0 24/48/72 h animal 3 is 1  No Effects are reversible

326 Guidance on the Appl ication of the CLP Criteria 326 5.0 – July 2017 Version – Conjunctiva Swelling: Evaluation after ... Positive responder? Animal Score ...  No. 2  72 hrs 21 days 48 hrs 24 hrs 1 hr 0 3 3 0 1 3  24/48/72 h animal 1 is 3 Yes 2 2 2 1 0 2  24/48/72 h animal 2 is 1.7 No 2 0 3 2 2 3 24/48/72 h animal 3 is 2.3  Yes Effects are reversible Classification according to CLP: Eye irritant Category 2 Rationale: 1: 3/3 animals Cornea ‘positive responder’   2: 2/3 animals and/or Conjunctiva ‘positive responder’ Iris ‘positive responder’  1: 3/3 animals and/or Example 2: Test carried out with more than 3 rabbits 3.3.5.1.2. Cornea: Evaluation after ... Positive responder? Anim  Score ... al No.  3  1 21d 14d 24h 48h 72h 1h 7d 1 1 2 3 3 1 1 0 24/48/72h = 2.7  no yes 2 1 2 2 3 1 1 0  24/48/72h = 2.3 no yes 1 2 3 3 2 1 0 3  24/48/72h = 2.7 no yes 4 1 2 4 4 2 1 0 24/48/72h = 3.3  yes yes Effects are reversible

327 Guidance on the Application of the CLP Criteria 5.0 – July 2017 327 Version Iris: Evaluation after ... Positive responder? Anim Score ...  al No. 1  72h 7d 14d 21d > 1.5 48h 24h 1h 0 0 0 0 0 0 0 1  24/48/72h = 0 no no 0 0 2 0 0 0 0 0 24/48/72h = 0  no no 1 1 1 1 0 0 0 3  24/48/72h = 1 no yes 0 0 0 4 0 0 0 0  24/48/72h = 0 no no Effects are reversible – Conjunctiva Erythema: Evaluation after ... Positive responder? Anim  Score ... al No.  2 48h 72h 7d 14d 21d 24h 1h 2 2 1 1 1 1 0 2 24/48/72h = 1.7  no 2 2 2 2 1 1 0 0  24/48/72h = 1.7 no 1 2 3 1 1 2 1 2  24/48/72h = 1.7 no 4 2 2 2 1 0 0 0 24/48/72h = 1.7  no Effects are irreversible

328 ication of the CLP Criteria Guidance on the Appl 328 Version 5.0 – July 2017 Conjunctiva Swelling: – Evaluation after ... Positive responder? Anim Score ...  al No.  2 72h 7d 14d 21d 24h 48h 1h 2 2 2 1 1 1 0 1  24/48/72h = 1.7 no 2 2 1 1 1 0 0 2  24/48/72h = 1.3 no 1 2 2 3 1 2 1 1  24/48/72h = 1.7 no 4 2 2 1 1 1 1 2  24/48/72h = 1.7 no Effects are irreversible Category 1 Classification according to CLP: Serious eye damage Rationale: Conjunctiva with irreversible effects 3.3.5.2. Examples of mixtures fulfilling the criteria for classification Example 3: Application of the additivity approach for mixtures containing 3.3.5.2.1. ingredients without SCLs ingredients with no assigned SCLs, then the appropriate Where the mixture is made up of summation(s) from CLP Annex I, Table 3.3.3 should be used. Skin / eye classification Concentration Ingredient SCL (% w/w) Eye Dam. 1 Substance A 1.8 Not assigned Substance B Irrit. 2 0.5 Not assigned Eye Dam. 1 Eye 5.4 Not assigned Substance C Substance D Not classified 4.0 Acid E Corr. 1A Skin 2.0 Not assigned Water Not classified 86.3 pH of the mixture is 9.0 – 10.0, thus extreme pH provisions do not apply. The mixture contains an acid but no surfactant. Additivity is considered to apply. Substance D and water can be disregarded as they are not classified for serious eye damage/eye irritation. Substance B can also be disregarded as present below 1%.

329 Guidance on the Application of the CLP Criteria 5.0 – July 2017 329 Version Dam. Mixture contains 7.2% Eye 1 ingredients as well as 2% acid E so the summation {Skin 1} applies and is > 3%, thus mixture is classified Eye corrosion Cat 1A, 1B, 1C + Eye Dam. Dam. 1. Example 4: Application of the additivity approach for mixtures containing 3.3.5.2.2. ingredients which may ha ve SCLs Skin / eye classification Concentration Ingredient SCL (% w/w) 2.0 Dam. 1 Not assigned Substance A Eye Eye Irrit. Substance B 0.5 Not assigned 2 Substance C Skin Corr. 1B 5.4 C ≥ 10 %: Skin Corr. 1B Irrit. 2 5 % ≤ C < 10 %: Eye Not classified 4.0 Substance D Substance E 1B 2.0 Not assigned Skin Corr. Water Not classified 86.1 pH of the mixture is 10.5 – 11.0, thus extreme pH provisions do not apply. Additivity is considered to apply. classified for serious eye Substance D and water can be disregarded as they are not damage/eye irritation. Substance B can also be disregarded as present below 1%. SCLs are not assigned to substance E or substance A, thus generic concentration limits (GCL) apply for these ingredients Eye 1 Dam. (% Substance A / GCL) + (% Substance C / SCL) + (% Substance E / GCL) = (2/3) + (5.4/10)  > 1 thus mixture is classified Eye Dam. 1 + (2/3) = 1.9 3.3.5.2.3. Example 5: Application of the additivity approach for mixtures containing ingredients which may have SCLs gredient Serious eye damage/ In Concentration SCL (% w/w) eye irritation classification Substance B Not assigned 0.7 1 Eye Dam. Irrit. 2 Substance C 74.9 Eye Not assigned Substance D 8.5 1 1 Eye C ≥ 25 %: Eye Dam. Dam. 10 % ≤ C < 25 %: Eye Irrit. 2 Substance E Not classified 15.9 pH of the mixture is 10.0 – 10.5 (10% solution), thus extreme pH provisions do not apply. Additivity is considered to apply. Substance E can be disregarded as it is not classified for serious eye damage/eye irritation. Substance B can also be disregarded as present below 1%.

330 ication of the CLP Criteria Guidance on the Appl 330 Version 5.0 – July 2017 SCLs are not assigned to substance C, thus GCL apply for this ingredient Eye 1 Dam. Mixture contains 8.5% substance D, the only ‘relevant’ ingredient classified as Eye Dam. 1. As Dam. 1 this is below the 25% SCL for sub stance D, the mixture is not classified Eye 2 Irrit. Eye (%substance D/ SCL) + (%substance C / GCL) = (8.5/10) + (74.9/10) which is > 1 thus mixture is classified Eye Irrit. 2 References 3.3.6. otoxicity/cell function based in ECVAM/ESAC (2009a) Statement on the scientific validity of cyt - http://ecvam.jrc.it/ vitro assays for eye irritation testing. Online: ECVAM/ESAC (2009b) Statement on the use of existing low volume eye test (LVET) data for weight of evidence decisions on classification and labelling of cleaning products and their main ingredients. Online: http://ecvam.jrc.it/ J.F., Nixon G.A., Bruce R.D., Reer P.J., Bannan Griffith E.A. (1980), Dose - response studies with chemical ir ritants in the albino rabbit eye as a basis for selecting optimum testing conditions for Toxicol Appl Pharmacol predicting hazard to the human eye. , 501 - 513. 55 ., Scott L ., Eskes C ., Hoffmann S ., Adriaens E ., Alepée N ., Bufo M ., Clothier R ., Facchini D Faller ., C Harbell J ., Hartung T ., Kamp H ., Varlet B.L ., Meloni M , McNamee P ., Osborne R ., Guest R ., ., Pfannenbecker U ., Prinsen M ., Seaman C ., Spielmann H ., Stokes W ., Trouba K ., Pape W Vinardell P ., ., Vassallo M ., Goethem F.V ., Zuang V . (2010), A proposed eye irritation Berghe C.V - - Up and Top Down testing strategy to reduce and replace in vivo studies using Bottom approaches. 24 , 1 - 9 . Toxicol in Vitro J.R., How M.J., Walker A.P., Worth W.M.H. (1988), Classification as corrosive or irritant Young to skin of preparations containing acidic or alkaline substances, without test on animals. Toxicol in Vitro 2 , 19 - 26. Young J.R., How M.J. (1994), Product cla ssification as corrosive or irritant by measuring pH and acid / alkali reserve. In Alternative Methods in Toxicology vol. 10 - In Vitro Skin Toxicology: Irritation, Phototoxicity, Sensitization, eds. A. Rougier, A.M. Goldberg and H.I Maibach, Mary Ann Lieb ert, Inc. 23 - 27.

331 Guidance on the Application of the CLP Criteria 5.0 – July 2017 331 Version SENSITISATION 3.4. RESPIRATORY OR SKIN Definitions and general considerations for respiratory or skin 3.4.1. sensitisation means a substance that will lead to hypersensitivity Respiratory sensitiser Annex I: 3.4.1.1. inhalation of the substance. of the airways following 3.4.1.2. Skin sensitiser Annex I: means a substance that will lead to an allergic response following skin contact. In terms of prevention it might be important to note that respiratory sensitisation may be induced not only by inhalation but also by skin contact (Dotson et al, 2015). Please refer also to IR&CSA , Section R.7.3. the Guidance on 3.4.1.3. For the p urpose of s ection 3.4 , Annex I: sensitisation includes two phases: the first phase is induction of specialised immunological memory in an individual by exposure to on of a cell mediated or antibody - an allergen. The second phase is elicitation, i.e. producti - mediated allergic response by exposure of a sensitised individual to an allergen. Annex I : 3.4.1.4. For respiratory sensitisation, the pattern of induction followed by elicitation phases is shared in common with skin sensi tisation. For skin sensitisation, an induction phase is required in which the immune system learns to react; clinical symptoms can then arise when subsequent exposure is sufficient to elicit a visible skin reaction edictive tests usually follow this pattern in which (elicitation phase). As a consequence, pr there is an induction phase, the response to which is measured by a standardised elicitation phase, typically involving a patch test. The local lymph node assay is the exception, directly duction response. Evidence of skin sensitisation in humans normally is measuring the in assessed by a diagnostic patch test. Annex I Usually, for both skin and respiratory sensitisation, lower levels are : 3.4.1.5. Provisions for alerting sensitised necessary for elicitation than are required for induction. individuals to the presence of a particular sensitiser in a mixture can be found in Annex II, section 2.8. The hazard class Respiratory or Skin Sensitisation is differentiated into: Annex I: 3.4.1.6. Respiratory Sensiti sation and; - - Skin Sensitisation. 3.4.2. Classification of substances for sensitisation 3.4.2.1. Classification of substances for respiratory sensitisation Identification of hazard information 3.4.2.1.1. There are no formally recognised and validated animal or in vitro tests for respiratory sensitisation. However there may be data from human observations indicating respiratory sensitisation in exposed populations or other sufficient evidence, including read across. - 3.4.2.1.1.1. Identification of human data Relevant information wit h respect to respiratory sensitisation may be available from case reports, epidemiological studies, medical surveillance, reporting schemes. For more details see the Guidance on IR&CSA , Section R.7.3.9 .2.

332 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 332 3.4.2.1.1.2. Identification of non human data in vitro tests currently e xist for respiratory No formally recognised and validated animal or sensitisation. However, data from some animal studies may be indicative of the potential of a substance to cause respiratory sensitisation in humans (CLP Annex I, 3.4.2.1.3) and may provide supportive evidence in case human evidence is available. These data may provide supportive evidence and should be used in a weight of evidence assessment. For further information see the Guidance on IR&CSA , Section R.7.3.9.1. 3.4.2.1.2. C lassification criteria for substances Annex I : 3.4.2.1. Respiratory sensitisers : 3.4.2.1.1. Annex I Hazard categories Annex I: 3.4.2.1.1.1. Respiratory sensitisers shall be classified in Category 1 where data are not sufficient for sub - categorisation. Ann 3.4.2.1.1.2. Where data are sufficient a refined evaluation according to 3.4.2.1.1.3 ex I: shall allow the allocation of respiratory sensitisers into sub - category 1A, strong sensitisers, or sub - category 1B for other respiratory sensitisers. Effects seen in either humans or animals will normally justify Annex I: 3.4.2. 1.1.3. classification in a weight of evidence approach for respiratory sensitisers. Substances may be allocated to one of the two sub categories 1A or 1B using a weight of evidence approach in - accordance with the criteria given in Table 3.4.1 and on the basis of reliable and good quality evidence from human cases or epidemiological studies and/or observations from appropriate studies in experimental animals. Substances shal l be classified as respiratory sensitisers in accordance : 3.4.2.1.1.4. Annex I with the criteria in Table 3.4.1: Table 3.4.1 - categories for respiratory sensitisers Hazard category and sub Category Criteria Substances shall be classified as respiratory sensitisers (Category 1) where data are not sufficient for sub - categorisation in accordance with the following criteria: Category 1 (a) if there is evidence in humans that the substance can lead to specific respiratory hypersensitivity; and /or if there are positive results from an appropriate animal test. (b) Substances showing a high frequency of occurrence in humans; or a category 1A: Sub probability of occurrence of a high sensitisation rate in humans based on - 1 ). Severity of reaction may also be considered. animal or other tests ( Substances showing a low to moderate frequency of occurrence in humans; or a probability of occurrence of a low to moderate sensitisation category 1B: Sub - 1 rate in humans based on animal or other tests ( ). Severity of reaction may also be considered. 1 ( ) At present, recognised and validated animal models for the testing of respiratory hypersensitivity are not available. Under certain circumstances, data from animal studies may provide valuable information in a weight of evidence assessment.

333 Guidance on the Application of the CLP Criteria 5.0 – July 2017 333 Version currently no clear way of establishing sub categories for respiratory sensitisation, There is - however if compelling evidence w ere available such as observations in the workplace, it may be category. possible to determine a sub - categories - Classification into sub is required when data are sufficient. When Category 1A cannot be excluded, Category 1 should be applied instead of Category 1B. High frequency and low to moderate frequency cannot be defined as specific concentrations or percentages for human study data be , when considering human evidence, it is necessary to take into account the cause size of the exposed population and the extent and conditions of exposure, including frequency. - It is necessary, therefore, to reach a view on a case - case basis. by of hazard information 3.4.2.1.3. Evaluation 3.4.2.1.3.1. Human data Substances shall be classified as respiratory sensitisers if there is evidence in humans or other - across that the substance can lead to specific respiratory sufficient evidence, including read hypersensitivity. : 3.4.2.1.2 Human evidence Annex I 3.4.2.1.2.1. Evidence that a substance can lead to specific hypersensitivity will Annex I: normally be based on human experience. In this context, hypersensitivity is normally seen as asthma, but other hypersensitivity reactions su ch as rhinitis/conjunctivitis and alveolitis are also considered. The condition will have the clinical character of an allergic reaction. However, immunological mechanisms do not have to be demonstrated. 3.4.2.1.2.2. When considering the human e vidence, it is necessary for a decision Annex I: on classification to take into account, in addition to the evidence from the cases: (a) the size of the population exposed; (b) the extent of exposure. [...] Annex I: 3.4.2.1.2.3. The evidence referred to above could be: (a) clinical history and data from appropriate lung function tests related to exposure to the substance, confirmed by other supportive evidence which may include: (i) in vivo immunological test (e.g. skin prick test) (ii) in vitro immunological test (e.g. serological analysis); (iii) studies that indicate other specific hypersensitivity reactions where - immunological mechanisms of action have not been proven, e.g. repeated low level irritation, pharmacologically mediated effects; cture related to substances known to cause respiratory (iv) a chemical stru hypersensitivity; (b) data from one or more positive bronchial challenge tests with the substance conducted according to accepted guidelines for the determination of a specific hypersensitivity reactio n. Annex I: 3.4.2.1.2.4. Clinical history shall include both medical and occupational history to determine a relationship between exposure to a specific substance and development of respiratory hypersensitivity. Relevant information includes aggravating fa ctors both in the home and workplace, the onset and progress of the disease, family history and medical history of the patient in question. The medical history shall also include a note of other allergic or airway disorders from childhood, and smoking hist ory.

334 ication of the CLP Criteria Guidance on the Appl 334 Version 5.0 – July 2017 Annex I : 3.4.2.1.2.5. The results of positive bronchial challenge tests are considered to provide sufficient evidence for classification on their own. It is however recognised that in en carried out. practice many of the examinations listed above will have already be 3.4.2.1.3.2. Non human data Annex I: 3.4.2.1.3. Animal studies * Data from appropriate animal studies ( ) which may be indicative of : 3.4.2.1.3.1. Annex I ** the potential of a substance to cause sensitisation by inhalation in humans ( ) may include: measurements of Immunoglobulin E (IgE) and other specific immunological parameters in (a) mice; (b) specific pulmonary responses in guinea pigs. (*) At present, recognised and validated animal models for the testing of respiratory available. Under certain circumstances, data from animal studies may hypersensitivity are not provide valuable information in a weight of evidence assessment. (**) The mechanisms by which substances induce symptoms of asthma are not yet fully known. For preventative measures, thes e substances are considered respiratory sensitisers. However, if on the basis of the evidence, it can be demonstrated that these substances induce symptoms of asthma by irritation only in people with bronchial hyper reactivity, they d respiratory sensitisers. should not be considere No formally recognised and validated animal tests currently exist for respiratory sensitisation. However data from some animal studies may be indicative of the potential of a substance to ns (CLP Annex I, 3.4.2.1.3) and may provide supportive cause respiratory sensitisation in huma vidence is available (see also S ection 3.4.2.1.2 above). This evidence in case human e ation may also be combined with information on structural alerts for respiratory inform sensitisation (see the Guidance on IR&CSA , Section R.7.3.9.1) and information on the skin sensitising properties of a substance and should be used in a weight of evidence asse ssment. Information on sensitizing activity of substances, such as that identified using contact sensitivity studies, may also be taken into consideration in a weight of evidence assessment. Based on a the LLNA, using BALB/c instead of assessment including mostly non - standrad versions of (at CBA/Ca strains mice, substance for which there were convincing negative data in the LLNA an appropriate test concentration and with the exception of large substances such as enzymes) most probably lacks the potential f or respiratory allergy (Dearman R.J., 2013). It should be noted that negative data on skin sensitisation cannot be used to negate data fulfilling the classification criteria for respiratory sensitisation. 3.4.2.1.4. Decision on classification According to CLP Ann ex I, Section 3.4.2.1.1.4 substances fulfilling the criteria for respiratory sensitisation will be classified as such in Category 1 (and in Sub - category 1A or 1B when sufficient data are available), Setting of specific concentration limits 3.4.2.1.5. Respiratory sens itisers cannot be identified reliably on the basis of animal tests yet, since no recognised validated test exists to determine sensitising potential and potency by inhalation. Therefore specific concentration limits (SCLs) cannot be set on the basis of ani mal data alone. Moreover, there is no concept available to set SCLs on the basis of human data for respiratory sensitisers.

335 Guidance on the Application of the CLP Criteria 5.0 July 2017 335 Version – Decision logic for classification of substances 3.4.2.1.6. It is strongly recommended that the person responsible for classification study the criteria for classification before and during use of the decision logic. No Are there data and/or information to evaluate C l assification not respiratory sensitisation ? possible Yes a. Is there evidence in humans that the substance can and/or lead to specific respiratory hypersensitivity, No l Not c assified b. Are there positive results from an appropriate animal test, and/or - c. Is there other sufficient evidence, including read across, that the substance can lead to specific respiratory hypersensitivity? Category 1 Yes No - Are data sufficient for sub categorisation? Danger Yes - category 1A Sub Based on weight of evidence, does the substance show a high frequency of occurrence of respiratory Yes sensitisation in humans; or a probability of occurrence of a high respiratory sensitisation rate in humans based on animal or other tests? Severity of tion may also be considered. reac Danger No category 1B Sub - Based on weight of evidence, does the substance show a low to moderate frequency of occurrence of Yes respiratory sensitisation in humans; or a probability of occurrence of a low to moderate respiratory sensitisation rate in huma ns based on animal or other tests. Severity of reaction may also be considered. Danger

336 ication of the CLP Criteria Guidance on the Appl 336 Version 5.0 – July 2017 3.4.2.2. Classification of substances for skin sensitisation Identification of hazard information 3.4.2.2.1. With respect to identification of relevant information for skin sensitisation see the Guidance on IR&CSA , Section R.7.3.4. Identification of human data 3.4.2.2.1.1. Relevant information with respect to skin sensitisation may be available from case reports, epidemiological studies, medical surveillance and reporting schemes based on human patch testing. For more details see the Guidance on IR&CSA , Section R.7.3.4.2 . Id entification of non human data 3.4.2.2.1.2. - At present no formally validated non testing systems exist to predict skin sensitising potential. However data such as structural alert data or data to show that the chemical structure of a wn sensitisers (e.g. QSARs or expert systems) may form part molecule is similar to that of kno , Section R.7.3.4). of the weight of evidence for classification (see also Guidance on IR&CSA The subject of in vitro testing for skin sensitisation has also been dealt with in the Guidance on IR&C in vitro/in chemico methods exist with the aim to identify a SA, Section R.7.3.4. Validated sensitising potential of a chemical. These include OECD TG442C (Peptide/protein binding), nse). The TG442D (keratinocyte response) and TG 442E (monocytic/dendritic cell respo in vitro/in chemico tests are not regarded as stand alone tests and the result from such a test should be used together with other data in an overall WoE assessment. Further, at present there is no agreed strategy on how to use in vitro/in chemic o methods for direct estimation of sensitising potency, but data from such tests can be used in a WoE assessment together with other data in order to assess skin sensitisation potency. See also the Guidance on IR&CSA, especially Section R.7.3.4.1. Informat ion on the current developments of in vitro tests and methodology can be found on the ECVAM website ( http://ihcp.jrc.ec.europa.eu/our_labs/eurl ). - ecvam ethods used to evaluate skin sensitisation for There are three standard animal test m substances: the mouse local lymph node assay (LLNA), the guinea pig maximisation test , Section IR&CSA (GPMT) and the Buehler assay. They are further described in the Guidance on classification in Section of this Guidance. R.7.3.4, and in the context of 3.4.3.2 Classification criteria for substances 3.4.2.2.2. Skin Sensitisers 3.4.2.2. Annex I: : 3.4.2.2.1 . Hazard categories Annex I 3.4.2.2.1.1. : in Category 1 where data are not Skin sensitisers shall be classified Annex I sufficient for sub - categorisation. : Where data are sufficient a refined evaluation according to section Annex I 3.4.2.2.1.2. - category 1A, strong sensitisers, or 3.4.2.2.1.3 allows the allocation of skin sensitisers into sub sub - category 1B for other skin sensitisers. Annex I : 3.4.2.2.1.3. Effects seen in either humans or animals will normally justify classification in a weight of evidence approach for skin sensitisers as described in section cated to one of the two sub - categories 1A or 1B using a 3.4.2.2.2. Substances may be allo weight of evidence approach in accordance with the criteria given in Table 3.4.2 and on the basis of reliable and good quality evidence from human cases or epidemiological studies and/or observations from appropriate studies in experimental animals according to the guidance values provided in sections 3.4.2.2.2.1 and 3.4.2.2.3.2 for sub - category 1A and in sections 3.4.2.2.2.2 and 3.4.2.2.3.3 for sub - category 1B.

337 Guidance on the Application of the CLP Criteria 5.0 Version July 2017 337 – : Substances shall b e classified as skin sensitisers in accordance with Annex I 3.4.2.2.1.4. the criteria in Table 3.4.2: Table 3.4.2 categories for skin sensitisers Hazard category and sub - Criteria Category Substances shall be classified as skin sensitisers (Category 1) where data are not sufficient for sub - categorisation in accordance with the following criteria: (a) if there is evidence in humans that the substance can lead to Category 1 sensitisation by skin contact in a substantial number of persons; or if there are positive results from an a ppropriate animal test (b) (see specific criteria in paragraph 3.4.2.2.4.1). Substances showing a high frequency of occurrence in humans and/or a high potency in animals can be presumed to have the - category 1A: Sub potential to produce significant sensitisati on in humans. Severity of reaction may also be considered. Substances showing a low to moderate frequency of occurrence in humans and/or a low to moderate potency in animals can be Sub - category 1B: n presumed to have the potential to produce sensitisation i humans. Severity of reaction may also be considered. Classification into sub - categories is required when data are sufficient. When Category 1A cannot be excluded, Category 1 should be applied instead of Category 1B. This is particularly important esponse after exposure to a high if only data are available from certain tests showing a high r concentration but where lower concentrations, which could show the presence of effects at lower doses, have not been tested (in line with some test protocols where a maximised dose should be used). When considering human evidence, it is necessary to take into account the size of the population exposed and the extent of exposure and frequency, and thus the consideration is on a case by case basis. Human data should be incorporated with animal data to decide on ` the sub - cate gorisation. Diagnostic patch testing is the gold standard in diagnos ing contact allergy in dermatitis patients (see e.g. Johansen et al, 2015). Patch test concentrations and substances must be suitable for the purpose, not causing false negatives, false positives, irritant reactions or induc ing contact allergy (skin sensitisation). The vehicle is important for the outcome of a diagnostic patch test, the most commonly used being petrolatum. Patch test concentrations are not based on products. The used concentrations may be too low and lead to a false concentrations used in negative reaction. Data from the testing of unselected, consecutive dermatitis patients is more standardised than testing which is undertaken on a specific patient group (e.g. those with facial eczema) or worker group (e.g. individuals with a particular type of exposure) and often involves patch testing with materials beyond those normally used, i.e. ‘the standard series’, as for example the European baseline series. To detect and confirm new sensitisers, suitable patch test concentrations have to be set, which is a laborious task. For many substances, standardised commercial patch tests are lacking.

