ATMOSPHERIC CHEMISTRY OF POLYFLUORINATED COMPOUNDS: LONG-LIVED GREENHOUSE GASES AND SOURCES OF PERFLUORINATED ACIDS

by

Cora J. Young

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Chemistry University of Toronto

© Copyright by Cora J. Young 2010 Atmospheric chemistry of polyfluorinated compounds: Long-lived greenhouse gases and sources of perfluorinated acids Doctor of Philosophy Degree, 2010 Cora J. Young Department of Chemistry, University of Toronto

ABSTRACT

Fluorinated compounds are environmentally persistent and have been demonstrated to bioaccumulate and contribute to climate change. The focus of this work was to better understand the atmospheric chemistry of poly- and per-fluorinated compounds in order to appreciate their impacts on the environment. Several fluorinated compounds exist for which data on climate impacts do not exist. Radiative efficiencies (REs) and atmospheric lifetimes of two new long-lived greenhouse gases (LLGHGs) were determined using smog chamber techniques: perfluoropolyethers and perfluoroalkyl amines. Through this, it was observed that RE was not directly related to the number of carbon-fluorine bonds. A structure-activity relationship was created to allow the determination of RE solely from the chemical structure of the compound. Also, a novel method was developed to detect polyfluorinated LLGHGs in the atmosphere. Using carbotrap, thermal desorption and cryogenic extraction coupled to GC-MS, atmospheric measurements can be made for a number of previously undetected compounds. A perfluoroalkyl amine was detected in the atmosphere using this technique, which is the compound with the highest RE ever detected in the atmosphere. Perfluorocarboxylic acids (PFCAs) are water soluble and non-volatile, suggesting they are not susceptible to long-range transport. A hypothesis was derived to explain the ubiquitous distribution of these compounds involving atmospheric formation of PFCAs from volatile precursors. Using smog chamber techniques with offline analysis, perfluorobutenes and fluorotelomer iodides were shown to yield PFCAs from atmospheric oxidation. Dehydrofluorination of perfluorinated alcohols (PFOHs) is poorly understood in the mechanism of PFCA atmospheric formation. Using density functional techniques, overtone-induced photolysis was shown to lead to dehydrofluorination of PFOHs. In the presence of water, this mechanism could be a sink of PFOHs in the atmosphere. Confirmation of the importance of volatile precursors was derived from examination of snow from High Arctic ice caps. This ii provided the first empirical evidence of atmospheric deposition. Through the analytes observed, fluxes and temporal trends, it was concluded that atmospheric oxidation of volatile precursors is an important source of PFCAs to the Arctic.

iii ACKNOWLEDGEMENTS

Without the assistance and support of numerous people, it would have been impossible for me to produce this thesis. My words cannot express the gratitude I feel toward my collaborators, mentors, friends and family, but I will do my best to communicate my appreciation here. First and foremost, I would like to thank my supervisor, Scott Mabury, from whom I have learned so much—both within and outside the realm of chemistry. I am also grateful to my committee members, Jon Abbatt and Frank Wania, as well as Jamie Donaldson and Jen Murphy. Each of you has been so supportive in providing advice on specific scientific problems as well as career issues. This work was not mine alone, but consisted of contributions from collaborators and co- authors. I would particularly like to thank Tim Wallington and Mike Hurley who were integral to most of my publications and who taught me so much. Thanks also to Jamie for helping me to think outside the box of my planned dissertation and assisting me in producing work using molecular modelling. Finally, thanks to my Arctic collaborators, Vasile Furdui, James Franklin, Derek Muir, Dan Walsh and the late Roy “Fritz” Koerner. The project was a huge undertaking and could never have been accomplished without your contributions. Members of the Mabury group and environmental chemistry, both past and present, have been colleagues as well as friends. I feel so lucky to have been part of such a wonderful group of people. Dr. Monica Lam, my original Mabury group mentor—thank you for teaching me so much about conducting experiments and being efficient. Derek, the man who always knows the answer—I hope you will be available when I have questions in the future! Erin, thanks for being such an amazing friend and for all the “water breaks”. Amy, Holly, Anne, Pablito, Naomi and Sarah—thanks for being always there to bounce ideas and vent! My compatriots, Amila, Craig, Jessica, Tara and Zam—thanks for always providing support, advice, a shoulder to cry on or a glass of wine when the going got tough. Dr. De Silva (aka “Abrupt Spice”), my GC-MS and life guru—your daily advice sessions are missed already! Craig, I can’t imagine a time when I won’t be able to roll back my chair and talk to you or ask you a question—expect to hear from me multiple times per day! Jess, my doppelganger, thanks for teaching me so much about life attitudes and balance—I aspire to be more like you. Tara, thank you for always smiling and for only needing to drink two glasses of wine! Zam, thanks for the inappropriate questions and for always being so positive. iv Of course, thanks to my family: my parents and Teresa, who inspired my curiosity and interest in science and have believed in me from the very beginning. It’s impossible to articulate how much this has meant to me over the years. Thank you for the love and support and for always making sure I wasn’t going hungry! My “non-chemistry” friends, Subha, Andrea and Erin—thank you for supporting me, even when you had no idea what I was talking about. Trevor, last but not least! I am so lucky to have you in my life—without your love, encouragement and engineering and management skills, I would never have made it.

v TABLE OF CONTENTS

CHAPTER ONE – Polyfluorinated Compounds as Long-Lived 1 Greenhouse Gases

1.1 Introduction 2 1.2 Assessing climate impact 3 1.2.1 Radiative forcing 3 1.2.2 Radiative efficiency 3 1.2.3 Global warming potential 4 1.2.4 Importance of halogenated compounds as long-lived 5 greenhouse gases 1.3 International agreements affecting long-lived greenhouse gases 6 1.3.1 Kyoto Protocol 6 1.3.2 Montreal Protocol 6 1.3.3 Other agreements 7 1.4 Atmospheric measurements of halogenated long-lived greenhouse gases 7 1.4.1 Methods 7 1.4.2 Observed levels 7 1.4.3 Global networks and trends 8 1.5 Goals and hypotheses 9 1.6 Sources cited 10

CHAPTER TWO – Atmospheric Perfluorinated Acid Precursors: 12 Chemistry, Occurrence and Impacts

2.1 Introduction 13 2.2 Mechanisms of atmospheric formation of perfluorinated acids (PFAs) 15 2.2.1 Perfluorocarboxylic acids (PFCAs) 15 2.2.1.1 Mechanisms for atmospheric formation of perfluoroacyl 15 halides 2.2.1.1.1 Chemistry of Perfluoroacyl halides 15 2.2.1.1.2 Mixed halide mechanism 16 2.2.1.1.3 Perfluorinated radical mechanism 17 2.2.1.2 Mechanisms for direct atmospheric formation of 18 perfluorocarboxylic acids (PFCAs) 2.2.1.2.1 Perfluoroacyl peroxy radical mechanism 18 2.2.1.2.2 Perfluorinated aldehyde (PFAL) hydrate 21 mechanism 2.2.2 Perfluorosulfonic acids (PFSAs) 21 2.3 Chemistry of perfluorinated acid (PFA) precursors 22 2.3.1 Fluorotelomer and related compounds 22 2.3.1.1 Perfluorinated aldehydes (PFALs) 24 2.3.1.2 Fluorotelomer aldehydes (FTALs) 30 2.3.1.3 Fluorotelomer alcohols (FTOHs) 32 2.3.1.4 Fluorotelomer olefins (FTOs) 35 2.3.1.5 Fluorotelomer acrylate (FTAc) 37

vi 2.3.2 Perfluoroalkanesulfonamides 39 2.3.2.1 N-alkyl-perfluoroalkanesulfonamides (NAFSA) 39 2.3.2.2 N-alkyl-perfluoroalkanesulfamidoethanols (NAFSE) 41 2.4 Atmospheric sources and levels 43 2.4.1 Volatile fluorinated anaesthetics 43 2.4.1.1 Potential sources to the atmosphere 43 2.4.2 Hydrochlorofluorocarbons (HCFCs) 44 2.4.2.1 Potential sources to the atmosphere 44 2.4.2.2 Atmospheric concentrations 44 2.4.3 Hydrofluorocarbons (HFCs, non-telomer based) 44 2.4.3.1 Saturated hydrofluorocarbons (HFCs) 44 2.4.3.1.1 Potential sources to the atmosphere 44 2.4.3.1.2 Atmospheric concentrations 45 2.4.3.2 Hydrofluoroolefins (HFOs) 45 2.4.3.2.1 Potential sources to the atmosphere 45 2.4.4 Fluorotelomer compounds 45 2.4.4.1 Potential sources to the atmosphere 45 2.4.4.2 Atmospheric concentrations 46 2.4.4.2.1 FTOHs 47 2.4.4.2.2 FTOs 47 2.4.4.2.3 FTAcs 47 2.4.5 Perfluorosulfonamides 48 2.4.5.1 Potential sources to the atmosphere 48 2.4.5.2 Atmospheric concentrations 49 2.4.5.2.1 NAFSA 49 2.4.5.2.2 NAFSE 49 2.5 Impact of precursors on environmental perfluorinated acid (PFA) levels 49 2.5.1 Trifluoroacetic acid (TFA) 49 2.5.2 Perfluorooctanesulfonic acid (PFOS), perfluorooctanoic acid 51 (PFOA) and perfluorononanoic acid (PFNA) 2.5.3 Long-chained perfluorocarboxylic acids (PFCAs) 54 2.6 Goals and Hypotheses 55 2.7 Sources Cited 56

CHAPTER THREE – Atmospheric Lifetime and Global Warming 67 Potential of a Perfluoropolyether

C.J. Young, M.D. Hurley, T.J. Wallington and S.A. Mabury Environ. Sci. Technol. 2006 40:2242-2246

3.1 Introduction 68 3.2 Experimental 69 3.2.1 Chemical preparation 69 3.2.2 Kinetics 69 3.3 Results and Discussion 70 3.3.1 Kinetics 70 3.3.2 Photolysis of PFPMIE 72 3.3.3 IR spectrum and global warming potential of PFPMIE 74 vii 3.4 Acknowledgements 78 3.5 Sources Cited 79

CHAPTER FOUR – Molecular Structure and Radiative Efficiency of 80 Fluorinated Ethers: A Structure-Activity Relationship

C.J. Young, M.D. Hurley, T.J. Wallington and S.A. Mabury J. Geophys. Res. 2008 113:D24301

4.1 Introduction 81 4.2 Methods 83 4.2.1 Experimental details 83 4.2.2 Computational details 83 4.2.3 Determination of radiative efficiency 84 4.3 Results and Discussion 85 4.3.1 Experimental infrared cross-sections 85 4.3.2 Validity of computed spectra 88 4.3.3 Development and utility of the SAR 88 4.3.4 Radiative efficiency and molecular structure 93 4.4 Conclusions 96 4.5 Acknowledgements 97 4.6 Sources Cited 98

CHAPTER FIVE – Perfluoroalkyl Amines: A New Class of Long-Lived 101 Greenhouse Gases

C.J. Young, M.D. Hurley, T.J. Wallington and S.A. Mabury

5.1 Introduction 102 5.2 Methods 102 5.2.1 Infrared absorption measurements 102 5.2.2 Determination of radiative efficiency 103 5.2.3 Physical properties and use of multi-species model 103 5.2.4 Air sample collection and analysis 103 5.3 Results and Discussion 104 5.3.1 Infrared spectrum and radiative efficiency 104 5.3.2 Physical properties and impact of ionization 105 5.3.3 Atmospheric lifetime of PFAms 106 5.3.4 Atmospheric detection of PFBAm 107 5.4 Environmental Implications 109 5.5 Acknowledgements 111 5.6 Sources Cited 112

viii CHAPTER SIX – Atmospheric Chemistry of Perfluorobutenes 114 (CF3CF=CFCF3 and CF3CF2CF=CF2): Kinetics and Mechanisms of Reactions with OH Radicals and Chlorine Atoms, IR Spectra, Global Warming Potentials, and Oxidation to Perfluorocarboxylic Acids

C.J. Young, M.D. Hurley, T.J. Wallington and S.A. Mabury Atmos. Env. 2009 43:3717-3724

6.1 Introduction 115 6.2 Methods 115 6.2.1 Chemicals 115 6.2.2 Kinetics 116 6.2.3 Products 117 6.3 Results and Discussion 117 6.3.1 Kinetics of reactions with Cl atoms 117 6.3.2 Kinetics of reactions with OH radicals 119 6.3.3 Products of Cl atom- and OH radical-initiated oxidation of 121 CF3CF=CFCF3 and CF3CF2CF=CF2 6.3.4 Proposed oxidation mechanisms 125 6.3.5 Infrared spectra and radiative efficiency of CF3CF=CFCF3 126 and CF3CF2CF=CF2 6.4 Atmospheric Implications 128 6.5 Acknowledgements 130 6.6 Sources Cited 131

CHAPTER SEVEN - Atmospheric Chemistry of 4:2 Fluorotelomer Iodide 134 (n-C4F9CH2CH2I): Kinetics and Products of Photolysis and Reaction with OH Radicals and Cl Atoms

C.J. Young, M.D. Hurley, T.J. Wallington and S.A. Mabury J. Phys. Chem. A 2008 112:13542-13548

7.1 Introduction 135 7.2 Experimental 136 7.2.1 Smog chamber methods 136 7.2.2 Offline sample collection and analysis 137 7.2.3 UV Spectral measurements and photolysis rate calculations 138 7.3 Results and Discussion 138 7.3.1 Kinetics of the Cl + 4:2 FTI reaction 138 7.3.2 Kinetics of the OH + 4:2 FTI reaction 139 7.3.3 Products of Cl + 4:2 FTI reaction 141 7.3.4 UV spectra and photolysis kinetics 143 7.3.5 Photolysis Products 144 7.4 Atmospheric Implications 146 7.4.1 FTI lifetime 146 7.4.2 Oxidation and photolysis products 147 7.4.3 Formation of PFCAs 148

ix 7.5 Acknowledgements 150 7.6 Sources Cited 151

CHAPTER EIGHT – Overtone-Induced Degradation of Perfluorinated 154 Alcohols in the Atmosphere

C.J. Young and D.J. Donaldson J. Phys. Chem A. 2007 111:13466-13471

8.1 Introduction 155 8.2 Methods 157 8.3 Results and Discussion 158 8.3.1 Alcohol-water complexes 158 8.3.2 Overall features of the reactions 160 8.3.3 Overtone-induced degradation 163 8.4 Atmospheric Implications 166 8.5 Acknowledgements 168 8.6 Sources Cited 169

CHAPTER NINE – Perfluorinated Acids in Arctic Snow: New 173 Evidence for Atmospheric Formation

C.J. Young, V.I. Furdui, J. Franklin, R.M. Koerner, D.C.G. Muir and S.A. Mabury Environ. Sci. Technol. 2007 41:3455-3461. Featured on May 15, 2007 cover

9.1 Introduction 174 9.2 Experimental 175 9.2.1 Chemicals 175 9.2.2 Sample collection 176 9.2.3 Sample preparation and analysis 176 9.3 Results and Discussion 178 9.3.1 QA/QC 178 9.3.2 Dating Arctic Snow 178 9.3.3 Concentrations of PFAs in arctic snow 178 9.3.4 Effects of annual melting 181 9.3.5 Fluxes of PFAs to the Arctic 181 9.3.6 Sources of PFAs to the Arctic 182 9.4 Acknowledgements 186 9.5 Sources Cited 187

CHAPTER TEN – Summary and Future Directions 190

10.1 Conclusions 191 10.2 Future Directions 193 10.2.1 Perfluoroalkyl amines 193 10.2.2 Atmospheric chemistry of new long-lived greenhouse gases 193 x 10.2.3 Atmospheric monitoring of fluorinated ethers 194 10.2.4 Atmospheric monitoring of volatile perfluorinated acid precursors 194 10.3 Sources Cited 195

xi LIST OF FIGURES

CHAPTER ONE Figure 1.1: Terrestrial emission spectrum shown in comparison to the region of 2 carbon-fluorine bond stretching vibrations and the atmospheric window. Figure 1.2: Radiative forcing of a) all long-lived greenhouse gases; and b) 5 halocarbons. Figure 1.3: Temporal evolution of the global average dry-air mole fractions 8 (ppt) of the major halogen-containing LLGHGs. These are derived mainly using monthly mean measurements from the AGAGE and NOAA/GMD networks. For clarity, the two network values are averaged with equal weight when both are available. While differences exist, these network measurements agree reasonably well with each other (except for CCl4 (differences of 2 – 4% between networks) and HCFC-142b (differences of 3 – 6% between networks)), and with other measurements where available.

CHAPTER TWO Figure 2.1: Mechanism of perfluoroacyl fluoride formation from mixed halides, 16 where X=F,Cl and Y=Cl,Br. Figure 2.2: Mechanism of PFCA formation via perfluorinated radicals. 18 Figure 2.3: Mechanism of PFCA formation via perfluoro acyl peroxy radicals. 20 Figure 2.4: Proposed mechanism for the formation of PFCAs from PFAL 21 hydrates. Figure 2.5: Proposed mechanism of PFBA formation from NMeFBSE. 22 Figure 2.6: Transformation pathways of perfluorinated acid precursors. Grey 23 compounds are commercially produced.

CHAPTER THREE Figure 3.1: General structure of PFPMIEs. 68 Figure 3.2: IR absorption cross section for PFPMIE (solid line) and n-C6F14 74 (light dashed line) shown with the radiative forcing per unit cross section of the atmosphere (dotted line).

CHAPTER FOUR Figure 4.1: Experimental infrared absorption cross-sections measured for (a) 86 CHF2CF2-O-CH3; (b) CHF2CF2-O-CH3; (c) CHF2CF2CH2-O-CH3; (d) CF3CF2CF2-O-CH3; (e) CF3CF2CH2OCHF2; (f) CHF2CF2CH2OCHF2; (g) CF3CHFCF2-O-CH2CH3; (h) CF3CHFCF2OCH2CF3; (i) CF3CF2CF2OCHFCF3; (j) CF3CHFCF2OCH2CF2CHF2; (k) CF3CHFCF2OCH2CF2CF3. Figure 4.2: Absorption wavenumber range for each type of C-F stretch in 92 hydrofluoroethers (HFE) functional groups shown with respect to the atmospheric window (…..).

xii Figure 4.3: Radiative forcing per C-F bond for each functional group, where 94 bars represent the observed range. The empirical correction factor for -O-CHF2 has been included in the CHF2 range. Figure 4.4: (a) Experimental infrared absorption cross-sections for 95 CHF2CF2CH2OCH3 and CHF2CF2OCH3 compared to the atmospheric window (…..); (b) total and radiative efficiency (RE) per C-F bond for each molecule.

CHAPTER FIVE Figure 5.1: Infrared absorption cross section for PFBAm (solid line) shown with 105 the radiative forcing per unit cross section of the atmosphere (dotted line). Figure 5.2: Chromatogram of GC-MS analysis (m/z 633) of September 28, 2009 108 sample (–) and blank sorbent tube (…). Figure 5.3: Comparison between September 28, 2009 sample (–) and authentic 108 standard (…) for a) m/z 633 and b) m/z 452.