338 ication of the CLP Criteria Guidance on the Appl 338 Version 5.0 – July 2017 For a newly identified skin sensitiser, which might also be a substance newly introduced on to the market, or a substance not included in the baseline diagnostic patch test series, the high severity of responses might be used as an indication that classification as Category 1A is appropriate. For example, where the substance has caused:  Hospital isation due to acute skin reaction Chronic dermatitis (lasting > 6 months)  Generalised (systemic/whole body) dermatitis  It should be noted that the severity/strength of diagnostic patch test reactions normally cannot be used for this purpose. It should be noted that in some cases a substance may autooxidise in contact with air or to a more hazardous form. This may warrant classification of the parent substance decompose - - even though it in itself is not or is less hazardous. A case case evaluation should b e done by considering available hazard information on humans or animals and/or the rate and extent of autoxidation or decomposition. Evaluation of hazard information 3.4.2.2.3. Human data 3.4.2.2.3.1. e data from The classification of a substance can be based on human evidence, such as positiv patch testing, epidemiological studies showing allergic contact dermatitis caused by the substance, positive data from experimental studies in man and/or well documented episodes of ch (see Section of this allergic contact dermatitis, using a weight of evidence approa 3.4.2.2.3.7 Guidance for details). Criteria for sub - categorisation are listed in CLP Annex I, 3.4.2.2.2.1 and 3.4.2.2.2.2: Annex I : 3.4.2.2.2.1. Human evidence for sub - category 1A can include: 2 (a) (HRIPT, HMT – induction threshold); positive responses at ≤ 500 μg/cm diagnostic patch test data where there is a relatively high and substantial incidence of (b) reactions in a defined population in relation to relatively low exposure; other epidemiological evidence where there is a relatively high and substantial (c) incidence of allergic contact dermatitis in relation to relatively low exposure. : 3.4.2.2.2.2. Human evid Annex I - category 1B can include: ence for sub 2 positive responses at > 500 μg/cm (a) (HRIPT, HMT – induction threshold); (b) diagnostic patch test data where there is a relatively low but substantial incidence of latively high exposure; reactions in a defined population in relation to re other epidemiological evidence where there is a relatively low but substantial (c) incidence of allergic contact dermatitis in relation to relatively high exposure. HRIPT: Human Repeat Insult Patch Test; HMT: Human Maximisation Test CLP Article 7 (3) states ‘Tests on humans shall not be performed for the purposes of this Regulation. However, data obtained from other sources, such as clinical studies, can be used for the purposes of this Regulation.’ Thus human induction studies such as HRIPT or HMT must not be performed, although historical data may be used as weight of evidence for the sub - categorisation. To provide further guidance on the types of human data that may be considered as data from other sources, please refer to the following table:

339 Guidance on the Application of the CLP Criteria 339 – July 2017 5.0 Version 1 Types of Human Studies 3 Table . Subjects Endpoint studied Comments Type Human Repeated Healthy volunteers Induction of This is not a clinical study and is only of historical relevance. New Insult Patch Test sensitisation (HRIPT) & Human studies for this regulation are not Maximization Test permitted. (HMT) Diagnostic patch test Eczema patients Primary source of clinical Elicitation (as an attending information on the occurrence of indicator of from individual clinics dermatology clinics skin sensitisation or collated clinic data previous sensitisation) Elicitation Not yet a standardised protocol, Sensitised Dose response study individuals (usually but provides an indication of the (e.g. patch test serial dilution; repeated open degree of sensitivity and of safe from diagnostic application test) patch tests) limits of exposure. Mainly used as confirmatory tests and in research. Eczema patients, Large general population studies Elicitation Epidemiology study selected are scarce; focused studies in occupational selected populations are more groups, other common and provide insights on selected groups, or frequency of sensitisation general population compared to expos ure The purpose of the material that follows is the provision of guidance concerning the evaluation of human data, particularly with respect to balancing considerations of exposure against the he concept of ‘guidance’ should clinical evidence regarding the frequency of skin sensitisation. T they represent indicators derived from be applied generally to all of the numeric criteria – expert opinion and are not to be taken as proven absolute values. Application of this guidance should permit sub categorisation whe re the human data on exposure and sensitisation is clear. - 3 . 2 Table Relatively high or low frequency of occurrence of skin sensitisation* Human diagnostic patch test data High frequency Low/moderate frequency ≥ 0.2 % < 0.2 % General population studies ≥ 1.0 % Dermatitis patients (unselected, consecutive) < 1.0 % Selected dermatitis patients (aimed testing, usually special test < 2.0 % ≥ 2.0 % series) Work place studies: ≥ 0.4 % < 0.4 % 1: all or randomly selected workers 2: selected workers with known exposure or dermatitis ≥ 1.0 % < 1.0 % Number of published cases ≥ 100 cases < 100 cases * Only one or two types of information may be sufficient for sub - categorisation.

340 ication of the CLP Criteria Guidance on the Appl 340 Version 5.0 – July 2017 The figure of 0.2% for the general population is intended to reflect that the frequency of contact 10) times higher than in the general allergy in dermatitis patients is approximately 5 (range 2 - population (Mirshahpanah and Maibach, 2007). The figure of 1% for consecutive (i.e. unselected) dermatitis patients is based on the generally ≥ 1% in such patients is of high agreed consideration that a contact allergy frequency of concern. The figure of 0.4% for unselected workers in a workplace is derived from the use in REACH of a 2 times higher assessment factor for the general population than for workers. It is important to note that the data from the testing of unselected, consecutive dermatitis patients is more standardised than testing which is undertaken on a specific patient group (e.g. those with facial eczema) or worker group (e.g. individuals with a particular type of exposure). Such clinical studies may be conducted on patients selected according to a particular type of eczema or based on their likelihood of occupational exposure and often involves patch testing et al , 2011). It is with materials beyond those normally used i.e. ‘the standard series’ (Andersen frequency related important to consider also that there may be variations in positive patch test to age, gender or region. 3 . 3 Relatively high or low exposure * Table Relatively high exposure Exposure data Relatively low exposure (weighting) (weighting) < 1.0% Concentration / dose ≥ 1.0% 2 2 500μg/cm < ≥ 500μg/cm (score 0) (score 2) Repeated exposure < once/daily (score 1) ≥ once/daily (score 2) Number of exposures (irrespective of ≥100 exposures (score 2) <100 exposures (score 0) concentration of sensitizer) exposure index (see text below) a response in each row is necessary. To achieve the * Table 3 . 3 represent weightings whose purpose is to enable an exposure index to The scores in be derived which best reflects our understanding of the relative importance of dose versus frequency of exposure. An additive exposure index of 1 4 equates to low exposure, whereas 5 - - 6 reflects high exposure. Careful consideration has to be given regarding the release (migration) of a sensitising substance from a solid object, and not the concentration. Ideally, skin exposure is best expressed in dose per unit area, but it is recognise d that this data is often not available, hence concentration may be used as a surrogate indicator of exposure. Table 3 . 4 Sub - categorisation decision table Relatively high frequency of Relatively low frequency of occurrence of skin occurrence of skin sensit isation sensitisation Sub - category 1B Category 1 Relatively high exposure 6) - or case by case evaluation (score 5 Relatively low exposure category 1A - Sub Category 1 or case by case evaluation 4) - (score 1

341 Guidance on the Application of the CLP Criteria 5.0 – July 2017 341 Version 3.4.2.2.3.2. Non human data Animal test results for sub - category 1A can include data with values Annex I : 3.4.2.2.3.2. indicated in Table 3.4.3 Table 3.4.3 Animal test results for sub - category 1A Criteria Assay Local lymph node assay EC3 value ≤ 2 % maximisation test ≥ 30 % responding at ≤ 0,1 % intradermal induction Guinea pig dose or ≥ 60 % responding at > 0,1 % to ≤ 1 % intradermal induction dose ≥ 15 % responding at ≤ 0,2 % topical induction dose or Buehler assay ≥ 60 % responding at > 0,2 % to ≤ 20 % topic al induction dose Annex I : 3.4.2.2.3.3. Animal test results for sub - category 1B can include data with values indicated in Table 3.4.4 below: Table 3.4.4 - category 1B Animal test results for sub Criteria Assay EC3 value > 2 % Local lymph node assay Guinea pig maximisation test ≥ 30 % to < 60 % responding at > 0,1 % to ≤ 1 % intradermal induction dose or ≥ 30 % responding at > 1 % intradermal induction dose Buehler assay ≥ 15 % to < 60 % responding at > 0,2 % to ≤ 20 % topical induction dose or ≥ 15 % responding at > 20 % topical induction dose The CLP Regulation allows classification of skin sensitisers in one hazard category, Category 1, which comprises two sub - categories, 1A and 1B. 3.4.2.2.1.1: Skin sensitisers shall be classified in Ca Annex I: tegory 1 where data are not sufficient for sub - categorisation. Classification into sub - categories is required when data are sufficient (CLP Annex I 3.4.2.2.1.1). When Category 1A cannot be excluded, Category 1 should be applied instead of Category 1B. This is particularly important if only data are available from the guinea pig tests or from the rLLNA showing a high response after exposure to a high concentration but where lower concentrations which could show the presence of such effects at lower doses are absent or in the absence of adequate dose - response information. Unless there is sufficient evidence to place such substances in sub category 1A or 1B, classification in category 1 should be the default

342 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 342 position. In other words, al in the T able 3.4.4 for classification to subcategory though the criteria 1B are fulfilled, the classification for subcategory 1A may not be excluded and therefore the substance should be classified as a Category 1 skin sensitiser (see also examples 6 & 7). The on requirements (as amended by Commission Regulation (EU) 2016/1688) for REACH informati skin sensitisation includes a requirement for a potency assessment, i.e. an assessment of on whether a substance "can be presumed to have the potential to produce significant sensitisati in humans (Cat. 1A)". The only exception to this is where there is existing animal information available (i.e. a study which was initiated or conducted before 11 October 2016) that does not allow an assessment of potency and thus only a conclusion in ca tegory 1 is possible. In such cases no further testing to assess potency is required (further details can be found in the Guidance on IR&CSA , Section R.7.3). Not all substances which need to be classified are registered under REACH, and thus for these subs tances the data base can be weaker and not sufficient to conclude on potency and therefore subcategorization is not possible and classification in category 1 is warranted. Since it is possible to refine the evaluation of skin sensitisers on the basis of th e potency of the sensitising effect, this guidance advises how to evaluate the potency on the basis of the recommended test methods. High potency is determined according to the results from the animal studies as given in CLP Annex I, Table 3.4.3 and l ow to moderate potency is determined according to the results from the animal studies as given in CLP Annex I, Table 3.4.4. The potency considerations may be used as a basis for setting specific concentration limits (see 3.4.2.2.5 of this Guidance). The three currently recognised and officially accepted Section animal test methods for skin sensitisation defined by OECD Test Guidelines are the Mouse Local Lymph Node Assay (LLNA) OECD TG 429 and its variations OECD TG 442A and 442B, Guinea Pig Maximisation Test by Magnusson & Kligman (GPMT) and the Buehler assay in the guinea pig OECD TG 406. The mouse and guinea pig methods differ fundamentally with respect to the endpoints used; whereas the mouse LLNA measures the responses provoked during the induction of sensitisation, the two guinea pig tests measure challenge induced elicitation he LLNA is now the reactions in previously sensitised animals. For new testing of substances t animal method of first choice, in case in vitro/in chemico assays are not considered relevant. In the exceptional circumstance that the LLNA is not appropriate, one of the alternative tests may be used (Buehler or GPMT), but justificatio , n shall be provided (see the Guidance on IR&CSA Section R.7.3.5.1). Test results from the LLNA, GPMT and the Buehler assay can be used directly for classification. They may also be used for potency evaluation. A sensitising potential of a substance is id entified if a significant effect has been obtained in an in vivo acceptable test. A significant skin sensitising effect in each of the three recognised animal tests is defined as follows: 3 . Table Definition of significant skin sensitising effect 5 Test Result Mouse local lymph node assay (LLNA) ( Stimulation Index ≥ 3 OECD TG 429)* LLNA: DA (OECD TG 442A),* Stimulation Index ≥ 1.8 LLNA: BrdU - ELISA (OECD TG 442B)* Stimulation Index ≥ 1.6 Redness (Score ≥ 1) in ≥ 30% of the Guinea pig maximisation test (GPMT) (OECD 406) test animals Redness (Score ≥ 1) in ≥ 15% of the Buehler assay (OECD 406) test animals *See further details in the test guidelines

343 Guidance on the Application of the CLP Criteria 5.0 – July 2017 343 Version on the basis of a positive test result in one of A substance may be classified as a skin sensitiser the above described animal tests. A positive result obtained by another test method not officially recognised may also justify classification as a skin sensitiser, but can normally not overrule a negative re sult obtained in one of the three recognised, animal tests described above. A new animal study should not be conducted in an attempt to negate a clearly positive response in a test method not officially recognised particularly where there is other supporti ng evidence that the substance is a skin sensitiser. 3.4.2.2.3.2.1. Mouse Local Lymph Node Assay The LLNA is used both for determination of skin sensitising potential (hazard identification) and for determination of relative skin sensitisation potency (hazard characterisation). In both lymph nodes following topical instances the metric is cellular proliferation induced in draining exposure to a chemical. Lymph node cell proliferation is causally and quantitatively correlated et al with the acquisition of skin sensitisation (Basketter . 2002a, 2002b). A correlation has been demonstrated between the conc entration of a chemical required for the acquisition of skin sensitisation in humans according to historical predictive data and skin sensitisation potency as measured in the mouse LLNA (Schneider and Akkan 2004, Basketter . 2005b). Potency is et al measured as a function of the derived EC3 - values. The EC3 - value is the amount of test chemical (% concentration, molar value or dose per unit area) required to elicit a stimulation et al . 2003). An inverse relationship exists between index of 3 in the standard LLNA (Kimber EC3 - - value and potency meaning that extremely potent sensitisers have extremely low EC3 values. The relevance of potency derives from an appreciation that skin sensitisers vary by up to four or five orders of magnitude with respect to the minimum concentration required inducing skin sensitisation. Potency is graded on the basis of these minimum concentrations each grade reflecting a concentration range of approximately one order of magnitude. However, it should be noted that if the dose interval for LLNA is too low so that all the stimulation indexes are below 3, it is not possible to know whether the higher doses would have generated a stimulation index above 3. Also, if only high doses would be used in an LLNA test, t he EC3 value may be associated with great uncertainty since the extrapolation is needed to low doses when the shape of the dose response curve is not known. It is also known that the choice of vehicle may - influence the EC3 value. when setting specific concentration limits (see Section Potency may be considered of 3.4.2.2.5 this Guidance). (rLLNA) described as an option in Different variants of the LLNA exist, namely the reduced LLNA - OECD TG 429, the LLNA: DA (OECD TG 442A), and the LLNA: BrdU ELISA (OECD TG 442B). The rLLNA uses fewer animals than the classical LLNA and should only be used in those not response information is circumstances where dose - required (e.g. to confirm a negative prediction of skin sensitising potential) and thus should not be used for sub - categorisation of skin - sensitisers. The last two variants avoid the use of DNA radiolabelling agent and provide quantitative data suitable fo - response assessment. However, the criteria for determining r dose the positive response is different from that of the traditional LLNA (OECD TG 429). Full details are given in the corresponding OECD Test Guidelines. There is no guidance for sub - categorisa tion. 3.4.2.2.3.3. Guinea Pig Maximisation Test (GPMT, OECD TG 406) This test has been used for over 40 years, to detect the sensitising potential of chemicals through a test system maximizing the sensitivity by both intradermal and epidermal induction and use of an ad juvant (Freund’s Complete Adjuvant). The intradermal induction is made by injection. Consequently the test is not suited for substances which cannot be made up into a liquid formulation. The GPMT was originally designed to maximise the ability to identify a sensitisation hazard, rather than to determine skin sensitisation potency. Yet, when only a GPMT test result is available, potency categorisation may be possible on the basis of the concentration of test

344 ication of the CLP Criteria Guidance on the Appl 344 Version 5.0 – July 2017 material used for intradermal induction and the percentage of guinea pigs sensitised. However, it should be recognised that there is often a degree of uncertainty associated with the derivation of allergenic potencies from the GPMT. hest induction dose It should be noted that the guinea pig tests should be conducted at hig to - moderate (GPMT) skin irritation. As a consequence, it is causing mild (Buehler Assay) or mild - at low concentration given in unlikely that substances (except strong irritants) would be tested T able 3.4.4 triggering classification as a skin sensitiser in sub category 1A. (see Section 3.4.2.2.5 of Potency may be considered when setting specific concentration limits this Guidance). Buehler assay (OECD TG 406) 3.4.2.2.3.4. This test has been in use for the last 40 years, although still a sensitive test to detect skin sensitisers using epidermal occluded exposure. The skin barrier of the test species (guinea pig) cy can be categorised using the results of the Buehler assay on is kept intact in this assay. Poten the basis of the number of animals sensitised and the concentration of the test material used for the epidermal induction. However, it should be recognised that there is often a degree of rtainty associated with the derivation of allergenic potencies from the Buehler assay. unce 3.4.2.2.5 Potency may be considered when setting specific concentration limits of (see Section this Guidance). It should be noted that the guinea pig tests should be conducted at highest induction dose causing mild (Buehler Assay) or mild - to - moderate (GPMT) skin irritation. As a con sequence, it is unlikely that substances (except strong irritants) would be tested at low concentration given the in T able 3.4.4 triggering classification as a skin sensitiser in sub category 1A. Non guideline skin sensitisation tests 3.4.2.2.3.5. - test methods In vivo which do not comply with recognised guidelines (see Article 8(3) of CLP) are strongly discouraged for the identification of skin sensitisers or assessment of skin sensitising potency. The results of such tests may provide supportive evidence when the tests are scientifically well justified and carefully evaluated. If doubts exist about the validity and the - of - evidence interpretation of the results, the evaluation needs to be done by using a weight 3.4.2.2.3.7 of this Guidance). approach as described below (see Section Animal test methods conducted for purposes other than sensitisation 3.4.2.2.3.6. s. These tests are often Occasionally signs of skin sensitisation occur in repeated dose test dermal toxicity tests on rats. Clearly, if signs of erythema/oedema occur in animals after repeated application, the possibility of skin sensitisation should be considered, and ideally assessed in an appropriate study. 3.4.2.2.3.7. idence Weight of ev 3.4.2.2.4. Annex I: Specific considerations 3.4.2.2.4.1. For classification of a substance, evidence shall include any or all of the following using a weight of evidence approach: (a) positive data from patch testing, normally obtained in more than one dermatology clinic; (b) epidemiological studies showing allergic contact dermatitis caused by the substance. Situations in which a high proportion of those exposed exhibit characteristic symptoms are to be looked at with special concern, even if the nu mber of cases is small; (c) positive data from appropriate animal studies

345 Guidance on the Application of the CLP Criteria 5.0 – July 2017 345 Version (d) positive data from experimental studies in man (see section 1.3.2.4.7); (e) well documented episodes of allergic contact dermatitis, normally obtained in more than one dermatolog y clinic; severity of reaction may also be considered. (f) Annex I: 3.4.2.2.4.2. Evidence from animal studies is usually much more reliable than evidence from human exposure. However, in cases where evidence is available from both sources, and there is conflict between the results, the quality and reliability of the evidence from both sources must be assessed in order to resolve the question of classification on a - by case basis. Normally, human data are not generated in controlled experiments with case - sment to volunteers for the purpose of hazard classification but rather as part of risk asses confirm lack of effects seen in animal tests. Consequently, positive human data on skin sensitisation are usually derived from case - control or other, less defined studies. Evaluation of frequency of cases reflect, in human data must therefore be carried out with caution as the addition to the inherent properties of the substances, factors such as the exposure situation, bioavailability, individual predisposition and preventive measures taken. Negative human data e positive results from animal studies. For both animal should not normally be used to negat and human data, consideration should be given to the impact of vehicle. If none of the abovementioned conditions are met, the substance need Annex I: 3.4.2.2.4.3. not be classified as a skin sensitise r. However, a combination of two or more indicators of skin sensitisation as listed below may alter the decision. This shall be considered on a case - by - case basis. Isolated episodes of allergic contact dermatitis; (a) epidemiological studies of limit ed power, e.g. where chance, bias or confounders have (b) not been ruled out fully with reasonable confidence; (c) data from animal tests, performed according to existing guidelines, which do not meet the criteria for a positive result described in section 3. 4.2.2.3, but which are sufficiently close to the limit to be considered significant; positive data from non - (d) standard methods; (e) positive results from close structural analogues. Annex I: 3.4.2.2.4.4. Immunological contact urticaria Substances meeti ng the criteria for classification as respiratory sensitisers may in addition cause immunological contact urticaria. Consideration should be given to classifying these substances also as skin sensitisers. Substances which cause immunological contact urtica ria without meeting the criteria for respiratory sensitisers should also be considered for classification as skin sensitisers. There is no recognised animal model available to identify substances which cause sification will normally be based on human immunological contact urticaria. Therefore, clas evidence which will be similar to that for skin sensitisation. Positive effects seen in either humans or animals for skin sensitisation will normally justify classification. Evidence from animal studies on skin sensitisation is usually more reliable than evidence from human exposure, although adequate reliable and representative human data are usually more relevant. In cases where evidence is available from both sources, and there is conflict between the results, the quality and reliability of the evidence from both sources must be assessed in order to decide on the classification on a case - by - case basis. Negative human data should not normally negate positive findings in animal studies (CLP Annex I, 3.4.2.2.4.2).

346 ication of the CLP Criteria Guidance on the Appl 346 Version 5.0 – July 2017 Since the data used in hazard or risk assessment should be relevant, reliable and sufficient for the regulatory purpose, it is necessary to base the assessment on the totality of available . information, i.e. to apply Weight of Evidence (WoE) considerations The WoE assessment can be based on the total of experimental data, as well as post market - surveys and/or occupational experience data. - testing data might be used to supplement and increase confidence in the available Non experimental data. In some cases , such data might be used to conclude on classification in line absence with the criteria in the of experimental data. WoE assessment can be divided into two stages: Assessment of each single test result and, if needed, of other data. It may be helpful to a. et al (1997). These criteria include apply criteria for reliability as defined by Klimisch details on the recognition of the test method, reporting detail, method relevance, test parameters, etc. Comparison of the weighed single test results. b. vitro/in chemico in Available tests cannot be considered as stand alone tests, but the results from such test s can be used together with other data in a weight of evidence assessment. There is currently no agreed strategy on how to use the results of these methods for potency assessment (see OECD TG442C E and Guidance on IR&CSA, R.7.3.4.1) - Good quality data on the substance itself have more weight than such data extrapolated from similar substances. Decision on classification 3.4.2.2.4. According to CLP Annex I, 3.4.2.2.1.4 sub stances fulfilling the criteria for skin sensitisation will category 1A or 1B when sufficient data are be classified as such in Category 1 (or in Sub - available). In addition substances classified for skin sensitisation can be allocated specific concentrati 3.4.2.2.5 of this Guidance. on limits as described in Section 3.4.2.2.5. Setting of specific concentration limits SCLs for skin sensitisation can be set based on the results from animal testing as reported below. SCLs are set on the basis of testing of the substance and never on the basis of testing of a mixture containing the sensitising substance (see CLP Annex I, 3.4.3.1.1). The setting of SCL tency; potency is already considered for the subcategorisation defining generic is based on po concentration limits. SCLs are generally applied for the most potent skin sensitisers classified in 1A. s for sensitisers. The The following schemes can be used for determination of potency categorie et al . (2005a). potency categories given in the 3 tables below are described in Basketter For the LLNA(OECD TG 429) Table 3 . 6 Skin Sensitisation Potency in the Mouse Local Lymph Node Assay EC3 Potency Resulting sub - category (*) value (% w/v) - ≤ 0.2 Extreme 1A > 0.2 - ≤ 2 Strong 1A > 2 Moderate 1B (*) based on Annex I Section 3.4.2.2.3.2. and Section 3.4.2.2.3.3. For the Guinea Pig Maximisation Test (OECD TG 406)

347 Guidance on the Application of the CLP Criteria 5.0 – July 2017 347 Version 3 7 Potency on basis of the Guinea Pig Maximisation Test . Table Concentration for - Incidence sensitised Resulting sub Potency category (*) guinea pigs (%) intradermal induction (% w/v) ≥ 60 Extreme 1A ≤ 0.1 Strong > - <60 30 1A ≤ 0.1 >0.1 ≤ 1.0 ≥60 Strong 1A - >0.1 ≤ 1.0 > 30 - <60 Moderate 1B(**) - Moderate > 1.0 1B(**) ≥ 30 (*) based on CLP Annex I Section 3.4.2.2.3.2. and Section 3.4.2.2.3.3. incidence of sensitised guinea pigs is very (**) If the concentration used for intradermal induction or the high, care should be taken to exclude the possibility of the substance being a Cat 1A (a strong or an extreme) sensitiser. For the Buehler Assay, (OECD TG 406) 3 . 8 Potency on basis of the Buehler assay Table Concentration for topical - Resulting sub Incidence Potency induction (% w/v) category (*) sensitised guinea pigs (%) Extreme 1A ≤ 0.2 ≥ 60 > 15 - <60 ≤ 0.2 1A Strong >0.2 ≤ 20 ≥ 60 - Strong 1A >0.2 - ≤ 20 (**) > 15 - <60 Moderate 1B (**) > 20 (**) Moderate 1B (**) ≥ 15 (*) based on CLP Annex I Section 3.4.2.2.3.2. and Section 3.4.2.2.3.3. (**) If the concentration used for topical induction or the incidence of sensitised guinea pigs is very high, be taken to exclude the possibility of the substance being a Cat 1A (a strong or an extreme) care should sensitiser. The generic concentration limits (GCLs) for the classification of sensitisers in mixtures are given 3.4.3.3.1 of this Guidance). In some cases, the GCL in CLP Annex I, Table 3.4.5 (see Section may not be sufficiently protective and an SCL shall be set in accordance with CLP Article 10, w hich will better reflect the hazard of mixtures containing that skin sensitiser. SCLs shall be set when there is adequate and reliable scientific information available showing e recommended that the specific hazard is evident below the GCL for classification. As such th SCL should normally be as given in Table 3 . 9 . However, supported by reliable data the SCL could have some other value below the GCL. Rel iable data could be human data from e.g. work place studies where the exposure is defined. It is more difficult to prove the absence of sensitising properties at certain concentration levels. Therefore an SCL above the GCL may only be set in exceptional ci rcumstances, if scientific information is adequate, reliable and conclusive for that particular skin sensitiser. However there is currently no guidance on how to set an SCL above the GCL.