CHAPTER SIX Figure 6.1: Loss of (a) CF3CF=CFCF3 and (b) CF3CF2CF=CF2 versus C2H2 (▲) 118 and C2H5Cl (●) following exposure to Cl atoms. Figure 6.2: Loss of (a) CF3CF=CFCF3 and (b) CF3CF2CF=CF2 versus C2H2 (▲) 120 and C2H5Cl (●) following exposure to hydroxyl radicals. Figure 6.3: IR spectra acquired before (A) and after (B) UV irradiation of a 122 mixture of 4.85 mTorr CF3CF=CFCF3 and 100 mTorr Cl2 in 700 Torr of air diluent. Panel (C) shows the product spectrum obtained after subtracting features attributable to CF3CF=CFCF3 from panel (B). Panel (D) is a reference spectrum of CF3C(O)F. Figure 6.4: Yield of CF3C(O)F following chlorine atom- (●) and hydroxyl 122 radical- (▲) initiated oxidation of CF3CF=CFCF3. The line has a slope of two. Figure 6.5: IR spectra acquired before (A) and after (B) UV irradiation of a 124 mixture of 7.64 mTorr CF3CF2CF=CF2 and 100 mTorr Cl2 in 700 Torr of air diluent. Panel (C) shows the product spectrum obtained after subtracting features attributable to CF3CF2CF=CF2 from panel (B). Panels (D) and (F) are reference spectra of COF2 and CF3C(O)F. Panel E is the residual spectrum obtained after subtracting features attributable to COF2 from the product spectrum (C). Figure 6.6: Yields of COF2 (a) and CF3CF2C(O)F (b) following chlorine atom- 125 initiated (●) and hydroxyl radical-initiated (▲) oxidation of CF3CF2CF=CF2. The lines have slopes of unity. Figure 6.7: Proposed mechanism for the atmospheric oxidation of (a) 126 CF3CF=CFCF3 and (b) CF3CF2CF=CF2 initiated by hydroxyl radicals. Stable products are given in boxes. In part (b), only addition to the terminal carbon is depicted for simplicity. Figure 6.8: Infrared spectra of (a) CF3CF=CFCF3 and (b) CF3CF2CF=CF2. 127

xiii CHAPTER SEVEN Figure 7.1: Decay of 4:2 FTI versus CH3Cl and CH3OCHO in the presence of Cl 139 atoms in 700 Torr of N2 at 295 ± 2 K. Figure 7.2: Decay of 4:2 FTI versus C2H4 and C3H8 in the presence of OH 140 radicals in 700 Torr of air diluent at 295 ± 2 K. Figure 7.3: FTIR spectra of a mixture of 12.2 mTorr 4:2 FTI and 100 142 mTorr Cl2 in 700 Torr of air before (A) and after (B) 10 minutes UV (blacklamps) irradiation. Panel C is the product spectrum. Reference spectra of the peracid C4F9CH2C(O)OOH and the aldehyde C4F9CH2CHO are given in panels D and E. Figure 7.4: UV-visible absorption cross sections for 4:2 FTI (gray) and 143 ethyl iodide (dashed) compared to actinic flux at the surface (dotted). Symbols represent literature data from Roehl et al. (x) and Rattigan et al. (+). Figure 7.5: FTIR spectra of a mixture of 18.5 mTorr 4:2 FTI in 50 146 Torr of O2 diluent before (A) and after (B) 137 min of UV (sunlamp) irradiation. Panel C shows the result of subtracting features attributable to 4:2 FTI from panel B. A reference spectrum of C4F9CH2CHO is given in panel D. Figure 7.6: Proposed mechanism for atmospheric oxidation of 4:2 FTI. 149 Shaded compounds were observed in FTIR spectra, compounds in boxes were observed in offline analyses. Subscript “x” represents 0-3 and subscript “y” represents 0-2, where the values will depend on the number of times through the degradation cycle. Thick arrows indicate steps that require a low-NOx environment (see text for details).

CHAPTER EIGHT Figure 8.1: Optimized structure of CF3CF2OH•H2O. See Table S2 for bond 158 lengths and angles. Figure 8.2: Calculated equilibrium constants for the reaction of CF3OH and 159 CF3CF2OH with water to form complexes. Figure 8.3: Optimized structures of the calculated transition states for reactions 161 ‡ ‡ (4) through (7): (a) [CF3OH] ; (b) [CF3OH•H2O] ; (c) ‡ ‡ [CF3CF2OH] ; and (d) [CF3CF2OH•H2O] . See Tables S1 and S2 for bond lengths and angles. Figure 8.4: Energetics of reactions (4) and (5) with all energies relative to 161 CF3OH. Also shown are calculated overtone vibrations for CF3OH (black) and CF3OH•H2O (blue). Figure 8.5: Energetics of reactions (6) and (7) with all energies relative to 162 CF3CF2OH. Also shown are calculated overtone vibrations for CF3CF2OH (black) and CF3CF2OH•H2O (blue).

xiv CHAPTER NINE Figure 9.1: Density-corrected concentrations of PFAs on Devon Ice Cap since 177 1996: (a) PFOA and PFNA; (b) PFDA and PFUnA; and (c) PFOS. Figure 9.2: PFA fluxes to Devon Ice Cap by year. Mean fluxes with standard 180 error of three replicates indicated by bars. Lines reflect a three-year moving average. Figure 9.3: Correlations between PFCA concentrations on Devon Ice Cap. 184

xv LIST OF TABLES

CHAPTER ONE Table 1.1: Lifetimes, radiative efficiencies and global warming potentials for 5 long-lived greenhouse gases.

CHAPTER TWO Table 2.1: List of acronyms. 14 Table 2.2: Product yields of reaction (10) and (11). 20 Table 2.3: Summary of chlorine atom-initiated kinetics for fluorotelomer and 25 related compounds. Table 2.4: Summary of hydroxyl radical-initiated kinetics for fluorotelomer 26 and related compounds at atmospheric pressure. Table 2.5: Photolysis properties of perfluorinated aldehydes (PFALs) and 28 fluorotelomer aldehydes (FTALs). Table 2.6: Fate of perfluoroacyl radicals. 30 Table 2.7: Summary of chlorine atom- and hydroxyl radical-initiated kinetics 43 for perfluorosulfonamides. Table 2.8: Estimated indirect and direct sources of trifluoroacetic acid (TFA). 50

CHAPTER THREE Table 3.1: Photolysis rate constants and lifetimes calculated between 162 and 71 210 nm for CHF2-O-CHF2, used to estimated those of PFPMIE, at different altitudes. Table 3.2: Lifetimes and GWPs of perfluorohexane and selected fluorinated 76 ethers.

CHAPTER FOUR Table 4.1: Measured integrated IR band strengths for hydrofluoroethers 85 (HFEs). Comparison between radiative efficiencies determined from measured cross-sections and by use of the structure-activity relationship (SAR). Table 4.2: Comparison between calculated vibrational frequencies in this work 87 and calculated and experimental frequencies of Good and Francisco.

Table 4.3: Radiative efficiencies for structural components of 90 hydrofluoroethers (HFEs) for use in the structure-activity relationship (SAR). Table 4.4: Structure-activity relationship (SAR) predicted versus published 91 radiative efficiencies (REs) determined from experimentally measured cross-sections. Values in brackets include the empirical correction factor for -O-CHF2.

xvi CHAPTER FIVE Table 5.1: Physical-chemical properties for PFBAm, PFBAmH+, PFOSH and 105 PFOS. Table 5.2: Atmospheric lifetimes of some perfluorinated compounds. 107 Table 5.3: Radiative efficiencies and global warming potentials for selected 110 perfluorinated compounds and PFBAm.

CHAPTER SEVEN Table 7.1: Twenty-four hour averaged photolysis rate constants and lifetimes 144 for ethyl iodide and 4:2 FTIs in Toronto, ON, Canada.

CHAPTER EIGHT Table 8.1: Calculated relative energies of species (kJ mol-1) at 273 K. 158 Table 8.2: Calculated energies and zero-point corrections for all species 160 investigated. Table 8.3: Comparison of calculated values (kJ mol-1 at 298 K) for reactions 164 (4) and (5) with literature values. Table 8.4: Band centres of PFOHs and PFOH-water complexes. 165

CHAPTER NINE Table 9.1: Mean fluxes of PFAs to the Arctic for 2004 and 2005. Note that 183 PFUnA concentrations were not measured for 2004 samples.

xvii LIST OF APPENDICES

APPENDIX A Supporting information for Chapter Four 196 APPENDIX B Supporting information for Chapter Five 200 APPENDIX C Supporting information for Chapter Seven 204 APPENDIX D Supporting information for Chapter Eight 207 APPENDIX E Supporting information for Chapter Nine 211

xviii PREFACE

This thesis is comprised of a series of manuscripts that have been published or are in preparation for submission to be published in peer-reviewed scientific journals. Consequently, repetition of introductory and experimental details was inevitable. It should be noted that Chapters One and Two together comprise the introduction to this thesis. All manuscripts were written by Cora J. Young with critical comments provided by Scott Mabury. The contributions of co-authors are detailed below.

Chapter One – Polyfluorinated Compounds as Long-Lived Greenhouse Gases Contributions – Prepared by Cora J. Young with editorial comments provided by Scott Mabury.

Chapter Two – Atmospheric Formation of Perfluorinated Acids Contributions – Prepared by Cora J. Young with editorial comments provided by Scott Mabury.

Chapter Three – Atmospheric Lifetime and Global Warming Potential of a Perfluoropolyether Published in: Environ. Sci. Technol. 2006 40:2242-2246 Author List – Cora J. Young, Michael D. Hurley, Timothy J. Wallington, Scott A. Mabury Contributions – Preparation of samples was done by Cora Young. Smog chamber experiments were performed by Cora Young and Michael Hurley. The manuscript was prepared by Cora Young with minor contributions from Timothy Wallington. The manuscript and research were conducted under the guidance of Scott Mabury.

Chapter Four – Molecular Structure and Radiative Efficiency of Fluorinated Ethers: A Structure-Activity Relationship Published in: J. Geophys. Res. 2008 113:D24301 Author List – Cora J. Young, Michael D. Hurley, Timothy J. Wallington and Scott A. Mabury Contributions – Laboratory and modelling experiments were conducted by Cora Young with critical comments provided by Michael Hurley and Timothy Wallington, under the guidance of Scott Mabury.

Chapter Five – Perfluoroalkyl Amines: A New Class of Long-Lived Greenhouse Gases Author List – Cora J. Young, Michael D. Hurley, Timothy J. Wallington and Scott A. Mabury xix Contributions – Laboratory studies, modeling, instrument development, sample collection and analysis were performed by Cora Young. Manuscript preparation was undertaken by Cora Young with critical comments provided by Michael Hurley, Tim Wallington and Scott Mabury.

Chapter Six – Atmospheric Chemistry of Perfluorobutenes (CF3CF=CFCF3 and

CF3CF2CF=CF2): Kinetics and Mechanisms of Reactions with OH Radicals and Chlorine Atoms, IR Spectra, Global Warming Potentials, and Oxidation to Perfluorocarboxylic Acids Published in: Atmos. Env. 2009 43:3717-3724 Author List – Cora J. Young, Michael D. Hurley, Timothy J. Wallington and Scott A. Mabury Contributions – Experiments were performed by Cora Young with the assistance of Michael Hurley. Data interpretation and manuscript preparation was conducted by Cora Young with critical comments provided by Michael Hurley and Timothy Wallington. All research and manuscript preparation was conducted under the guidance of Scott Mabury.

Chapter Seven – Atmospheric Chemistry of 4:2 Fluorotelomer Iodide (n-C4F9CH2CH2I): Kinetics and Products of Photolysis and Reaction with OH Radicals and Cl Atoms Published in: J. Phys. Chem. A 2008 112:13542-13548 Author List – Cora J. Young, Michael D. Hurley, Timothy J. Wallington and Scott A. Mabury Contributions – Smog chamber experiments were undertaken by Cora Young with assistance by Michael Hurley. Sample extraction, analysis, modeling experiments and data interpretation were performed by Cora Young. Manuscript preparation was conducted by Cora Young with assistance from Timothy Wallington. Research and manuscript preparation was conducted under the guidance of Scott Mabury.

Chapter Eight – Overtone-Induced Degradation of Perfluorinated Alcohols in the Atmosphere Published in: J. Phys. Chem. A 2007 111:13466-13471 Author List – Cora J. Young and D.J. Donaldson Contributions – Modeling experiments, data interpretation and manuscript preparation were performed by Cora Young under the guidance of Jamie Donaldson.

xx Chapter Nine – Perfluorinated Acids in Arctic Snow: New Evidence for Atmospheric Formation Published in: Environ. Sci. Technol. 2007 41:3455-3461 Author List – Cora J. Young, Vasile I. Furdui, James Franklin, Roy M. Koerner, Derek C.G. Muir and Scott A. Mabury Contributions – Sample collection, extraction and analysis was undertaken by Cora Young and Vasile Furdui. Sample dating was done by Roy Koerner. Data interpretation and manuscript preparation was conducted by Cora Young with critical comments provided by James Franklin and Derek Muir. The research and manuscript preparation was performed under the guidance of Scott Mabury.

Chapter Ten – Conclusions and Future Directions Contributions – Prepared by Cora Young with editorial comments provided by Scott Mabury.

xxi Other Publications During PhD:

Young, C.J.; De Silva, A.O.; Donaldson, D.J. and Mabury S.A. Polarizability as a predictor of physical properties for poly-and perfluorinated chemicals. To be submitted to Environmental Science and Technology. Young, C.J.; Hurley, M.D.; Wallington, T.J. and Mabury, S.A. Atmospheric Chemistry of CF3CF2H and CF3CF2CF2CF2H: Kinetics and Products of Gas-Phase Reactions with Cl atoms and OH Radicals, Infrared Spectra, and Formation of Perfluorocarboxylic Acids. 2009 Chemical Physics Letters 473:251-256. Butt, C.M.; Young, C.J.; Mabury, S.A.; Hurley, M.D. and Wallington, T.J. Atmospheric Chemistry of 4:2 Fluorotelomer Acrylate (C4F9CH2CH2OC(O)CH=CH2): Kinetics, Mechanisms and Products of Chlorine Atom and OH Radical Initiated Oxidation. 2009 Journal of Physical Chemistry A 113:3155-3161. Young, C.J.; Gόmez Biagi, R.F.; Hurley, M.D.; Wallington, T.J. and Mabury, S.A. Paint solvent to food additive: An environmental route of dehalogenation for 4- chlorobenzotrifluoride. 2008 Environmental Toxicology and Chemistry 27:2233-2238.

xxii

CHAPTER ONE

Polyfluorinated Compounds as Long-Lived Greenhouse Gases

1 2 1.1 Introduction

In order for a compound to be considered a long-lived greenhouse gas (LLGHG) it must have a long atmospheric lifetime and be capable of absorbing outgoing infrared radiation within the atmospheric window. Poly- and perfluorinated compounds are often characterized by these two properties. Fluorination drastically reduces reactivity toward atmospheric oxidants and photolysis due to its electron-withdrawing properties. Thus, polyfluorinated compounds tend to be much longer-lived in the atmosphere. Perfluorinated alkyl compounds tend to be highly unreactive, as they contain no abstractable hydrogens or sites for oxidant addition reactions. The major fate for perfluorinated compounds is typically reaction in the upper atmosphere (1,2), leading to atmospheric lifetimes of hundreds of years.

The stretching vibration of the carbon-fluorine bond absorbs within the spectral region of the atmosphere where little absorption by other major atmospheric components occurs. This region, between 750 – 1250 cm-1, is termed the atmospheric window (Figure 1.1). Compounds that absorb within this window, such as polyfluorinated chemicals, have the potential to impact climate. This potential, combined with a long lifetime, causes most poly- and perfluorinated compounds to be classed as LLGHGs.

100 Atmospheric Window ) -1 )

-1 80 (cm CO2 -2 60 C-F stretch Wm

-3 O3 40

CH4 20 H2O Radiance (10

0 800 1000 1200 1400 -1 Wavenumber (cm )

Figure 1.1: Terrestrial emission spectrum shown in comparison to the region of carbon-fluorine bond stretching vibrations and the atmospheric window.

3 1.2 Assessing Climate Impact

1.2.1 Radiative forcing

The climate impact of different anthropogenic and natural chemicals is estimated using radiative forcing (RF). This metric is defined by the IPCC as “the change in net irradiance at the tropopause after allowing for stratospheric temperatures to readjust to radiative equilibrium, but with surface and tropospheric temperatures and state held fixed at the unperturbed values” (3). An overall surface temperature change is related to RF by the following equation:

ΔTs = λRF where ΔTs is the mean global surface temperature change and λ is the climate sensitivity parameter. This simple equation relates two equilibrium climate states in a linear way, and does not accurately characterize the overall climate response, which would include other factors such as humidity. Despite this drawback, RF is readily used, because it can be easily calculated and compared between chemicals. It is particularly utilized in the assessment of LLGHGs, for which it has been shown to be effective (3).

It is common for RFs to be determined in a different manner than indicated by the IPCC definition. One that is often used for LLGHGs is the instantaneous RF. For this determination, the temperature is fixed everywhere in the atmosphere, and the stratospheric temperatures are not allowed to re-adjust (3). For halocarbons, this typically results in an underestimate of the RF by approximately 5 – 9 % (4). Compounds, such as halocarbons, that absorb within the atmospheric window can absorb outgoing radiation in the stratosphere, thereby increasing the heating rate of the stratosphere. Thus, if the stratosphere were allowed to adjust to the new, warmer, temperatures, downward emission into the troposphere would intensify, and the calculated RF of the compound would appear to increase (4).

1.2.2 Radiative efficiency

The RF of a LLGHG is determined through its radiative properties, along with its atmospheric concentration. The radiative properties are assessed through the radiative efficiency (RE).

RF (W m-2) = C (ppb) * RE (W m-2 ppb-1) 4 Typically, REs are determined using measured infrared absorption cross-sections. Input of a known amount of the LLGHG into the atmosphere is then assumed and the change to radiative balance assessed. Although differences in cross-section measurements can impact the determined RE, typically the radiative transfer model selected to determine the RE has the greatest impact on the determined value (5). As mentioned above, the determination of an instantaneous versus an adjusted RF can lead to differences in RE of close to 10 %. Additionally, assumptions of atmospheric properties, such as the degree of cloudiness and mixing ratios of other LLGHGs can greatly impact the determined RF (4,5). Differences of up to 30 – 35 % were observed in RE determinations for the same compound between instantaneous, cloudy-sky and adjusted, clear-sky conditions (4). Unfortunately, it is often the case that these assumptions are not reported along with their determined RE value, making it difficult to compare RE values from different sources. Pinnock et al. (4) developed a simple method for estimating RE directly from the measured absorption cross-section, without the use of a model. This method has been shown to be of comparable accuracy to full radiative transfer models and more readily allows comparison between the REs of halocarbons.

1.2.3 Global warming potential

The RF metric does not give an indication of the time-scale of climate effects, such as the atmospheric lifetime of a LLGHG or changes in emissions. Determination of a global warming potential (GWP) is a method to partially account for these factors. Calculation of a GWP involves the integrated RF over a given time-scale of a 1 kg emission of a given LLGHG compared to that of a reference gas. Typically, a GWP is referenced to CO2, but for halocarbons, CFC-11 (CFCl3) is often used as a reference gas, yielding a halocarbon global warming potential (HGWP):

⎛ IF ⎞⎛τ M ⎞⎛ 1− exp(− t τ )/ ⎞ =⎜ X ⎟⎜ X CFC−11 ⎟⎜ X ⎟ HGWPX ⎜ ⎟⎜ ⎜ IF ⎟ τ M 1 exp(−− t τ )/ ⎟ ⎝ CFC−11 ⎠⎝ CFC−11 X ⎠⎝ CFC−11 ⎠ where IFX,IFCFC-11, MX, MCFC-11, τX, and τCFC-11 are the instantaneous radiative efficiencies, molecular weights, and atmospheric lifetimes of compound X and CFC-11, respectively, and t is the time horizon over which the forcing is integrated.