348 ication of the CLP Criteria Guidance on the Appl – 5.0 348 July 2017 Version The concentration limits for skin sensitisers categorised according to their sensitisation potency in Table 3 . 9 are based on the recommendations from an EU expert group on skin sensitisation (Basketter et al ., 2005a). . 9 Skin sensitising potency for substances and recommendations on concentration 3 Table limits Concentration Limit (% w/v) Potency Extreme 0.001 (SCL) Strong 0.1 (GCL) Moderate 1 (GCL)

349 Guidance on the Application of the CLP Criteria 5.0 – July 2017 349 Version 3.4.2.2.6. Decision logic for classification of substances It is strongly recommended that the person responsible for classification study the criteria for classification before and during use of the decision logic. No Classification “Are there data and/or information to evaluate skin not possible sensitisation ? Yes Is there evidence in humans that the substance can a . lead to sensitisation by skin contact in a substantial No number of persons, or Not classified b Are there positive results from an appropriate . in vitro/in chemico test animal test ? or Category 1 Yes No categorisation? Are data sufficient for sub - Warning Yes category 1A Sub - Based on weight of evidence, does the substance Yes show a high frequency of skin sensitisation in humans and/or a high potency in animals? Severity of reaction may also be considered. Warning No category 1B - Sub Based on weight of evidence, does the substance show a low to moderate frequency of skin Yes sensitisation in humans and/or a low to moderate potency in animals? Severity of react ion may also be considered. Warning

350 ication of the CLP Criteria Guidance on the Appl 350 Version 5.0 – July 2017 3.4.3. Classification of mixtures for respiratory or skin sensitisation Identification of hazard information for respiratory sensitisation 3.4.3.1. of this Guidance). The same principles apply as for substances (see Section 3.4.2.1.1 Identification of hazard information for skin sensitisation 3.4.3.2. Fo r identification of the sensitisation potential of a mixture the following information may be available: a. test results on one or more, preferably all of its potentially sensitising components; or b. test results on the mixture itself; or c. test results of a similar mixture. 3.4.2.2.1 of this Guidance. Test methods are outlined in Section However, these animal tests have been developed to identify sensitising substances and not mixtures. Therefore the results obtained on mixtures need to be evaluated with car e. For a mixture the cut - off in the mouse LLNA should be seen as a threshold for identification of a sensitiser rather than as a threshold for sensitisation. A conclusion on the absence of sensitising potential of a mixture based on the negative outcome in a test must be taken with great caution. On the other hand test data on a mixture takes into account effects of possible interactions of its components. For instance, it is known that the presence of a vehicle may significantly ing potency, by influencing the penetration of the sensitising influence the skin sensitis 2001, Dearman et al . 1996, Heylings et al. . component(s) through the skin, (Basketter et al et 1996) or through other mechanisms involved in the acquisition of sensitisation (Cumberbatch al 93; Dearman et al . 1996). . 19 Repeated exposure to mixtures, that are non - sensitising under standard LLNA exposure conditions, might induce skin sensitisation, if the sensitising component in the mixture has sufficient accumulation potential in the skin to re ach the minimum concentration for a positive effect (De Jong et al. 2007). Uncertainty also exists about the effect of such a mixture after exposure on a larger skin area. Therefore additional information is important, if the outcome of sensitisation tests on mixtures contrasts with the classification based on the content of sensitising component(s). For example, the validity of a well conducted LLNA on a mixture with a negative outcome can scientifically be confirmed by spiking the test mixture with anothe r sensitiser (positive control) at different concentrations, or by showing a dose response relationship. Such LLNA tests could have been designed to provide such information without use of extra animals. Additional animal testing for the purpose of classif ication and labelling shall be undertaken only where no other alternatives, which provide adequate reliability and quality of data, are possible (CLP Article 7(1)) . Limitations apply to in chemico and in vitro methods (see the specific OECD test guidelines ). 3.4.3.3. Classification criteria for mixtures When mixtures are classified as sensitizing based on the presence of a sensitizing substance at a concentration at or above the generic or specific concentration limit, no sub - categorisation is required.

351 Guidance on the Application of the CLP Criteria 5.0 – July 2017 351 Version When data a 3.4.3.3.1. re available for all ingredients or only for some ingredients The mixture shall be classified as a respiratory or skin sensitiser when at : Annex I 3.4.3.3.1. least one ingredient has been classified as a respiratory or skin sensitiser and is present at or above the appropriate generic concentration limit as shown in Table 3.4.5 below for solid/liquid and gas respectively. 3.4.5 Table Generic concentration limits of components of a mixture classified as either sensitisers or skin sensitisers that trigger classification of the mixture respiratory Concentration triggering classification of a mixture as: Respiratory sensitiser Skin sensitiser Component classified as: Category 1 Category 1 Gas Solid/Liquid All physical states Respiratory sensitiser ≥ 1,0 % ≥ 0,2 % Category 1 Respiratory sensitiser % ≥ 0,1 % ≥ 0,1 Sub category 1A - Respiratory sensitiser % ≥ 1,0 % ≥ 0,2 category 1B - Sub Skin sensitiser ≥ 1,0 % Category 1 Skin sensitiser ≥ 0,1% category 1A Sub - Skin sensitiser ≥ % 1,0 category 1B Sub - All sensitising components of a mixture at or above their generic or specific concentration limit should be taken into consideration for the purpose of classification. Specific concentration limits (see Section of this Guidance ) will always take precedence over the generic 3.4.2.2.5 concentration limits. The additivity concept is not applicable for respiratory or skin sensitisation, i.e. if one single classified substance is present in the mixture above the generic or specific concentration limit, the mixture must be classified for that hazard. If the mixture contains two substances each below the generic or specific concentration limits, the mixture will not be classified.

352 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 352 3.4.3.3.2. Some substances that are classified as sensitisers may elicit a response, Annex I: when present in a mixture in quantities below the concentrations established in Table 3.4.5, sensitised to the substance or mixture (see Note 1 to Table in individuals who are already 3.4.6). 3.4.6 Table Concentration limits for elicitation of components of a mixture Concentration limits for elicitation Respiratory sensitiser Skin sensitiser Component classified as: Category 1 Category 1 Solid Gas All physical states /Liquid Respiratory sensitiser % ≥ 0,1 % 0,1 ≥ (Note 1) (Note 1) Category 1 Respiratory sensitiser 0,01 ≥ 0,01 % ≥ % (Note 1) (Note 1) category 1A Sub - % Respiratory sensitiser ≥ 0,1 % ≥ 0,1 (Note 1) (Note 1) Sub - category 1B Skin sensitiser ≥ 0,1 % (Note 1) Category 1 Skin sensitiser % (Note 1) 0,01 ≥ category 1A Sub - % (Note 1) Skin sensitiser ≥ 0,1 Sub category 1B - Note 1: application of the special labelling This concentration limit for elicitation is used for the II to protect already sensitised individuals. A SDS is requirements section 2.8 of Annex required for the mixture containing a component at or above this concentration. For sensitising substances with specific concentra tion limit lower than 0,1 %, the concentration limit for elicitation should be set at one tenth of the specific concentration limit. Further details on the additional labelling provisions to protect already sensitised individuals are provided in Section 3.4.4.1 of this Guidance.

353 Guidance on the Application of the CLP Criteria 5.0 – July 2017 353 Version 3.4.3.3.2. When data are available for the complete mixture When reliable and good quality evidence from human experience or Annex I: 3.4.3.1.1. appropriate studies in experimental animals, as described in the criteria for substances, is available for the mixture, then the mixture can be classified by weight of - evidence evaluation - of these data. Care shall be exercised in evaluating data on mixtures, that the dose used does not render the results inconclusive. In case classification of a mixture is based on test results for the mixture as a whole, this data the must be shown to be conclusive. Especially it should be taken into account that in case of skin sensitisation current test methods are based on applicati on of a maximised dose, which can only be obtained using a substance by itself and not diluted in a mixture. not showing sensitisation in a test, It is recognised that mixtures may still contain a low concentration of sensitising component. idance on the test methods and evaluation of the results see Section For specific gu of 3.4.3.2 this Guidance and CLP Annex I, 3.4.3.1.1. 3.4.3.3.3. When data are not availabl e for the complete mixture: Bridging Principles Annex I: 3.4.3.2.1. Where the mixture itself has not been tested to determine its sensitising properties, but there are sufficient data on the individual ingredients and similar tested mixtures to adequately characterise the hazards of the mixture, these data shall be used in accordance with the bridging rules out in section 1.1.3. In order to apply bridging principles, there needs to be sufficient data on similar tested mixtures as well as the ingredients of the mixture. of similar The same limitations apply for the use of existing test results tested mixtures generated with current test methods as those described for any mixture in sections 3.4.3.2. Care must be exercised in evaluating data on mixtures, that the dose used does not render the results inconclusive. Note that the following bridging principles are not applicable to this hazard class:  concentration of highly hazardous mixtures  interpolation within one hazard category 1.1.3.4). (see CLP Annex 1, 1.1.3.3 and When the available identified information is inappropriate for the application of the bridging principles then the mixture should be classified using the method described in Section 3.4.3.3.3 of this Guidance. 3.4.3.4. Decision logic for classification of mixtures It is strongly recommended that the person responsible for classification study the criteria for classification before and duri ng use of the decision logic.

354 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 354 3.4.3.4.1. Decision logic for classification of mixtures for respiratory sensitisation No Does the mixture as a whole or its ingredients have Classification not possible respiratory sensitisation data? Yes Does the mixture as a whole have respiratory sensitisation data? Category 1 (*) Yes No a. Is there evidence in humans that the mixture can lead to specific respiratory hypersensitivity, Yes and/or b. Are there positive results from an appropriate Danger animal test? No Classify in Yes bridging principles be Can appropriate applied? category No Care shall be exercised in evaluating data on No mixtures, that the dose used does not render the results inconclusive. Not classified of See Section this Is this the case? 3.4.2.1.3 Guidance. Yes Category 1 Does the mixture contain one or more ingredients classified as a respiratory sensitiser at: a. ≥ 0.1% w/w (solid/liquid)?, b. ≥ 1.0% w/w (solid/liquid)?; Yes or c. ≥ 0.1% v/v (gas)?, d. ≥ 0.2% v/v (gas)?; Danger or above a SCL set for the ingredient(s)? No Not classified - (*) can be sub 3.4.2.1.6 of this Guidance . categorised into 1A or 1B according to decision logic in Section

355 Guidance on the Application of the CLP Criteria July 2017 – 5.0 Version 355 Decision logic for classification of mixtures for skin sensitisation 3.4.3.4.2 No Does the mixture as a whole or its ingredients have skin Classification sensitisation data? not possible Yes Does the mixture as a whole have skin sensitisation data? Category 1 (*) Yes No a. Is there evidence in humans that the mixture can lead to sensitisation by skin contact in a Yes substantial number or persons, or b. Are there positive results from an appropriate Warning animal test? No Classify in Yes Can bridging principles be appropriate applied? category No Care shall be exercised in evaluating data on No mixtures, that the dose used does not render the results inconclusive. Not classified and Is this the case? See Section 3.4.3.2 3.4.3.3.2 o f this Guidance. Yes Category 1 Does the mixture contain one or more ingredients classified as a skin sensitiser at: Yes a. ≥ 0.1%?, b. ≥ 1.0%? Warning or above a SCL set for the ingredient(s)? No Not classified - (*) can be sub 3.4.2.2.6 of this Guidance. categorised into 1A or 1B according to decision logic in Section

356 ication of the CLP Criteria Guidance on the Appl 356 Version 5.0 – July 2017 3.4.4. Hazard communication for respiratory or skin sensitisation Pictograms, signal words, hazard statements and precautionary 3.4.4.1. statements Annex I: Label elements s hall be used for substances or mixtures meeting the criteria 3.4.4.1. for classification in this hazard class in accordance with Table 3.4.7 Table 3.4.7 Respiratory or skin sensitisation label elements Respiratory sensitisation Skin sensitisation Classification Category 1 and Category 1 and - categories 1A and 1B sub sub categories 1A and 1B - GHS Pictograms No Warning Danger Signal Word H334: May cause allergy or H317: May cause an Hazard Statement skin reaction asthma symptoms or breathing allergic difficulties if inhaled P261 Precautionary Statement P261 Prevention P272 P285 P280 P261 P261 Precautionary Statement Prevention P272 P284 P280 P304 + P341 Precautionary Statement P302 + P352 Response P333 + P313 P342 + P311 P321 P363 P304 + P340 Precautionary Statement P302 + P352 Response P333 + P313 P342 + P311 P321 P362 + P364 Precautionary Statement Storage

357 Guidance on the Application of the CLP Criteria 5.0 – July 2017 357 Version P501 P501 Precautionary Statement Disposal Article 26 1 (d) If the hazard pictogram ‘GHS08’ applies for respiratory sensitisation, the hazard pictogram ‘GHS07’ shall not appear for skin sensitisation or for skin and eye irritation. Additional labelling provisions 3.4.4.2. 2.8. Annex II: Mixtures containing at least one sensitising substance The label on the packaging of mixtures not classified as sensitising but containing at least one substance classified as sensitising and present in a concentration equal to or greater than that specified in Table 3.4.6 of Annex I shall bear the statement: ‘Contains (name of sensitising substance). May p EUH208 – roduce an allergic reaction’. Mixtures classified as sensitising containing other substance(s) classified as sensitising (in addition to the one that leads to the classification of the mixture) and present in a concentration equal to or greater than that s pecified in Table 3.4.6 of Annex I shall bear the name(s) of that/those substance(s) on the label. Where a mixture is labelled in accordance with section 2.4 or 2.5, the statement EUH208 may be omitted from the label for the substance concerned. Examples of classification for skin sensitisation 3.4.5. 3.4.5.1. Example of substances and mixtures fulfilling the criteria for classification for skin sensitisation Example 1 3.4.5.1.1. value of 10.4%. As this EC3 - - Substance X gave a positive result in the LLNA with an EC3 value is above th e cut - off of 2%, the substance is considered to be a moderate skin sensitiser, and should be classified as a Category 1 (Sub category 1B) skin sensitiser. The GCL for classification - of mixtures containing substance X is 1%. Example 2 3.4.5.1.2. Substance Y tested positive in the LLNA with an EC3 - value of 0.5%. In the GPMT a dermal induction concentration of 0.375% produced a positive response in 70% of the animals. On the basis of both these positive results, the substance is considered to be a strong sensitiser re - category 1A) skin sensitiser. The GCL for quiring classification as a Category 1 (Sub classification of mixtures containing substance Y is 0.1%. 3.4.5.1.3. Example 3 Herby is a herbicide formulation containing 28 g/l substance X, a Sub - category 1B skin sensitiser (se e example 1). There is no sensitisation data for the formulation itself. As Herby contains more than the GCL (1%) of this sensitising substance, and in the absence of any additional information, it should be classified as a Category 1 skin sensitiser. 3.4.5.1.4. Exa mple 4 Substance Z being an extreme sensitiser, is classified as a Sub - category 1A. It has a specific concentration limit with regard to skin sensitisation of 0.001%, and due to this property any

358 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 358 mixture containing the substance at a concentration ≥ 0.001% must be classified as a Category 1 skin sensitiser. Example 5 3.4.5.1.5. Woody is a wood preservative containing strong sensitising substances (Sub - category 1A): two substance A is present at 1% and substance B is present at 0.05%. There are no data for the formula tion itself. The mixture will be classified as cat 1 H317, due to the content of substance A (present above the GCL of 0.1%). Substance B is present below the classification limit. The name of both substances should appear on the label, substance A because it determines the classification of the mixture, and substance B because it is present in a concentration above the elicitation level (1/10 of the GCL of 0.1%). Example 6 3.4.5.1.6. Substance C was tested in a reduced LLNA test in accordance with OECD 429 using a c oncentration of 25%. This resulted in a stimulation index (SI) of 20 compared to the concurrent control. This is clearly above the SI of 3 required for classification. Therefore, classification as a skin sensitiser is required. However, the available infor mation does not allow determin calculati an EC3 value required to on of e the sub - categorisation. Although the substance was clearly positive at a high concentration of 25%, it cannot be excluded that also Therefore, there is not sufficient data for at a concentration of 2% or lower the SI will be 3. sub - categorisation. The substance is classified as Skin Sens Cat 1. Example 7 3.4.5.1.7. Substance D gave a positive response in a guinea pig maximisation test with 90 % responding uehler assay 70% responded at 30 % topical at 50 % intradermal induction dose. In a B induction dose. The response in both GPMT and Buehler assay was > 60% and the substance was not tested at ≤ 1 % intradermal induction dose in the guinea pig maximisation test or at ≤ 20 % topical induction dose in the Buehler assay. Although the criteria for classification to subcategory 1B are fulfilled, the classification for subcategory 1A cannot be excluded and therefore the substance should be classified as a Category 1 skin sensitiser. 3.4.5.1.8. Example 8 contradictory results from two or more skin sensitisation tests, the following If there are examples will give guidance for the classification. Since these are ideal cases, the weight of evidence approach should be applied if studies indicate shortcomings/are not cons idered fully reliable. 8(a): Substance E was tested in three separate animal tests performed with different test methods. In a Buehler assay no responses were observed with a topical induction dose of 70%. In the LLNA the EC3 value was 0.8%, indicating cla ssification for subcategory 1A. In GPMT, 30 % response was observed with an intradermal induction dose of 0,5 %, indicating classification for subcategory 1B. The substance should be classified for Skin Sens. 1A unless there is sufficient information to di scount some of the results. 8(b): Substance F is a skin sensitiser in humans indicating classification for sub - category 1A and in animals indicating classification for sub - category 1B. The substance should be classified for Skin Sens. 1A. 8(c): Substance G is a skin sensitiser in animal test s indicating classification for sub - category 1A and in humans indicating classification for category 1. The substance should be classified for Skin Sens. 1A.

359 Guidance on the Application of the CLP Criteria 5.0 – July 2017 359 Version a for 3.4.5.2. Example of substances or mixtures not fulfilling the criteri classification for skin sensitisation Example 9 3.4.5.2.1. Substance H was tested at concentrations up to 50% in the LLNA using a recommended and appropriate vehicle. It gave a maximum stimulation index of 2.6 and evidence of a positive dose response. On the b asis that the stimulation index was below 3 at a high dose, the substance does not require classification. However, had the highest concentrations been lower, e.g. 10%, - standard vehicle used, then further information would have be en and/or a non required b efore a classification decision could be reached. 3.4.5.2.2. Example 10 Insecto super is an insecticide formulation containing 9 g/l substance X (see Example 1). Substance X is a Sub - category 1B skin sensitiser (generic concentration limit in mixtures 1%). Based on t he classification of substance X, the insecticide formulation shall not be classified as as the concentration of the substance is below the GCL of 1%. The label must bear sensitising the statement EUH208. classification for 3.4.5.3. Examples of substances fulfilling the criteria for respiratory sensitisation 3.4.5.3.1. Example 11 Five case studies describe the fact that work - related exposure to substance P is associated with asthma or rhinitis. In all of these cases blinded specific bronchial challenge tests with substance P provoked the respiratory symptoms, confirming that substance P is the causal substance. In a cohort of 51 workers exposed to substance P, 26 (51%) were diagnosed with occupational asthma and 12 of those also suffered from occupational rhinitis. The diagnos is was based on specific bronchial challenge tests with substance P. There is sufficient human evidence to conclude that substance P should be classified as a category 1 respiratory sensitizer. Sub - categorization was not considered as there is currently - categories. no clear way to establish sub 3.4.5.3.2. Example 12 - related exposure to substance Q was associated with occupational asthma and rhinitis in Work several case studies. In those studies specific bronchial challenges were performed with substance Q and respiratory allergy symptoms could be reproduced, demonstr ating that substance Q is the causal agent. In addition, a large retrospective analysis of nine longitudinal studies involving 2,689 persons exposed occupationally to substance Q in a period of 35 years, showed that the incidences of occupational asthma ca used by substance Q were 2.7 - 5.5% in the earliest studies and decreased to 0.3 - 0.7% in the latest studies. Guinea pigs were exposed to substance Q by inhalation for 3 hours a day for 5 consecutive days 3 to concentrations of 4, 12, 24, and 48 mg/m eeks after the first encounter with the . Three w 3 inducing agent, animals were challenged with substance Q at a concentration of 2 mg/m . During challenge breathing patterns were affected already at the lowest test concentration in guinea pigs that were sensitized and challenged to substance Q and not in control animals. Additionally, pulmonary inflammation and increased specific IgG1 levels were observed in guinea pigs sensitized and challenged with substance Q. On the basis of human evidence supported by data fro m an animal study, substance Q should be classified as a Category 1 respiratory sensitizer. Sub - categorization was not considered as there is currently no clear way to establish sub - categories.

360 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 360 3.4.6. References Andersen KE, White IR and Goossens A (2011) Allerge ns from the European Baseline Series. In th edition Johansen JD, Frosch PJ and Lepoittevin J - P, Springer “contact Dermatitis”, 5 Verlag, - - Heidelberg, pp545 590 Basketter DA, Andersen KE, Liden C, Van Loveren H, Boman A, Kimber I, Alanko K, Berggren E. (2005a ): Evaluation of the skin sensitizing potency of chemicals by using the existing methods Contact Dermatitis 52 :39 and considerations for the relevance of elicitation. 43. - Basketter DA, Clapp C, Jefferies D, Safford R, Ryan C, Gerberick G, Dearman R, Kimber I. (2005b): Predictive identification of human skin sensitisation thresholds. Contact Dermatitis 53 :260 - 267. Basketter DA, Evans P, Fielder RJ, Gerberick GF, Dearman RJ, Kimber I. (2002a): Local lymph – validation and use in practice. Food Chem Toxicol 40 :593 - 598. node assay Basketter DA, Wright ZM, Colson NR, Patlewicz GY, and Pease CK. (2002b): Investigation of the Contact Dermatiti s 47 (3): 161 - 164. skin sensitizing activity of linalool. Basketter DA, Gerberick GF, Kimber I. (2001): Skin sensitisation, vehicle effects and the local lymph node assay. 39 : 621 - 627. Food Chem Toxicol Cumberbatch M, Scott RC, Basketter DA, Scholes EW, Hilton J, Dearman RJ, Kimber I (1993): Inf luence of sodium lauryl sulfate on 2.4 - dinitrochlorobenzene - induced lymph node activation. Toxicology 77 : 181 - 191. De Jong WH, De Klerk A, Beek MT, Veenman C, Van Loveren H. (2007): Effect of prolonged repeated exposure to formaldehyde donors with doses be low the EC3 value on draining lymph node responses. - 46. (3):239 4 J Immunotoxicol relationship between different classes of - Dearman RJ, Basketter DA, Kimber I (2013): Inter 33(7):558 - 65. chemical allergens. J Appl Toxicology n J, Clowes HM, Heylings JR, Kimber I (1996): Influence of Dearman RJ, Cumberbatch M, Hilto Fundam Appl Toxicol 33 : dibutyl phthalate on dermal sensitization to fluorescein isothiocyanate. - 30. 24 Dotson GS, Maier A,Siegel PD, Anderson SE, Green BJ, Stefaniak AB, Codispoti CD, Kimber I (20 15): Setting Occupational Exposure Limits for Chemical Allergens — Understanding the - S98. Challenges. Journal of occupational and environmental hygiene 12:S82 Heylings JR, Clowes HM, Cumberbatch M, Dearman RJ, Fielding I, Hilton J, Kimber I (1996): Sensitiza tion to 2,4 - dinitrochlorobenzene: influence of vehicle on absorption and lymph node activation. Toxicology 109 : 57 - 65. Johansen et al. European Society of Contact Dermatitis guideline for diagnostic patch testing – - 221 recommendations on best practice. Contact Dermatitis 2015; 73, 195 Kimber I, Basketter DA, Butler M, Gamer A, Garrigue JL, Gerberick GF, Newsome C, Steiling W, Vohr HW. (2003): Classification of contact allergens according to potency: Proposals. Food Chem Toxicol 41 : 1799 - 1809. Klimisch HJ, A ndreae M, Tillmann U (1997): A Systematic Approach for Evaluating the Quality of Experimental Toxicological and Ecotoxicological Data. 25 :1 - 5. Regul Toxico. Pharmacol Mirshahpanah P and Maibach HI (2007) Relationship of patch positive in a general versus a n eczema population. Contact Dermatitis, 56: 125 - 130. Robinson MK, Nusair TL, Fletcher ER, Ritz HL. (1990): A review of the Buehler guinea pig skin sensitization test and its use in a risk assessment process for human skin sensitization. Fundam Appl Toxico l 17 : 103 - 119.