5 1.2.4 Importance of halogenated compounds as long-lived greenhouse gases

As of 2005, the total RF of LLGHGs is 2.63 W m-2, of which the halocarbons make up 0.337 W m-2 (Figure 1.2a) (3). The majority of the halocarbon RF is comprised of

a) b)

N2O PFCs 1% CH4 18% 6%

Halocarbons HCFCs 11% 13% CFCs SF 1% 80% 6 Chlorinated solvents 4%

CO2 HFCs 3% 63%

Figure 1.2: Radiative forcing of a) all long-lived greenhouse gases; and b) halocarbons (3).

Table 1.1: Lifetimes, radiative efficiencies and global warming potentials for greenhouse gases (3).

Radiative Global warming efficiency Lifetime potential (100 Common name Chemical formula (W m-2 ppb-1) (years) year timescale) -5 Carbon dioxide CO2 1.4 × 10 1 -4 Methane CH4 3.7 × 10 12 25 -3 Nitrous oxide N2O 3.03 × 10 114 298 CFC-11 CCl3F 0.25 45 4750 HFC-134a CH2FCF3 0.16 14 1430 HFC-125 CHF2CF3 0.23 29 3500 HFE-125 CHF2OCF3 0.44 136 14900 a a HG-1040x CHF2OCF2OC2F4OCHF2 1.02 6.3 1240 PFC-5-1-14 C6F14 0.49 3200 9300 Sulfur SF 0.52 3200 22800 hexafluoride 6 Nitrogen NF 0.21 740 17200 trifluoride 3 Trifluoromethyl sulfur SF5CF3 0.57 800 17700 pentafluoride a RE taken from Wallington et al. (6); GWP calculated using revised RE value. 6 chlorofluorocarbons (CFCs) and hydrochlorofluorocarbons (HCFCs), with smaller contributions from hydrofluorocarbons (HFCs), perfluorocarbons (PFCs), sulfur hexafluoride and chlorinated solvents (Figure 1.2b). The concentration of carbon dioxide is approximately 386 parts-per- million (ppm) (7), while halocarbons levels are in the range of parts-per-trillion (ppt) (3). The high impact of halocarbons, despite their low atmospheric concentrations is a result of their absorption within the atmospheric window and consequent high REs (Table 1.1). The halocarbons have REs many orders of magnitude higher than those of carbon dioxide, requiring low atmospheric concentrations to yield a high RF. Impacts of structure on RE are not well elucidated, though it is clear that compounds with more carbon-fluorine bonds tend to have higher REs.

1.3 International Agreements Affecting Long-Lived Greenhouse Gases

1.3.1 Kyoto Protocol

In the early 1990s, as climate change became recognized as a global environmental problem, many countries joined an international treaty, The United Nations Framework Convention on Climate Change. The Kyoto Protocol was approved by a number of nations as a legally binding addition to the treaty in 1997. Within the Kyoto Protocol are binding targets for the reduction of LLGHG emissions, averaging 5 % less than emissions of 1990 by 2012, though targets differ between countries. For example, target emissions for the European Union and Canada are 1990 – 8 % and 6 %, respectively (8). Six major LLGHG compound or classes are included in the protocol: carbon dioxide, methane, nitrous oxide, PFCs, HFCs and SF6 (8). An estimate of the effective LLGHG reduction affected by the Kyoto Protocol target can be determined by estimating business-as-usual emissions to 2012, combined with the target reductions. This yields a total reduction of about 2 Gt of equivalent carbon dioxide emissions -1 per year (GtCO2-eq yr ) (9).

1.3.2 Montreal Protocol

The Montreal Protocol on Substances that Deplete the Ozone Layer was ratified in 1987 for the purpose of protecting the ozone layer by limiting stratospheric concentrations of chlorine and bromine. Many of the compounds regulated under the Montreal Protocol are also powerful LLGHGs. Compounds such as CFCs, HCFCs and chlorinated solvents, which currently contribute 0.32 W m-2 or 12 % of the total RF of LLGHGs, are regulated by the Montreal 7 Protocol and subsequent amendments and adjustments. It has been estimated that without the Montreal Protocol, concentrations of CFCs and HCFCs would have currently contributed an -2 -1 additional 0.60 – 0.65 W m of RF, equivalent to 15 – 18 GtCO2-eq yr (9). Over the period of the Kyoto Protocol, it is estimated the Montreal Protocol has reduced LLGHGs equivalent to -1 about 8 GtCO2-eq yr , including offsets from halogenated CFC-replacement compounds and ozone recovery (9). This indicates the Montreal Protocol has had a greater impact on climate protection than the Kyoto Protocol.

1.3.3 Other agreements

Although the Montreal and Kyoto Protocols represent global initiatives that have worked to decrease climate change, other actions have also led to a reduction in climate impact from halocarbons. Many halocarbons, such as hydrofluoroethers (HFEs) and HFCs, are used in refrigeration and air conditioning as replacements for ozone-depleting chemicals. One of the simplest means of decreasing atmospheric concentrations of these chemical is through improved leakage rates. For example, in the Netherlands, containment schemes have reduced leakage rates from 30 % in the early 1990s to 4.8 % in 1999 (10). Similar programs have been implemented more recently by the United States Environmental Protection Agency (11). More strict action has been taken by the European Union, through a planned phase-out of HFC-134a from mobile air conditioning units (12). As a result, alternatives are being proposed that have lower REs, including polyfluorinated olefins and carbon dioxide (13).

1.4 Atmospheric Measurements of Halogenated Long-Lived Greenhouse Gases

1.4.1 Methods

Typical methods for measuring halogenated LLGHGs involve analysis by gas chromatography (GC). Collection methods differ from grab samples typically collected in vacuum flasks (eg. (14,15)) to in situ methods (eg. (16,17)). Samples collected in flasks are often extracted with cryofocusing and desorbed into a GC (14,15). Collection for in situ samples can be done a number of ways, including cryo trapping (17).

1.4.2 Observed levels

Concentrations of halogenated LLGHGs in the atmosphere are typically in the ppt range, with the most abundant compounds having concentrations in the hundreds of ppt. The 8 halogenated LLGHG with the highest current atmospheric concentration is CFC-12 (CCl2F2), with a global mean level 538 ppt (in 2005) (3). This high concentration also causes it to have the highest RF of the halogenated LLGHGs, at 0.17 W m-2 (3). The most radiatively potent gas -2 -1 yet detected in the atmosphere is SF5CF3, which has a RE of 0.57 W m ppb . However, the observed low concentration of 0.12 ppt, leads to an overall contribution of 6.8 × 10-5 W m-2 of RF to climate change (14).

1.4.3 Global networks and trends

For a number of the compounds, global networks exist to monitor variations in atmospheric levels. For example, the “Advanced Global Atmospheric Gases Experiment” (AGAGE) as well as the National Ocean and Atmospheric Administration’s “Halocarbon and other Atmospheric Trace Gases” (HATs) group have measured global long-term trends of

LLGHG concentrations. Both groups have been monitoring CFCs, HCFCs, HFCs and SF6 and have sampling stations set up all over the world (17-19). These sampling networks have

Figure 1.3: From IPCC 2007 report (3); Temporal evolution of the global average dry-air mole fractions (ppt) of the major halogen-containing LLGHGs. These are derived mainly using monthly mean measurements from the AGAGE and NOAA/GMD networks. For clarity, the two network values are averaged with equal weight when both are available. While differences exist, these network measurements agree reasonably well with each other (except for CCl4 (differences of 2 – 4% between networks) and HCFC-142b (differences of 3 – 6% between networks)), and with other measurements where available.

9 allowed the determination of accurate global trends of LLGHGs (Figure 1.3). The impact of the Montreal Protocol is evident in the plateaued or decreasing concentrations of chlorinated compounds, with the exception of the HCFCs, which continue to increase slowly due to their delayed phase-out relative to CFCs. However, concentrations of CFC-replacement compounds, such as HFCs, are increasing rapidly. As these compounds continue to increase and Montreal Protocol-regulated compounds stabilize or decrease, they will likely comprise a more significant fraction of the RF of halogenated LLGHGs.

1.5 Goals and Hypotheses

Halogenated LLGHGs are clearly important contributors to the total anthropogenic RF. We hypothesize that a number of radiatively important perfluorinated compounds are LLGHGs and have not yet been identified. Although each of these compounds is likely to be present in the atmosphere at low levels, the sum effect of these multiple new LLGHGs could have an impact on climate.

In Chapters 3 and 5, we identify two new halogenated LLGHGs. The relationship between chemical structure and RE for polyfluorinated chemicals is currently believed to be a function of the number of C-F bonds. We hypothesize in Chapter 4 that C-F bonds are not all equal and that the chemical environment of these bonds affects their ability to contribute to the RE of a molecule. The development of a structure-activity relationship to predict the RE from chemical structure for hydrofluoroethers will allow the intelligent design of future products to minimize climate impacts. Concern about climate change has prompted measures to reduce LLGHG usage. Poly- and perfluorinated olefins are suggested replacement compounds that require study before large-scale usage. In Chapter 5, we examine the fate of two perfluorinated olefins, demonstrating they are not LLGHGs, but could be the source of persistent perfluorinated carboxylic acids. 10 1.6 Sources Cited

(1) Ravishankara, A.R.; Solomon, S.; Turnipseed, A.A.; Warren, R.F. Atmospheric lifetimes of long-lived halogenated species. Science 1993, 259, 194-199.

(2) Prather, M.J.; Hsu, J. NF3, the greenhouse gas missing from Kyoto. Geophysical Research Letters 2008, 35, L12810.

(3) Forster, P.; Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R.; Fahey, D.W.; Haywood, J.; Lean, J.; Lowe, D.C.; Myhre, G.; Nganga, J.; Prinn, R.; Raga, G.; Schulz, M.; Van Dorland, R. In Climate Change 2007: The Physical Science Basis; Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L., Eds.; Cambridge University Press: Cambridge, United Kingdom, 2007.

(4) Pinnock, S.; Hurley, M.D.; Shine, K.P.; Wallington, T.J.; Smyth, T.J. Radiative forcing of climate by hydrochlorofluorocarbons and hydrofluorocarbons. Journal of Geophysical Research 1995, 100, 23227-23238.

(5) Christidis, N.; Hurley, M.D.; Pinnock, S.; Shine, K.P.; Wallington, T.J. Radiative forcing of climate change by CFC-11 and possible CFC replacements. Journal of Geophysical Research 1997, 102, 19,597-519,609.

(6) Wallington, T.J.; Hurley, M.D.; Nielsen, O.J. The radiative efficiency of HCF2OCF2OCF2CF2OCF2H (H-Galden 1040x) revisited. Atmospheric Environment 2009, 43, 4247-4249.

(7) Tans, P. 2009. NOAA/ESRL (www.esrl.noaa.gov/gmd/ccgg/trends/). Accessed September 30, 2009.

(8) Secretariat for the United Nations Framework Convention on Climate Change. 1998. Kyoto Protocol, United Nations Environmental Programme.

(9) Velders, G.J.M.; Andersen, S.O.; Daniel, J.S.; Fahey, D.W.; McFarland, M. The importance of the Montreal Protocol in protecting climate. Proceedings of the National Academy of Sciences 2007, 104, 4814-4819.

(10) Lindley, A.A.; McCulloch, A. Regulating to reduce emissions of fluorinated greenhouse gases. Journal of Fluorine Chemistry 2005, 126, 1457-1462.

(11) Mobile Air Conditioning Climate Protection Partnership. http://www.sae.org/news/releases/mobileac.htm, 2004. Ambitious Mobile Air Conditioning Climate Protection Goal Announced. Accessed October 1, 2009.

(12) The European Parliament and the Council of the European Union In Directive 2006/40/EC, 2006.

(13) Reisch, M.S. New refrigerant takes heat. Chemical and Engineering News 2008, 46, 35- 36. 11

(14) Sturges, W.T.; Wallington, T.J.; Hurley, M.D.; Shine, K.P.; Sihra, K.; Engel, A.; Oram, D.E.; Penkett, S.A.; Mulvaney, R.; Brenninkmeijer, C.A.M. A potent greenhouse gas identified in the atmosphere: SF5CF3. Science 2000, 289, 611-613.

(15) Montzka, S.A.; Hall, B.D.; Elkins, J.W. Accelerated increases observed for hydrochlorofluorocarbons since 2004 in the global atmosphere. Geophysical Research Letters 2009, 36, L03804, doi: 03810.01029/02008GL036475.

(16) Elkins, J.W.; Fahey, D.W.; Gilligan, J.M.; Dutton, G.S.; Baring, T.J.; Volk, C.M.; Dunn, R.E.; Myers, R.C.; Montzka, S.A.; Wamsley, P.R.; Hayden, A.H.; Butler, J.H.; Thompson, T.M.; Swanson, T.H.; Dlugokencky, E.J.; Novelli, P.C.; Hurst, D.F.; Lobert, J.M.; Ciciora, S.J.; McLaughlin, R.J.; Thompson, T.L.; Winkler, R.H.; Fraser, P.J.; Steele, L.P.; Lucarelli, M.P. Airborne gas chromatograph for in situ measurements of long-lived species in the upper troposphere and lower stratosphere. Geophysical Research Letters 1996, 23, 347-350.

(17) Miller, B.R.; Weiss, R.F.; Salameh, P.K.; Tanhua, T.; Greally, B.R.; Mühle, J.; Simmonds, P.G. Medusa: A sample preconcentration and GC/MS detector system for in situ measurements of atmospheric trace halocarbons, hydrocarbons, and sulfur compounds. Analytical Chemistry 2008, 80, 1536-1545.

(18) Montzka, S.A.; Myers, R.C.; Butler, J.H.; Elkins, J.W.; Cummings, S.O. Global tropospheric distribution and calibration scale of HCFC-22. Geophysical Research Letters 1993, 20, 703-706.

(19) Prinn, R.G.; Weiss, R.F.; Fraser, P.J.; Simmonds, P.G.; Cunnold, D.M.; Alyea, F.N.; O'Doherty, S.; Salameh, P.; Miller, B.R.; Huang, J.; Wang, R.H.J.; Hartley, D.E.; Harth, C.; Steele, L.P.; Sturrock, G.; Midgley, P.M.; McCulloch, A. A history of chemically and radiatively important gases in air deduced from ALE/GAGE/AGAGE. Journal of Geophysical Research 2000, 105, 17751-17792.

CHAPTER TWO

Atmospheric Formation of Perfluorinated Acids

Cora J. Young and Scott A. Mabury

Excerpts from the invited review: Cora J. Young and Scott A. Mabury. Atmospheric Perfluorinated Acid Precursors: Chemistry, Occurrence and Impacts. Rev. Environ. Contam. Toxicol. Submitted

12 12 2.1 Introduction

Interest in perfluorinated acids (PFAs) began over the past decade with the realization that PFAs were present in organisms (1), including humans (2,3). Further studies indicated long-chain congeners were bioaccumulative (4,5) and could magnify within a food chain (6). Global monitoring for PFAs revealed these compounds were found ubiquitously in ocean water (7,8), precipitation (9), biota (10) and human blood (11).

Perfluorocarboxylic acids (PFCAs) and perfluorosulfonic acids (PFSAs) are both strong acids and are likely to be ionized at environmental pH, suggesting they will be present primarily in the aqueous phase. Long-range transport through water occurs slowly, on the order of decades. The ubiquitous distribution of PFAs suggests a faster, atmospheric dissemination mechanism. In addition, not all PFAs observed in the environment have been commercially produced.

It is well-established that PFCAs can be formed in the atmosphere from the hydrolysis of acyl halides from compounds such as hydrochlorofluorocarbons (HCFCs) and hydrofluorocarbons (HFCs). However, in the past five years, a number of studies have examined the potential for other compounds to form acyl halides or PFCAs through more complex mechanisms. These mechanisms are of particular interest due to the potential to form bioaccumulative long-chain PFCAs.

PFA precursors are exclusively anthropogenic products, manufactured primarily for two major categories of products. The first is the surfactant and fluoropolymer industry, which includes compounds that are used in the synthesis of water- and oil-repellent compounds. PFA precursors are also marketed as CFC replacements, for uses including coolants, solvents and fire suppressors. PFA precursors are found in other industries, though not as frequently.

This chapter will discuss the known atmospheric formation mechanisms of PFAs, the chemistry of selected compounds that can form PFAs, atmospheric levels of precursors and the significance of atmospheric formation of PFAs.

13 Table 2.1: List of acronyms.

Acronym Name Structure

PFCA perfluorocarboxylic acid CF3(CF2)xC(O)OH TFA trifluoroacetic acid CF3C(O)OH PFPrA perfluoropropionic acid CF3CF2C(O)OH PFBA perfluorobutanoic acid CF3(CF2)2C(O)OH PFPeA perfluoropentanoic acid CF3(CF2)3C(O)OH PFHxA perfluorohexanoic acid CF3(CF2)4C(O)OH PFHeA perfluoroheptanoic acid CF3(CF2)5C(O)OH PFOA perfluorooctanoic acid CF3(CF2)6C(O)OH PFNA perfluorononanoic acid CF3(CF2)7C(O)OH PFDA perfluorodecanoic acid CF3(CF2)8C(O)OH PFUnA perfluoroundecanoic acid CF3(CF2)9C(O)OH PFDoA perfluorododecanoic acid CF3(CF2)10C(O)OH PFTrA perfluorotridecanoic acid CF3(CF2)11C(O)OH PFTA perfluorotetradecanoic acid CF3(CF2)12C(O)OH PFPA perfluoropentadecanoic acid CF3(CF2)13C(O)OH

PFSA perfluorosulfonic acid CF3(CF2)xSO3H - PFBS perfluorobutane sulfonate CF3(CF2)3SO3 - PFOS perfluorooctane sulfonate CF3(CF2)7SO3

HFO hydrofluoroolefin CxFyH(2x-y)

HFC hydrofluorocarbon CxFyH(2x+2-y) HFC-134a 1,1,1,2-tetrafluoroethane CF3CH2F HFC-125 1H-perfluoroethane CF3CF2H

HFC-329 1H-perfluoropropane CF3(CF2)2H

HFC-227 2H-perfluoropropane CF3CHFCF3

CFC chlorofluorocarbon CxFyCl(2x+2-y)

HCFC hydrochlorofluorocarbon CxFyClzH(2x+2-y-z) HCFC-124 1-chloro-1,2,2,2-tetrafluoroethane CF3(CF2)xCHFCl HCFC-123 1,1-dichloro-2,2,2-trifluoroethane CF3CHCl2

HCFC-225ca 1,1-dichloro-2,2,3,3,3-pentafluoropropane CF3CF2CHCl2

PFAL perfluorinated aldehyde CF3(CF2)xC(O)H PFAL hydrate perfluorinated aldehyde hydrate CF3(CF2)xCH(OH)2 FTAL fluorotelomer aldehyde CF3(CF2)xCH2C(O)H FTOH fluorotelomer alcohol CF3(CF2)xCH2CH2OH FTO fluorotelomer olefin CF3(CF2)xCH=CH2 FTI fluorotelomer iodide CF3(CF2)xCH2CH2I FTAc fluorotelomer acrylate CF3(CF2)xCH2CH2OC(O)CH=CH2 FTCA fluorotelomer carboxylic acid CF3(CF2)xCH2C(O)OH

NAFSA N-alkyl-perfluoroalkanesulfonamide CF3(CF2)xSO2N(H)CyH(2y+1) NAFBSA N-alkyl-perfluorobutanesulfonamide CF3(CF2)3SO2N(H)CyH(2y+1) NAFOSA N-alkyl-perfluorooctanesulfonamide CF3(CF2)7SO2N(H)CyH(2y+1)

NEtFBSA N-ethyl-perfluorobutanesulfonamide CF3(CF2)3SO2N(H)CH2CH3 NAFSE N-alkyl-perfluoroalkanesulfamido ethanol CF3(CF2)xSO2N(CyH(2y+1))CH2CH2OH NAFBSE N-alkyl-perfluorobutanesulfamido ethanol CF3(CF2)3SO2N(CyH(2y+1))CH2CH2OH NAFOSE N-alkyl-perfluorooctanesulfamido ethanol CF3(CF2)7SO2N(CyH(2y+1))CH2CH2OH

NMeFBSE N-methyl-perfluorobutanesulfamido ethanol CF3(CF2)3SO2N(CH3)CH2CH2OH

NMeFOSE N-methylperfluoroocanesulfamido ethanol CF3(CF2)7SO2N(CH3)CH2CH2OH

14

2.2 Mechanisms of atmospheric formation of perfluorinated acids (PFAs)

2.2.1 Perfluorocarboxylic acids (PFCAs)

2.2.1.1 Mechanisms for atmospheric formation of perfluoroacyl halides

2.2.1.1.1 Chemistry of Perfluoroacyl halides

The atmospheric fate of perfluoroacyl fluorides can be limited by photolysis or hydrolysis:

CF3(CF2)xC(O)F + hv → products (1)

CF3(CF2)xC(O)F + H2O(l) → products (2)

Absorption cross-sections for CF3C(O)F have been measured and a quantum yield for reaction (1) of approximately unity determined (12). Using this information, photolysis lifetimes of

3000 years (12) and 7500 years (13) were estimated. In contrast, hydrolysis of CF3C(O)F occurs on a short timescale. Atmospherically relevant hydrolysis measurements are difficult (14), and have led to vastly differing results. Kanakidou et al. (15) determined lifetimes for in- cloud loss by including time in the gas phase, probability of collision with a droplet, sticking coefficient and the actual hydrolysis using a model. Hydrolysis rates used in the model spanned over two orders of magnitude that encompassed measured values. Determined lifetimes ranged from 5.9 – 7.6 days, suggesting little sensitivity to the actual hydrolysis rate. A hydrolysis lifetime of 13.6 days has also been estimated by Wild et al. (13). Although no studies have focused on the fate of longer-chained perfluoroacyl fluorides, it is expected that hydrolysis will still dominate. Increasing perfluorinated chain length has been observed to lead to a slight red- shift of the absorption spectrum, as well as an increase in absorption cross-section (16). However, these differences are unlikely to have a large impact on the photolytic lifetime. This suggests the principal fate of CF3(CF2)xC(O)F is hydrolysis, which occurs with a lifetime on the order of days to weeks. The exclusive product of hydrolysis of perfluoroacyl fluorides is the corresponding PFCA.

Similarly, the fate of perfluoroacyl chlorides can be determined through photolysis and hydrolysis:

15

CF3(CF2)xC(O)Cl + hv → products (3)

CF3(CF2)xC(O)Cl + H2O(l) → products (4)

Rattigan et al. (12) and Meller and Moortgat (17) have measured the absorption cross-section of

CF3C(O)Cl, and also determined a quantum yield of dissociation of near unity. Utilizing these measurements, photolysis lifetimes of 33 days (12), 40 – 60 days (18), <86 days (19) and 56.4 days (13) were estimated. Kanakidou et al. (15) used various hydrolysis rates to determine lifetimes with respect to in-cloud hydrolysis, which ranged from 5.4 to 6.9 days. A similar estimate of 11.1 days was made by Wild et al. (13). Although hydrolysis proceeds faster than photolysis, photolytic degradation will likely play a role in the fate of perfluoroacyl chlorides.

Analysis of CF3C(O)Cl in a two-dimensional model demonstrated that hydrolysis was the major fate (19). Photolysis was observed to be less significant, with potential importance primarily in the upper troposphere or lower stratosphere, forming C(O)F2, HCl and HF. The degree to which photolysis is important may be dependent on perfluorinated chain length, as Chiappero et al.

(16) observed a slight red-shift on increasing from –CF3 to a longer perfluorinated chain, along with an increased absorption cross-section. For CF3C(O)Cl, the fraction reacting via hydrolysis was estimated as 0.6 (20). The relative importance of these two processes has not been studied for other chain-length perfluoroacyl chlorides, but it is clear both pathways will be important in the overall atmospheric fate. The sole product of hydrolysis of CF3(CF2)xC(O)Cl is the corresponding PFCA.

2.2.1.1.2 Mixed halide mechanism

Mixed halides of the structure CF3(CF2)xCHXY, where X is fluorine or chlorine and Y is chlorine or bromine, can form PFCAs. Abstraction of the hydrogen atom is followed by reaction with molecular oxygen and NO to yield an acyl radical. The radical decomposes, eliminating the Y atom and leaving a perfluoroacyl fluoride or chloride (Figure 2.1). This has

16

O H O OH O Y 2 Y CxF2x+1 Y CxF2x+1 CxF2x+1 X -H2O X X NO

-NO2

O O -Y

CxF2x+1 Y CxF2x+1 X X

Figure 2.1: Mechanism of perfluoroacyl fluoride formation from mixed halides, where X=F,Cl and Y=Cl,Br. been observed to occur for CF3CHFCl (21), CF3CHCl2 (22) and CF3CHClBr (23). Given that gas-phase radical mechanisms have been shown to be independent of perfluorinated chain length (24), this mechanism should apply to all volatile fluorinated chemicals of this class.

2.2.1.1.3 Perfluorinated radical mechanism

● Perfluorinated radicals (CF3(CF2)xCF2 , where x≥0) are formed from the atmospheric oxidation of many poly- and perfluorinated compounds. The most common environmental fate of these perfluorinated radicals is “unzipping” to yield a carbon-equivalent number of molecules of carbonyl fluoride (COF2) (14). However, a lower-yield degradation pathway initially described by Ellis et al. (25) can lead to the formation of PFCAs (Figure 2.2). The initial step in the atmospheric oxidation of perfluorinated radicals is reaction with oxygen to yield a perfluoro alkyl peroxy radical:

● ● CF3(CF2)xCF2 + O2 → CF3(CF2)xCF2OO (5)

This radical has a number of possible fates. The most common is reaction with the abundant radical, nitric oxide (NO), to yield a perfluoro alkoxy radical, which loses COF2 to yield a perfluorinated radical containing one fewer carbon atoms:

● ● CF3(CF2)xCF2OO + NO → CF3(CF2)xCF2O (6a)

● ● CF3(CF2)xCF2O → C(O)F2 + CF3(CF2)x-1CF2 (7)

17 The cycling of reactions (5), (6a) and (7) characterize the unzipping pathway that dominates the fate of perfluorinated radicals under typical atmospheric conditions. In remote areas where levels of NO are lower, reaction with peroxy radicals (ROO●) can become important. The reaction of a perfluoro peroxy radical with ROO● can yield a perfluoro alkoxy radical that goes on to react via reaction (7) and form COF2:

● ● ● ● CF3(CF2)xCF2OO + ROO → CF3(CF2)xCF2O + RO + O2 (6b)

If the ROO● contains an alpha hydrogen (RR’CHCOO●), the perfluoro peroxy radical can react to yield a perfluorinated alcohol:

● ● CF3(CF2)xCF2OO + RR’CHCOO → CF3(CF2)xCF2OH + R’C(O)R (6c)

Perfluorinated alcohols are inherently unstable, and quickly lose HF to yield perfluoro acyl fluorides by a mechanism that has not been entirely elucidated:

CF3(CF2)xCF2OH → CF3(CF2)xC(O)F + HF (8)

Perfluoro acyl fluorides readily hydrolyze, producing PFCAs.

CF3(CF2)xC(O)F + H2O → CF3(CF2)xC(O)OH (9)

Yields of PFCAs from perfluorinated radicals have not been directly studied, but will be related to the ratio of RR’CHOO● to other reactive species. In urban areas, where NO is abundant, it is unlikely that reaction (6c) will be significant. However, in remote, low-NOx environments, such as the Arctic and over the ocean, the production of PFCAs is likely to occur more readily (26).

Although the impact of perfluorinated chain length on the yield of PFCAs has not been explicitly determined, it has been shown that under NOx-free conditions, perfluorinated radicals ● (CF3(CF2)xCF2 , x = 7, 5, 3) form a homologous series of shorter-chain PFCAs (25). This occurs as a result of the removal of COF2 through the unzipping mechanism, and the formation of perfluorinated radicals of decreasing chain lengths. Thus, from the perfluorinated radical, ● CF3CF2 , trifluoroacetic acid (TFA) will be formed, while perfluorinated radicals of the ● structure CF3(CF2)xCF2 , where x≥1, will form PFCAs (CF3(CF2)nC(O)OH) of multiple chain lengths, from n = x to TFA (n = 0).

18

CF3(CF2)x

O2

O RR’CHOO OH CF3(CF2)x O CF3(CF2)x + RC(O)R’

NO Heterogeneous reaction or Overtone-induced photolysis x-1 O O CF3(CF2)x

CF3(CF2)x-1 F

Hydrolysis

CF (CF ) + C(O)F 3 2 x-1 2 O “unzipping” CF3(CF2)x-1 OH

Figure 2.2: Mechanism of PFCA formation via perfluorinated radicals (adapted from (25)).

2.2.1.2 Mechanisms for direct atmospheric formation of perfluorocarboxylic acids (PFCAs)

2.2.1.2.1 Perfluoroacyl peroxy radical mechanism

A mechanism that has been observed to lead directly to formation of PFCAs is via ● reaction of perfluoro acyl peroxy radicals (CF3(CF2)xC(O)OO , where x ≥ 0) with HO2 (25). ● Acetyl peroxy radicals (CH3C(O)OO ) are known to degrade by one of three channels, forming peracetic acid, acetic acid or acetoxy radicals (27):

● CH3C(O)OO + HO2 → CH3C(O)OOH + O2 (10a)

● CH3C(O)OO + HO2 → CH3C(O)OH + O3 (10b)

● ● CH3C(O)OO + HO2 → CH3C(O)O + O2 + OH (10c)

● Studies with CF3(CF2)xC(O)OO (where x = 0 – 3) demonstrated that perfluoroacyl peroxy radicals can undergo analogous reactions, yielding perfluorocarboxylic peracids, PFCAs or perfluoro acyl oxy radicals (28,29):

● CF3(CF2)xC (O)OO + HO2 → CF3(CF2)xC(O)OOH + O2 (11a)

● CF3(CF2)xC(O)OO + HO2 → CF3(CF2)xC(O)OH + O3 (11b)

19 ● ● CF3(CF2)xC(O)OO + HO2 → CF3(CF2)xC(O)O + O2 + OH (11c)

By analogy with the hydrocarbon reactions, these reactions are hypothesized to take place via a tetroxide intermediate (Figure 2.3) (17), though this intermediate has never been observed experimentally. In general, a trend of increased PFCA and decreased radical yields with increasing chain length was observed. The dominant products were dependent on chain length ● (Table 2.2), with radical propagation by reaction (11c) dominating for CF3C(O)OO , with a ● yield of 0.52 ± 0.05 and PFCA formation by reaction (11b) dominating for C4F9C(O)OO , with a yield of 0.73 ± 0.18 (30). The product of reaction (11a), the perfluoroperacid, was detected ● only for CF3C(O)OO (29). In reactions of longer chain length perfluoroacyl peroxy radicals, the corresponding perfluorocarboxylic peracid was below detection limits (28,29). These product distributions are in contrast to those observed for analogous hydrocarbon reactions. Specifically, in observations of acetyl peroxy radicals, the peracetic acid is a more dominant product (27).

O

O CF3(CF2)x O

+ HO2

O C O O CF3(CF2)x O O H

AB C

H O O O O H O C O O O O CF3(CF2)x O O H O CF3(CF2)x O O CF3(CF2)x O

-O2 -O3 -O2, - OH

O OH O

OH CF3(CF2)x O CF3(CF2)x O CF3(CF2)x O

Figure 2.3: Mechanism of PFCA formation via perfluoro acyl peroxy radicals (adapted from (28)). Under atmospheric conditions, the reactions of NOx with perfluoroacyl peroxy radicals will compete with the reactions described above:

20 ● ● CF3(CF2)xC(O)OO + NO → CF3(CF2)xC(O)O + NO2 (11d)

● CF3(CF2)xC(O)OO + NO2 → CF3(CF2)xC(O)O2NO2 (11e)

It is difficult to determine the fraction of perfluoroacyl peroxy radicals that will react with HO2 in the atmosphere. In urban areas, levels of NOx greatly exceed levels of HO2, leading to reactions (11d) and (11e) dominating over reaction with HO2. Reactions with HO2 are expected to be more prevalent in remote areas where HO2 and NOx levels are similar.

Table 2.2: Product yields of reaction (10) and (11) (adapted from (30))

Product Yields ● R RC(O)OOH + O2 RC(O)OH + O3 RC(O)O + O2 + OH Reference

CH3 0.40 ± 0.16 0.20 ± 0.08 0.40 ± 0.16 (27)

CF3 0.09 ± 0.04 0.39 ± 0.04 0.52 ± 0.05 (30)

CF3CF2 <0.12 0.50 ± 0.08 0.50 ± 0.08 (30)

CF3CF2CF2 <0.16 0.53 ± 0.11 0.47 ± 0.11 (30)

CF3CF2CF2CF2 <0.27 0.73 ± 0.18 0.27 ± 0.18 (30)

2.2.1.2.2 Perfluorinated aldehyde (PFAL) hydrate mechanism

The formation of PFCAs was observed from the atmospheric oxidation of perfluorinated aldehyde (PFAL) hydrates (31). Reaction of CF3CH(OH)2 with chlorine atoms in the absence of NOx was demonstrated to form TFA as the primary product in a yield indistinguishable from 100%.

CF3CH(OH)2 + Cl → CF3C(O)OH (12)

The formation of CF3C(O)OH was also seen through reaction with hydroxyl radicals, with a yield close to 100%, though this was not independently quantified. This reaction is expected to occur via rearrangement of the peroxy radical as shown in Figure 2.4. The production of PFCAs from PFAL hydrates in the presence of NOx has not been investigated. Mechanisms of gas- phase reactions have been observed to be independent of perfluorinated chain length (24), suggesting that PFAL hydrates of all chain lengths will form PFCAs.

21 H O

OH CxF2x+1 OH CxF2x+1 OH ● OH -HO2

-H2O O O● ● O C F OH 2 x 2x+1 CxF2x+1 O H OH OH

Figure 2.4: Proposed mechanism for the formation of PFCAs from PFAL hydrates.

2.2.2. Perfluorosulfonic acids (PFSAs)

Perfluorosulfonic acids (PFSAs, CF3(CF2)xSO3H) can be formed from atmospheric reactions of perfluorosulfonamides (PFSAms). Perfluorobutane sulfonate (PFBS) was observed from the atmospheric oxidation of N-methyl perfluorobutane sulfonamidoethanol (NMeFBSE,

C4F9SO2N(CH3)CH2CH2OH) (32). The proposed mechanism to account for this observation is shown in Figure 2.5 and involves the addition of hydroxyl radicals to the sulfone double bond. This results in a sulfonyl radical that can cleave the S-N bond to yield PFBS and a nitrogen- centred radical. Alternatively, the C-S bond can be broken, which yields a sulfamic acid and a perfluorinated radical that subsequently can decompose into PFCAs as discussed in section 2.1.1.3. The ratio of PFBS to PFCAs formed from NMeFBSE was approximately 1:10, suggesting higher probability of C-S bond breakage. The degradation of NMeFBSE is the sole observation of a PFSA from atmospheric oxidation. The class of PFSAms is diverse, with multiple perfluorinated chain lengths and substitutions on the amine group. It is possible to extrapolate the results from NMeFBSE to N-alkyl fluorosulfonamides (NAFSAs) with different perfluorinated chain lengths, as this has been shown to have little effect on atmospheric chemistry (24,33). However, it is difficult to speculate on how substitution at the amine might affect the possible production of PFSAs from the atmospheric oxidation of PFSAms.

22 O

C4F9 S N OH O

●OH

O OH

C4F9 S N OH O

O O N C4F9 S OH C4F9 HO S N OH OH O O

Figure 2.5: Proposed mechanism of PFBA formation from NMeFBSE (adapted from (32))

2.3 Chemistry of perfluorinated acid (PFA) precursors

A simplified scheme of the chemistry of perfluorinated acid precursors is shown in Figure 2.6. The chemistry of each precursor compound or class is discussed in detail in Young et al. (34). For brevity only selected fluorotelomer compounds and PFSAms are included here.

2.3.1 Fluorotelomer and related compounds

Numerous fluorotelomer compounds have been produced commercially, including fluorotelomer iodides (FTIs), fluorotelomer olefins (FTOs), fluorotelomer alcohols (FTOHs) and fluorotelomer acrylates (FTAcs). These compounds are named with the prefix x:y, where x is the number of perfluorinated carbons and y is the number of hydrogenated carbons. For example, 6:2 FTOH refers to C6F13CH2CH2OH. Among the compounds that are confirmed

23

FTOH FTAc

NAFSA FTAL

NAFSE PFAL FTO CF3(CF2)xH PFAL Hydrate

PFSA Perfluoroacyl Peroxy Perfluorinated Radical Radical

Perfluoroacyl CF (CF ) CHFCF PFCA Fluoride 3 2 x 3 CF3(CF2)xCHClBr

Perfluoroacyl HFO Chloride

CF3(CF2)xCHFCl

CF3(CF2)xCHCl2 CF3(CF2)xCHClOCHF2 CF (CF ) CH F 3 2 x 2

Figure 2.6: Transformation pathways of perfluorinated acid precursors. Grey compounds are commercially produced.

high-production volume chemicals are 4:2 to 18:2 FTIs, 4:2 FTO, 2:2 to 18:2 FTOHs and 6:2 to 14:2 FTAcs (35). Other compounds described below, including perfluorinated aldehydes and fluorotelomer aldehydes, are known products of the degradation of these commercially produced compounds and have, themselves, been intensively studied. For many of these compounds, chlorine radicals have been used as a proxy to understand reaction mechanisms for hydroxyl radicals, because they can be produced easily and cleanly in a smog chamber setup. Although concentrations of chlorine radicals are usually too low to impact the overall fate of organic compounds, they can shed light on hydroxyl radical mechanisms. In particular, when the mechanism for reaction with hydroxyl radical is hydrogen abstraction, chlorine atom reactions provide a reasonable proxy. Some concern has been raised about stability of radical products following reaction of chlorine atoms compared to hydroxyl radicals (36); however, in most cases chlorine atoms remain a useful tool for understanding atmospheric degradation pathways for organic compounds.

24 2.3.1.1 Perfluorinated aldehydes (PFALs)

Atmospheric lifetime

The atmospheric fate of PFALs involves contributions from a number of processes. The first is reaction with chlorine atoms or hydroxyl radicals:

CF3(CF2)xC(O)H + Cl → products (13)

CF3(CF2)xC(O)H + OH → products (14)

The kinetics of reaction (13) have been studied using relative rate methods (Table 2.3) (37-41) and fitting techniques (24). A slight increase in rate constant was observed with increasing perfluorinated chain length, although this did not exceed the error associated with the determinations. The reaction of PFALs of all chain lengths with chlorine atoms proceed with -12 3 -1 -1 rate constant, k(Cl + CF3(CF2)xC(O)H) of (1.8 – 2.8) × 10 cm molecule s . The kinetics of reaction (14) for CF3C(O)H have been studied in detail using a number of techniques (Table -13 2.4) (37,39,42,43). An IUPAC preferred value is given as k(OH + CF3C(O)H) = 5.7 × 10 cm3 molecule-1 s-1 at 298 K (44), which is the average of values reported by Scollard et al. (37), Sellevåg et al. (43) and Sulbaek Andersen et al. (39). Longer-chain PFALs have been studied using relative rate techniques for CF3(CF2)xC(O)H (x = 0 – 3) (39,40) (Table 2.4) and laser photolysis-laser-induced fluorescence for CF3(CF2)xC(O)H (x = 2,3) (45). Apart from one measurement for CF3C(O)H (42) that was excluded from the IUPAC-recommended value due to large uncertainty (44), rate constants for PFALs of different chain lengths are indistinguishable from one another, within error. Thus, the rate of reaction of PFALs with hydroxyl radicals is -12 3 -1 -1 k(OH + CF3(CF2)xC(O)H) = (0.58 – 1.1) × 10 cm molecule s .