361 Guidance on the Application of the CLP Criteria 5.0 – July 2017 Version 361 Schlede E, Eppler R. (1995): Testing for skin sensitisation according to the notification procedure for new chemicals. The Magnusson and Kligman test. Contact Dermatitis 32 : 1 - 4. Schneider K, Akkan Z. (2004). Quantitative relationship betwee n the local lymph node assay and human skin sensitization assays. Regul Toxicol Pharmacol 39 : 245 - 255.

362 ication of the CLP Criteria Guidance on the Appl 362 Version 5.0 – July 2017 3.5. TY GERM CELL MUTAGENICI Definitions and general considerations for classification for germ cell 3.5.1. mutagenicity : 3.5.1.1. A mutation means a permanent change in the amount or structure of the Annex I genetic material in a cell. The term ‘mutation’ applies both to heritable genetic changes that may be manifested at the phenotypic level and to the underlying DNA modifications when wn (including specific base pair changes and chromosomal translocations). The term kno ‘mutagenic’ and ‘mutagen’ will be used for agents giving rise to an increased occurrence of mutations in populations of cells and/or organisms. : 3.5.1.2. The more ge neral terms ‘genotoxic’ and ‘genotoxicity’ apply to agents or Annex I processes which alter the structure, information content, or segregation of DNA, including those which cause DNA damage by interfering with normal replication processes, or which in a - physio logical manner (temporarily) alter its replication. Genotoxicity test results are non usually taken as indicators for mutagenic effects. Germ cell mutations are those that occur in the egg or sperm cells (germ cells) and therefore anism's offspring. Somatic mutations are those that happen in cells can be passed on to the org other than the germ cells, and they cannot be transmitted to the next generation. This is an important distinction to keep in mind in terms of both the causes and the effects of mutation. Annex I: 3.5.2.1 This hazard class is primarily concerned with substances that may cause mutations in the germ cells of humans that can be transmitted to the progeny. However, the results from mutagenicity or genotoxicity tests in vitro and in mammalian so matic and germ cells in vivo are also considered in classifying substances and mixtures within this hazard class. 3.6.2.2 Specific considerations for classification of substances as carcinogens Annex I: Annex I: [...] Mutagenicity: It is recognised that genetic events are central in the 3.6.2.2.6. overall process of cancer development. Therefore evidence of mutagenic activity in vivo may indicate that a substance has a potential for carcinogenic effects. n for germ cell mutagenicity primarily aims to identify substances causing Hazard classificatio heritable mutations or being suspected of causing heritable mutations. A secondary aim is that the hazard class germ cell mutagenicity offers supporting information with respect to the classification of carcinogenic substances. This is expressed by the broad meaning of the hazard May cause genetic defects’ and ‘H341: Suspected of causing genetic defects’ statements ‘H340: ll mutagenicity. Thus, which comprises heritable genetic damage as well as somatic ce classification as a germ cell mutagen (Category 1A, 1B, and 2) classifies for the hazard heritable genetic damage as well as providing an indication that the substance could be carcinogenic. It is also warranted that where there is e vidence of only somatic cell genotoxicity, substances Classification as a suspected germ cell mutagen are classified as suspected germ cell mutagens. may also have implications for potential carcinogenicity classification. This holds true especially ose genotoxicants which are incapable of causing heritable mutations because they cannot for th reach the germ cells (e.g. genotoxicants only acting locally, ‘ site of contact’ genotoxicants). This means that if positive results in vitro are supported by at least one positive local in vivo , somatic cell test, such an effect should be considered as enough evidence to lead to classification in Category 2. If there is also negative or equivocal data, a weight of evidence approach using expert judgement has to be appli ed.

363 Guidance on the Application of the CLP Criteria – Version 363 July 2017 5.0 Classification of substances for germ cell mutagenicity 3.5.2. Identification of hazard information 3.5.2.1. Identification of human data 3.5.2.1.1. Occasionally, studies of genotoxic effects in humans exposed by, for example, accident, udies (e.g. from case reports or epidemiological studies) occupation or participation in clinical st may be available. Generally, cells circulating in blood are investigated for the occurrence of Guidance on IR&CSA , Section R.7.7.3.2. various types of genetic alterations; see also the Identificatio n of non human data 3.5.2.1.2. Animal data in vivo assays for genotoxicity/mutagenicity testing, with or without OECD There is a number of TGs. Modifications to OECD protocols have been developed for various classes of substances and may serve to enhance the accuracy of test results. Use of such modified protocols is a matter of expert judgement and will vary as a function of the chemical and physical properties of the substance to be evaluated. Commonly used in vivo tests employ methods by which any al can be examined for effects on the genetic material, giving the possibility to tissue of an anim intestinal - - contact tissues ( i.e., skin, epithelium of the respiratory or gastro - of examine site tract) in genotoxicity testing. In addition, test methods developed over the past decades in Drosophila and in various species of plants and fungi are available; see also the Guidance on 66 , Sect ion R.7.7.3 IR&CSA . These latter tests have, however, been deleted as OECD TGs as of 2014. tests in somatic cells which provide information on genotoxicity include, for example, the In vivo 67 for DNA strand breaks. Assays such as gene Comet single cell gel electrophoresis assay 68 mutations in transgenic rodent (TGR) models using reporter genes or mammalian erythrocyte micronucleus test for chromosome aberrations can be used for mutagenicity assessment. Please note that of these assays TG R is suitable for germ cells. data In vitro Typically, in vitro tests are performed with cultured bacterial cells, human or other mammalian cells. The sensitivity and specificity of tests will vary with different classes of substances; see also the Guidan ce on IR&CSA , Section R.7.7.3. Use of other data IR&CSA , Section R. 7.7.3.1. See the Guidance on Existing test methods See the Guidance on IR&CSA , Section R. 7.7.3.1. 66 The Guidance on IR/CSA, Chapter R.7a (version 4.1). 67 OECD TG 489 In Vivo Mammalian Alkaline Comet Assay (26 September 2014). 68 OECD TG 488 Transgenic Rodent Somatic and Germ Cell Gene Mutation Assays (26 July 2013).

364 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 364 3.5.2.2. Classification criteria for substances For the purpose of classificatio n for germ cell mutagenicity, substances are Annex I: 3.5.2.2. allocated to one of two categories as shown in Table 3.5.1. 3.5.1 Table Hazard categories for germ cell mutagens Categories Criteria CATEGORY 1: Substances known to induce heritable mutations or to be regarded as if they induce heritable mutations in the germ cells of humans. Substances known to induce heritable mutations in the germ cells of humans. The classification in Category 1A is based on positive evidence from 1A: Category human epidemiological studies. Substances to be regarded as if they induce heritable mutations in the germ cells of humans. The classification in Category 1B is based on: Category 1B: – positive result(s) from in vivo heritable germ cell mutagenicity tests in mammals; or pos itive result(s) from in vivo somatic cell mutagenicity tests in – mammals, in combination with some evidence that the substance has potential to cause mutations to germ cells. It is possible to derive this ests in germ cells supporting evidence from mutagenicity/genotoxicity t in vivo, or by demonstrating the ability of the substance or its metabolite(s) to interact with the genetic material of germ cells; or positive results from tests showing mutagenic effects in the germ cells – of humans, without demonstrati on of transmission to progeny; for example, an increase in the frequency of aneuploidy in sperm cells of exposed people. CATEGORY 2: Substances which cause concern for humans owing to the possibility that they may induce heritable mutations in the germ c ells of humans. 2 is based on: The classification in Category Positive evidence obtained from experiments in mammals and/or in – some cases from in vitro experiments, obtained from: – Somatic cell mutagenicity tests in vivo, in mammals; or – Other in vivo somatic cell genotoxicity tests which are supported by positive results from in vitro mutagenicity assays. Note: Substances which are positive in in vitro mammalian mutagenicity assays, and which also show chemical structure activity relationship to known germ cell mutagens, shall be considered for classification as Category 2 mutagens.

365 Guidance on the Application of the CLP Criteria 5.0 – July 2017 365 Version 3.5.2.3. Evaluation of hazard information Annex I: 3.5.2.3.3 Classification for heritable effects in human germ cells is made on the Regulation (EC) basis of well conducted, sufficiently validated tests, preferably as described in No 440/2008 adopted in accordance with Article 13(3) of Regulation (EC) No 1907/2006 (‘Test Method Regulation’) such as those listed in the following paragraphs. Evaluation of the using expert judgement and all the available evidence shall be test results shall be done weighed in arriving at a classification. Evaluation of human data 3.5.2.3.1. Human data have to be assessed carefully on a case by - case basis. The interpretation of such - data requires considerable expertise. Attention should be paid especially to the adequacy of the - exposure information, confounding factors, co exposures and to sources of bias in the study design or incident. The statistical power of the test may also be considered ( see the Guidance on IR&CSA , Section R.7.4.4.2). Evaluation of non human data 3.5.2.3.2. Evaluation of genotoxicity test data should be made with care. Regarding findings, positive responses generated only at highly toxic/cytotoxic concentrations should be interpreted with caution, a - response relationship should be considered. In nd the presence or absence of a dose case of negativ e findings in vivo toxicokinetic and other available information should be considered e.g. to verify whether the substance has reached the target organ (for detailed IR&CSA , Section R.7.7.4.1). guidance see the Guidance on across and (Q)SARs can be used as part of a WoE approach for germ cell mutagenicity - Read in vitro data from mammalian mutagenicity assays, structural classification. If there are positive milarities not sufficient for grouping/read - si across may still warrant classification. 3.5.2.4. Decision on classification 3.5.2.3.1. Annex I: To arrive at a classification, test results are considered from experiments determining mutagenic and/or genotoxic effects i n germ and/or somatic cells of exposed animals. Mutagenic and/or genotoxic effects determined in in vitro tests shall also be considered. Annex I: 3.5.2.3.9. The classification of individual substances shall be based on the total weight of evidence available, using expert judgement (See 1.1.1). In those instances where a single well conducted test is used for classification, it shall provide clear and unambigu ously - positive results. If new, well validated, tests arise these may also be used in the total weight of evidence to be considered. The relevance of the route of exposure used in the study of the substance compared to the most likely route of human exposu re shall also be taken into account. Classification as a Category 1A mutagen Epidemiological studies have been to date unable to provide evidence to classify a substance as a Category 1A mutagen. Hereditary diseases in humans for the most part have an un known origin and show a varying distribution in different populations. Due to the random distribution of mutations in the genome it is not expected that one particular substance would induce one specific genetic disorder. Therefore, it is unlikely that suc h evidence may be obtained by epidemiological studies to enable classification of a substance as a Category 1A mutagen. Classification as a Category 1B mutagen

366 ication of the CLP Criteria Guidance on the Appl 366 Version 5.0 – July 2017 Classification in Category 1B may be based on positive results of at least one valid in vivo malian germ cell mutagenicity test. In case there are also negative or equivocal data, a mam weight of evidence approach using expert judgement has to be applied. It could be argued that in a case where in vivo mutagenicity/genotoxicity is proven and the subst ance under consideration is systemically available, then that substance should also be considered as a Category 1B mutagen. Germ cell such as the spermatogonia are generally not - testes barrier formed by the Se rtoli cells. In protected from substance exposure by the blood such circumstances the relevant criteria are as follows: 3.5.2.2. (extract from Table 3.5.1) Annex I: Category 1B [...] – positive result(s) from in vivo somatic cell mutagenicity tests in mammals, in combination with some evidence that the substance has potential to cause mutations to germ cells. It is possible to derive this supporting evidence from mutagenicity/genotoxicity tests in germ cells in vivo, or by demonstrating the ability of the substance or its metabolite(s) to interact with t he genetic material of germ cells; [...] Supporting evidence in addition to positive results of a valid in vivo somatic cell mutagenicity test in mammals is needed to be able to classify a substance as a Category 1B mutagen when no data on mammalian germ cells are available. In the examples provided in the second sentence in the green box, mutagenicity/genotoxicity in germ cells or data showing that t he substance or its metabolite(s) interact with the genetic material of germ cells is mentioned. , Moreover, genetic damage to germ cells in exposed humans related to substance exposure , may offer additional information . somatic Thus, in such circumstances, in addition to an in vivo cell mutagenicity test, further experimental evidence is needed to be able to classify a substance as a Category 1B mutagen . by application of a WoE approach using expert judgement Classifica tion as a Category 2 mutagen in vivo valid Classification in Category 2 may be based on positive results of at least one mammalian somatic cell mutagenicity test, indicating mutagenic effects in somatic cells. A Category 2 mutagen classification may also b in vivo e based on positive results of a least one in vitro mutagenicity valid mammalian somatic cell genotoxicity test, supported by positive results. Genetic damage to somatic cells in exposed humans shown to be caused by substance exposure supported by positive in vitro mutagenicity results may also offer information warranting classification as a Category 2 mutagen. In vitro results can only lead to a Category 2 mutagen classification in a case where there is support by chemical structure activity relat ionship to known germ cell mutagens. In the case where there are also negative or equivocal data, a weight of evidence approach using expert judgement has to be applied. In general, mutations can be differentiated into gene mutations (e.g. point or frame s hift mutation), chromosome mutations (structural chromosome changes) and genome mutations (loss or gain of whole chromosomes). Different mutagenicity tests may detect different types of the weight of evidence mutations and genotoxic effects which have to be taken into account in determination. For instance, a substance which only causes chromosome mutations may be negative in a test for detecting point mutations. A complex data situation with positive and negative results might still lead to classification. This is because all tests detecting a certain type of mutation (e.g. point mutations) have been positive and all tests detecting chromosome mutations have been negative. Such circumstances clearly warrant classification although several tests have been ne gative which is plausible in this case.

367 Guidance on the Application of the CLP Criteria 5.0 – July 2017 367 Version A positive result for somatic or germinal mutagenicity in a test using intraperitoneal administration only shows that the tested substance has an intrinsic mutagenic property, and xhibited by other routes of dosage may be related to factors the fact that negative results are e influencing the distribution/ metabolism of the substance which may be characteristic to the tested animal species. It cannot be ruled out that a positive test result in intraperitoneal studies Note that intraperitoneal injection is since 2016 i n rodents may be relevant to humans. generally not recommended for new testing without specific scientific justification because it is not an intended route of human exposure. However, existing studies with intraperitoneal injection should be assessed as described in this and the next paragraph If there are positive results in at least one valid in vivo mutagenicity test using intraperitoneal in vivo genotoxicity test application, or from at least one valid using intraperitoneal application plus supportive data, classification is warranted. In cases where there are additional in vitro in vivo tests with oral, dermal or inhalative substance application, a weight of data from further evidence approach using exp ert judgement has to be applied in order to come to a decision. For instance, it may be difficult to reach a decision on whether or not to classify in the case where in vivo data from at least one in vivo test using intraperitoneal appli there are positive cation but (only) negative test data from (an) test(s) using oral, dermal, or inhalative application. in vivo In such a case, it could be argued that mutagenicity/genotoxicity can only be shown at internal body substance concentrations which cannot be achi eved using application routes other than intraperitoneal. However, it also has to be taken into account that there is generally no threshold for mutagenicity unless there is specific proof for the existence of such a threshold as may be the case for aneuge ns. Thus, if mutagenicity/genotoxicity can only be demonstrated for the intraperitoneal route exclusively, then this may mean that the effect in the in vivo tests using application routes other than intraperitoneal may have been present, but it may not hav e been detected because it was below the detection limit of the oral, dermal, or inhalative test assays. In summary, classification as a Category 2 mutagen would generally apply if only in vivo tests show mutagenicity/genotoxicity and the n intraperitoneal egative test results from the in vivo tests using other routes of application are plausible. Factors influencing plausibility - are e.g. the doses tested and putative kinetic data on the test substance. However, on a case - by vidence approach and expert judgement, non - classification case analysis using a weight of e may also result. Classification of substances containing CMR constituents, additives 3.5.2.5. or impurities From a compositional and a toxicological point of view the situation for substances containing CMR constituents, additives or impurities is the same as for mixtures containing components classified for these endpoints. For this reason the classification procedure for CMR endpoints that is foreseen by CLP for mixtures containing CMR components, is consid ered applicable also mpurities (see S tion 1.1.6.1 ). As to substances containing CMR constituents, additives or i ec discussed in S 3.5.3 below, mixtures containing components classified as germ cell ection mutagens shall be normally classified using only the relevant available informa tion for the individual substances in the mixture. Further, in cases where the available test data on the mixture itself demonstrate CMR effects which have not been identified from the information on n into account. For CMR endpoints the the individual substances, those data shall also be take lowest incidence possible to detect in the tests may be by far unacceptable in humans. Thus a dose as high as possible (such as maximal tolerated dose, MTD dose) is needed to be able to detect CMR hazards. Dilution, as would be the case if mixtures or substances containing CMR constituents were tested, would increase the risk that CMR hazards would not be detected. According to article 10 (1) substances in other substances and substances in mixtures are treated in the same way regarding the use of GCLs and SCLs.

368 ication of the CLP Criteria Guidance on the Appl – 5.0 368 July 2017 Version 3.5.2.6. Setting of specific concentration limits There is no detailed and accepted guidance developed for the setting of specific concentration limits (SCLs) for mutagenicity, as is the case for carcinogenic substances and substances toxic concept for carcinogens covering all relevant aspects to reproduction. Guidance such as the T 25 would need to be developed in order to derive SCLs for mutagens in a standardized manner. There are several reasons why it is considered i mpossible to set SCLs for mutagens without a comprehensive guidance, one of them being that mutagenicity tests have not been specifically developed for the derivation of a quantitative response. Moreover, different mutagenicity tests have different sensiti vities in detecting mutagens. Thus, it is very difficult to describe the minimum data requirements which would allow a standardized SCL derivation. Another drawback in practice is that the results obtained for the most part do not offer sufficient ion on dose informat tests. In conclusion, the - response, especially in the case for in vivo possibility to set SCL for germ cell mutagenicity is therefore not considered possible in the process of self - classification as there is no standardized methodical approach available which adequately takes into account all relevant information.

369 Guidance on the Application of the CLP Criteria 5.0 July 2017 369 Version – 3.5.2.7. Decision logic for classification of substances The decision logic which follows is provided as additional guidance. It is strongly recommended that the person responsible for classif ication study the criteria before and during use of the decision logic. No Classification Does the substance have data on mutagenicity? not possible Yes Category 1 According to the criteria, is the substance: (a) Known to induce heritable mutations in germ cells of humans, or Yes (b) regarded as if it induces heritable Should it be mutations in the germ cells of humans? Danger Application of the criteria needs expert judgement in a weight of evidence approach. No Category 2 According to the criteria, does the substance cause concern for humans owing to the possibility that it Yes may induce heritable mutations in the germ cells of humans? Application of the criteria needs expert judgement Warning . in a weight of evidence approach No Not classified

370 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 370 3.5.3. Classification of mixtures for germ cell mutagenicity 3.5.3.1. Classification criteria for mixtures ingredients Classification of mixtures will be based on the available test data for the individual of the mixture, using concentration limits for those ingredients. Under rare circumstances, the - by - case basis based on the available test data for the classification may be modified on a case mixture as a whole or based on bridging principles ( see CLP Article 6(3) and CLP Annex I, 3.5.3.2 and 3.5.3.3). When data are available for the complete mixture 3.5.3.1.1. Annex I: Classification of mixtures will be based on the available test data for the 3.5.3.2.1. individual ingredients of the mixture using concent ration limits for the ingredients classified as germ cell mutagens. On a case - by - case basis, test data on mixtures may be used for classification when demonstrating effects that have not been established from the evaluation based on the individual ingredie nts. In such cases, the test results for the mixture as a whole must be shown to be conclusive taking into account dose and other factors such as duration, observations, sensitivity and statistical analysis of germ cell mutagenicity test systems. Adequate documentation supporting the classification shall be retained and made available for review upon request. 3.5.3.1.2. When data are not available for the complete mixture: bridging principles ne its germ cell 3.5.3.3.1. Annex I: Where the mixture itself has not been tested to determi mutagenicity hazard, but there are sufficient data on the individual ingredients and similar tested mixtures (subject to paragraph 3.5.3.2.1), to adequately characterise the hazards of th the applicable bridging rules set out the mixture, these data shall be used in accordance wi in section 1.1.3. Bridging principles will only be used on a case by case basis. Note that the following bridging principles are not applicable to this hazard class: concentration of highly hazardous mixtures  interpolation within one hazard category  (see CLP Annex 1, 1.1.3.3 and 1.1.3.4) Note that the bridging priciples are relevant only in case of comparable tested mixtures showing mutagenic effects not established from the evaluation of the individual ingredi ents. Classification for CMR hazards is based on tests with the ingredients. 3.5.3.2. Generic concentration limits for substances triggering classification of mixtures Annex I : 3.5.3.1.1. The mixture shall be classified as a mutagen when at least one ingredient ha s been classified as a Category 1A, Category 1B or Category 2 mutagen and is present at or above the appropriate generic concentration limit as shown in Table 3.5.2 for Category 1A, Category 1B and Category 2 respectively. Table 3.5.2 Generic concentration limits of ingredients of a mixture classified as germ cell mutagens that trigger classification of the mixture. Concentration limits triggering classification of a mixture as:

371 Guidance on the Application of the CLP Criteria 5.0 – July 2017 371 Version Ingredient classified as: Category 1 mutagen Category 2 mutagen Category 1B Category 1A Category 1A mutagen 0,1 % — — ≥ Category 1B mutagen — ≥ 0,1 % — Category 2 mutagen — — ≥ 1,0 % Note The concentration limits in the table above apply to solids and liquids (w/w units) as well as gases (v/v units). The option to set SCL for germ cell mutagenicity is not considered possible in the process of - classification as there is no standardized methodical approach available which adequately self 3.5.2.6 of this Guidance). takes into account all relevant information (see Section For germ cell mutagenicity it is reasonable to assume additivity for mutagens, unless there are specific reasons not to do so. 3.5.3.3. Decision logic for classification of mixtures The decision logic which follows is provided as additional guidance. It is strongly recommended that the person responsible for classification study the criteria before and during use of the decision logic. Thi s decision logic deviates (slightly) from the original GHS guidance, to meet CLP requirements. Classification based on individual ingredients of the mixture

372 ication of the CLP Criteria Guidance on the Appl July 2017 Version 5.0 – 372 Category 1 Yes Does the mixture contain one or more ingredients 0.1%?  classified as a Category 1 mutagen at Danger No Category 2 Yes Does the mixture contain one or more ingredients classified as a Category 2 mutagen at 1.0%?  Warning No Not classified Modified classification on a case - - case basis by be used for classification when demonstrating effects that have not Test data on mixtures may been established from the evaluation based on the individual ingredients (CLP Annex I, 3.5.3.2.1, see also CLP Article 6(3)).

373 Guidance on the Application of the CLP Criteria – July 2017 373 5.0 Version Classify in Are the test results on the appropriate mixture conclusive taking Are test data available into account dose and category for the mixture itself other factors such as demonstrating a Yes Yes duration, observations mutagenic effect not and analysis (e.g. identified from the data statistical analysis, test on individual sensitivity) of germ cell substances? mutagenicity test Danger systems? or No Warning No Yes Can bridging principles be applied? No See above: Classification based on individual ingredients of the mixture.

374 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 374 3.5.4. Hazard communication in form of labelling for germ cell mutagenicity Pictograms, signal words, hazard statements and precautionary 3.5.4.1. statements 3.5.4.1. Annex I: Label elements shall be used in accordance with Table 3.5.3, for substances or mixtures meeting the criteria for classification in this hazard class. Table 3.5.3 Label elements of germ cell mutagenicity Classification Category 1 Category 2 1A, 1B) (Category GHS Pictograms Signal Word Danger Warning Hazard Statement H340: May cause genetic H341: Suspected of causing defects (state route of exposure genetic defects (state route of exposure if it is conclusively if it is conclusively proven that proven that no other routes of no other routes of exposure exposure cause the hazard) cause the hazard) P201 Precautionary Statement P201 P202 Prevention P202 P280 P280 Precautionary Statement P308 + P313 P308 + P313 Response Precautionary Statement P405 P405 Storage Precautionary Statement P501 P501 Disposal The hazard statement to be applied for the classification germ cell mutagenicity has to be amended to state the route of exposure if it is conclusively proven that no other routes of exposure will lead to the respective effect. A conclusive proof means tha t valid in vivo test data need to be available for all three exposure routes clearly indicating that only one exposure route leads to positive results. Moreover, such findings should be plausible with respect to the mode h circumstances rarely, if ever, exist. Therefore, amending the of action. It is estimated that suc hazard statement with the route of exposure generally does not have to be considered. 3.5.4.2. Additional labelling provisions There are no additional labelling provisions for substances and mixtures classified for germ cell mutagenicity under the CLP Regulation. However entry 29 of Annex XVII to REACH addresses as such substances and mixtures. The packaging of substances with a harmonised classification

375 Guidance on the Application of the CLP Criteria 5.0 – Version 375 July 2017 Muta 1A or 1B and that are included in Appendice s 3 and 4 of Annex VII of REACH,as well as the packaging of mixtur e s containing those substances above the concentration limits leading to the classification of the mixture, ‘must be marked visibly, legibly and indelibly as follows: “Restricted to professional users”.’ Derogations from this obligation are outlined in the same provision.

376 ication of the CLP Criteria Guidance on the Appl 376 Version 5.0 – July 2017 3.6. CARCINOGENICITY Definitions and general considerations for classification for 3.6.1. carcinogenicity Carcinogen means a substance or a mixture of substa Annex I: 3.6.1.1. nces which induce cancer or increase its incidence. Substances which have induced benign and malignant tumours in well performed experimental studies on animals are considered also to be presumed or suspected human carcinogens unless there is strong eviden ce that the mechanism of tumour formation is not relevant for humans. More explicitly, chemicals are defined as carcinogenic if they induce tumours, increase tumour incidence and/or malignancy or shorten the time to tumour occurrence. Benign tumours that are considered to have the potential to progress to malignant tumours are generally considered along with malignant tumours. Chemicals can potentially induce cancer by any route of t carcinogenic exposure (e.g. when inhaled, ingested, applied to the skin or injected), bu potential and potency may depend on the conditions of exposure (e.g., route, level, pattern and duration of exposure). Carcinogenic chemicals have conventionally been divided according to the presumed mode of - genotox ic, see Section 3.6.2.3.2.( k ) of this Guidance. action; genotoxic or non Classification of a substance as a carcinogen is based on consideration of the strength of the evidence of available data for classification with considerations of all other relevant information (weight of evidence) being taken into account as appropriate. Strength of evidence involves the enumeration of tumours in human and animal studies and determination of their leve l of statistical significance. A number of other factors need to be considered that influence the overall likelihood that a substance poses a carcinogenic hazard in humans (weight of evidence determination). The list of factors for additional consideration is long and requires the most up - - to date scientific knowledge. It is recognised that, in most cases, expert judgement is necessary to be able to determine the most appropriate category for classification for carcinogenicity. Classification of substances f 3.6.2. or carcinogenicity Identification of hazard information 3.6.2.1. Carcinogens may be identified from epidemiological studies, from animal experiments and/or other appropriate means that may include (Quantitative) Structure - Activity Relationships d/or extrapolation from structurally similar substances (read ((Q)SAR) analyses an across). In - addition some information on the carcinogenic potential can be inferred from and in vitro in vivo in vitro cell transformation assays , and gap germ cell and somatic cell mutagenicity studies, junction intercellular communication (GJIC) tests. Extensive guidance on data requirements, information sources and strategies for the identification of potential carcinogens are given in the Guidance on IR&CSA , Section R.7.7.9 rements on carcinogenicity) and Section R.7.7.10 ( Information and its (Information requi sources on carcinogenicity) and for potential mutagens Section R.7.7.3 (Information and its sources on mutagenicity) . For more about non testing data see 3.6.2.3.4 of this Guidance. Section 3.6.2.2. Classification criteria for substances Substances are classified according to their potential to cause cancer in humans. In some cases there will be direct evidence on the carcinogenicity to humans from epidemiological studies. However, in most cases the available information on carcinogenicity will be primarily from

377 Guidance on the Application of the CLP Criteria – Version 377 July 2017 5.0 s must be animal studies. In this case the relevance of the findings in animals to human considered. Annex I: 3.6.2.1. For the purpose of classification for carcinogenicity, substances are allocated to one of two categories based on strength of evidence and additional considerations - specific classification may be warranted, if it (weight of evidence). In certain instances, route can be conclusively proved that no other route of exposure exhibits the hazard. Table 3.6.1 Hazard categories for carcinogens Criteria Categories presumed human carcinogens Known or CATEGORY 1: A substance is classified in Category 1 for carcinogenicity on the basis of epidemiological and/or animal data. A substance may be further distinguished as: Category 1A: Category 1A, known to have carcinogenic potential for humans, classific ation is largely based on human evidence, or Category 1B, presumed to have carcinogenic potential for humans, Category 1B: classification is largely based on animal evidence. The classification in Category 1A and 1B is based on strength of onal considerations (see section 3.6.2.2). evidence together with additi Such evidence may be derived from: human studies that establish a causal relationship between human – exposure to a substance and the development of cancer (known human carcinogen); or 1 animal experiments for which t here is sufficient ( – ) evidence to demonstrate animal carcinogenicity (presumed human carcinogen). In addition, on a case - by - case basis, scientific judgement may warrant a decision of presumed human carcinogenicity derived from studies showing limited evid ence of carcinogenicity in humans together with limited evidence of carcinogenicity in experimental animals. CATEGORY 2: Suspected human carcinogens The placing of a substance in Category 2 is done on the basis of animal studies, but which is not evidence obtained from human and/or sufficiently convincing to place the substance in Category 1A or 1B, based on strength of evidence together with additional considerations (see section 3.6.2.2). Such evidence may be derived either from 1 limited( ) evidence of carcinogenicity in human studies or from limited evidence of carcinogenicity in animal studies. 1 ( ) Note: See 3.6.2.2.4.

378 ication of the CLP Criteria Guidance on the Appl 378 Version 5.0 – July 2017 3.6.2.3. Evaluation of hazard information Classification as a carcinogen is made on the basis of evidence from 3.6.2.2.1. Annex I: reliable and acceptable studies and is intended to be used for substances which have an intrinsic property to cause cancer. The evaluations shall be based on all existing data, peer - reviewed published studies and additional acceptable data. 3.6.2. 2.2. Classification of a substance as a carcinogen is a process that involves Annex I: two interrelated determinations: evaluations of strength of evidence and consideration of all rd other relevant information to place substances with human cancer potential into haza categories. Classification of a substance as a carcinogen requires expert judgement and consideration of many different factors (weight and strength of evidence) included in the hazard information on o data analysis rather than hard and fast carcinogenicity. The guidance provides an approach t rules. A stepwise approach to the classification can be taken where all the factors, both weight and strength of evidence, that may influence the outcome are considered systematically. Such ideration of these factors is outlined, in McGregor et al, 2009 and approach, including cons et al , 2006. Also the IPCS ‘Conceptual Framework for Evaluating a Mode of Action for Boobis Chemical carcinogenesis’ (2001), ILSI ‘Framework for Human Relevance Analysis of Information on et al ., 2003; Cohen et al Carcinogenic Modes of Action’ (Meek , 2003, 2004) and the International Agency for Research on Cancer (IARC, 2006 - Preamble Section B) provide a basis for systematic assessments which may be performed in a consistent fashion inter nationally; however they are not intended to provide lists of criteria to be checked off. Specific considerations that are necessary are outlined in CLP Annex I, 3.6.2.2.3 (see Section 3.6.2.3.1 of this Guidance) and other important factors to consider in CLP Annex I, 3.6.2.2.6 (see Section of this Guidance). Further guidance on these important factors is given in 3.6.2.3.2 this document. Specific considerations for classification 3.6.2.3.1. There is a strong link between CLP and the IARC classification criteria. The definitions for sufficient and limited evidence are part of the criteria (CLP Annex I, as defined by IARC limited’ as follows: ‘ It is 3.6.2.2.3). IARC, however, understands the criteria of ‘sufficient’ and ‘ recognized that the criteria for these evaluations, described below, cannot encompass all of the factors that may be relevant to an evaluation of carcinogenicity. In considering all of the relevant scientific data, the Working G roup may assign the agent to a higher or lower category than a strict interpretation of these criteria would indicate.’ (IARC 2006 preamble Section 6, Evaluation and rationale). This sentence emphasises that in certain circumstances expert errule the strict interpretation of the IARC criteria for ‘ sufficient’ and ‘limited’. judgement may ov These same limitations apply with the current criteria in that expert judgement is necessary and can override the strict interpretation of the definitions. Annex I: 3.6.2.2.3. Strength of evidence involves the enumeration of tumours in human and animal studies and determination of their level of statistical significance. Sufficient human ncer, evidence demonstrates causality between human exposure and the development of ca whereas sufficient evidence in animals shows a causal relationship between the substance and an increased incidence of tumours. Limited evidence in humans is demonstrated by a positive association between exposure and cancer, but a causal relationshi p cannot be stated. Limited evidence in animals is provided when data suggest a carcinogenic effect, but are less than sufficient. The terms 'sufficient' and 'limited' have been used here as they have been defined by the International Agency for Research o n Cancer (IARC) and read as follows: (a) Carcinogenicity in humans

379 Guidance on the Application of the CLP Criteria 5.0 – July 2017 379 Version The evidence relevant to carcinogenicity from studies in humans is classified into one of the following categories: sufficient evidence of carcinogenicity: a causal relationship has been es tablished – between exposure to the agent and human cancer. That is, a positive relationship has been observed between the exposure and cancer in studies in which chance, bias and confounding could be ruled out with reasonable confidence; – limited evidence of carcinogenicity: a positive association has been observed between exposure to the agent and cancer for which a causal interpretation is considered to be credible, but chance, bias or confounding could not be ruled out with reasonable confidence. nogenicity in experimental animals (b) Carci Carcinogenicity in experimental animals can be evaluated using conventional bioassays, bioassays that employ genetically modified animals, and other in - vivo rcinogenesis. In the bioassays that focus on one or more of the critical stages of ca - term bioassays or from assays with neoplasia absence of data from conventional long as the end - point, consistently positive results in several models that address several stages in the multistage process of carcinogenesis should be con sidered in evaluating the degree of evidence of carcinogenicity in experimental animals. The evidence relevant to carcinogenicity in experimental animals is classified into one of the following categories: – elationship has been established sufficient evidence of carcinogenicity: a causal r between the agent and an increased incidence of malignant neoplasms or of an appropriate combination of benign and malignant neoplasms in (a) two or more ies carried out at species of animals or (b) two or more independent studies in one spec different times or in different laboratories or under different protocols. An increased incidence of tumours in both sexes of a single species in a well - conducted study, sufficient ideally conducted under Good Laboratory Practices, can also provide evidence. A single study in one species and sex might be considered to provide sufficient evidence of carcinogenicity when malignant neoplasms occur to an unusual degree with regard to incidence, site, type of tumour or age at onset, or when th ere are strong findings of tumours at multiple sites; – limited evidence of carcinogenicity: the data suggest a carcinogenic effect but are limited for making a definitive evaluation because, e.g. (a) the evidence of carcinogenicity is restricted to a single experiment; (b) there are unresolved questions regarding the adequacy of the design, conduct or interpretation of the studies; (c) the agent increases the incidence only of benign neoplasms or lesions of uncertain neoplastic potential; or (d) the evidence of carcinogenicity is restricted to studies that demonstrate only promoting activity in a narrow range of tissues or organs. For human studies, the quality and power of the epidemiology studies require expert consideration and would normally lead to a Category 1A classification if data of adequate quality shows causality of exposure and cancer development. The Guidance on IR&CSA , Section R.7.7.10.2, further discusses the types of human epidemiology data available and the limitations of the data. Where there is sufficient doubt in the human data then classification in Category 1B may be more appropriate. On the other hand epidemio logical studies may fail, because of uncertainties in the exposure assessment and/or limited sensitivity and statistical power, to confirm the carcinogenic properties of a substance as identified in animal studies (WHO Working group, 2000).

380 ication of the CLP Criteria Guidance on the Appl 380 Version 5.0 – July 2017 3.6.2.3.2. siderations for classification Additional con 3.6.2.2.4. Additional considerations (as part of the weight of evidence approach Annex I: (see 1.1.1)). Beyond the determination of the strength of evidence for carcinogenicity, a number of other factors need to be considered that influence the overall likelihood that a substance poses a carcinogenic hazard in humans. The full list of factors that influence this determination would be very lengthy, but some of the more important ones are considered here. : 3.6.2.2.5. Th e factors can be viewed as either increasing or decreasing the level of Annex I concern for human carcinogenicity. The relative emphasis accorded to each factor depends upon the amount and coherence of evidence bearing on each. Generally there is a more complete information to decrease than to increase the level of concern. requirement for Additional considerations should be used in evaluating the tumour findings and the other factors in a case by - case manner. - : 3.6.2.2.6. Some important factors which may be taken into consideration, when Annex I assessing the overall level of concern are: tumour type and background incidence; (a) (b) multi - site responses; (c) progression of lesions to malignancy; (d) reduced tumour latency; whether responses are in single or bot (e) h sexes; whether responses are in a single species or several species; (f) (g) structural similarity to a substance(s) for which there is good evidence of carcinogenicity; (h) routes of exposure; comparison of absorption, distribution, metabolism and e xcretion between test animals (i) and humans; (j) the possibility of a confounding effect of excessive toxicity at test doses; (k) mode of action and its relevance for humans, such as cytotoxicity with growth stimulation, mitogenesis, immunosuppression, mutage nicity. [...] As indicated above, the evaluation of animal carcinogenicity data requires consideration of a number of important additional factors which may increase or decrease the level of concern and the classification category. The list in CLP Annex I, 3.6.2.2.6 is not exhaustive. Each of these factors is discussed individually below. a. Tum our type and background incidence Knowledge about the tumour type including its tumour biology is indispensable to decide on the relevance of observed tumours for humans. By default, carcinogenic effects in experimental animals are considered relevant to humans and are considered for classification as carcinogens. Only when there is sufficient evidence showing that a certain type of tumour is not relevant to humans should this tumour type be excluded for classification. Certain tumour types observed in an imal carcinogenicity studies are of questionable or no relevance to humans. In case of multiple tumours anticipated to have no relevance for humans

381 Guidance on the Application of the CLP Criteria 5.0 – July 2017 381 Version justification should be given for each tumour type. The justification for dismissing any particular tumour s hould be presented as a scientifically robust and transparent argument. There are several reasons why a tumour observed in animals may be judged to be not relevant s for humans or may be judged to be of lower concern. In most of these cases the tumour arise via a mode of action which does not occur in humans (see this Section part k). In some cases the tumour may arise in a tissue known to be overly susceptible in the species tested to development of certain tumours and consequently may be judged to be less relevant for humans. In a few cases a tumour may occur in a tissue with no equivalent in humans. Tumours occurring in tissues with no human equivalent Some of the commonly used animal species have some tissues with no equivalent in humans. Tumours occurri ng in these tissues include the following  Forestomach tumours in rodents following administration by gavage of irritating or corrosive, non mutagenic substances. In rodents, the stomach is divided into two parts by the muco - epidermoid junction separating s quamous from glandular epithelium. The - proximal part, or forestomach, is non glandular, forms a continuum with the oesophagus, and is lined by keratinized, stratified squamous epithelium. While humans do not have a forestomach, they do have comparable squa mous epithelial tissues in the oral cavity and the upper two - thirds of the oesophagus. See also this Section (k), IARC (2003), and RIVM (2003).  Tumours in the Zymbal’s glands. Zymbal’s glands are located beneath squamous epithelium at the anterior and post erior aspect of the ear canal. The external portion of the gland in rats is 3 to 5 millimetres in diameter. Tumours in the Harderian glands. Harderian glands are found in all vertebrates that  d behind the eyeball in possess a nictitating membrane, or third eyelid. They are locate the orbit nictitating membrane, encircling the optic nerve. Humans have a rudimentary one. Tumours occurring in such tissues indicate that the substance has the potential to induce carcinogenic effects in the species tested. It cann ot automatically be ruled out that the substance could cause similar tumours of comparable cell/tissue origin (e.g. squamous cell tumours at other epithelial tissues) in humans. Careful consideration and expert judgement of these tumours in the context of the complete tumour response (i.e. if there are also tumours at other sites) and the assumed mode of action is required to decide if these findings would support a classification. However, tumours observed only in these tissues, with no other observed tumo urs are unlikely to lead to classification. However, such determinations must be evaluated carefully in justifying the carcinogenic potential for humans; any occurrence of other tumours at distant sites must also be considered. Considering the background i ncidence and use of historical control data - Any statistically significant increase in tumour incidence, especially where there is a dose response relationship, is generally taken as positive evidence of carcinogenic activity. However, ults involve an increase incidence of tumours in treated animals which lies in some cases the res at the borderline of biological and/or statistical significance or there is an increase in a spontaneous tumour type, then comparison of the tumour incidence with historical control tumour data is strongly encouraged. Historical control data provide useful information on the normal pattern and range of tumour types and incidences for a particular strain/species, which may not be reflected by the tumour findings in the concurre nt controls in any individual study. This can be particularly relevant for animal strains which have a propensity to develop a particular type of tumour spontaneously with variable and potentially high incidence. In such a case the tumour incidence in the treated group may be significantly above the concurrent control but could still be within the historical

382 ication of the CLP Criteria Guidance on the Appl 382 Version 5.0 – July 2017 incidence range for that tumour type in that species and therefore may not be providing reliable evidence of treatment related carcinogenicity. amples of animal tissues with a high spontaneous tumour incidence are: Some ex Adrenal pheochromocytoma in male F344 rats (NTP, 2007a), Sprague - Dawley rats (NTP,  ., 2002); et al 2005; RIVM, 2001; Ozaki Pituitary adenomas in F344 rats (NTP, 2007a), Sprague - Dawley ra  ts (NTP 2005; RIVM 2005); - Dawley rats  Mammary gland tumours (adenomas and carcinomas) in female Sprague (NTP, 2005); Mononuclear cell leukaemia in F344 rats (NTP, 2007a; RIVM, 2005);   Liver tumours in B6C3F1 mice (NTP, 2007b; Haseman et al . 1998; Battershil l, J.M. and Fielder, R.J., 1998); ., 2002; RIVM,  Leydig cell adenomas in male F344 rats (Cook et al ., 1999; Mati et al 2004; EU Specialised Experts Report, 2004). Historical control data can also be useful to judge the biological significance of marginal in creases in uncommon tumours. If there is a small increase in a particular tumour type which historical data shows to be very uncommon and unlikely to have occurred by chance then this may support a conclusion of carcinogenicity without the requirement for a statistically significant increase. Use of historical control data should be on a case by case basis with due consideration of the appropriateness and relevance of the historical control data for the study under evaluation. In a d as closely as possible to the general sense, the historical control data set should be matche study being evaluated. The historical data must be from the same animal strain/species, and ideally, be from the same laboratory to minimise any potential confounding due to variations in ons, animal suppliers, husbandry etc. It is also known that laboratory conditions, study conditi tumour incidences in control animals can change over time, due to factors such as genetic drift, changes in diagnostic criteria for pathological changes/tumour types, and husbandry factors ing the standard diet used), so the historical data should be contemporary to the study (includ being evaluated (e.g. within a period of up to around 5 years of the study). Historical data older than this should be used with caution and acknowledgement of its lowe r relevance and reliability. (RIVM, 2005; Fung et al , 1996; Greim et al , 2003). Even when a particular tumour type may be discounted, expert judgment must be used in assessing the total tumour profile in any animal. However, appearance of only spontaneous tumours, especially if they appear only at high dose levels, may be sufficient to downgrade a classification from Category 1B to Category 2, or even no classification. Where the only available tumour data are liver tumours in certain sensitive strains of m ice, without any other supplementary evidence, the substance may not be classified in any of the categories, (Battershill and Fielder, 1998). Expert judgment is required to evaluate the relevance of the results. b. Multi - site responses In general, chemicals a - year bioassays conducted re evaluated for carcinogenic potential in two in mice and rats. The chemicals produce a spectrum of responses ranging from no effects in either species to induction of malignant neoplasms in multiple tissues in both species. Betwe en these two extremes, there are variable responses in tissues, sexes and species, which demonstrate that there are important differences among the carcinogens, as well as between the species in which they are tested. The tumour profile observed with a sub stance should be taken into account when considering the most appropriate classification.

383 Guidance on the Application of the CLP Criteria 5.0 – July 2017 383 Version Evidence shows that substances which cause tumours in either multiple sites and/or multiple species tend to be more potent carcinogens than those causing tumours at o nly one site in one , 1997). This is often true for substances which are mutagenic. Also, species (Dybing et al. where human carcinogens have been tested in two or more species, the majority have caused substance causes tumours at multiple sites cancer in several species (Tennant, 1993). Thus, if a and/or in more than one species then this usually provides strong evidence of carcinogenicity. Typically such a tumour profile would lead to a classification in category 1B. c. Progression of lesions to malignancy In general, if a substance involves a treatment related increase in tumours then it will meet the criteria for classification as a carcinogen. If the substance has been shown to cause malignant tumours this will usually constitute sufficient evidence of carc inogenicity supporting Category 1B (CLP Annex I, 3.6.2.2.3) The induction of only benign tumours usually provides a lower strength of evidence for carcinogenicity than the induction of malignant tumours and will usually support Category 2 (CLP Annex I, 3.6 .2.2.3). However, benign tumours may also be of significant concern and the strength of evidence for carcinogenicity that they provide should be considered using expert judgement. For instance, some benign tumours may have the potential to progress to ignant tumours and therefore any indication that the observed tumours have the potential to mal progress to malignancy may increase the level of concern. Also, some benign tumours, for example brain tumours, may be of concern in themselves. d. Reduced tumour late ncy The latency of tumour development i.e. how quickly a substance induces tumours, often reflects the potency of a carcinogen. This is particularly true for mutagenic substances which often induce tumours with relatively short latency and usually more rap idly than non - genotoxic agents. Tumour latency is not generally investigated in detail in standard carcinogenicity studies, although some information may be provided if the study used serial sacrifices. ffect the classification and hazard The latency of tumour formation does not materially a category. Any substance causing cancer will attract classification regardless of the latency for tumour development. This also includes tumour responses at late treatment/life periods if substance - related. However unusua l tumour types or tumours occurring with reduced latency may add to the weight of evidence for the carcinogenic potential of a substance, even if the tumours are not statistically significant. Whether responses are in single or both sexes e. In general, in st andard carcinogenicity studies both male and female animals are tested. There may be cases where tumours are only observed in one sex. Tumours in one sex only may arise for two broad reasons. The tumours may occur in a gender - specific tissue, for instance the uterus or testes (sex - specific tissue), or in a non sex - specific tissue, in one sex only. Tumours may also be induced by a mechanism that is gender (or sex) - - mediated mechanism or one involving gender (or sex) - specific, for instance a hormonally speci fic differences in toxicokinetics. As with all cases the strength of evidence of carcinogenicity should be assessed based on the totality of the information available using a weight of evidence type approach. A default position is that such tumours are sti ll evidence of carcinogenicity and should be evaluated in light of the total tumorigenic response to the substance observed at other sites (multi - site responses or incidence above background) in determining the carcinogenic potential and the classification category. If tumours are seen only in one sex of an animal species, the mode of action should be carefully evaluated to see if the response is consistent with the postulated mode of action. Effects seen only in one sex in a test species may be less convin cing than effects seen in both sexes, unless there is a clear patho - physiological difference consistent with the mode of action to explain the

384 ication of the CLP Criteria Guidance on the Appl 384 Version 5.0 – July 2017 single sex response. However, there is no requirement for a mechanistic understanding of tumour induction in ord er to use these findings to support classification. If there is clear specific tumour then classification in Cat 1B - evidence for induction of either a gender (or a sex) eria may be appropriate. However, it has to be taken into account that according to the crit additional data are required to provide sufficient evidence for animal carcinogenicity (1B). f. Whether responses are in single species or several species that carcinogenicity in a single animal species (both sexes, ideally in a GLP The criteria indicate sufficient evidence and could therefore lead to a Category 1B classification in study) could be - the absence of any other data. This represents a change compared to the previous EU system 1B classification. where such a study would rarely lead to the equivalent of a Category s defined under ‘sufficient’ evidence (CLP Annex I, 3.6.2.2.3 (b)), a single study in However, a one species and sex might be considered to provide sufficient evidence of carcinogenicity when malignant neoplasms occur to an unusual degr ee with regard to incidence, site, type of tumour or age at onset, or when there are strong findings of tumours at multiple sites. Moreover a - vivo mutagenicity data would single study in one species and sex in combination with positive in provide sufficient evidence of carcinogenicity. be considered to Positive responses in several species add to the weight of evidence, that a chemical is a carcinogen. Structural similarity or not to a chemical(s) for which there is good evidence of g. carcinogenicity ect See S ion 3.6.2.3.4 of this Guidance. h. Routes of exposure; 3.6.2.2.8. Annex I: The classification shall take into consideration whether or not the substance is absorbed by a given route(s); or whether there are only local tumours at the site of administration for the tested route(s), and adequate testing by other major route(s) show lack of carcinogenicity. The classification for carcinogenicity generally does not specify specific routes of exposure. If a chemical has been shown to cause tumours by any route of administration then it may require ssing the findings from a particular classification, unless there is a robust justification for dismi route. However, a specific hazard statement has been established in CLP , H350i; May cause cancer by inhalation. Most standard carcinogenicity studies use physiological routes of exposure for humans, namely ral or dermal exposure. The findings from such routes are usually considered inhalation, o directly relevant for humans. Studies using these routes will generally take precedence over similar studies using other routes of exposure. Sometimes other non - physiological rout es are used, such as intra - muscular, sub - cutaneous, intra - - tracheal injections or instillations. Findings from studies using these peritoneal and intra routes may provide useful information but should be considered with caution. Usually dosing via tes provides a high bolus dose which gives different toxicokinetics to normal routes these rou and can lead to atypical indication of carcinogenicity. For instance, the high local concentration can lead to local tumours at the site of injection. These would not norm ally be considered reliable indications of carcinogenicity as they most likely arose from the abnormally high local concentration of the test substance and would lead to a lower category classification or no classification. Where findings are available fro m studies using standard routes and non - physiological routes, the former will generally take precedence. Usually studies using non - standard routes provide supporting evidence only.