A further atmospheric sink of PFALs is reaction with water to form stable hydrates:

CF3(CF2)xC(O)H + H2O CF3(CF2)xC(OH)2 (15)

An investigation into the likelihood of hydration of PFALs was undertaken by Sulbaek Andersen et al. (31). Observations indicated that reaction (15) did not occur under homogeneous gas-phase conditions, with an upper-limit for reaction calculated as -23 3 -1 -1 k(CF3C(O)H(g) + H2O(g)) < 2 × 10 cm molecule s . However, it was demonstrated that gas- phase PFALs were lost rapidly upon contact with liquid water and that some of the PFALs

25 formed PFAL hydrates. In humid smog chamber experiments, gas-phase CF3C(O)H slowly formed CF3C(OH)2. Rates between replicate experiments were inconsistent, which suggests the mechanism of hydration is heterogeneous. Because information on surfaces within the smog chamber is not available, these experiments cannot be quantitatively applied to the atmosphere, though they do shed some light on the nature of equilibrium between PFALs and PFAL hydrates.

Table 2.3: Summary of chlorine atom-initiated kinetics for fluorotelomer and related compounds.

Rate Constant Structure x (cm3 molecule-1 s-1) T (K) Method Reference 2.28 × 10-12 298 Relative rate (37) -12 (1.8 ± 0.4) × 10 295 Relative rate (38) 0 (1.90 ± 0.25) × 10-12 296 Fitting (24)

(1.85 ± 0.26) × 10-12 296 Relative rate (39)

(1.96 ± 0.28) × 10-12 296 Relative rate (40) PFAL 1 -12 CF3(CF2)xC(O)H (2.35 ± 0.42) × 10 296 Fitting (24) (2.56 ± 0.35) × 10-12 296 Fitting (24) 2 (2.03 ± 0.23) × 10-12 296 Relative rate (39) (2.48 ± 0.31) × 10-12 296 Fitting (24) 3 (2.34 ± 0.25) × 10-12 296 Relative rate (39) 5 (2.8 ± 0.7) × 10-12 298 Relative rate (46) (2.57 ± 0.04) × 10-11 298 Relative rate (47) 0 (1.81 ± 0.27) × 10-11 296 Relative rate (48) FTAL 3 (1.84 ± 0.30) × 10-11 296 Fitting (49) CF3(CF2)xCH2C(O)H 7 (1.9 ± 0.2) × 10-11 296 Relative rate (50) Very low-pressure 0 (6.78 ± 0.63) × 10-13 303 reactor mass (51) spectrometry (2.24 ± 0.04) × 10-11 296 Relative rate (47) (1.59 ± 0.20) × 10-11 296 Relative rate (48) FTOH 0 Very low-pressure CF3(CF2)xCH2CH2OH (1.90 ± 0.17) × 10-11 303 reactor mass (51) spectrometry 3,5,7 (1.61 ± 0.49) × 10-11 296 Relative rate (52)

FTO -11 0,1,3,5,7 (9.07 ± 1.08) × 10 296 Relative rate (53) CF3(CF2)xCH=CH2

FTAc -10 3 (2.21 ± 0.16) × 10 296 Relative rate (54) CF3(CF2)xCH2CH2OC(O)CHCH2

26

Table 2.4: Summary of hydroxyl radical-initiated kinetics for fluorotelomer and related compounds at atmospheric pressure. Rate Constant Structure x (cm3 molecule-1 s-1) T (K) Method Reference

-12 Discharge flow-resonance (1.1 ± 0.7) × 10 299 (42) fluorescence -13 Pulsed laser photolysis- 0 (6.5 ± 0.5) × 10 298 (37) resonance fluorescence (5.4 ± 1.2) × 10-13 298 Relative rate (37) -13 PFAL (4.80 ± 0.31) × 10 298 Relative rate (43) -13 CF3(CF2)xC(O)H 1 (5.26 ± 0.80) × 10 296 Relative rate (40) 0, 2, 3 (6.5 ± 1.2) × 10-13 296 Relative rate (39) -13 Pulsed laser photolysis-laser 2 (5.8 ± 0.6) × 10 298 (45) induced fluorescence

-13 Pulsed laser photolysis-laser 3 (6.1 ± 0.5) × 10 298 (45) induced fluorescence (3.30 ± 0.08) × 10-12 298 Relative rate (43)

-12 Pulsed laser photolysis-laser 0 (2.96 ± 0.04) × 10 298 (47) FTAL induced fluorescence CF (CF ) CH C(O)H -12 3 2 x 2 (2.57 ± 0.44) × 10 296 Relative rate (48) 7 (2.0 ± 0.4) × 10-12 296 Relative rate (50) 0,2,3 (1.02 ± 0.10) × 10-13 296 Relative rate (24) (1.08 ± 0.05) × 10-12 298 Relative rate (47)

-13 Pulsed laser photolysis-laser 0 (8.9 ± 0.3) × 10 298 (47) FTOH induced fluorescence -13 CF3(CF2)xCH2CH2OH (6.91 ± 0.91) × 10 296 Relative rate (48) 3,5,7 (1.07 ± 0.22) × 10-12 296 Relative rate (52) 5 (7.9 ± 0.8) × 10-13 298 Relative rate (47) -12 Flash photolysis-resonance 0 (1.54 ± 0.05) × 10 298 (55) fluorescence FTO -12 Pulsed laser photolysis-laser 5 (1.35 ± 0.11) × 10 298 (56) CF3(CF2)xCH=CH2 induced fluorescence 0,1,3,5,7 (1.36 ± 0.25) × 10-12 296 Relative rate (53)

FTAc -11 3 (1.13 ± 0.12) × 10 296 Relative rate (54) CF3(CF2)xCH2CH2OC(O)CHCH2

27 Finally, PFALs can also be subject to photolysis:

CF3(CF2)xC(O)H + hν → products (16)

A few studies have examined the potential of photolysis to limit the atmospheric lifetime of PFALs (16,43,45). Observations regarding the UV absorption cross-sections for

CF3(CF2)xC(O)H (x = 0 – 3) are in good agreement (Table 2.5). The measurements suggest a red-shift in absorption for PFALs in moving from CF3C(O)H to CF3CF2C(O)H, but no effect in further increasing perfluorinated chain length. There is also a clear increase in UV absorption cross-section with increasing perfluorinated chain length. Chiappero et al. (16) determined quantum yields of dissocation for PFALs of four perfluorinated chain lengths. A decrease in quantum yield of dissociation was observed with increasing perfluorinated carbons, presumably as a result of the greater degrees of freedom in the larger molecules. However, measurements of photolysis quantum yields and ultimate photolysis lifetimes are not consistent between studies. Chiappero et al. (16) determined photolysis lifetimes by interpolating quantum yields over the absorbance range. This, along with measured UV absorption cross-sections, was entered into the Tropospheric Ultraviolet-Visible (TUV) model to yield approximate lifetimes of <2 days for

CF3(CF2)xC(O)H (x = 1,2,3) and <6 days for x = 0. Similar lifetimes were determined by

Solignac et al. (46), with lifetimes on the order of hours to days measured for CF3(CF2)xC(O)H (x = 2,3,5) in the Euphore chamber in Valencia, Spain under natural sunlight conditions. There is significant discrepancy between these results and those of Sellevåg et al. (43), who determined an atmospheric lifetime for CF3C(O)H of >27 days, also using the Euphore chamber. The source of this inconsistency is not clear.

Although PFALs are much more reactive with chlorine atoms than hydroxyl radicals, typical concentrations of chlorine are too low to affect overall atmospheric fate (57). Assuming an average concentration of hydroxyl radicals of 1 × 106 molecules cm-3, the lifetime of PFALs with respect to reaction with atmospheric oxidants ranges from about 10 – 20 days. Of the processes that are currently understood for PFALs, photolysis dominates with the majority of studies suggesting degradation occurs over the timescale of less than 2 days. More studies are required to determine the relative importance of photolysis and formation of PFAL hydrates in the overall atmospheric fate of PFALs.

28 Products of atmospheric degradation

The products of atmospheric degradation have been well-studied for PFALs (16,28- 30,39-41). Reaction of PFALs with chlorine atoms and hydroxyl radicals is known to yield perfluoroacyl radicals:

● CF3(CF2)xC(O)H + Cl/OH → CF3(CF2)xC(O) + HCl/H2O (17)

Two major fates exist for the perfluoroacyl radical: reaction with oxygen and loss of carbon monoxide to yield a perfluorinated radical:

Table 2.5: Photolysis properties of perfluorinated aldehydes (PFALs) and fluorotelomer aldehydes (FTALs).

Quantum Yield of Dissociation Cross Section at with Measured Absorption Absorption Max Wavelength in Atmospheric x Max (nm) (x10-20 cm2 molecule-1) Brackets (nm) Lifetime Ref

0.79 (254); a 300 2.89 <6 days (16) 0.17 (308) 0 <0.02 (full range of 301 3.2 >27 daysb (43) atmospheric absorption) 1 308 5.86 0.81 (254) <2 daysa (16) 308 8.15 0.63 (254) <2 daysa (16) PFAL 0.023 (range of 2 b CF3(CF2)xC(O)H 309 8.1 ± 0.6 atmospheric 21 ± 10 hr (45) absorption) 0.60 (254); 308 9.49 <2 daysa (16) 0.08 (308) 0.029 (full 3 range of b 309 9.4 ± 0.7 15 ± 7 hr (45) atmospheric absorption) 5 46 ± 23 hrb (45) 0.74 (254); 290 3.52 <40 daysa (16) 0.04 (308) <0.04 (full 0 range of FTAL 292 3.845 >15 daysb (43) CF3(CF2)xCH2C(O)H atmospheric absorption) 300 13.3 0.55 (254) <20 daysa (16) 5 283 5.4 ± 0.4 (45) a determined using TUV model b measured in Euphore chamber

29 ● ● CF3(CF2)xC(O) + O2 → CF3(CF2)xC(O)OO (18a)

● ● CF3(CF2)xC(O) + M → CF3(CF2)x + CO + M (18b)

Reaction (18a) has been demonstrated to occur through the observation of PFCAs of equal carbon length formed from PFALs (see also Section 2.1.2.1.) (25,29,30,49). The prevalence of reactions (18a) and (18b) has been studied both experimentally (30,41) and computationally (36,58). As expected, experimental yields of reaction (18a) increase with increasing concentration of oxygen (30). Product yields of reaction (18b) (Table 2.6) in air were shown to increase with increasing perfluorinated chain length (30). Computational studies demonstrated that increasing perfluorinated chain length weakens the C-CO bond, leading to an increased prevalence of the decomposition reaction (18b) (36). Regardless of the mechanism, chlorine atom- or hydroxyl radical-initiated reaction of PFALs can lead to the formation of PFCAs, whether through reaction (18a) and the perfluoroacyl peroxy mechanism or through reaction (18b) and the perfluorinated radical mechanism.

Products of photolysis of PFALs were studied by Chiappero et al. (16). Two products were observed from photolysis reactions of PFALs (CF3(CF2)xC(O)H, x = 0 – 3) at 254 nm: ● CF3(CF2)x radicals and CF3(CF2)xH, suggesting occurrence of the following reactions:

● CF3(CF2)xC(O)H + hν → CF3(CF2)x + HCO (19a)

CF3(CF2)xC(O)H + hν → CF3(CF2)xH + CO (19b)

Both reactions appeared to occur with approximately equal prevalence at 254 nm. However, actinic radiation is not available at 254 nm in the lower atmosphere. Experiments at 308 nm, which is close to the absorption maximum for PFALs and within the actinic spectrum, did not observe the formation of CF3(CF2)xH from photolysis of PFALs (CF3(CF2)xC(O)H, x = 0 – 3).

Irradiation experiments CF3C(O)H at 308 nm used nitric oxide (NO) as a radical scavenger to prevent radical cross-reactions. Under these conditions, a 98 ± 7% yield for CF3NO was formed from the reaction of CF3 with NO, suggesting reaction (19a) is prevalent at this wavelength.

The dominant product of PFAL degradation is likely the photolysis product. Photolysis of PFALs yields primarily the corresponding perfluorinated radical, which has been shown to form PFCAs of all chain-lengths.

30 Table 2.6: Fate of perfluoroacyl radicals.

Product Yields RC(O) → R + RC(O) + O2 → R CO RC(O)OO Ref CF 0.02 0.98 (58) 3 0.02 0.98 (30) 0.52 0.48 (30) CF CF 3 2 0.61 0.39 (46) CF3CF2CF2 0.81 0.19 (30) CF3CF2CF2CF2 0.89 0.11 (30)

2.3.1.2 Fluorotelomer aldehydes (FTALs)

Atmospheric lifetime

The atmospheric lifetime of FTALs can be limited by reaction with chlorine atoms and hydroxyl radicals:

CF3(CF2)xCH2C(O)H + Cl → products (20)

CF3(CF2)xCH2C(O)H + OH → products (21)

Reaction (20) has been measured by relative rate techniques (47,48,50) and determined through fitting product formation and degradation curves (49) (Table 2.3). There is good agreement between the values, with the exception of that measured by Kelly et al. (47). Hurley et al. (48) suggest this discrepancy may be due to the use of a single reference compound and errors associated with the reference rate. In addition, there does not appear to be a chain-length effect on the rate of the reaction. A final rate constant can be determined from the values in Table 2.3, excluding the outlying value of Kelly et al., and including extremes of the individual -11 3 -1 -1 measurements of k(Cl + CF3(CF2)xCH2C(O)H) = (1.85 ± 0.31) × 10 cm molecule s . The kinetics of reaction (21) have been studied by relative rate techniques (43,48,50) and pulsed laser photolysis-laser induced fluorescence techniques (47) (Table 2.4). Within the three measurements of CF3CH2C(O)H, there is significant discrepancy, the source of which is unclear. The quoted rate for longer perfluorinated-chain FTALs is somewhat lower, though within error of some of the rates for CF3CH2C(O)H, suggesting that, as with other polyfluorinated compounds, perfluorinated chain-length does not impact kinetics.

The photolysis of FTALs may also contribute to their atmospheric fate:

31

CF3(CF2)xCH2C(O)H + hv → products (22)

Three studies have examined the potential for FTALs to undergo photolysis (16,43,45) (Table

2.5). Those examining the fate of CF3CH2C(O)H are in agreement, both in regards to the absorption cross-section and the quantum yield of dissociation. Chiappero et al. (16) used the

TUV model to determine a lifetime for CF3CH2C(O)H of less than 40 days, while Sellevåg et al. (43) measured a lifetime of greater than 15 days in the Euphore chamber. The photolysis of

CF3(CF2)5CH2C(O)H has also been examined, where measurements show poor agreement. It can be seen from Table 2.5 that the absorption cross-section and location of maximum absorption vary significantly between the two studies. Chiappero et al. (16) determined a quantum yield, but at a wavelength well into the uv and below the absorption maximum for the chromophore. Nevertheless, they used the TUV model to estimate a lifetime with respect to photolysis of less than 20 days.

Atmospheric concentrations of chlorine atoms are too small to impact the atmospheric fate of FTALs (57). Assuming an average concentration of hydroxyl radicals of 9.4 × 105 molecules cm-3, Sellevåg et al. (43) determined a lifetime of approximately 4 days for reaction of CF3CH2C(O)H. Although there are discrepancies within the limited available data concerning the atmospheric fate of FTALs, it appears as though the atmospheric lifetime will be determined by reaction with hydroxyl radical and will be on the order of a few days.

Products of atmospheric degradation

There have been few studies examining the products of atmospheric reaction of FTALs.

The products of reaction (20) have been studied for CF3CH2C(O)H and CF3(CF2)3CH2C(O)H under smog chamber conditions in the absence of NOx (48,49). A dominant primary product for both FTALs was the corresponding PFAL, with a yield of 46 ± 3 % in the case of

CF3(CF2)3C(O)H formed from CF3(CF2)3CH2C(O)H. The other easily identifiable primary product was the corresponding fluorotelomer carboxylic acid (FTCA, CF3(CF2)xCH2C(O)OH). Another primary product was formed, for which standards were not available, but this was determined to be the corresponding fluorotelomer carboxylic peracid

(CF3(CF2)xCH2C(O)OOH). There have not been any specific studies of reaction (20) in the presence of NOx, but indirect evidence suggests the mechanism and products are different under these conditions (59).

32 The atmospheric fate of FTALs is expected to be dominated by reaction with hydroxyl radical (reaction (21)), but the products of this reaction have not been studied. Since the reaction is initiated by hydrogen abstraction, the productions of chlorine atom-initiated oxidation are a reasonable proxy for those expected from reaction with hydroxyl radicals. Thus, under low-NOx conditions, productions of reaction (21) would consist of the corresponding PFAL, FTCA and fluorotelomer carboxylic peracid. The products in the presence of NOx require further study.

The products of photolysis of FTALs (reaction (22)) have not been studied, but by analogy to PFALs and hydrocarbon aldehydes, it is likely that photolysis occurs via C-C bond scission:

● CF3(CF2)xCH2C(O)H + hv → CF3(CF2)xC( )H2 + HCO (23)

The first stable product of the radical formed in reaction (23) would be the corresponding PFAL.

Through photolysis and hydroxyl radical-initiated atmospheric oxidation, FTALs form PFALs, which are known PFCA precursors. Consequently, FTALs can be considered sources of PFCAs.

2.3.1.3 Fluorotelomer alcohols (FTOHs)

Atmospheric lifetime

The atmospheric fate of fluorotelomer alcohols (CF3(CF2)xCH2CH2OH, FTOHs) has been the subject of numerous studies. FTOHs can react with atmospheric oxidants:

CF3(CF2)xCH2CH2OH + Cl → products (24)

CF3(CF2)xCH2CH2OH + OH → products (25)

Reaction (24) has been studied for CF3CH2CH2OH using relative rate techniques (47,48) and has been measured directly using very low-pressure reactor mass spectrometry (51). As shown in Table 2.3, the rate of Kelly et al. (47) appears to be unusually high and reasons for this discrepancy are unclear. Rates for longer-chained FTOHs (CF3(CF2)xCH2CH2OH, x = 3,5,7) were studied using relative rate techniques (52). Apart from the anomalous value, measurements of the reaction of FTOHs with chlorine atoms are in good agreement and appear to be independent of chain-length. Thus, the rate of reaction of FTOHs with chlorine atoms is

33 -11 3 -1 -1 k(Cl + CF3(CF2)xCH2CH2OH) = (1.59 – 1.90) × 10 cm molecule s . The kinetics of reaction (25) for CF3CH2CH2OH have been studied using relative rate techniques (47,48) and pulsed laser photolysis-laser induced fluorescence (47). As shown in Table 2.3, values are reasonably similar, but are not within experimental error of each other. The cause of the inconsistency is not clear. Kinetics for longer-chained FTOHs (CF3(CF2)xCH2CH2OH, x = 3,5,7) have been determined using relative rates (47,52) and studies are in agreement. Chain length does not appear to affect the kinetics of reaction with hydroxyl radicals, although measurements are not all within experimental error. Taking into account the values in Table

2.4, the rate of reaction of FTOHs with hydroxyl radicals is k(OH + CF3(CF2)xCH2CH2OH) = (0.69 – 1.07) × 10-12 cm3 molecule-1 s-1.