385 Guidance on the Application of the CLP Criteria 5.0 – July 2017 385 Version The hazard statement allows for identifying the route of exposure ‘if it is conclusively proven that no other routes of exposure cause the hazard’ (CLP Annex I, Table 3.6.3). In this case the carcinogens are generally suspected hazard statement may be modified accordingly. Genotoxic to be carcinogenic by any route. Comparison of absorption, distribution, metabolism and excretion between test animals i. and humans; Annex I: It is important that whatever is known of the physico - chemical, 3.6.2.2.9. toxicokinetic and toxicodynamic properties of the substances, as well as any available relevant information on chemical analogues, i.e. structure activity relationship, is taken into consideration when undertaking classification. Consideration of absorption, distribution, metabolism and excretion (toxicokinetics) of the substance in the te st animal species and in humans is one important consideration, including where a substance is metabolised to an active carcinogenic metabolite. Toxicokinetic behaviour is normally assumed to be similar in animals and humans, at least from a qualitative rspective. On the other hand, certain tumour types in animals may be associated with pe toxicokinetics or toxicodynamics that are unique to the animal species tested and may not be predictive of carcinogenicity in humans. Where significant qualitative and qua ntitative differences in toxicokinetics exist between animals and humans this can impact on the relevance of the animal findings for humans and in certain instances may influence the category of classification. Where a carcinogenic metabolite identified in animals is demonstrated not to be produced in humans, no classification may be warranted where it can be shown that this is the only mechanism of action for carcinogenicity. The use of physiologically based pharmacokinetic ( PB/PK) modelling requires more validation - and while it may not lead directly to a modification of classification, however expert judgement in conjunction with PB/PK modelling may help to modify the concern for humans. The possibility of a confounding effect of excessive toxicity at test doses j. In lifetime bioassays compounds are routinely tested using at least three dose levels to enable hazard identification and hazard characterisation as part of risk assessment. Of these doses, the highest dose needs to induce minimal toxicity, such as characterised by an approximately 10% reduction in body weight gain (maximal tolerated dose, MTD dose). The MTD is the highest dose of the test agent during the bioassay that can be predicted not to alter the animal’s normal longevity from effects other t han carcinogenicity. Data obtained from a sub - chronic or other repeated dose toxicity study are used as the basis for determining the MTD. Excessive toxicity, for instance toxicity at doses exceeding the MTD, can affect the carcinogenic responses in bioass ays. Such toxicity can cause effects such as cell death (necrosis) with associated regenerative hyperplasia, which can lead to tumour development as a secondary consequence unrelated to the intrinsic potential of the substance itself to cause tumours at wer less toxic doses. lo Tumours occurring only at excessive doses associated with severe toxicity generally have a more doubtful potential for carcinogenicity in humans. In addition, tumours occurring only at sites of contact and/or only at excessive doses n eed to be carefully evaluated for human relevance for carcinogenic hazard. For example, as indicated in this Section (a) ‘Tumour type and background incidence’, forestomach tumours, following administration by gavage of an irritating or corrosive, non - muta genic chemical, may be of questionable relevance, both due to the lack of a corresponding tissue in humans, but importantly, due to the high dose direct effect on the tissue. However, such determinations must be evaluated carefully in justifying the carcin ogenic potential for humans; any occurrence of other tumours at distant sites must also be considered.

386 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 386 The proceedings of a WHO/IPCS workshop on the Harmonization of Risk Assessment for - A Scoping Meeting (IPC S, 1995; Ashby et al , Carcinogenicity and Mutagenicity (Germ cells) 1996), points to a number of scientific questions arising for classification of chemicals, e.g. mouse liver tumours, peroxisome proliferation, receptor - mediated reactions, chemicals which are carcinogenic only at toxic doses and which do not demonstrate mutagenicity. If a test compound is only found to be carcinogenic at the highest dose(s) used in a lifetime bioassay, and the characteristics associated with doses exceeding the MTD as outlined above are present, this could be an indica tion of a confounding effect of excessive toxicity. This may support a classification of the test compound in Category 2 or no classification. k. Mode of action and its relevance for humans, such as mutagenicity, cytotoxicity with growth stimulation, mitogene sis, immunosuppression Carcinogenic chemicals have conventionally been divided into two categories according to the presumed mode of action; genotoxic or non - genotoxic. Genotoxic modes of action involve genetic alterations caused by the chemical interactin g directly with DNA to possibly result in a change in the primary sequence of DNA after cell division. A chemical can also cause genetic alterations indirectly following interaction with other cellular processes (e.g. secondary to the induction of oxidativ e stress). Non - genotoxic modes of action include epigenetic changes, i.e. effects that do not involve alterations in DNA but that may influence gene expression, altered cell - cell communication, or other factors involved in the carcinogenic process. For exa mple, chronic cytotoxicity with subsequent regenerative cell proliferation is considered a mode of action by which tumour development can be enhanced: the induction of urinary bladder tumours in rats may, in certain cases, be due to persistent irritation/i nflammation, tissue erosion and regenerative hyperplasia of the urothelium following the formation of bladder stones. Other modes of non - genotoxic action can involve specific receptors (e.g., peroxisome activated receptor - alpha (PPARα) which i s associated with liver tumours in rodents; proliferator - or tumours induced by various hormonal mechanisms). More detail is given in the Guidance on IR/CIS Section R7.7.8. Some modes of action of tumour formation are considered to be not relevant to humans. Where such a mechanism is identified then classification may not be appropriate. Only if a mode of action of tumour development is conclusively determined not to be operative in humans may the carcinogenic evidence for that tumour be discounted. However, a weight of evidence evaluation for a substance calls for any other tumorigenic activity to be evaluated as well. In addition, the existence of a secondary mechanism of action with the implication of a practical threshold above a certain dose level (e.g., hormonal eff ects on target organs or on mechanisms of physiological regulation, chronic stimulation of cell proliferation) may lead to a downgrading of a Category 1 to Category 2 classification. The various international documents on carcinogen assessment all note tha t mode of action in - by - and of itself, or consideration of comparative metabolism, should be evaluated on a case case basis and are part of an analytic evaluative approach. One must look closely at any mode of action in animal experiments taking into consid eration comparative toxicokinetics/toxicodynamics between the animal test species and humans to determine the relevance of the results to humans. This may lead to the possibility of discounting very specific effects of certain types of chemicals. Life stag e - dependent effects on cellular differentiation may also lead to qualitative differences between animals and humans. To establish a mode of action will usually require specific investigative studies over and above the standard carcinogenicity study. All av ailable data must be considered carefully to judge if it can be concluded with confidence that the tumours are being induced through that specific mechanism. The IPCS Framework for Analyzing the Relevance of a Cancer Mode of Action for Humans (2007) can be a useful way to construct and present a robust and transparent assessment of such data. Some mechanisms of tumour formation considered not relevant for humans:

387 Guidance on the Application of the CLP Criteria 5.0 – July 2017 387 Version - globulin  Kidney tumours in male rats associated with substances causing α2μ nephropathy (IARC, 1999) Pheochromocytomas in male rats exposed to particulates through inhalation secondary  to hypoxemia (Ozaki et al, 2002)  - releasing Leydig cell adenomas induced by dopamine antagonists or gonadotropin 04) hormone (GnRH) (EU Specialised Experts, 2004; RIVM, 20  Urinary bladder tumours due to crystals in the bladder (IARC, 1999)  Forestomach tumours in rodents following administration by gavage of irritating or corrosive, non genotoxic substances (RIVM, 2003; IARC 2003) - ed by UDP glucuronyltransferase (UGT)  Certain thyroid tumours in rodents mediat induction (IARC, 1999; EU Specialised Experts, 1999)  Liver tumours in rodents conclusively linked to peroxisome proliferation (IARC, 1994) Consideration of mutagenicity 3.6.2.3.3. 3.6.2.2.6. [...] Mutagenicity: It is recog nised that genetic events are central in the Annex I: overall process of cancer development. Therefore evidence of mutagenic activity in vivo may indicate that a substance has a potential for carcinogenic effects. 3.6.2.1 of this Guidance and above, carcinogenic chemicals have As indicated in Section - conventionally been divided according to the presumed mode of action; genotoxic or non genotoxic. Evidence of genotoxic activity is gai ned from studies on mutagenic activity. It should be noted that in general if a substance is mutagenic then it will be considered to be potentially carcinogenic in humans however mutagenicity data alone are insufficient information to justify a carcinogen in vitro and in vivo mutagenicity classification. In some cases where only are present without carcinogenicity data, a Category 2 classification can be considered when all factors have been considered such as type and quality of the mutagenicity data, stru cture activity relationships etc. A single positive carcinogenicity study in one species and sex in in - vivo mutagenicity data would be considered to provide sufficient combination with positive evidence of carcinogenicity. tor that other mechanisms are in operation as indicated in Lack of genotoxicity is an indica k Section 3.6.2.3.2.( ) of this Guidance. Thus careful analysis based on all available information is d to identify the mechanism and derive a classification category taking into account the require factors leading to the tumours observed, in the animals. 3.6.2.3.4. Non testing data Annex I: 3.6.2.2.7. A substance that has not been tested for carcinogenicity may in certain i nstances be classified in Category 1A, Category 1B or Category 2 based on tumour data from a structural analogue together with substantial support from consideration of other important factors such as formation of common significant metabolites, e.g. for b enzidine congener dyes. A chemical that has not been tested for carcinogenicity may in certain instances be classified as a carcinogen based on tumour data from a structurally similar chemical with which it is always be based on a predicted to have similar carcinogenic activity. Such an approach must robust and transparent argument to support this supposition. There may also be evidence demonstrating similarity in terms of other important factors such as toxicokinetics or mutagenic activity etc. (OECD 2004, 2005, 2007; Guidance on IR&CSA , Section R.6, QSARs and grouping of chemicals ). In the absence of carcinogenicity data, read - across can be used to support a classification for carcinogenicity when the chemical in question is similar to a known or suspected carcinogen (Category 1A , 1B or 2). The similarity between chemicals is considered in terms of structural features, physico - chemical properties and overall toxicological profile.

388 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 388 In general the chemicals will share a common structural element or functional group ( ., a i.e e) that has been shown to be integral to the underlying mechanism of carcinogenicity toxophor for chemicals with this toxiphore in well conducted studies. These toxiphores can be identified through expert judgement or through automated systems such as (Q)SARs. The - across read should also consider the physico chemical properties of the chemical and data from other - toxicity studies to judge the similarity between the chemicals in terms of bioavailability by relevant routes of exposure and toxicokinetics. The toxicity profile from other studies should also be compared ( e.g., acute and repeated - dose toxicity and mutagenicity) and should share similarities in nature and severity. Data from shorter term toxicity studies may be useful, particularly for non genotoxic carcinogens, to indicate that the chemicals cause the same - lying pathological changes ( e.g., under hyperplasia), and act via a common mode of action. Any predictions made on the basis of read - across should take into account the totality of data on the chemicals in question, including the physico - chemical properties, tox icological profile, toxicokinetics, structural analogy and the performance of any (Q)SAR models used, in a weight of evidence approach driven by expert judgement. The final decision must be clear, scientifically defensible and transparent. The specific cat egory depends on the category of the known carcinogen and the degree of confidence in the robustness of the read - across prediction. The category will not be higher than the chemical used to read - across from, but normally may be the same. However a lower across highlights a possible carcinogenic hazard, and thus ca tegory may be applied if the read - supports a classification, but there is uncertainty as to the robustness of the read - across studies, that the chemical prediction or there is evidence, for instance from mechanistic or other may be of lower concern for carcinogenicity. If a chemical is similar to a substance known to be carcinogenic and shares the toxiphore that is considered to be causally related to carcinogenicity, then it is unlikely that there will be sufficient confidence in a prediction of no hazard (for instance based on arguments relating to - chemical or steric properties), to justify no classification in the absence of differences in physico supporting negative experimental data. However, th e bioavailability of the toxiphore will need IR&CSA R.6). evaluation (Guidance on Decision on classification 3.6.2.4. As mentioned throughout, classification as a carcinogen is based on consideration of the strength of evidence with additional considerations (weigh t of evidence) being taken into account as appropriate. It is recognised that, in most cases, expert judgment is necessary to determine the classification category. 3.6.2.5. Classification of substances containing CMR constituents From a compositional and a toxicol ogical point of view the situation for substances containing CMR constituents, additives or impurities is the same as for mixtures containing components classified for these endpoints. For this reason the classification procedure for CMR endpoints that is foreseen by CLP for mixtures containing CMR components, is considered applicable also to substances containing CMR constituents ec tion 1.1.6.1 ). As , additives or impurities (see S discussed in S ection 3.6.3 below, mixtures containing components classified as carcinigenic shall be normally classi fied using only the relevant available information for the individual substances in the mixture. Further, in cases where the available test data on the mixture itself demonstrate CMR effects which have not been identified from the information on the indivi dual substances, those data shall also be taken into account. For CMR endpoints the lowest incidence possible to detect in the tests is by far unacceptable in humans. Thus a dose as high as possible (such as maximal tolerated dose, MTD dose) is needed to b e able to detect CMR hazards. Dilution, as would be the case if mixtures or substances containing CMR constituents were tested, would increase the risk that CMR hazards would not be detected.

389 Guidance on the Application of the CLP Criteria – Version 389 5.0 July 2017 substances in mixtures are According to article 10 (1) substances in other substances and treated in the same way regarding the use of GCLs and SCLs. 3.6.2.6. Setting of specific concentration limits Experimental studies have revealed large variations in the doses of various carcinogenic animals. Thus, the amounts of chemical carcinogens substances needed to induce tumours in 9 8 required to induce tumours vary with a factor of up to 10 - 10 for different compounds. It is reasonable to assume that there is similar variation in the potency of substances carcinogenic to humans (Sanne r and Dybing, 2005). The carcinogenic properties of mixtures are normally not tested. The classification and labelling of mixtures for carcinogenicity is therefore based on the classification of the ingredients and the 3.6.3 of this Guidance, the percentage of each ingredient in the mixture. As indicated in Section criteria contain default percentages for classification of mixtures with carcinogenic properties but CLP, Article 10.1 allows the use of specific concentration limits (SCL) based on the potency of the carcinogen(s). The EU has adopte et al ., d the T25 concept for carcinogenicity (Dybing 1997) with additional considerations as a measure for intrinsic potency and a guidance document (EC, 1999) to assist in establishing SCLs for carcinogens. By using this approach the be reduced or raised from the default generic concentration limits. SCL may occasionally Decision logic for classification of substances 3.6.2.7. The decision logic which follows is taken from the GHS Guidance. It is strongly recommended that the person responsible for classification, study the criteria for classification before and during use of the decision logic. No Classification not possible Does the subststance have carcinogenicity data? Yes Category 1 According to the criteria, is the substance: a Known . to have carcinogenic potential for humans, or Yes to have carcinogenic potential for humans? . Presumed b Application of the criteria needs expert judgement in a Danger strength and weight of evidence approach. No Category 2 According to the criteria (see ), is the S ection 3.6.2 substance a suspected human carcinogen? Yes Application of the criteria needs expert judgement in a e approach. strength and weight of evidenc Warning No Not classified

390 ication of the CLP Criteria Guidance on the Appl 390 Version 5.0 – July 2017 3.6.3. Classification of mixtures for carcinogenicity Classification criteria for mixtures 3.6.3.1. Classification of mixtures will be based on the available test data for the individual of the mixture, using cut - off values/concentration limits for those ingredients and ingredients se - by - case basis be taking into account potency consideration. The classification may on a ca 3.6.3.1.2 of this based on the available test data for the mixture as a whole (see Section 3.6.3.1.3 of this Guidance). Guidance) or based on bridging principles (see Section When data are available for all ingredients or only fo 3.6.3.1.1. r some ingredients : 3.6.3.1.1. Annex I The mixture will be classified as a carcinogen when at least one ingredient has been classified as a Category 1A, Category 1B or Category 2 carcinogen and is present at or above the appropriate generic concentration l imit as shown in Table 3.6.2 below for Category 1A, Category 1B and Category 2 respectively. Table 3.6.2 Generic concentration limits of ingredients of a mixture classified as carcinogen that trigger classification of the mixture Generic concentration limits triggering classification of a mixture as: Ingredient classified as: Category 1 carcinogen Category 2 carcinogen Category 1A Category 1B Category 1A carcinogen 0,1 % — —  —  0,1 % — Category 1B carcinogen 1,0 % [Note 1] —  — Category 2 carcinogen Note The concentration limits in the table above apply to solids and liquids (w/w units) as well as gases (v/v units). Note 1 If a Category 2 carcinogen is present in the mixture as an ingredient at a concentration ailable for the mixture upon request. ≥ 0,1% a SDS shall be av In case a SCL has been established for one or more ingredients these SCLs have precedence 3.6.2.6 of this Guidance for the setting of SCLs for over the respective GCLs. See Section substances. 3.6.3.1.2. When data are available for the complete mixture Annex I: 3.6.3.2.1. Classification of mixtures will be based on the available test da ta for the individual ingredients of the mixture using concentration limits for the ingredients classified as carcinogens. On a case - by - case basis, test data on mixtures may be used for classification when demonstrating effects that have not been establish ed from the evaluation based on the individual ingredients. In such cases, the test results for the mixture as a whole must be shown to be conclusive taking into account dose and other factors such as duration, lysis of carcinogenicity test systems. Adequate observations, sensitivity and statistical ana

391 Guidance on the Application of the CLP Criteria 5.0 – July 2017 391 Version documentation supporting the classification shall be retained and made available for review upon request. When data are not available for the complete mixture: bridging principles 3.6.3.1.3. Where the mixture itself has not been tested to determine its Annex I: 3.6.3.3.1. carcinogenic hazard, but there are sufficient data on the individual ingredients and similar tested mixtures (subject to the provisions of paragraph 3.6.3.2.1) to adequately characterise f the mixture, these data shall be used in accordance with the applicable the hazards o bridging rules set out in section 1.1.3. Bridging principles will only be use d on a case by case basis (see S ection 3.6.3.1 of this guidance). Note that the following bridging principles are not applicable to this hazard class:  concentration of highly hazardous mixtures  interpolation within one hazard category (see CLP Annex 1, 1.1.3.3 and 1.1.3.4) 3.6.3.2. Decision logic for classification of mixtures The decision logic which is based on the GHS Guidance is revised to meet CLP requirements. It is strongly recommended that the person responsible for classification, study the criteria for classification before and during use of the decision logic. Classification based on individual ingredients of the mixture Category 1 Does the mixture contain one or more ingredients Yes classified as a Category 1 carcinogen at  0.1 %, or a SCL set for the ingredient(s)? above Danger No Category 2 Does the mixture contain one or more ingredients Yes 1.0 %, or classified as a Category 2 carcinogen at  above a SCL set for the ingredient(s)? Warning No Not classified

392 ication of the CLP Criteria Guidance on the Appl 392 Version 5.0 – July 2017 by - case basis - Modified classification on a case Test data on mixtures may be used for classification when demonstrating effects that have not been established from the evaluation based on the individual ingredients (CLP Annex I, )). 3.6.3.1.1, see also CLP Article 6(3 Are the test results on the Classify in mixture conclusive taking appropriate Are test data available into account dose and category for the mixture other factors such as demonstrating a Yes Yes ions duration, observat carcinogenic effect not and analysis (e.g. identified from the data statistical analysis, test on individual sensitivity) of substances? carcinogenicity test Danger systems? or No Warning No Yes Can bridging principles be applied? No See above: Classification based on of the mixture. individual ingredients

393 Guidance on the Application of the CLP Criteria 5.0 – July 2017 393 Version 3.6.4. Hazard communication in form of labelling for carcinogenicity 3.6.4.1. Pictograms, signal words, hazard statements and precautionary statements 3.6.4.1 Table 3.6.3, for substances Annex I: Label elements shall be used in accordance with or mixtures meeting the criteria for classification in this hazard class. Table 3.6.3 Label elements for carcinogenicity Category 1 Classification Category 2 (Category 1A, 1B) GHS Pictograms Signal Word Danger Warning H350: May cause cancer H351: Suspected of causing Hazard Statement (state route of exposure if cancer (state route of exposure it is conclusively proven if it is conclusively proven that exposure that no other routes of no other routes of exposure cause the hazard) cause the hazard) Precautionary Statement P201 P201 P202 Prevention P202 P281 P281 P201 Statement P201 Precautionary P202 P202 Prevention P280 P280 Statement Precautionary P308 + P313 P308 + P313 Response Precautionary P405 P405 Statement Storage Statement P501 P501 Precautionary Disposal The wording of the Precautionary Statements is found in CLP Annex IV, Part 2. Where there is conclusive proof that cancer is caused only by certain route(s), then this route may be stated in the hazard sta tement. In case of Category 1 carcinogens where there is conclusive proof that cancer is caused only by inhalation, the hazard phrase ‘H350i: May cause cancer by inhalation’ applies (CLP Annex VII, Table 1.1).

394 ication of the CLP Criteria Guidance on the Appl 394 Version 5.0 – July 2017 3.6.4.2. Additional labelling provisions There are no a dditional labelling provisions for carcinogenic substances and mixtures in CLP, however there are provisions laid out in Annex XVII to REACH. The packaging of substances with harmonised classification as carcinogenic Category 1A or Category 1B, or mixtures containing such substances at concentrations warranting classification of the mixture as carcinogenic Category 1A or Category 1B, ‘must be marked visibly legibly and indelibly as , follows: “Restricted to professional users”.’ (REACH, Annex XVII, point 28. Derogations from this obligation are outlined in the same provision). Some additional considerations for re 3.6.4.3. classification - There are only few situations where the direct translation may lead to different results, however, these are likely to be very rare. The first difference in applying the CLP criteria is that sufficient evidence (Carc. 1B) for from two or more independent studies in one carcinogenicity in animals can also be derived species carried out at different times or in different laboratories or under different protocols. The second difference applying the CLP criteria is that sufficient evidence (Carc. 1B) for carcinogenicity in animals c an be derived from an increased incidence of tumours in both sexes - conducted study, ideally conducted under GLP. The criteria of a single species in a well according to DSD allowed classification in Carc. Cat. 2 (analogous to CLP Carc. 1B) where there positive results in two animal species or clear positive evidence in one species, together were with supporting evidence such as genotoxicity data, metabolic or biochemical studies, induction of benign tumours, structural relationship with other known carcinog ens, or data from epidemiological studies suggesting an association. Another difference can be derived from the IARC classification as ‘ possibly carcinogenic to sufficient humans (IARC 2B)’. This category is used for substances for which there is less than in experimental animals. According to IARC, classification as evidence of carcinogenicity possibly carcinogenic to humans’ may be derived from solely strong evidence from mechanistic ‘ and other relevant data. This means that no in vivo carcinogenicity nor (Q)SAR data need to be available to arrive at classification for limited evidence of carcinogenicity. Examples of classification for carcinogenicity 3.6.5. Classification for carcinogenicity involves the consideration of many different factors, as outlined above , and is a complex task which needs expert judgement. Therefore no examples of classification for carcinogenicity are included in this guidance document. 3.6.6. References Ashby J, Waters M.D., Preston M.D., J., Adler d I. - D., Douglas G.R., Fielder R., Shelby - Anderson T., Gopalan H.N.B., Becking G. and Sonich ofuni Mullin C.: IPCS harmonization D., S of methods for the prediction and quantification of human carcinogenic/mutagenic hazard, and for indicating the probable mechanism of action of carcinogens. Mutation Research/Fundamental 352 (1 - 2), 153 - 157, 1996. and Molecular Mechanisms of Mutagenesis R.J.: Mouse - specific carcinogens: an assessment of hazard and J.M. and Fielder Battershill - term carcinogenicity bioassays in transgenic mice. Hum. Exp. significance for validation of short 17 (4), 193 - 205, 1998. Toxicol. A.R., Cohen Boobis S.M., Dellarco V. , McGregor D., Meek M.E., Vickers C., Willcocks D., Farland W.: IPCS framework for analyzing the relevance of a cancer mode of action for humans. Crit. Rev. Toxicol. 36 (10), 781 - 92, 2006. Cohen S.M.; Meek M.E., Klaunig J.E.; Patton D.E., Fenner - Crisp P.A.: The Human Relevance of Information on Carcinogenic Modes of Action: Overview . Critical Reviews in Toxicology 33 (6), 581 - 589, 2003.