The low abundance of chlorine atoms in the atmosphere (57) suggests that reaction with chlorine will not be a dominant sink for FTOHs. Ellis et al. (52) scaled the rate of the FTOH reaction with hydroxyl radicals to the lifetime of CH3CCl3 and determined an overall rate of approximately 20 days. Using a globally averaged concentration of hydroxyl radicals of 1 × 106 molecules cm-3, a lifetime for FTOHs of about 12 days was estimated, which may be underestimated due to the anomalously high rate constant (47). Saturated compounds, such as

FTOHs, do not react appreciably with ozone. Hurley et al. (48) confirmed that CF3CH2CH2OH was unreactive with ozone and determined a minimum lifetime of 5900 days, assuming a background ozone concentration of 40 ppb. Photolysis of FTOHs is also unlikely, because alcohols typically do not absorb within the actinic spectrum. This was confirmed by determining the UV absorption of FTOHs using computational methods (60), where absorption was demonstrated to occur in the region 140 – 175 nm, with no absorption in the actinic region. In addition, potential for wet and dry deposition were examined using simple expressions and 6 lifetimes for CF3(CF2)7CH2CH2OH of 2.5 × 10 and 8.4 years were determined, respectively (52). Consequently, the atmospheric fate of FTOHs is limited by reaction with hydroxyl radicals, with lifetimes on the order of 10 – 20 days.

Products of atmospheric degradation

The observation that chlorine atom-initiated atmospheric oxidation of fluorotelomer alcohols (FTOHs) results in PFCA production (25) led to in-depth research into the underlying mechanism, as well as the overall atmospheric fate of FTOHs and intermediate species, much of

34 which has been discussed in previous sections. As a consequence, the products of atmospheric oxidation of FTOHs are well understood.

A number of studies have observed the corresponding FTAL to be the sole primary product of reaction of FTOHs of various chain-lengths with chlorine atoms in the absence of NOx (25,47-51). The same trend has also been observed for reaction of FTOHs with hydroxyl radicals in the absence of NOx (47). The formation of the FTAL suggests that hydrogen abstraction occurs at the carbon adjacent to the alcohol group. This is corroborated by structure- activity relationship calculations (52), which indicate that greater than 90% of reaction occurs at the alcohol-adjacent group and by computational studies that show the weakest C-H bonds are adjacent to the alcohol (51).

The formation of FTALs from FTOHs has been suggested to occur through the initial formation of an α-hydroxyl alkyl radical that reacts with molecular oxygen to form a chemically excited peroxy radical:

● CF3(CF2)xCH2CH2OH + Cl/OH → CF3(CF2)xCH2C( )HOH + HCl/H2O (26)

● ● CF3(CF2)xCH2C( )HOH + O2 → [CF3(CF2)xCH2C(OO )HOH]* (27)

The radical product of reaction (27) can either collisionally deactivate or can rapidly decompose:

● ● [CF3(CF2)xCH2C(OO )HOH]* + M → CF3(CF2)xCH2C(OO )HOH + M (28a)

● [CF3(CF2)xCH2C(OO )HOH]* → CF3(CF2)xCH2C(O)H + HO2 (28b)

Decomposition typically occurs on a timescale faster than deactivation and yields the corresponding FTAL and HO2. Collisionally deactivated peroxy radicals have two possible fates: decomposition to yield FTAL and HO2 or reaction with NOx:

● CF3(CF2)xCH2C(OO )HOH → CF3(CF2)xCH2C(O)H + HO2 (29a)

● CF3(CF2)xCH2C(OO )HOH + NO → products (29b)

In the absence of NOx, reaction (29b) does not occur and the sole product is the FTAL. In smog chamber experiments performed in the presence of abundant NOx, FTAL was not the exclusive primary product (48,59). However, when rates were scaled to more realistic NOx

35 concentrations of [NO] < 40 ppb, which encompass all but highly polluted urban conditions, it was determined that reaction (29b) would be of minimal importance. This was supported by further studies (47) that observed FTAL as the sole product of chlorine atom and hydroxyl radical-initiated atmospheric oxidation of FTOHs in the presence and absence of NOx. Thus, the atmospheric fate of FTOHs can be understood as atmospheric oxidation through reactions (28b) and (28a) followed by (29a) to form the corresponding FTAL:

CF3(CF2)xCH2CH2OH + Cl/OH → CF3(CF2)xCH2C(O)H (30)

The FTALs are known precursors of PFCAs, which suggests the formation of PFCAs from FTOHs. In addition, PFCAs have been directly observed in smog chamber experiments of the atmospheric oxidation of FTOHs, both by in situ FTIR (49) and offline sampling (25), demonstrating their importance as precursors.

2.3.1.4 Fluorotelomer olefins (FTOs)

Atmospheric lifetime

The atmospheric reactivity of fluorotelomer olefins (CF3(CF2)xCH=CH2, FTOs) has been well documented. The lifetime of these compounds is limited by reaction with atmospheric oxidants:

CF3(CF2)xCH=CH2 + Cl → products (31)

CF3(CF2)xCH=CH2 + OH → products (32)

CF3(CF2)xCH=CH2 + O3 → products (33)

Rate constants for reaction (31) were determined for various FTO chain-lengths

(CF3(CF2)xCH=CH2, x = 0,1,3,5,7) using smog chamber techniques at 296 K (53). There was no discernable effect of perfluorinated chain-length on chlorine atom reactivity and a final rate -11 3 constant for FTOs was cited as k(Cl + CF3(CF2)xCH=CH2) = (9.07 ± 1.08) × 10 cm molecule-1 s-1 (Table 2.3). Studies of chlorine atom-initiated oxidation were also performed at low pressures and various temperatures using pulsed laser photolysis-resonance fluorescence (56). Rate constants were shown to be independent of pressure and chain length for

CF3(CF2)xCH=CH2 (x = 3,5) and were in good agreement with those measured at atmospheric pressure.

36 Reaction (32) has been the subject of three separate studies (Table 2.4) (53,55,56). A study under smog chamber conditions at 296 K demonstrated that perfluorinated chain-length did not affect the rate for reaction of FTOs with hydroxyl radicals and a final rate of k(OH + -12 3 -1 -1 CF3(CF2)xCH=CH2) = (1.36 ± 0.25) × 10 cm molecule s was determined (53). This rate is in good agreement with a measurement made by pulsed laser photolysis-laser induced fluorescence for CF3(CF2)5CH=CH2 at atmospheric pressure and 298 K (56). A measurement for CF3CH=CH2 under flash photolysis-resonance fluorescence conditions (55) is slightly higher than rate constants from the other two studies and may indicate a slight chain-length effect on moving from CF3CH=CH2 to FTOs with longer perfluorinated tails.

Studies of reaction (33) showed the rate of CF3CH=CH2 with ozone was much faster than for CF3CF2CH=CH2, but no further decrease in rate was observed with increasing chain -19 3 length. Rate constants were determined as k(O3 + CF3CH=CH2) = (3.5 ± 0.25) × 10 cm -1 -1 -19 3 -1 -1 molecule s and k(O3 + CF3(CF2)xCH=CH2, x ≥ 1) = (2.0 ± 0.4) × 10 cm molecule s .

Atmospheric concentrations of chlorine atoms are too low (57) to impact the fate of FTOs. Assuming a global average hydroxyl radical concentration of 1 × 106 molecules cm-3, the atmospheric lifetime of FTOs with respect to reaction with hydroxyl radicals is about 8.5 days. Using the global background concentration of ozone of 35 ppb, the lifetime of FTOs

(CF3(CF2)xCH=CH2, x ≥ 1) for reaction with ozone is about 70 days. Thus, the overall atmospheric lifetime of FTOs is 7.6 days, with hydroxyl radical reactions contributing 90% and ozone reactions contributing about 10% (53).

Products of atmospheric degradation

Two detailed studies have examined the products of FTO degradation via reactions (33) and (32) (56,61).

Initial studies of FTO reaction with chlorine atoms demonstrated the formation of a carbonyl-containing primary product that was tentatively identified as CF3(CF2)xC(O)CH2Cl (x

= 3,5) (56). The production of CF3C(O)CH2Cl was confirmed from CF3CH=CH2, and the yield was quantified as 70 ± 5 % in the absence of NOx (61). A small amount of CF3C(O)H was also observed as a primary product with a yield of 6.2 ± 0.5 %. Using the yields of these two primary products, which occur due to reaction at the internal carbon atom, it was determined that chlorine atom addition occurs 74 % to the terminal and 26 % to the internal carbon atoms

37

(61). Secondary reaction products, including COF2 and CO were observed and attributed to degradation of the primary carbonyl products (56). In the presence of NOx, chlorine atom- initiated reaction was observed to form additional nitrogen-containing species. In short experiments with CF3CH=CH2, the production of two primary species, identified as a nitrite

(CF3C(ONO)HCH2Cl) and nitrate (CF3C(ONO2)HCH2Cl), was observed (61). Experiments of greater duration with the longer-chain FTOs (CF3(CF2)xC(O)CH2Cl (x = 3,5)) also led to the production of PAN-like species, CF3(CF2)xC(O)O2NO2, which are likely formed as a secondary products from degradation of the primary carbonyl species (56). However, NOx concentrations used in the studies were much higher than typical atmospheric concentrations and resulting yields of NOx-containing compounds were likely inflated.

The reaction of FTO with hydroxyl radicals has been shown to yield the corresponding

PFAL as the primary product for CF3(CF2)xC=CH2 (x = 0,3,5) (56,61). A yield of 88 ± 9 % was determined for PFALs from FTOs (CF3(CF2)xC=CH2, x = 0,3) (61). The only other carbon- containing species observed was trace amounts of COF2, which was attributed to the degradation of PFALs. The production of PFALs from FTOs in a yield close to unity suggests the fate of FTOs with hydroxyl radicals can be represented as:

CF3(CF2)xCH=CH2 + OH → CF3(CF2)xC(O)H (34)

There have not been any studies that have examined the fate of FTOs with ozone (reaction (33)). However, given that atmospheric oxidation of FTOs is dominated by reaction with hydroxyl radicals, the principal product of FTO degradation is the corresponding PFAL. As such, the atmospheric oxidation of FTOs could be a source of PFCAs.

2.3.1.5 Fluorotelomer acrylate (FTAc)

Atmospheric lifetime

The atmospheric fate of fluorotelomer acrylates (FTAc,

CF3(CF2)xCH2CH2OC(O)CH=CH2) has been examined in one study, focusing on a representative compound, C4F9CH2CH2OC(O)CH=CH2 (54). The likely fate of this chemical is reaction with chlorine atoms, hydroxyl radicals or ozone:

C4F9CH2CH2OC(O)CH=CH2 + Cl → products (35)

C4F9CH2CH2OC(O)CH=CH2 + OH → products (36)

38

C4F9CH2CH2OC(O)CH=CH2 + O3 → products (37)

Rate constants for reactions (35) and (36) were determined using competition kinetics at 296 K

(54). The rate for reaction with chlorine is k(Cl + C4F9CH2CH2OC(O)CH=CH2) = (2.21 ± 0.16) × 10-10 cm3 molecule-1 s-1 (Table 2.3), while that for reaction with hydroxyl radical is k(OH + -11 3 -1 -1 C4F9CH2CH2OC(O)CH=CH2) = (1.13 ± 0.12) × 10 cm molecule s (Table 2.4) (54). Chlorine atoms and hydroxyl radicals are presumed to react primarily at the double bond, and this is consistent with the observed rate constants. Rate constants for reaction (36) were indistinguishable within experimental error from non-fluorinated acrylates, such as methyl acrylate, suggesting the fluorinated tail does not impact the reactivity. Measurements were not undertaken to determine the rate of reaction with ozone, but by comparison to methyl -18 methacrylate, reaction (37) can be estimated as k(O3 + C4F9CH2CH2OC(O)CH=CH2) = 1 x 10 cm3 molecule-1 s-1. Kinetics of reaction for fluorinated species have been shown to be independent of perfluorinated chain length (24,33), so the rate constants determined here for

C4F9CH2CH2OC(O)CH=CH2 should apply to FTAcs of all chain lengths.

The concentrations of chlorine atoms in the atmosphere are not sufficient to impact the lifetime of FTAcs (57). Assuming a hydroxyl radical concentration of 1 x 106 molecules cm-3, an atmospheric lifetime of FTAcs of approximately one day was determined. Assuming a concentration of 50 ppb for ozone yields a lifetime of approximately 9 days. Thus, the atmospheric lifetime of FTAcs are limited by reaction with hydroxyl radical and are on the order of a day.

Products of atmospheric degradation

Products of chlorine atom-initiated degradation of C4F9CH2CH2OC(O)CH=CH2 were studied using FTIR under smog chamber conditions in the presence and absence of NOx (54).

Evidence of the production of C4F9CH2C(O)H as a primary product was observed, with yields of 10% and 18% in the presence and absence of NOx, respectively. Although reaction of chlorine atoms occurs primarily by addition at the double bond, this suggests the occurrence of hydrogen abstraction as an additional loss mechanism:

● C4F9CH2CH2OC(O)CH=CH2 + Cl → C4F9CH2C(O)H + OC(O)CH=CH2 + HCl (38)

39 Other FTIR features that could not be explicitly identified were attributed to a single product, C4F9CH2CH2OC(O)C(O)HCH2Cl, formed from the addition of chlorine to the double bond:

C4F9CH2CH2OC(O)CH=CH2 + Cl → C4F9CH2CH2OC(O)C(O)HCH2Cl (39)

A single product resulted from reaction of C4F9CH2CH2OC(O)CH=CH2 with hydroxyl radicals corresponding to the fluorotelomer glyoxalate (FTGly, C4F9CH2CH2OC(O)CHO). Perfluorinated chain length is not expected to affect the products of atmospheric oxidation, so the reaction of FTAcs with hydroxyl radicals can be described by the following reaction:

C4F9CH2CH2OC(O)CH=CH2 + OH → C4F9CH2CH2OC(O)CHO + HCHO (40)

The atmospheric fate of FTGlys was not studied explicitly, but was suggested to be dominated by photolysis. The chromophores in FTGlys are similar to those in CH3C(O)C(O)H, which has an atmospheric lifetime with respect to photolysis of about 3 hours at a latitude of 45º. Thus,

FTGlys likely degrade via photolysis on a similar timescale. By analogy to CH3C(O)C(O)H, photolysis is expected to break the carbon-carbon bond between the dicarbonyl

(CF3(CF2)xCH2CH2OC(O)–C(O)H), which would ultimately lead to the formation of FTAL:

CF3(CF2)xCH2CH2OC(O)C(O)H + hν → CF3(CF2)xCH2C(O)H (41)

The data suggests the fluorotelomer aldehyde, CF3(CF2)xCH2C(O)H, is formed as the secondary product of atmospheric oxidation of FTAcs (54). The FTAL then reacts to form PFCAs in small yields. Consequently, FTAcs are potential sources of PFCAs to the environment.

2.3.2 Perfluoroalkanesulfonamides

Many perfluoroalkanesulfonamides with varied structures were used commercially. Prior to 2001, eight-carbon congeners were the dominant products, while more recently four- carbon congeners have been primarily produced.

2.3.2.1 N-alkyl-perfluoroalkanesulfonamides (NAFSA)

Atmospheric lifetime

The atmospheric lifetime of a representative compound for this class, N-ethyl perfluorobutanesulfonamide (NEtFBSA, C4F9SO2N(H)CH2CH3) has been studied by Martin et

40 al. (62). The fate of this compound is governed by reaction with chlorine atoms and hydroxyl radicals:

C4F9SO2N(H)CH2CH3 + Cl → products (42)

C4F9SO2N(H)CH2CH3 + OH → products (43)

Kinetics of reactions (42) and (43) were examined using competition kinetics, with FTIR and offline LC-MS-MS detection. The rate constant for reaction with chlorine atoms was -12 3 -1 -1 determined as k(Cl + C4F9SO2N(H)CH2CH3) = (8.37 ± 1.44) × 10 cm molecule s and the rate constant for reaction with hydroxyl radicals was measured to be k(OH + -13 3 -1 -1 C4F9SO2N(H)CH2CH3) = (3.74 ± 0.77) × 10 cm molecule s (Table 2.7) (62).

Since atmospheric concentrations of chlorine atoms are very low (57), the gas-phase atmospheric lifetime is expected to be dominated by reaction with hydroxyl radicals. Reactions are also likely independent of perfluorinated chain length (24,33). Assuming an average OH concentration of 1 × 106 molecules cm-3, the lifetime for NAFSAs are 20 – 50 days, depending on location, time of year and temperature. It is important to note that rate constants and lifetimes determined in this study are for gas-phase reactions only. The low volatility of these compounds suggests that gas-particle partitioning could play a role in the overall atmospheric fate of NAFSAs; however, the degree of partitioning remains unclear (63,64).

Products of atmospheric degradation

The products of the reaction of NEtFBSA with chlorine atoms have been studied experimentally (62) and theoretically (65). Experiments demonstrated the formation of two identifiable primary stable products, formed from reaction at the secondary (reaction (44)) and primary (reaction (45)) carbons on the ethane moiety (62):

C4F9SO2N(H)CH2CH3 + Cl → C4F9SO2N(H)C(O)CH3 (44)

C4F9SO2N(H)CH2CH3 + Cl → C4F9SO2N(H)CH2C(O)H (45)

Two additional partially resolved peaks were observed in LC-MS-MS chromatograms of reacted NEtFBSA. These were assumed to be additional primary products, but were not unambiguously identified. Theoretical studies indicate these products arise due to hydrogen abstraction from the nitrogen atom, and go on to form one of three isomeric stable products (65):

41

C4F9SO2N(H)CH2CH3 + Cl → C4F9SO2N(OH)CH2CH2 (46a)

C4F9SO2N(H)CH2CH3 + Cl → C4F9SO2OON(H)CH2CH2 (46b)

C4F9SO2N(H)CH2CH3 + Cl → C4F9SO2N(H)OCH2CH2 (46c)

It is suggested that the product of reaction (46a) is dominant, with products of (46b) and (46c) playing a minor role, if formed at all. Following reactions in the smog chamber, COF2 and SO2 were observed by FTIR and the homologous series of PFCAs (CF3(CF2)xC(O)OH, x = 0 – 2) were observed by LC-MS-MS (62). The PFCAs are likely formed from the perfluorobutyl ● radical, CF3CF2CF2CF2 , by the mechanism described in section 2.1.1.3. This radical could be ● created through degradation of the products of reactions (44) and (45) to form C4F9SO2 , which loses SO2 to yield the perfluorobutyl radical. Computational studies also suggest that products ● ● of reactions (46b) and (46c) can decompose into C4F9SO2 or directly into C4F9 and SO2 (65). These studies suggest that NAFSAs can act as PFCA precursors.