395 Guidance on the Application of the CLP Criteria 5.0 – July 2017 395 Version year S.M.: Human carcinogenic risk evaluation: an alternative approach to the two Cohen - 80 - 229, 2004. Toxicol. Sci. rodent bioassay. , 225 J.C., Klinefelter G.R., Hardisty J.F., Sharpe R.M., Foster P Cook .M.: Rodent Leydig cell tumorigenesis: a review of the physiology, pathology, mechanisms, and relevance to humans. 29 (2), 169 - 261, 1999. Crit. Rev. Toxicol . T., Roelfzema H., Kroese D, Tennant R.W.: T25: A simplified carcinogenic Dybing E., Sanner potency index. Description of the system and study correlation between carcinogenic potency and species/site specificity and mutagenicity. 80 , 272 - 279, 1997. Pharmacol. Toxicol. f directive EC: Guidelines for setting specific concentration limits for carcinogens in Annex I o 67/548/EEC. Inclusion of potency considerations. Commission working group on the classification and labelling of dangerous substances. Office for the Official Publications of the 5, 1999. 828 - European Communities - , Luxembourg, ISBN 92 - 7443 mmission Group of Specialised Experts in the fields of carcinogenicity, mutagenicity and EU Co reprotoxicity: Non genotoxic thyroid carcinogens in the rodent bioassay, ECBI/49/99 Add. 1 Rev. 2 excerpt of agenda point 3.1, 1999. d Experts in the fields of carcinogenicity, mutagenicity and EU Commission Group of Specialise reprotoxicity: Leydig tumours 2004, ECBI/08/04 Rev. 2, 2004 D., and Smythe R.T: A comparison of tests for trend with historical controls Fung K.Y., Krewski Canadian J. of in carcinogen bioassay. Statist. 24, 431 - 454, 1996. Greim H., Gelbke H. - P., Reuter U., Thielmann H.W. and Edler L.: Evaluation of historical control data in carcinogenicity studies. Human Exp. Toxicol. 22: 541 - 549, 2003. J.K, Hailey J.R., Morris R.W.: Spontaneous neoplasm incidences in Fischer 344 rats Haseman 1 mice in two - year carcinogenicity studies: A national toxicology program update. and B6C3F - 26 - 441, 1998. Toxicol Pathol. (3) 428 Peroxisome. Proliferation and its role in carcinogenesis views and expert opinions IARC (1994) - of an IARC Working Group, Lyon, 7 11 December 1994, Technical Publication No. 24, Lyon, France. IARC (1999) Scientific Publications No. 147 Species Differences in Thyroid, Kidney and Urinary Bladder Carcinogenesis. IARC (2003). Predictive value of rodent f orestomach and gastric neuroendocrine tumours in evaluating carcinogenic risks to humans. Views and expert opinions of an IARC Working Group, – Lyon, 29 November - 1 December 1999. IARC Technical Publication No. 39, Lyon, France, p.1 73. IARC (2006) Monograp hs on the Evaluation of Carcinogenic Risks to Humans, Preamble Section B, Lyon, France. http://monographs.iarc.fr/ENG/Preamble/CurrentPreamble.pdf IPCS (1995). Harmonization of the Assessment of Risk for Carcinogenicity and Mutagenicity (Germ Cells), Carshalton, UK 1995. IPCS/95.x. A.R., Cohen S.M., Dellarco V., McGregor IPCS (2007) Boobis M.E., Vickers C., D., Meek Willcocks D., Farland W.: IPCS Framework for Analyzing the Relevance of a Cancer Mode of Action for Humans in IPCS Harmonization Project Document No. 4, Part 1, IPCS fram ework for analysing the relevance of a cancer mode of action for humans and case studies. - http://www.who.int/ipcs/methods/harmonization/areas/cancer_mode.pdf Mati W., Lam G., Dahl C., Thorup Andersen J., Balslev E.: Leydig cell tumour – a rare testicular tumour. Int. Urol. Nephrol. 33 (1), 103 - 106, 2002.

396 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 396 McGregor D., Binaglia M., Boobis A., Botham P., Hennes E.C., Hoffstadt L., Hubbard S., Petry T., Riley D.: 2008 Guidance for the Classification of Carcinogens under the Globally A., Schwartz Harmonised System of Classification and Labelling of Chemicals (GHS). Submitted for publication, 2009. M.E., Bucher J.R., Cohen S.M., Dellarco Meek V., Hill R.N., Lehman - McKeeman L.D., Longfellow D.G., Pastoor T., Seed J., Patton D.E.: A Framework for Human Relevance Analysis of Information on Carcinogenic Modes of Action . 33 (6), 591 - 653, Critical Reviews in Toxicology 2003. rn oil, Sprague - Dawley, Females, 2005. NTP (2005) Historical Control Report, Gavage, Co http://ntp.niehs.nih.gov/files/HIST_2005SD_TUMOR_RATS1.pdf NTP (2007a) Historical Controls Report All Routes And Vehicles. RATS, October 2007. http://ntp.niehs.nih.gov/files/HistCont - 2007 - 10 - 17 - Rats_All_Routes.pdf NTP (2007b) Historical Controls Report All Routes And Vehicles. MICE, October 2007. http://ntp.niehs.nih.gov/files/HistCont - 2007 - 10 - 17 - Mice_All_Routes.pdf OECD (2004). OECD Principles for the Validation, for Regulatory Purposes, of (Quantitative) Relationship Models. - Activity Structure http://www.oecd.org/document/23/0,2340,en_2649_34379_33957015_1_1_1_1,00.html OECD (2005). 3.2 Guidance on the Development and Use of Chemic al Categories in HPV Chemicals Programme http://www.oecd.org/document/7/0,2340,en_2649_34379_1947463_1_1_1_1,00.html OECD (2007). Report on the Regulatory Uses and Applications in OECD Member Countries of (Q)SAR Models in the Assessment of New and Existing Chemicals. ENV/JM/MONO(2006)25 http://www.oecd.org/dataoecd/55/22/38131728.pdf J.K., Hailey, J.R., Maronpot, R.R., Nyska, A.: Association of adrenal Ozaki, K., Haseman, pheochromocytoma and lung pathology in inhalation studies with particulate compounds in the male F344 rat . – the National Toxicology Program Experience, 2002 (2001) report 601516007 : Factsheets for the (eco)toxicological risk assessment strategy RIVM of the National Institute of Public Health and the Environment (RIVM). Editors: R. Luttik and M.T.M. van Raaij. (2003) report 601516012: Factsheets for the (eco)toxicological risk assess ment strategy RIVM - of the National Institute for Public Health and the Environment Part IV. Editors: C.E. Smit and M.T.M. van Raaij. RIVM (2004) report 601516012: Factsheets for the (eco)toxicological risk assessment strategy lic Health and the Environment - Part IV. Editors: C.E. Smit and of the National Institute for Pub M.T.M. van Raaij. Factsheets for the (eco)toxicological risk assessment strategy (2005) report 601516013: RIVM Public Health and the Environment Part V. J.W.A. S cheepmaker, of the National Institute for C.E. Smit, M.T.M. van Raaij. Sanner T. and E. Dybing: Comparison of Carcinogen Hazard Characterisation Based on Animal Basic & Clinical Pharmacology & Toxicology Studies and Epidemiology , 66 – 70, 2005. 96 Sonich - Mullin C., Fielder R., Wiltse J., Baetcke K., Dempsey J., Fenner - Crisp P., Grant D., Hartley M., Knaap A., Kroese D., Mangelsdorf I., Meek E., Rice J.M., Younes M.: IPCS Conceptual Framework for Evaluating a Mode of Action for Chemical Carcinogenesis ., Regulatory Toxicology and Pharmacology 34 (2), 146 - 152, 2001.

397 Guidance on the Application of the CLP Criteria 5.0 July 2017 Version – 397 http://www.who.int/ipcs/methods/harmonization/areas/cancer/en/ Tennant RW. Stratification of rodent carcinogenicity bioassay results to reflect relative human hazard. Mutation Research 286, 111 - 118, 1993. WHO working group: Evaluation and use of epidemiological evidence for environmental health - Env. Health Perspect. 108 , 997 risk assessment: Who Guideline Document. 1002 , 2000.

398 ication of the CLP Criteria Guidance on the Appl 398 Version 5.0 – July 2017 3.7. Y REPRODUCTIVE TOXICIT Definitions and general considerations for reproductive toxicity 3.7.1. : 3.7.1.1. Reproductive toxicity includes adverse effects on sexual function and Annex I toxicity in the offspring. The fertility in adult males and females, as well as developmental definitions presented below are adapted from those agreed as working definitions in IPCS/EHC Document N°225, Principles for Evaluating Health Risks to Reproduction Associated with poses, the known induction of genetically based Exposure to Chemicals. For classification pur heritable effects in the offspring is addressed in Germ Cell Mutagenicity (section 3.5), since in the present classification system it is considered more appropriate to address such effects under the separate hazard class of germ cell mutagenicity. In this classification system, reproductive toxicity is subdivided under two main headings: Adverse effects on sexual function and fertility; (a) (b) Adverse effects on development of the offspring. Some reproductive toxic effects cannot be clearly assigned to either impairment of sexual function and fertility or to developmental toxicity. Nonetheless, substances with these effects, or mixtures containing them, shall be classified as reproductive toxicants . 3.7.1.2. Annex I: For the purpose of classification the hazard class Reproductive Toxicity is differentiated into: adverse effects – on sexual function and fertility, or – on development; – effects on or via lactation – : 3.7.1.3. Adverse effects on sexua l function and fertility Annex I Any effect of substances that has the potential to interfere with sexual function and fertility. This includes, but is not limited to, alterations to the female and male reproductive system, adverse effects on onset of puberty, gam ete production and transport, reproductive cycle normality, sexual behaviour, fertility, parturition, pregnancy outcomes, premature reproductive senescence, or modifications in other functions that are dependent on the integrity of the reproductive systems . : 3.7.1.4. Annex I Adverse effects on development of the offspring Developmental toxicity includes, in its widest sense, any effect which interferes with normal development of the conceptus, either before or after birth, and resulting from exposure of eit her parent prior to conception, or exposure of the developing offspring during prenatal development, or postnatally, to the time of sexual maturation. However, it is considered that classification under the heading of developmental toxicity is primarily in tended to provide a hazard warning for pregnant women, and for men and women of reproductive capacity. Therefore, for pragmatic purposes of classification, developmental toxicity essentially means adverse effects induced during pregnancy, or as a result of parental exposure. These effects can be manifested at any point in the life span of the organism. The major manifestations of developmental toxicity include (1) death of the developing organism, (2) structural abnormality, (3) altered growth, and (4) func tional deficiency.

399 Guidance on the Application of the CLP Criteria 5.0 – July 2017 399 Version 3.7.1.1. Special considerations on effects on or via lactation This classification is intended to indicate when a substance may cause harm due to its effects on or via lactation. This can be due to the substance being absorbed by women and adversely affecting milk production or quality, or due to the substance (or its metabolites) being present in breast milk in amounts sufficient to cause concern for the health of a breastfed child. Annex I: included under reproductive toxicity, 3.7.1.5. Adverse effects on or via lactation are but for classification purposes such effects are treated separately. This is because it is desirable to be able to classify substances specifically for an adverse effect on lactation so bout this effect can be provided for lactating mothers. that a specific hazard warning a Therefore, if the adverse effects that lead to impaired development in the offspring also occur after exposure then the substance would also be classified for developmental toxicity. in utero In othe is independent of consideration of r words, the classification for effects on or via lactation the reproductive toxicity of the substance, and a substance can be classified for effects on or via lactation whether or not the substance is also classified for reproductive toxicity. Classification for effects on or via lactation alone is not sufficient for a substance to be subject to harmonised classification and labelling in accordance with CLP Article 36 (1). Classification of substances for reproductive t oxicity 3.7.2. 3.7.2.1. Identification of hazard information 3.7.2.1.1. Identification of human data Epidemiological studies as well as clinical data and case reports may be available as stated in CLP Annex I, 3.7.2.2.3 and further in the Guidance on IR&CSA , Section R.7.6.3.2. 3.7.2.1.2. Identification of non human data - animal data and non - testing information used for classification is outlined in CLP Annex In vitro I, 3.7.2.5. and further specific references to different testing methods are listed in the Guidance IR&CSA , Section R.7.6 on .3.1. 3.7.2.2. Classification criteria Annex I : 3.7.2.1.1 . For the purpose of classification for reproductive toxicity, substances are allocated to one of two categories. Within each category, effects on sexual function and fertility, and on development, are considered separately. In addition, effects on lactation are allocated to a separate hazard category. Table 3.7.1 (a) Hazard categories for reproductive toxicants Criteria Categories CATEGORY 1 Known or presumed human reproductive toxicant Substances are classified in Category 1 for reproductive toxicity when they are known to have produced an adverse effect on sexual function and fertility, or on development in humans or when there is evidence from animal studies, poss ibly supplemented with other information, to provide a strong presumption that the substance has the capacity to interfere with reproduction in humans. The classification of a substance is further

400 ication of the CLP Criteria Guidance on the Appl 400 Version 5.0 – July 2017 sification is distinguished on the basis of whether the evidence for clas primarily from human data (Category 1A) or from animal data (Category 1B). Category 1A Known human reproductive toxicant The classification of a substance in this Category 1A is largely based on evidence from humans. Category 1B Presumed human reproductive toxican t The classification of a substance in this Category 1B is largely based on data from animal studies. Such data shall provide clear evidence of an adverse effect on sexual function and fertility or on development in the absence of f occurring together with other toxic effects the other toxic effects, or i - adverse effect on reproduction is considered not to be a secondary non specific consequence of other toxic effects. However, when there is mechanistic information that raises doubt about the relevance of th e effect for humans, classification in Category 2 may be more appropriate. CATEGORY 2 Suspected human reproductive toxicant Substances are classified in Category 2 for reproductive toxicity when there is some evidence from humans or experimental animals, possibly supplemented with other information, of an adverse effect on sexual function and fertility, or on development, and where the evidence is not sufficiently convincing to place the substance in Category 1. If deficiencies in the study make the qualit y of evidence less convincing, Category 2 could be the more appropriate classification. Such effects shall have been observed in the absence of other toxic effects, or if occurring together with other toxic effects the adverse effect on reproduction is con - specific consequence of sidered not to be a secondary non the other toxic effects. 3.7.2.2.1. Classification in the presence of parental toxicity 3.7.2.2.1.1. Effects to be considered in the presence of marked systemic effects In general all findings on reproductive toxicity should be considered for classification purposes irrespective of the level of parental toxicity. A comparison between the severity of the effects on fertility/development and the severity of other toxicological findings must be performed. Fertility effects Adverse effects on fertility and reproductive performance seen only at dose levels causing marked systemic toxicity (e.g. lethality, dramatic reduction in absolute body weight, coma) are not relevant for classification purposes. There is no established re lationship between fertility effects and less marked systemic toxicity. Therefore it should be assumed that effects on fertility seen at dose levels causing less marked systemic toxicity are not a secondary consequence of this toxicity. However, mating beh aviour can be influenced by parental effects not directly related to reproduction (e.g. sedation, paralysis), and such effects on mating behaviour may not warrant classification. Developmental effects:

401 Guidance on the Application of the CLP Criteria 5.0 – July 2017 401 Version . Maternal toxicity Annex I: 3.7.2.4 3.7.2.4.1. Development of the offspring throughout gestation and during the early Annex I: specific - postnatal stages can be influenced by toxic effects in the mother either through non mechanisms related to stress and the disruption of maternal homeostasis, or by specific maternally - mediated mechanisms. In the interpretation of the developmental outcome to decide classification for developmental effects it is important to consider the possible influence tainties surrounding the of maternal toxicity. This is a complex issue because of uncer relationship between maternal toxicity and developmental outcome. Expert judgement and a weight of evidence approach, using all available studies, shall be used to determine the degree of influence that shall be attributed to mater nal toxicity when interpreting the criteria for classification for developmental effects. The adverse effects in the embryo/foetus shall be first considered, and then maternal toxicity, along with any other factors which are likely to have influenced these effects, as weight of evidence, to help reach a conclusion about classification. : 3.7.2.4.2. Annex I Based on pragmatic observation, maternal toxicity may, depending on severity, influence development via non specific secondary mechanisms, producing effec ts - such as depressed foetal weight, retarded ossification, and possibly resorptions and certain malformations in some strains of certain species. However, the limited number of studies which have investigated the relationship between developmental effects and general maternal toxicity have failed to demonstrate a consistent, reproducible relationship across species. Developmental effects which occur even in the presence of maternal toxicity are considered to be evidence of developmental toxicity, unless it can be unequivocally demonstrated on a case - by - case basis that the developmental effects are secondary to maternal toxicity. Moreover, classification shall be considered where there is a significant toxic effect in the offspring, e.g. irreversible effects such as structural malformations, embryo/foetal lethality, significant post - natal functional deficiencies. Annex I : 3.7.2.4.3. Classification shall not automatically be discounted for substances that produce developmental toxicity only in association with maternal toxicity, even if a specific - maternally mediated mechanism has been demonstrated. In such a case, classification in Category 2 may be considered more appropriate than Category 1. However, when a substance is so toxic that maternal death or severe inanition results, or the dams are prostrate and incapable of nursing the pups, it is reasonable to assume that developmental toxicity is produced solely as a secondary consequence of maternal toxicity and discount the developmental effects. Classification is not necessarily the outcome in the case of minor developmental changes, when there is only a small reduction in foetal/pup body weight or retardation of ossification when seen in association with maternal toxicity. survival and growth seen only at dose levels causing maternal Adverse effects on postnatal toxicity may be due to lack of maternal care or other causes such as adverse effects on or via lactation or developmental toxicity. In case post - natal effects are caused by lack of maternal care classification for developmental effects may not be warranted. 3.7.2.2.1.2. Relevance of specific effects in the parent All types of reproductive toxic effects may be considered as secondary to parental toxicity. With pecific effects indicating toxicity in parental current knowledge it is not possible to identify s animals which do not have any relevance to reproductive toxicity (e.g. peroxisome proliferation). However parental toxicity that is less than marked should not influence the classification for reproductive to xicity independent of the specific parental effects observed. In general it is very difficult to prove a causal relationship between a parentally mediated mechanism and adverse effects in the offspring. Usually data are insufficient to conclude if an effec t on the offspring is a direct effect or secondary to parental toxicity. In order to determine whether a reproductive toxic effect is independent or secondary to a parental effect, it would be

402 ication of the CLP Criteria Guidance on the Appl 402 Version 5.0 – July 2017 most appropriate to correlate individual data for offspring and their parents. Nevertheless, associations between parental and offspring effects do not by default prove a causal relationship. In cases where a causal relationship is established between reproductive and parental toxicity and the effects on the offspring can be proved to be secondary to maternal toxicity, they may still be relevant for developmental classification, dependent on the severity of the effects. A comparison between the severity of the maternal toxicity and the severity of the findings in ffspring must be performed. There are several examples showing that the developing the o - organism can be more susceptible and the long term consequences can be more severe than in ted. the adult. The mother might recover while the offspring could be permanently affec 3.7.2.4.4 Annex I: . Some of the end points used to assess maternal effects are provided below. Data on these end points, if available, need to be evaluated in light of their statistical or biological significance and dose response relationship. al mortality: Matern an increased incidence of mortality among the treated dams over the controls shall be considered evidence of maternal toxicity if the increase occurs in a dose - related manner and can be attributed to the systemic toxicity of the test material . Maternal mortality greater than % is considered excessive and the data for that dose level shall not normally be considered 10 for further evaluation. Mating index 1 ) (no. animals with seminal plugs or sperm/no. mated x 100)( Fertility index: als with implants/no. of matings x 100) (no. anim Gestation length (if allowed to deliver) Body weight and body weight change: Consideration of the maternal body weight change and/or adjusted (corrected) maternal body weight shall be included in the evaluation of m aternal toxicity whenever such data are available. The calculation of an adjusted (corrected) mean maternal body weight change, which is the difference between the initial and terminal body weight minus the gravid uterine of the weights of the foetuses), may indicate whether the weight (or alternatively, the sum effect is maternal or intrauterine. In rabbits, the body weight gain may not be useful indicators of maternal toxicity because of normal fluctuations in body weight during pregnancy. Food and water consumption (if relevant): The observation of a significant decrease in the average food or water consumption in treated dams compared to the control group is useful in evaluating maternal toxicity, particularly when the test material is administered in t he diet or drinking water. Changes in food or water consumption need to be evaluated in conjunction with maternal body weights when determining if the effects noted are reflective of maternal toxicity or more simply, unpalatability of the test material in feed or water. Clinical evaluations (including clinical signs, markers, haematology and clinical chemistry studies):

403 Guidance on the Application of the CLP Criteria 5.0 – July 2017 403 Version The observation of increased incidence of significant clinical signs of toxicity in treated dams relative to the control group is useful in evaluating maternal toxicity. If this is to be used as the basis for the assessment of maternal toxicity, the types, incidence, degree and duration of clinical signs shall be reported in the study. Clinical signs of maternal intoxication include: ostration, hyperactivity, loss of righting reflex, ataxia, or laboured breathing. coma, pr Post - mortem data: - mortem findings may be indicative of maternal Increased incidence and/or severity of post ngs or organ weight data, toxicity. This can include gross or microscopic pathological findi - - body weight ratio, or organ - to - brain weight ratio. including absolute organ weight, organ to When supported by findings of adverse histopathological effects in the affected organ(s), the observation of a significant change in the average weight of suspected target organ(s) of treated dams, compared to those in the control group, may be considered evidence of maternal toxicity. 1 ( ) It is recognised that the Mating index and the Fertility index can also be affected by the male. 3.7.2.2.2. Sub stances causing effects on or via lactation Annex I: Table 3.7.1 (b) Hazard category for lactation effects EFFECTS ON OR VIA LACTATION Effects on or via lactation are allocated to a separate single category. It is recognised that for many substances there is no information on the potential to cause adverse effects on the offspring via lactation. However, substances which are absorbed by women and have been shown to interfere with lactation, or which may be present (including metabolites) in breast milk in amounts sufficient to cause concern for the health of a breastfed child, shall be classified and labelled to indicate this property hazardous to breastfed babies. This classification can be assigned on the: (a) human evidence indicating a hazard to babies during the lactation period; and/or (b) results of one or two generation studies in animals which provide clear evidence of adverse effect in the offspring due to transfer in the milk or adverse effect on the quality of the milk; and/or (c) absorption, met abolism, distribution and excretion studies that indicate the likelihood that the substance is present in potentially toxic levels in breast milk. There are the two general criteria for this classification. i. ...are absorbed by women and have been shown to in terfere with lactation. This relates to effects in the mother that impact adversely on the breast milk, either in terms of the quantity produced or the quality of the milk produced (i.e. the composition). Any effect on the quantity or quality of the breas t milk is likely to be due to systemic effects in the mother. However, overt maternal toxicity may not be seen (e.g. the substance may just affect the transfer of a nutrient into the milk with no consequence for the mother). The type and magnitude of the m aternal effects and their potential influence on lactation/milk production

404 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 404 need to be considered on a case by - case basis to determine whether classification for effects on - or via lactation is necessary. in the mother at the same dose level then If a substance causes marked overt systemic toxicity - it is possible that this may indirectly impair milk production or impair maternal care as a non specific secondary effect. The type and magnitude of the maternal effects and their potential milk production needs to be considered on a case - by - case basis using influence on lactation/ expert judgment. If there is robust evidence to indicate that the effects on lactation are not caused directly by the substance then it should not be classified as such. does not cause overt toxicity in the mother but which interferes with milk A substance which production or quality will normally be classified for effects on or via lactation because in this case the effect on lactation is most likely a direct substance - related effect. ii. ... may be present (including metabolites) in breast milk in amounts sufficient to cause concern for the health of a breastfed child. This relates to the ability of the substance (including metabolites), to enter the breast milk in amounts sufficient to cause a concern. When the effect on the offspring is caused by the substance (or metabolite) after transport through the milk then the maternal toxicity has no relevance for classification. In general, positive data should usually be available to show that a sub stance leads to an adverse effect in offspring due to effects on lactation to support classification. However, in exceptional circumstances, if there are substantiated grounds for concern that the substance may have an adverse effect via lactation then it may be classified as such in the absence of direct evidence. This should be based on a quantitative comparison of the estimated transfer via the milk and the threshold for toxicity in the pups. This might apply in cases where the substance has the capacity to bioaccumulate which would lead to a potentially higher burden in the offspring, or where there is evidence that the offspring may be more sensitive to the substance’s toxicity than adult. The mere presence of the substance in the milk alone, without a strong justification for a concern to offspring, would normally not support classification for effects on or via lactation. Evaluation of hazard information 3.7.2.3. Appropriate classification will always depend on an integrated assessment of all available data and their interrelationship using a weight of evidence approach. Individual datasets should be analysed case by case using expert judgment. 3.7.2.3.1. Use of data from standard repeat dose tests Fertility effects: Toxicological effects, including marked effects, obs erved in a standard repeat dose study could be considered valid for the pre - mating phase for adult females and the pre - and post - mating phase for adult males. However in case of contradictions between the standard repeat dose studies and reproductive studi es, the result from the latter should be considered more relevant. For pregnant and lactating females and juveniles data from standard repeat dose studies cannot easily be extrapolated. Developmental effects: A detailed assessment of toxicity in pregnant animals cannot be extrapolated from studies with non - pregnant animals. However information from general toxicity studies might give an indication of the maternal toxicity that could be anticipated in a subsequent developmental toxicity study. 3.7.2.3.2. Study desig n Assessment of the dose - response relationships of parental and reproductive toxicity end points and their possible interrelationship require study designs where the dose intervals are not too