2.3.2.2 N-alkyl-perfluoroalkanesulfamidoethanols (NAFSE)

Atmospheric lifetime

The atmospheric lifetime for a representative compound from this class, N-methyl perfluorobutane sulfamidoethanole (NMeFBSE, C4F9SO2N(CH3)CH2CH2OH) was determined using competition kinetics (32). The lifetime of this chemical is limited by atmospheric oxidation initiated by hydroxyl radicals:

C4F9SO2N(CH3)CH2CH2OH + OH → products (47)

The rate of this reaction was determined as k(OH + C4F9SO2N(CH3)CH2CH2OH) = (5.8 ± 0.8) × 10-12 cm3 molecule-1 s-1 (Table 2.7). Using a globally-averaged concentration of hydroxyl radicals of 1 × 106 molecules cm-3, the gas-phase atmospheric lifetime of NMeFBSE is approximately 2 days. The relatively low volatility of NAFSEs suggests that gas-particle partitioning may play a role in the overall atmospheric fate of this class of compounds, but the degree to which this occurs is not well defined (63,64).

Products of atmospheric degradation

D’eon et al. (32) studied the products of reaction (47) under smog chamber conditions, with in-situ FTIR detection and offline LC-MS-MS and GC-MS analysis. The similarity in

42 reactivity of NMeFBSE and n-C3H7OH with hydroxyl radicals suggests reaction takes place primarily on the –CH2CH2OH portion of NMeFBSE. This is supported by the observation of an aldehyde product formed early in the reaction:

C4F9SO2N(CH3)CH2CH3OH + OH → C4F9SO2N(CH3)CH2C(O)H (48a)

An N-dealkylation product was also observed:

C4F9SO2N(CH3)CH2CH3OH + OH → C4F9SO2N(H)CH3 (48b)

These types of dealkylation reactions have been observed in the gas-phase (66), but there have not been any mechanistic explanations made to date. Offline samples taken after reaction showed the homologous series of PFCAs (CF3(CF2)xC(O)OH, x = 0 – 2) and C4F9SO3H (perfluorobutane sulfonate, PFBS). The mechanism proposed to explain these observations involves addition of the hydroxyl radical to the sulfone double bond and subsequent breakage of the C-S or C-N bond to form PFCAs, or PFBS, respectively. This is described in detail in section 2.2.

Experiments to determine the products of NMeFBSE reaction with chlorine atoms were also undertaken (32):

C4F9SO2N(CH3)CH2CH2OH + Cl → products (49)

The products observed were similar to those observed for reaction with hydroxyl radical, including the N-dealkylation product (C4F9SO2N(H)CH3), PFCAs and PFBS. Chlorine atoms are often used as surrogates for hydroxyl radicals when oxidation is initiated by hydrogen abstraction, as in reactions (48a) and (48b). It is surprising that PFBS was observed from the chlorine atom-initiated oxidation of NMeFBSE, as this cannot be explained by the mechanism described above and in section 2.2. An alternative mechanism can be found in computational work on reaction of chlorine atoms with NEtFBSA (reaction (42)), which is similar in structure to the N-dealkylation product formed from NMeFBSE in reaction (48b) (65). Abstraction of a hydrogen atom from the nitrogen, followed by reaction with molecular oxygen can lead to insertion of oxygen into the S-O bond (see also reaction (46b)):

● CxF2x-1SO2N(H)R + Cl → CxF2x-1SO2N( )R (50)

● CxF2x-1SO2N( )R + O2 → CxF2x-1SO2OONR (51)

43 This product may decompose, forming the corresponding sulfonic radical:

● ● CxF2x-1SO2OONR + Cl → CxF2x-1SO2O + ONR (52)

The fate of this sulfonic radical would presumably be reaction with HO2 to yield the sulfonic acid:

● CxF2x-1SO2O + HO2 → CxF2x-1SO2OH + O2 (53)

Although the activation energy of reaction (51) is high, the reaction is overall exoergetic in nature, suggesting this may be a feasible mechanism, especially given the highly exoergetic nature of reaction (52) (65). These observations indicate that NAFSEs can act as sources of both PFCAs and PFSAs.

Table 2.7: Summary of chlorine atom- and hydroxyl radical-initiated kinetics for perfluorosulfonamides Rate Constant Structure x (cm3 molecule-1 s-1) T (K) Method Ref NAFSA Cl 3 (8.37 ± 1.44) × 10-12 296 Relative rate (62) CF3(CF2)xSO2N(H)CH2CH3 NAFSA 3 (3.74 ± 0.77) × 10-13 301 Relative rate (62) CF3(CF2)xSO2N(H)CH2CH3 OH NAFSE 3 (5.8 ± 0.8) × 10-12 296 Relative rate (32) CF3(CF2)xSO2N(CH3)CH2CH3

2.4 Atmospheric sources and levels

2.4.1 Volatile fluorinated anaesthetics

2.4.1.1 Potential sources to the atmosphere

Both halothane (CF3CHClBr) and isoflurane (CF3CHClOCHF2) are gaseous and are commonly used as anaesthetics. Exposure to these compounds by medical personnel is of concern (67), where operating room levels can be up to hundreds of parts-per-billion (68). This suggests the compounds will be ventilated into the atmosphere unless specific precautions are taken to prevent their release. Despite the likelihood of their presence, no measurements have been made of halothane and isoflurane in the atmosphere.

44 2.4.2 Hydrochlorofluorocarbons (HCFCs)

2.4.2.1 Potential sources to the atmosphere

Hydrochlorofluorocarbons (HCFCs) have been used as replacements for chlorofluorocarbons (CFCs) because of the lower ozone-depletion potential of the former. Uses of these chemicals are predictably increasing as the Montreal Protocol and related documents come into effect (69). These compounds are gases at environmentally relevant temperatures and can enter the environment through intentional and fugitive emissions. HCFCs are also slated for phase-out through the Montreal Protocol, with production scheduled to be frozen in the next few years (69).

2.4.2.2 Atmospheric concentrations

Concentrations of HCFC-124 were below detection limits used by contemporary methods in the late 1990s. However, more recently, HCFC-124 has been measured in the atmosphere at concentrations between 1.34 and 1.67 pptv in the troposphere, with a growth rate of 0.06 to 0.35 pptv year-1 (70). The most recent measurements available for HCFC-123 suggest that it remains at the sub-pptv level (70). Measurements of HCFC-225ca have not been made.

2.4.3 Hydrofluorocarbons (HFCs, non-telomer based)

2.4.3.1 Saturated hydrofluorocarbons (HFCs)

2.4.3.1.1 Potential sources to the atmosphere

Hydrofluorocarbons (HFCs) are primarily used as replacements for ozone-depleting CFCs, since they contain no chlorine and have no impact on stratospheric ozone. These compounds are gases employed in the coolant industry that can be released by fugitive emissions. The emission of HFC-134a has been observed at elevated concentrations in a road traffic tunnel, presumably as a result of release from car air conditioners (71). As CFCs are phased out by the Montreal Protocol (69), production of HFCs have been increasing to compensate. However, there is concern about the ability of HFCs to act as long-lived greenhouse gases (72), which in some cases has prompted directives to eliminate their use. Legislation to replace HFC-134a as the major coolant in mobile air conditioners has been introduced by the European Union (73), suggesting production of this compound is likely to decrease in the coming years.

45 2.4.3.1.2 Atmospheric concentrations

Concentrations of saturated HFCs are expected to be well-mixed in the atmosphere due to their long lifetimes. Few measurements of HFC-125 have been made, but levels of 1.4 to 5.1 pptv have been detected from diverse locations around the world (70,74,75). Consistent with expectations regarding usage, atmospheric concentrations of HFC-125 were observed to be increasing at a rate of 0.43 to 0.56 pptv year-1 (70,74).

Atmospheric levels of HFC-134a have received a great deal of attention due to high usage (Table 12). Levels have been steadily increasing from sub-pptv levels in the late 1980s to tens of pptv in the late 1990s to early 2000s. Northern hemisphere concentrations appear to be higher than southern hemisphere concentrations, presumably due to the larger source region present in the north. It also appears that concentrations may be higher in urban areas, which is consistent with population-driven usage and emissions.

Concentrations of other saturated HFCs have not yet been detected in the atmosphere.

2.4.3.2 Hydrofluoroolefins (HFOs)

2.4.3.2.1 Potential sources to the atmosphere

Hydrofluoroolefins (HFOs) are used in the synthesis of fluorinated polymers and are proposed as replacements for CFC replacements with a high climate impact. These compounds are also known pyrolysis products of perfluoroalkyl ethers (76) and fluoropolymers (77). The incineration of waste containing these polymers could act as an important environmental source. Jordan and Frank (1999) determined an approximate yield of TFA of 200 t year-1 from European incineration of fluoropolymers and subsequent release of hexafluoropropene. Currently, HFOs for use as coolants are in the development stage and are not produced in large quantities for this purpose. There have been no measurements made of these compounds in the atmosphere to date.

2.4.4 Fluorotelomer compounds

2.4.4.1 Potential sources to the atmosphere

Surfactant-based polyfluorinated chemicals used for their stain-repellent properties are synthesized by either the telomerization or electrochemical process. Fluorotelomer compounds

46 are synthesized using telomerization, where products are typically characterized by linear molecules with even numbers of carbon atoms, though odd-carbon chain lengths are also possible. The initial product of telomerization is the perfluorinated iodide (CF3(CF2)xI), from which a number of volatile fluorotelomer chemicals, such as fluorotelomer iodides (FTIs), fluorotelomer alcohols (FTOHs), fluorotelomer olefins (FTOs) and fluorotelomer acrylates (FTAcs) are synthesized. The initial product of the perfluorinated iodide is the FTI, which is then used to synthesize FTOHs, leaving ≤2% FTI as residual and forming 2-5% FTO as a byproduct (78). These compounds can be incorporated directly into consumer products. Alternatively, FTOHs or FTIs can be used to synthesize FTAcs, which are used to create a type of polymer that makes up >80% of the fluoropolymer market (78). The use of FTOHs to form FTAcs results in 0.1 – 0.5 % by weight FTOH residuals, while use of FTIs as a reaction precursor to FTAcs yields 3 – 8 % by weight FTOs as byproducts (78). Approximately 0.4 % by mass residual FTI, FTO, FTOH and FTAc was detected in a commercial fluoropolymer (79). The presence of unreacted FTOHs of multiple chain-lengths was also observed at a few percent by mass in a number of commercial fluorotelomer products (80).

The presence of unreacted materials or residuals, in commercial products could be a major source of PFA precursors to the atmosphere. This is corroborated by high levels of fluorotelomer chemicals in indoor air (81-84). Indoor concentrations were consistently at least ten times greater than outdoor air, with some environments exceeding a hundred times difference (81,82). In response to this potential source to the atmosphere, some producers of fluorotelomer products have committed to reducing the residuals present in their products (85).

2.4.4.2 Atmospheric concentrations

Atmospheric concentrations of volatile PFA precursors, such as FTOHs, FTOs, FTAcs, NAFSAs and NAFSEs have been widely measured. Due to their relatively short lifetimes (< 1 month), they are not uniformly distributed in the atmosphere and spatial measurements gain importance. These measurements have all been made by collection of the chemicals on a sorbent, followed by solvent extraction and analysis by gas chromatography coupled to mass spectrometry (GC-MS). Methods by which these compounds have been measured were comprehensively reviewed elsewhere (86).

47 2.4.4.2.1 FTOHs

These compounds were first measured in 2001 by Martin et al. (87). Measurements have since been made in a number of locations, but have not been distributed globally. Most measurements were made in the Northern Hemisphere and have been concentrated in North America, Europe and Japan. Studies have primarily focused on the 6:2, 8:2 and 10:2 FTOHs

(CF3(CF2)xCH2CH2OH, x = 5,7,9, respectively), though some measurements of 4:2 and 12:2 FTOHs (x = 3,11) have been made. Measured levels show large variability, from < 1 to 117 pg m-3 for 4:2 FTOH, < 1 to 196 pg m-3 for 6:2 FTOH, 2.4 to 4585 pg m-3 for 8:2 FTOH, < 1 to 518 pg m-3 for 10:2 FTOH and 1.8 to 20.9 pg m-3 for 12:2 FTOH (81,82,87-96). Despite this variability, some clear trends emerge. Levels of 8:2 FTOH are almost always highest, followed by 6:2 FTOH in North American and European measurements and 10:2 FTOH in Japanese measurements. Levels of 4:2 and 12:2 FTOHs tend to be lowest. In general, urban > semi- urban > rural > remote concentrations, with some exceptions that are probably due to air mass source. This trend is not surprising, given the source of these compounds appears to be directly related to human population. The high range of observed concentrations, from sub-pg m-3 to >1000 pg m-3, illustrate the importance of source regions and atmospheric transport in determining the concentrations of these compounds.

2.4.4.2.2 FTOs

The FTOs have been included in only a few measurement campaigns. The higher volatility of these compounds with respect to FTOHs precludes them from the commonly used methods for less volatile fluorinated compounds. Sorbent-based collection with solvent extraction and subsequent evaporation has typically led to poor recoveries for FTOs, particularly the shorter-chain congeners (96). The few measurements that are available show similar trends as those seen for FTOHs. Urban concentrations are the highest, while rural and remote measurements show low levels of FTOs. In addition, the 8:2 FTO (CF3(CF2)xCH=CH2, x = 7) is the dominant FTO observed in the atmosphere. Observed concentrations of FTOs are in the low (<25) pg m-3 range (82,96).

2.4.4.2.3 FTAcs

Measurements of FTAcs have been attempted by a few studies since 2005. Of the studies that included FTAcs, only a few resulted in detections. This is likely attributed to the

48 short atmospheric lifetime of FTAcs. Concentrations tend to be less than 100 pg m-3, with levels of 8:2 FTAc (CF3(CF2)xCH2CH2OC(O)CH=CH2, x = 7) apt to be higher (89,90,93-97). Detections are too sparse to determine specific spatial trends; however, levels in Japan appear to be elevated. In particular, a measurement from Higashiyodogawa in Osaka City, Japan, shows exceptionally high levels of 8:2 FTAc (up to 2953 pg m-3) (94). This suggests the proximity of a point source and demonstrates the need for high spatial resolution for short-lived precursors, such as FTAcs.

2.4.5 Perfluorosulfonamides

2.4.5.1 Potential sources to the atmosphere

Polyfluorinated chemicals used for their hydro- and lipophobic properties are synthesized by either the telomerization or electrochemical process. Perfluorosulfonamides have been produced by the electrochemical process, where products are characterized by the presence of a mixture of linear and branched carbon chains. This process was primarily used to produce perfluorosulfonyl fluoride (CF3(CF2)xSO2F). This is the synthetic precursor to NAFSAs, which can subsequently be converted to NAFSEs. Both NAFSAs and NAFSEs are typically converted into other products for commercial use, including phosphate esters for use in food packaging and polymers for fabric and carpet stain treatment (98). Incorporation of volatile compounds into polymers could lead to the presence of unreacted compounds in commercial products. Dinglasan-Panlilio and Mabury (80) observed unreacted N- methylperfluoroocanesulfamido ethanol (NMeFOSE) in a product produced by electrochemical fluorination.

The presence of these unreacted materials, or residuals, in commercial products is a potential source of PFA precursors to the atmosphere. High levels of perfluorosulfonamides have been observed in indoor air (63,81,82,99). Indoor concentrations were consistently much higher than those observed outdoors (63,81,82,99).

Industry has responded to the presence of perfluorinated acids in wildlife. In 2001-2002, the largest producer of perfluorooctanesulfonyl fluoride, the precursor to PFOS and eight-carbon perfluorosulfonamido alcohols, voluntarily removed the products from the market (100). These compounds have largely been replaced with their four-carbon equivalents, which as a result, should be observed in progressively higher concentrations in the environment.

49 2.4.5.2 Atmospheric concentrations

2.4.5.2.1 NAFSA

The NAFSAs have been a subject of a number of monitoring studies from various locations, typically alongside FTOHs. However, as with the FTOHs, studies have been concentrated in North America and Europe. NAFSAs with a perfluorinated chain of eight carbons (NAFOSAs) were first measured in 2001 by Martin et al. (87) and have been detected many times subsequently, while those with a perfluorinated chain of four (NAFBSAs) have been included since 2005. NAFSAs are found at low concentrations, typically less than 100 pg m-3 range (81,82,87-91,95,96,99,101). Levels of NAFOSAs and NAFBSEs in the atmosphere are comparable, despite the differences in production trends. NAFSAs appear to partition between the gas and particulate phase, though NAFOSAs appear more frequently in the particulate phase. The low levels detected in the atmosphere make it difficult to discern trends, while it is clear that concentrations in remote areas are lower than those of other areas.

2.4.5.2.2 NAFSE

The NAFSEs have typically been measured alongside NAFSAs in a number of studies conducted primarily in North America and Europe. NAFSEs with a perfluorinated chain of eight carbons (NAFOSEs) have been the focus of studies since 2001, while those containing a four-carbon perfluorinated chain have been incorporated since 2005 (NAFBSEs). Spatial variability is evident for NAFOSEs, with levels in urban areas typically higher (up to 393 pg m- 3) than those in remote areas (<100 pg m-3) (64,81,82,87-92,95,96,99,101).

2.5 Impact of precursors on environmental perfluorinated acid (PFA) levels

2.5.1 Trifluoroacetic acid (TFA)

Atmospheric TFA is primarily formed through perfluoroacyl halide hydrolysis, presumably occurring in cloud droplets. TFA is very soluble and likely to stay in solution and to rain-out, on a similar timescale to nitric acid (approximately 9 days) (102). However, it is possible that TFA could be liberated to the gas phase where it can react with hydroxyl radicals on a timescale of 100 to 230 days (33,103). This process has been estimated to reduce the amount of TFA present in rainwater by less than 5% (15). As a result, almost all TFA formed from precursors in the atmosphere will be rained out into the aqueous environment.

50 TFA is present ubiquitously in the aqueous environment, including in deep ocean water (8) and precipitation (9). It has been observed to be slightly phytoaccumulative to higher plants (104,105) and phytotoxic to some species at low concentrations (106). Although TFA itself was produced as a commercial product, production levels are relatively small, on the order of 1000 tonnes per year (107) A summary of estimated direct and indirect contributions to TFA are shown in Table 2.8. A number of studies have examined the impact of atmospheric sources on environmental TFA levels. One study has calculated that under very specific conditions, TFA derived from CFC-replacement compounds could accumulate in wetlands to levels that have been observed to be toxic (108). However, Boutonnet et al. (107) suggested these conditions were unlikely to ever occur in the environment and such accumulation was improbable. The study of Tromp et al. (108) only took into account degradation of HFCs and HCFCs. Other studies that have examined TFA levels derived from saturated HFCs, HCFCs and anaesthetics were unable to account for observed environmental levels (20,102,109). In addition, Jordan and Frank (109) observed that TFA levels were much higher in rivers in industrialized regions than in remote rivers. This suggests the potential contributions of shorter-lived precursors, such as HFOs. The thermolysis of fluoropolymers during use of commercial products (77) or incineration of those products (109) may also be adding to the environmental burden. More information regarding source levels and atmospheric concentrations is required in order to better understand the true impact of atmospheric oxidation of volatile chemicals on environmental levels of TFA.

Table 2.8: Estimated indirect and direct sources of trifluoroacetic acid (TFA).

Estimated TFA Chemical production (t yr-1) Ref

Halothane CF3CHClBr 520 (20) Isoflurane CF3CHClOCHF2 280 (20) HCFC-123 CF3CHCl2 266 (70) Indirect HCFC-124 CF3CHFCl 4440 (70) Sources HFC-134a CF3CH2F 4560 (70) Pyrolysis of 200 (109) fluoropolymers Total 10266 Direct Direct ~1000 (107) production Sources Total 1000

51 2.5.2 Perfluorooctanesulfonic acid (PFOS), perfluorooctanoic acid (PFOA) and perfluorononanoic acid (PFNA)

A great deal of attention has been paid to the impact of atmospheric formation of perfluorooctanesulfonic acid (PFOS), perfluorooctanoic acid (PFOA) and perfluorononanoic acid (PFNA). All three of these compounds can be formed indirectly through atmospheric oxidation of volatile precursors, but have also been intentionally produced directly. As such, detailed studies are required to attribute the importance of direct and indirect sources.