405 Guidance on the Application of the CLP Criteria 5.0 – July 2017 405 Version - also reduce the chance of far apart. This will improve dose response assessment and will masking malformations by severe toxicity (e.g. resorptions, lethality) at high dose levels. This may lead to experimental designs in which more than the standard three dose groups and a se toxicity studies may be considered useful for control are tested. Endpoints from repeat do inclusion in subsequent reproductive toxicity studies. These endpoints should be evaluated both in parental animals and in offspring. 3.7.2.3.3. Evaluation of evidence relating to effects on or via lactation Human evid l. ence indicating a hazard to babies during the lactation period; This criterion acknowledges that human data, e.g. from epidemiological studies or case reports, indicating a hazard to babies during the lactation period can also be used to support classifica - explanatory and any tion for effects on or via lactation. The use of human data is self study should be assessed on its merits for which expert judgment may be required. Observations in humans that give evidence of adverse effects in breastfed babies of mo thers exposed to the chemical in question should be taken to provide clear evidence supporting classification. Such studies which do not show an adverse effect need to be considered carefully. s of exposure, and a negative Human studies investigate the risk under the specific condition finding may just reflect inadequate methods to detect effects or insufficient exposures rather than prove the absence of a hazard. In practice, useful human data are likely to be rare due to the nature of the endpoint. More l ikely are survey type studies which measure the levels of the chemical in breast milk. Such studies may provide useful information on the potential for maternal exposure to lead to the presence of the chemical in the breast milk and so they may be of use i n assessing the need for classification for effects on or via lactation. m. Results of one or two generation studies in animals which provide clear evidence of adverse effect in the offspring due to transfer in the milk or adverse effect on the quality of th e milk; Ideally, studies will be available which inform directly on whether the substance causes adverse effects in the offspring due to an adverse effect on lactation. One generation or multi generation - reproductive toxicity studies, which involve direct exposure or exposure via the milk of the offspring postnatally, usually provide information on this. The most common study performed - generation study, but one - generation studies with new study designs, like the today is the two 422 or the developmental neurotoxicity study OECD TG 426, also screening study OECD TG 421/ exist. The value of these studies is that they directly observe the pups during lactation and any adverse effects, such as deaths, decreased viability, clinical signs such as reduced bodyweight gain etc, can be directly observed and quantified. However, expert judgement is required to decide whether these effects in pups are due to a direct adverse effect on lactation, or are due to impaired nursing behaviour which is a non specific secondary co nsequence of maternal toxicity. If the impaired nursing behaviour is proven to be a substance related specific effect on behaviour, then classification for effects on or via lactation may be appropriate. It should also be noted that some developmental effe cts resulting from exposure in utero would only manifest - natally and those should not be used for classification for effects on or via lactation. Cross - post fostering studies, where available, may help establish whether effects are due to in utero or lacta tional exposure. If there is sufficient data that animal results are not relevant to humans, they should not be taken into account. n. Absorption, metabolism, distribution and excretion studies that indicate the likelihood that the substance is present in pot entially toxic levels in breast milk; The criterion indicates that toxicokinetic studies showing that the substance can be present at potentially toxic levels in breast milk can support classification. The implicit assumption behind this clause is that the pups may receive a body burden of the toxic entity through suckling that is sufficient to cause toxicity when the level of the toxic entity in the milk is above a certain threshold level (‘a level to cause concern’). There is no robust way to estimate wha t this

406 ication of the CLP Criteria Guidance on the Appl 4 Version 5.0 – July 2017 06 threshold is, although the likely body burden expected in the breastfed child may be compared to the toxicity data in adults (e.g. an appropriate NOAEL or BMD) to indicate whether toxicity is likely. The mere presence of a substance in the milk, wi thout a robust argument that these levels may be potentially toxic to offspring would not normally support classification. The toxicokinetics of a substance and the likelihood that it will enter the breast milk may be - chemical properties of the chemical (e.g. using pKa, logP, predicted on the basis of the physico water solubility, and molecular weight etc) and this information could be used as part of the argumentation outlined above. The potential of a substance to bioaccumulate following repeated exposure may also be an important factor to consider as this may contribute to the body burden reaching a potentially toxic level in the offspring. Studies where the offspring/neonates have extended exposure, such as multi generation studies, implicitly allow for bioaccumulation and - so findings from these studies can, in themselves, be taken to provide information on the potential effects of bioaccumulation. Where these types of studies are not available, potential part of the toxicokinetic assessment using bioaccumulation can be taken into consideration as expert judgement. There may be toxicokinetic and toxicodynamic reasons why neonates may potentially be more or less vulnerable to a particular adverse effect than adults due to the fact that certain systems the immune and metabolic systems) and tissues/organs are immature and are still (e.g. developing. Whether the neonate is more or less vulnerable than adults will depend on the specific chemical and will be determined by factors such as the hazardous properties of the chemical, its’ physico - chemical properties and how it is metabolised. Therefore, the relative sensitivity of neonates and adults to a substance must be judged on a case by case basis using information to inform on this, it expert judgement. In the absence of any reliable and robust should be assumed that neonates and adults are equivalent in terms of sensitivity to the substance. Overall, classification for effects on or via lactation can be assigned on the basis of toxicokinetic stantiated estimate of the exposure through the milk alone provided that it is data or a well sub supported by an argument clearly justifying that the level present in the breast milk would be likely to harm developing offspring. 3.7.2.4. Decision on classification Annex I, Section 3.7.2.1.1, reproductive toxic substances are allocated to According to CLP either Category 1A, 1B or 2. Effects on lactation are allocated to a separate hazard category and should be ascribed to a substance irrespective if it classified in any other categ ory for reproductive toxicity or not. Classification of substances containing CMR constituents 3.7.2.5. From a compositional and a toxicological point of view the situation for substances containing CMR constituents, additives or impurities is the same as for mixtu res containing components classified for these endpoints. For this reason the classification procedure for CMR endpoints that is foreseen by CLP for mixtures containing CMR components, is considered applicable also to substances containing CMR constituents , additives or impurities (see S ec tion 1.1.6.1 ). As discussed in S ection 3.7.3 below, mixtures containing components classified as germ cell mutagens shall be normally classified using only the relevant available information for the Further, in cases where the available test data on the individual substances in the mixture. mixture itself demonstrate CMR effects which have not been identified from the information on the individual substances, those data shall also be taken into account. For CMR endpoints the lowest incid ence possible to detect in the tests is by far unacceptable in humans. Thus a dose as high as possible (such as maximal tolerated dose, MTD dose) is needed to be able to detect CMR hazards. Dilution, as would be the case if mixtures or substances containin g CMR constituents were tested, would increase the risk that CMR hazards would not be detected.

407 Guidance on the Application of the CLP Criteria 5.0 – July 2017 407 Version According to article 10 (1) substances in other substances and substances in mixtures are treated in the same way regarding the use of GCLs and SCLs. Setting o f specific concentration limits 3.7.2.6. Specific concentration limits and generic concentration limits are limits Article 10(1) assigned to a substance indicating a threshold at or above which the presence of that substance in another substance or in a mixture as an identified impurity, additive or individual constituent leads to the classification of the substance or mixture as hazardous. Specific concentration limits shall be set by the manufacturer, importer or downstream user where adequate and reliable scient ific information shows that the hazard of a substance is evident when the substance is present at a level below the concentrations set for any hazard class in Part 2 of Annex I or below the generic concentration limits set for any hazard class in 4 and 5 of Annex I. Parts 3, In exceptional circumstances specific concentration limits may be set by the manufacturer, importer or downstream user where he has adequate, reliable and conclusive scientific information that a hazard of a substance classified as haza rdous is not evident at a level above the concentrations set for the relevant hazard class in Part 2 of Annex I or above the generic concentration limits set for the relevant hazard class in Parts 3, 4 and 5 of that Annex. Procedure 3.7.2.6.1. The available data fr om animal and human studies are evaluated to establish the reproductive toxicity dose descriptor, ED (effective dose with a 10% effect level above the background), as 10 described below. A preliminary conclusion as to whether the substance shows high, mediu m or low potency is taken based on the ED data. The preliminary potency evaluation may be 10 modified after due consideration of a number of modifying factors as described in Chapter 3.7.2.6.5 . This results in the final potency group. Each final potency group is connected with a generic concentration limit ( GCL) or a specific concentration limit ( SCL). In this way SCLs are taking into account all relevant considerations. See Figure then set . 6 . A background document 3 containing the justification of the boundaries of the potency groups and the SCLs is available in Annex VI to this document. It is noted that there may be alternative approaches to assess potency, such as basing it on the BMD Methodology (Bench Mark Dose). However such alternative methods are not elaborated in this c urrent guidance, although this does not exclude their use. If alternative approaches are used, they have to be clearly justified from a scientific and regulatory point of view (see Article 10, CLP) and they must be able to provide robust scientific propos als and justifications. Figure 3 Procedure for setting SCL for reproductive toxicity 6 . Determine ED using the available data 10 Determine preliminary potency group Determine final potency group considering the modifying factors Determine SCL

408 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 408 3.7.2.6.2. Cases where potency evaluation is difficult or unfeasible The process for evaluating potency assumes the availability of certain types of data. However, lassification of substances as reproductive these data may not always be available. Also, the c - toxicants may be based on information such as grouping, read across and the use of QSARs IR&CSA , sections R.6 and R.7.2.3.1 ). In such cases, no direct estimate of the (Guidance reproductive toxicity potency based on an ED value is possible. While there are often good 10 reasons for extrapolation of the hazardous properties from one or more substances to another, the expected potency of the individual substances within the group may vary. In these cases a potency ev aluation may be difficult or impossible. However, determination of the classification and the potency using non - testing methods is possible in some cases. These cases could include within a group of substances with comparable struc tures and effects or interpolation of an ED 10 correction for molecular weight in case of extrapolation between different salts with comparable availability. If the classification of a substance in Category 2 is done on the basis of ‘limited data will in such cases determine whether a potency evidence’, the quality of the available assessment is possible. In cases where no further evaluation is possible, the generic concentration limits of CLP apply. In general, more conclusive evidence is required when moving a substance to a lower potency group than to a higher potency group. 3.7.2.6.3. Determination of the ED value 10 The ED value (as used for reprotoxicity SCLs) is the lowest dose which induces reproductive 10 toxic effects which fulfil the criteria for classification for reproductive toxicit y with an incidence or magnitude of 10% after correction for the spontaneous incidence (see in Section 3.7.2.6.3.2 ). Determining exactly which effe ct or combination of effects is the one that fulfils the However, for the majority of substances in the classification criteria may seem difficult. database, the developmental effect(s) observed at the lowest dose level was(/were) an increase ons and/or lethalities of the offspring. The ED in malformati for effects on sexual function and 10 fertility is mainly based on effects on fertility and histopathological changes of the reproductive organs. These effects clearly fulfil the classification requirements. Also, allocation to the final imited number of potency groups and not on the exact ED SCLs is based on a l value. 10 values for several different effects fall into the Therefore, in practice, it is likely that the ED 10 same potency grouping, resulting in the same SCL. The ED may be obtained either directl y or by linear interpolation from experimental data or 10 estimated using Bench Mark Dose (BMD) software. The use of BMD software will result in a more precise estimate of the ED response curve are used. The because all data from the dose - 10 cannot be determined using linear interpolation e is needed when an ED use of BMD softwar 10 due to the absence of a NOAEL when the LOAEL has an effect size above 10%. In general, however, the use of BMD software is not required because of the wide potency groups used for set ting the SCLs. However, it could be important for substances which are close to the boundary of a potency group. When an ED cannot be calculated by direct or linear interpolation from 10 experimental data or by the use of BMD software, interpolation between the control group and the LOAEL should be used to determine the ED . In such cases, only SCLs below the GCL can 10 be determined and not those above the GCL, if no other reliable information is available, the absence of effects at lower dose levels. because it may be difficult in these cases to prove 3.7.2.6.3.1. Determination in practice In practice, often several effects on reproduction are observed in various studies, and the classification is based on the weight of evidence of all results. As a first step, it should be determined whether the classification is for effects on development, for effects on sexual function and fertility or both. The effects used for classification for developmental toxicity should be used to determine the potency for developmental toxicity only. The same applies to effects on sexual function and fertility. This means that for substances fulfilling the criteria for classification for both developmental effects and effects on sexual function and fertility, two

409 Guidance on the Application of the CLP Criteria 5.0 – July 2017 409 Version differ and lead eventually to different SCLs. For both ED values are derived which may 10 developmental effects and effects on sexual function and fertility, the lowest ED for the 10 effect(s) that fulfil the criteria for classification in the different studies, is then used as the ED 10 th at determines the potency of that substance. Where there are doubts as to whether a specific effect fulfils the classification criteria, ED values for different effects could be taken forward to 10 the next step, when modifing factors are considered, to det ermine the impact. by linear interpolation requires a different approach depending on The calculation of the ED 10 whether the effect is measured as an incidence (quantal data, non parametric data), a - magnitude (continuous data, parametric data) or both. Q uantal or non - parametric data 3.7.2.6.3.2. For effects that are measured as changes in incidence, such as an increase in the number of malformations or resorptions, the ED is defined as the dose level at which 10% of the test 10 current control shows the effect. There may be population above the incidence in the con occasions where the historical control data have to be taken into account (for example when the concurrent control data are atypical and close to the extremes of the historical data). In the 3 . 10 , the ED example in is 90 mg/kg bw/day because at this dose level 12% - 2% Table 10 he control (control) = 10% of the test population shows the effect above the incidence in t group. 3 10 Example of the calculation of the ED Table . 10 Dose 10 mg/kg 30 mg/kg 90 mg/kg 0 mg/kg Malformations 2% 3% 7% 12% For some effects the results of the calculation of the ED based on the incidence in pups may 10 be different from that based on the incidence in litters. Scientific evidence may indicate which parameter is more appropriate, but in the absence of such information it is not possible to is more app ropriate for a specific effect. In such cases, both the incidence estimate which ED 10 value should be in offspring and the incidence in litters should be calculated, and the lower ED 10 used. Continuous or parametric data 3.7.2.6.3.3. For effects that are measured as changes in magnitude s uch as mean pup weight or testis is defined as the dose at which a change of 10%, compared to the concurrent weight, the ED 10 Table is 19.3 mg/kg bw/day . 11 , the ED 3 control group, is observed. In the example in 10 because at this dose level the mean foetal bodyweight is calculated to be 90% of the control value. A 10% reduction of the control value of 6.2 g gives 5.58 g. Interpolation between 10 and 30 mg/kg bw/day to a dose level which would be expected to result in a foetal bodyweight of 5.58 g gives a value of 19.3 mg/kg bw/day. Calculations: (30 – 10)/(6 - 5.1) = 22.2; 6.0 – 5.58 = 0.42; 0.42 x 22.2 = 9.3; 10 + 9.3 = 19.3 mg/kg bw/day. Table 3 . 11 Example on the calculation of the ED 10 Dose 0 mg/kg 10 mg/kg 30 mg/kg 90 mg/kg 4.5 Mean foetal bodyweight (g) 6.2 6.0 5.1 NOAEL LOAEL

410 ication of the CLP Criteria Guidance on the Appl 410 Version 5.0 – July 2017 3.7.2.6.3.4. Data combining incidence and magnitude Some effects such as histopathological changes in the testis are a combination of effects on incidence and magnitude (grading of the effect by a pathologist). However, calculation of an taking both the incidence and the magnitude into account is not possible or at least more ED 10 should therefore be based on the incidence of the effect below or above a complex. The ED 10 certain magnitude. The magnitude of the effects that will be selected as a starting point has to be chosen carefully. Normally the partic ular effect size would be the lowest relevant for the respective classification. The ED is then determined as the dose level at which the incidence, 10 of effects with a magnitude above that of the starting point, is 10% above the incidence in the control g roup. In practice this means that the grading system is converted into a simplified system where only percentages of animals in each dose group with an effect with a magnitude above the starting point are regarded as positive. However, it is recognised tha t this approach uses only a part of the actual data and is imprecise, and it may be appropriate that other effects . also be considered in determining the ED 10 Table . 12 Example on the calculation of the ED 3 for testicular effects (N=10) 10 Testicular degeneration (n) Dose (mg/kg) none slight moderate marked severe 4 5 0 0 0 1 0 10 5 5 0 0 NOAEL 5 4 1 0 0 30 0 0 4 LOAEL 2 4 90 Table 3 . 12 , the effects observed in the 10 mg/kg and 30 mg/kg dose groups For the example in have to be considered as equivalent to the effects of the control group so the NOAEL is 30 mg/kg. The magnitude of the testicula r effect in the control group and the 10 and 30 mg/kg bw/day groups is slight or less. Because of the incidence observed in these three groups, the level of damage estimated as the starting point magnitude is ‘slight’. The ED is then defined as 10 a 10% inc rease of moderate effects or more above the control. In this example the incidences for moderate testicular degeneration or more are 10%, 0%, 10% and 100% at respectively 0, 10, 30 and 90 mg/kg bw/day. The ED ontrol plus is then defined as the dose level with 20% (c 10 10%) of moderate testicular effects. The ED would be 36.6 mg/kg bw/day based on 10 interpolation between 30 and 90 mg/kg bw/day to a dose with 20% animals with moderate testicular degeneration or higher. Specific data types 3.7.2.6.3.5. - oral studies Non In m ost cases only oral studies will be available and used for determination of the potency. However, if the classification is based on the effects seen in non - oral studies or only non - oral studies are available, then these data should also be used to determin e the potency. This requires route - to - route extrapolation of the external dermal or inhalatory dose to a corresponding oral dose. This should be done as described in the ECHA G uidance on information requirements and chemical safety assessment in REACH ( IR& CSA , section R.8). Extrapolation from dermal exposure to oral exposure should only be done when there are sufficient kinetic data on dermal availability because assuming a high dermal availability is not a worst case assumption. In cases where such data ar e not available a direct comparison of the dermal dose with the oral potency ranges could be performed in exceptional cases. However, such comparison should not result in moving the substance to a lower potency group (higher

411 Guidance on the Application of the CLP Criteria 5.0 – July 2017 411 Version ) should be only moving the substan ce to a higher potency group (lower ED ) – ED 10 10 considered. Extrapolation from inhalatory exposure to oral exposure can only be done when there are sufficient kinetic data on inhaled availability because assuming a high inhaled availability is not rst case assumption. If no inhalatory information on availability is available then it should a wo be assumed that the inhalation and oral availability are comparable. However, such comparison – higher ED only moving ) should not result in moving the substance to a lower potency group ( 10 ) should be considered. the substance to a higher potency group (lower ED 10 Human data The use of human data for ED calculation has several drawbacks including limited data on 10 exposure, limited data on the size of the exposed population and limited information on whether the exposure included the window of sensitivity. For all these reasons, it is difficult to determin e based on human data. Therefore, and because in most instances animal data are also an ED 10 available for determining an ED , these data are evaluated together on a case by case basis. 10 Guidance on the use of human data for the derivation of DNELs and DME Ls has been developed by ECHA and is available at the ECHA website, see http://guidance.echa.europa.eu/guidance4_en.htm 3.7.2.6.4. Provisional evaluation of the potency classification potency evaluation applying the ED value is made at this stage. A preliminary 10 ED values can be used to place substances classified as a reproductive toxicant into selected 10 ranges that define potency groups. In this way, it is possible to identify reproductive toxica nts of high, medium and low potency. For the purpose of determining the preliminary potency group, the boundaries in Table 3 . 13 are used. 69 Table 13 Boundaries of the potency groups 3 . . Boundaries Potency group ED value ≤ 4 mg/kg bw/day High potency group 10 value 4 mg/kg bw/day < ED 400 mg/kg bw/day  Medium potency group 10 ED value  400 mg/kg bw/day. Low potency group 10 3.7.2.6.5. Modifying factors Modifying factors are a means to account for case - specific data situations which indicate that the potency group for a substance as obtained by the preliminary assessment, should be changed. While most modifying factors would result in a higher potency group than the - specific knowledge is available preliminary one, also the opposite could occur: If substance (such as e.g. toxicokinetic information on a higher bioavailability in test animals vs. humans), er potency class might be assigned. also a low While some modifying factors should always be taken into account, other modifying factors could be more relevant when the potency is close to the boundary between two groups (see Table 3 . 13 above). Some modifying factors are of a more qualitative nature. When applied, they will simply point to assessment. Other a potency group different from the one resulting from the preliminary modifying factors might be quantifiable, at least on a semi - quantitative scale. In such cases, a potency group higher (or lower) than the preliminary one should be chosen if the estimated size 69 See Annex VI of this guidance document for more details.

412 Guidance on the Appl ication of the CLP Criteria Version 5.0 – July 2017 412 of the modifying factor exceeds the distance of the preliminary ED to the border of the 10 relevant (higher or lower) adjacent potency group. Furthermore, for some substances more than one modifying factor will apply. It will then take expert judgement to decide on how to reasonably combine all of the se individual factors into one overall modifying factor. In exceptional cases, such a combination of individual factors might even result in a change of two potency classes (e.g. assignment of the high potency sulted in the low potency class). class, where the preliminary assessment had re In this context, it should be noted that several of the modifying factors may be interrelated. Moreover, some factors may have already been taken into account in deciding on the classification as a reproductive toxicant. Where such considerations have been made, care should be taken not to use that information again when determining the potency. For example, when the effects determining the ED were observed at dose levels also causing maternal 10 toxicity, this should alre ady have been taken into consideration during the classification and should not be used again to set a higher SCL. 3.7.2.6.5.1. Type of effect / severity The type of effect(s) resulting in the same classification as reproductive toxicant differs between substances. S ome effects could be considered as more severe than others, however, ranking different effects based on their severity is controversial and difficult to establish criteria. m variations Further, the effects of a developmental toxicant can differ between dose levels fro via malformations to death of the foetuses. The adverse effects on fertility and sexual function of a substance can differ between dose levels from small changes in testes histopathology through effects on fertility to an irreversible and comp lete absence of fertility. As the difference between the dose levels is often smaller than the proposed potency groups (factor 10 - 100) this will make no difference in most cases. Also classification is in most cases based on severe effects like malformatio ns or death of the foetuses for developmental toxicants and effects on fertility or histopathological changes of the reproductive organs for fertility toxicants. For most th classified substances such severe effects were already observed at the lowest dose wi reproductive effects (Muller et al , 2012). Therefore, differentiation between types of effect is considered to have limited added value. Exceptions can be dealt with on a case by case basis. For example, results in a preliminary conclusion f or the medium potency group but is if the ED 10 close to the border for the high potency group and the ED is based on a severe effect like 10 malformations or irreversible effects on sexual function and fertility then using the higher potency group (lower ED ) for that substance should be considered. To determine what is ‘close 10 to the border’ is to compare the distance to the next category border with the significance of modifying factors. Data availability 3.7.2.6.5.2. ich are: There are several aspects to this modifying factor, some of wh  limited data availability where certain test protocols are lacking and therefore certain parameters have not been evaluated; limited data availability where the spectrum of evaluated parameters is sufficient, but  only studies with limited duratio n are available; and  limited data availability where only a LOAEL, but no NOAEL could be identified. Where only limited data are available, such as a screening study (OECD 421 and 422), a 28 - day repeated dose toxicity study or non OECD studies which do not exclude the presence of - reproductive effects at lower dose levels, the calculated ED should not be used to set a SCL 10 above the GCL. Furthermore it should be considered to assign a modifying factor accounting for the limitations in the database in a sim ilar approach to the one used in deriving DNELs under REACH. Guidance regarding the potential size of such a factor can be obtained from ECHA’s Guidance on IR&CSA

413 Guidance on the Application of the CLP Criteria 5.0 – July 2017 413 Version - R.8 (‘Characterisation of dose [concentration] response for human health’). Section R.8.4.3.1 ‘Assessment of factors relating to extrapolation’, gives recommendations on how to set factors for extrapolating to longer study durations as well as for compensation of the lack of a NOAEL or of the generally poor quality of a database. If there are onl y limited data which result in an ED in the medium potency group which is 10 close to the border for the high potency group, then using the higher potency group should be of 8 mg/kg bw/day considered. For example an ED estimated based on a might have been 10 for malformations in the absence of a NOAEL, LOAEL is only higher by a factor of 2 This ED 10 mg/kg bw/d : see. Table 3.7.2.5.4 (i.e 2 times the border of the high potency group of 4 additional data at and assigning the high potency group should be considered until above), . Thus, there is uncertainty, if the ED lower dose levels are available based on extrapolation 10 from and below the LOAEL in the absence of a NOAEL and a correction may be justified. The size of this uncertainty could be determined by the BMDL (Benchmark dose lower 95% - confidence bound). In such cases, the BMDL could be used as a potency estimate instead of the . ED 10 3.7.2.6.5.3. Dose - response relationship The ED will in most cases probably be in the same range as the NOAEL and LOAEL. However, 10 in cas es of a shallow dose effect relationship curve, the LOAEL may sometimes be clearly below . In such situations, if a substance would fall into a lower potency group based on the the ED 10 potency group should ED but into a higher potency group based on the LOAEL then the higher 10 be used for that substance. Mode or mechanism of action 3.7.2.6.5.4. It is assumed that effects observed in animal studies are relevant to humans. Where it is known relevance to that the mode or mechanism of action is not relevant for humans or is of doubtful humans, this should have been taken into account in the classification and should not be used again as a modifying factor for potency. However, quantitative differences in toxicodynamics count in the classification. In cases can be taken into account when not already taken into ac where mechanistic information shows a lower sensitivity in humans than in experimental animals, this may move substances which are close to the potency boundaries to a lower potency group. In cases where mechanistic in formation indicates a higher sensitivity in humans than in experimental animals, this may move substances near the potency boundaries to a higher potency group. In general, more conclusive evidence is required when moving a substance to a lower potency gro up than to a higher potency group. 3.7.2.6.5.5. Toxicokinetics The toxicokinetics of a substance can differ between the tested animal species and humans. Where a difference is known this should be taken into account when determining the potency is should be based on a comprehensive knowledge of all involved group of a substance. Th toxicokinetic factors and not only on a single parameter. Also differences in kinetics between pregnant and non - pregnant animals and transport to the foetus should be taken into account. Based on the available data, quantification of this modifying factor has to be performed on a case by case basis. This modifyi