Relative significance of direct and indirect sources depends on the proximity to either a source of directly emitted compound or to a source of precursor compounds. Potential mechanisms of transport are expected to be very different for PFOS, PFOA and PFNA and their respective precursors. The low pKa of PFOS, PFOA and PFNA causes them to be ionized under most environmental conditions and to have low volatility and high water solubility. Thus, the primary repository for these compounds is the aqueous environment, which has been substantiated by ubiquitous observations of these compounds in ocean water (7). Movement of compounds via water is very slow compared to air, reducing the speed of long-range transport. Emissions data from a PFOA producer indicate that a small amount (5%) is emitted to air (78). It has often been assumed that PFOA cannot undergo long-range atmospheric transport because of its low volatility and relatively high water solubility. This was made under the assumption that the pKa of PFOA was low and that it would be present exclusively in the anionic form in the environment. Further recent studies have suggested the pKa could range from 0 (110) to 3.8 (111), while a recent study observed that PFOA behaved similarly to the highly-acidic PFOS and has a pKa <1 (112). Some experiments have suggested that PFOA could be liberated into the gas phase by the production of aerosols and subsequent aerosol-gas partitioning, thus facilitating long-range transport through the atmosphere of the directly emitted chemical (113). A subsequent modeling study examined this problem and determined that long-range atmospheric transport of PFOA would be insignificant unless the pKa was at least 3.5 (114). If indeed directly-emitted PFOA was being transported through the air, it would be expected that PFNA might be transported via a similar mechanisms and that precipitation levels would reflect their respective direct production levels. Despite the fact that PFOA is produced directly at levels at least an order of magnitude higher than PFNA, it is observed that PFOA and PFNA concentrations are similar in samples from mid- and high-latitudes (9,115). In addition, the linear isomer of PFOA has been suggested to be more readily liberated from the aerosol to the

52 gas phase than branched PFOA isomers (113). If this were the case, the rain-out of isomer- enriched aerosols should lead to enrichment of branched isomers in precipitation near source regions relative to precipitation in remote areas (116). Isomer analysis from temperate precipitation and a remote arctic lake that receives input solely from the atmosphere demonstrated similar ratios of branched to linear PFOA isomers (117). This implies the mechanism of transport does not discriminate between isomers, which is inconsistent with the hypothesis of aerosol-mediated transport of directly produced PFOA. Thus, we can suggest that the dominant mechanism of transport of PFOA and PFNA is the same as the highly-acidic PFOS and occurs through water.

The yield of PFOS, PFOA and PFNA from atmospheric oxidation of precursor chemicals is dependent on the ratio of NOx to HO2 and RO2 compounds. Typically, NOx levels are high in urban environments and low in remote environments. As a result, the highest yields of PFOS, PFOA and PFNA from precursor compounds are likely to be in remote regions. However, atmospheric concentrations of precursor compounds are also higher at mid-latitudes, close to production facilities and greater population density. Indirect production of PFOA at mid-latitudes is evident from the presence of isomers of PFOA as well as perfluoroheptanoic acid (PFHpA) and perfluorohexanoic acid (PFHxA) in rainwater (117). While branched PFHpA and PFHxA are not directly produced, volatile precursors to PFOA that are known to contain branched isomers, N-alkylperfluoroocylsulfonamides (NAFOSAs), can degrade via the perfluorinated radical mechanism described in Section 2.2.1.1.3. to form branched PFCAs of shorter chain-lengths.

The predicted deposition flux of PFOA and PFNA formed from 8:2 FTOH is highest in areas close to the sites of manufacture and use of the compounds, while the flux in remote regions is approximately one to two orders of magnitude lower (118). Overall contamination in remote regions is much lower, so indirect formation of PFOS, PFOA and PFNA could be an important contributor. A number of modelling studies have attempted to elucidate the importance of direct and indirect sources of PFOA (26,119-121). The study of Armitage et al. (119) considers only direct sources, Wallington et al. (26) and Schenker et al. (121) consider only indirect sources, while Wania (120) considers the impacts of both. Fluxes of PFOA through direct emissions and transport through ocean currents are estimated as 8 – 22 tonnes yr-1 by Armitage et al. (119) and 9 – 20 tonnes yr-1 by Wania (120). Predicted surface ocean concentrations (119) were observed to be in reasonable agreement with, though slightly higher

53 than, measurements in the North and Greenland Seas (122). These predicted concentrations were shown to be most sensitive to emissions input to the model, leading to varied results depending on assumptions made regarding emissions. All three studies that examine indirect formation include the impact of FTOHs, while Schenker et al. (121) also include formation from NAFSEs. The studies exclude numerous precursors discussed in Section 3, most notably FTOs, which are estimated to be present in consumer products at levels equal to FTOHs (78). As a result, indirect formation is probably underestimated by all three studies. A PFOA flux to the Arctic of 400 kg yr-1 was determined by Wallington et al. (26), while fluxes for 2005 of 154 kg yr-1 and 113-226 kg yr-1 were determined by Wania (120) and Schenker et al. (121), respectively. Differences in results appear to reflect differing assumptions regarding emissions of FTOHs as well as the level of atmospheric chemistry included. While Wallington et al. (26) used FTOH emissions (1000 t yr-1) that would be required to maintain the observed atmospheric FTOH concentrations, Wania (120) and Schenker et al. (121) included emission estimates provided by industry (≤200 t yr-1) that are not available to the general public or confirmable. The atmospheric chemistry included in the models varies dramatically. Schenker et al. (121) use a simplified chemical mechanism and Wania (120) uses zonally-averaged yields of PFOA from FTOHs. Wallington et al. (26) include a full chemical mechanism of FTOH degradation, with measured kinetics included where available. It should be noted that the ability of perfluoroacyl radicals to lose CO (reaction 58b) was not recognized when the Wallington et al. (26) model was developed, which may have lead to an underestimate of the PFOA flux. These broad differences between the models of Wallington et al. (26) and Wania (120) and Schenker et al. (121) may serve to explain the discrepancy in calculated fluxes from indirect sources. Although predicted fluxes of directly emitted PFOA are higher than those of indirectly formed and deposited PFOA, these estimated numbers do not necessarily reflect the relative importance of each process with respect to biota and human exposure. More measurements are required to determine accurate fluxes and to elucidate source apportionment.

Few studies have examined the importance of direct and indirect sources of PFNA. A single estimate of PFNA flux to the Arctic due to direct sources of 2 – 9 t yr-1 was determined (123), based on production estimates given in Prevedouros et al. (78). The importance of PFNA deposition to the Arctic from FTOHs was studied by Wallington et al. (26). Although a specific flux was not given, the calculated molar yield was similar to PFOA, suggesting a flux on the same order (approximately 400 kg yr-1). Again, the relative fluxes estimated for direct and

54 indirect sources are influenced by emission estimates and require validation from further monitoring data.

Compounds require decades to travel through the ocean to reach remote areas, such as the Arctic (124), while transport through the atmosphere is much faster, on the order of days to weeks. Directly formed PFOS moves through the ocean, while the indirect source of PFOS, NAFSEs, move through the atmosphere. Phase-out of PFOS and all its precursors by the largest manufacturer occurred in 2001 – 2002. Response in environmental levels to this production change could be indicative of the relative importance of direct and indirect sources. Evidence from the western Canadian Arctic revealed that levels of PFOS have declined rapidly in recent years in both biota (125). Such a fast decrease in concentration suggests the importance of atmospheric sources to the Arctic.

Further evidence of the importance of atmospheric sources to remote regions is the observation of PFOS, PFOA and PFNA in areas impacted only by atmospheric sources, such as land-locked lakes (126).

Oceanic transport of directly-produced PFOS, PFOA and PFNA and atmospheric transport of precursors and subsequent degradation and deposition are both sources of these compounds to the environment. At present, it is difficult to accurately determine the relative importance of each of these processes to environmental contamination of PFOS, PFOA and PFNA.

2.5.3 Long-chained perfluorocarboxylic acids (PFCAs)

Perfluorocarboxylic acids (PFCA) with chains equal to or greater than ten carbons have received relatively less attention. None of these compounds have ever been intentionally produced in large quantities. They have been observed as impurities in a commercial product containing PFNA as the primary component (78). However, it is not known how representative this single measurement is of global production. Long-chain PFCAs can also be formed from volatile precursors, as indicated in Table 19. Precipitation collected at mid-latitudes has been observed to contain PFCAs up to twelve carbons in length (9), likely as a result of atmospheric formation. Observations of PFCAs from perfluorodecanoic acid (PFDA) to perfluoropentadecanoic acid (PFPA) in Arctic wildlife (127-129), suggests these compounds are reaching remote regions and being bioaccumulated.

55 A single modelling study has examined the propensity for contamination of the Arctic by direct production of long-chain PFCAs (123). Using emissions determined by extrapolating the measurement of long-chain impurities of the single-product measurement reported by Prevedouros et al. (78), fluxes of perfluoroundecanoic acid (PFUnA) and perfluorotridecanoic acid (PFTrA) were calculated as 200 – 1400 and 4 – 145 kg yr-1, respectively.

Support for the importance of indirect sources is derived from the distinctive even-odd pattern observed in Arctic biota (129). Assuming approximately equal exposure to a given even PFCA and odd-chain-length congener one carbon longer, a higher concentration of the odd PFCA is observed in biota due to higher bioaccumulation of the longer chain-length PFCA. This pattern has been observed in virtually all arctic biota (10,127,129).

Observations and trends of long-chain PFCAs suggest volatile precursors could be sources of these compounds to the environment.

2.6 Summary

Perfluorocarboxylic acids can be formed from the hydrolysis of perfluoroacyl fluorides and chlorides, which can be produced in the atmosphere. Alternatively, PFCAs can be formed directly through reaction of perfluoroacyl peroxy radicals or perfluorinated aldehyde hydrates. Each of the mechanisms has been elucidated using smog chamber techniques, where yields of PFCAs vary from less than 10% to 100% depending on the mechanism in question. The formation of PFSAs in the atmosphere can also occur, though the mechanism has not been entirely elucidated. A large number of compounds have been confirmed as perfluorinated acid precursors, including CFC replacement compounds, fluorotelomer compounds and perfluorosulfonamides. From current environmental monitoring data, atmospheric oxidation of volatile precursors appears to contribute to the overall burden of PFAs.

2.7 Goals and Hypotheses

PFAs are formed in the atmosphere through a number of known mechanisms. We hypothesize that volatile precursors are important contributors to environmental PFCA contamination. We also believe that unidentified chemicals can degrade via known and as-yet- unrecognized pathways to form PFCAs. In Chapter 6, PFCA formation is observed from CFC- replacement compounds via a new mechanism. Chapter 7 examines a class of compounds, the fluorotelomer iodides, which we show can form PFCAs through the perfluoroperoxy radical and

56 perfluorinated radical mechanisms. Within the perfluorinated radical mechanism, the one step that remains unclear is the reaction to form the perfluorinated acid fluoride from the perfluorinated alcohol. In Chapter 8, a mechanism involving overtone-induced photolysis is proposed to partially account for this reaction. Finally, in order to determine if volatile precursors are a significant contributor to perfluorinated acid contamination in the Arctic, snow samples were collected from a High Arctic ice cap. Chapter 9 describes the fluxes and trends observed in these samples.

57 2.8 Sources Cited

(1) Giesy, J.P.; Kannan, K. Distribution of perfluorooctane sulfonate in wildlife. Environmental Science and Technology 2001, 35, 1339-1342.

(2) Taves, D.R. Evidence that there are two forms of fluoride in human serum. Nature 1968, 217, 1050-1051.

(3) Hansen, K.J.; Clemen, L.A.; Ellefson, M.E.; Johnson, H.O. Compound-specific, quantitative characterization of organic fluorochemicals in biological matrices. Environmental Science and Technology 2001, 35, 766-770.

(4) Martin, J.W.; Mabury, S.A.; Solomon, K.R.; Muir, D.C.G. Dietary accumulations of perfluorinated acids in juvenile rainbow trout (Oncorhynchus mykiss). Environmental Toxicology and Chemistry 2003, 22, 189-195.

(5) Martin, J.W.; Mabury, S.A.; Solomon, K.R.; Muir, D.C.G. Bioconcentration and tissue distribution of perfluorinated acids in rainbow trout (Oncorhynchus mykiss). Environmental Toxicology and Chemistry 2003, 22, 196-204.

(6) Tomy, G.; Budakowski, W.; Halldorson, T.; Helm, P.A.; Stern, G.A.; Friesen, K.; Pepper, K.; Tittlemier, S.A.; Fisk, A.T. Fluorinated organic compounds in an Eastern Arctic food web. Environmental Science and Technology 2004, 38, 6475-6481.

(7) Yamashita, N.; Kannan, K.; Taniyasu, S.; Horii, Y.; Petrick, G.; Gamo, T. A global survey of perfluorinated acids in oceans. Marine Pollution Bulletin 2005, 51, 658-668.

(8) Scott, B.F.; Macdonald, R.W.; Kannan, K.; Fisk, A.; Witter, A.; Yamashita, N.; Durham, L.; Spencer, C.; Muir, D.C.G. Trifluoroacetate profiles in the Arctic, Atlantic and Pacific Oceans. Environmental Science and Technology 2005, 39, 6555-6560.

(9) Scott, B.F.; Spencer, C.; Mabury, S.A.; Muir, D.C.G. Poly and perfluorinated carboxylates in North American precipitation. Environmental Science and Technology 2006, 40, 7167-7174.

(10) Houde, M.; Martin, J.W.; Letcher, R.J.; Solomon, K.R.; Muir, D.C.G. Biological monitoring of polyfluoroalkyl substances: A review. Environmental Science and Technology 2006, 40, 3463-3473.

(11) Kannan, K.; Corsolini, S.; Falandysz, J.; Fillmann, G.; Kumar, K.S.; Loganathan, B.G.; Mohd, M.A.; Olivero, J.; Van Wouwe, N.; Yang, J.H.; Aldous, K.M. Perfluorooctanesulfonate and related fluorochemicals in human blood from several countries. Environmental Science and Technology 2004, 38, 4489-4495.

(12) Rattigan, O.V.; Wild, O.; Jones, R.L.; Cox, R.A. Temperature-dependent absorption cross-sections of CF3COCl, CF3COF, CH3COF, CCl3CHO and CF3COOH. Journal of Photochemistry and Photobiology A: Chemistry 1993, 73, 1-9.

(13) Wild, O.; Rattigan, O.V.; Jones, R.L.; Pyle, J.A.; Cox, R.A. Two-dimensional modelling of some CFC replacement compounds. Journal of Atmospheric Chemistry 1996, 25, 167-199.

58 (14) Wallington, T.J.; Schneider, W.F.; Worsnop, D.R.; Nielsen, O.J.; Sehested, J.; Debruyn, W.J.; Shorter, J.A. The environmental impact of CFC replacements - HFCs and HCFCs. Environmental Science and Technology 1994, 28, 320A-326A.

(15) Kanakidou, M.; Dentener, F.J.; Crutzen, P.J. A global three-dimensional study of the fate of HCFCs and HFC-134a in the troposphere. Journal of Geophysical Research 1995, 100, 18781-18801.

(16) Chiappero, M.S.; Malanca, F.E.; Arguello, G.A.; Wooldridge, S.T.; Hurley, M.D.; Ball, J.C.; Wallington, T.J.; Waterland, R.L.; Buck, R.C. Atmospheric chemistry of perfluoroaldehydes (CxF2x+1CHO) and fluorotelomer aldehydes (CxF2x+1CH2CHO): Quantification of the important role of photolysis. Journal of Physical Chemistry A 2006, 110, 11944-11953.

(17) Moortgat, G.K.; Veyret, B.; Lesclaux, R. Kinetics of the reaction of HO2 with CH3C(O)O2 in the temperature range 253 - 368 K. Chemical Physics Letters 1989, 160, 443- 447.

(18) Meller, R.; Moortgat, G.K. CF3C(O)Cl: Temperature-dependent (223-298 K) absorption cross-sections and quantum yields at 254 nm. Journal of Photochemistry and Photobiology A: Chemistry 1997, 108, 105-116.

(19) Hayman, G.D.; Jenkin, M.E.; Murrells, T.P.; Johnson, C.E. Tropospheric degradation chemistry of HCFC-123 (CF3CHCl2): A proposed replacement chlorofluorocarbon. Atmospheric Environment 1994, 28, 421-437.

(20) Tang, X.; Madronich, S.; Wallington, T.; Calamari, D. Changes in tropospheric composition and air quality. Journal of Photochemistry and Photobiology B: Biology 1998, 46, 83-95.

(21) Edney, E.O.; Driscoll, D.J. Chlorine initiated photooxidation studies of hydrochlorofluorocarbons (HCFCs) and hydrofluorocarbons (HFCs): Results for HCFC-22 (CHClF2); HFC-41 (CH3F); HCFC-124 (CClFHCF3); HFC-125 (CF3CHF2); HFC-134a (CF3CH2F); HCFC-142b (CClF2CH3); and HFC-152a (CHF2CH3). International Journal of Chemical Kinetics 1992, 24, 1067-1081.

(22) Edney, E.O.; Gay Jr., B.W.; Driscoll, D.J. Chlorine initiated oxidation studies of hydrochlorofluorocarbons: Results for HCFC-123 (CF3CHCl2) and HCFC-141b (CFCl2CH3). Journal of Atmospheric Chemistry 1991, 12, 105-120.

(23) Bilde, M.; Wallington, T.J.; Ferronato, C.; Orlando, J.J.; Tyndall, G.S.; Estupiñan, E.; Haberkorn, S. Atmospheric chemistry of CH2BrCl, CHBrCl2, CHBr2Cl, CF3CHBrCl, and CBr2Cl2. Journal of Physical Chemistry A 1998, 102, 1976-1986.

(24) Hurley, M.D.; Ball, J.C.; Wallington, T.J.; Sulbaek Andersen, M.P.; Ellis, D.A.; Martin, J.W.; Mabury, S.A. Atmospheric chemistry of fluorinated alcohols: Reaction with Cl atoms and OH radicals and atmospheric lifetimes. Journal of Physical Chemistry A 2004, 108, 1973-1979.

59 (25) Ellis, D.A.; Martin, J.W.; De Silva, A.O.; Mabury, S.A.; Hurley, M.D.; Sulbaek Andersen, M.P.; Wallington, T.J. Degradation of fluorotelomer alcohols: A likely atmospheric source of perfluorinated carboxylic acids. Environmental Science and Technology 2004, 38, 3316-3321.

(26) Wallington, T.J.; Hurley, M.D.; Xia, J.; Wuebbles, D.J.; Sillman, S.; Ito, A.; Penner, J.E.; Ellis, D.A.; Martin, J.W.; Mabury, S.A.; Nielsen, C.J.; Sulbaek Andersen, M.P. Formation of C7F15COOH (PFOA) and other perfluorocarboxylic acids during the atmospheric oxidation of 8:2 fluorotelomer alcohol. Environmental Science and Technology 2006, 40, 924-930.

(27) Hasson, A.S.; Tyndall, G.W.; Orlando, J.J. A product yield study of the reaction of HO2 radicals with ethyl peroxy (C2H5O2), acetyl peroxy (CH3C(O)O2), and acetonyl peroxy (CH3C(O)CH2O2) radicals. Journal of pHysical Chemistry A 2004, 108, 5979-5989.

(28) Sulbaek Andersen, M.P.; Hurley, M.D.; Wallington, T.; Ball, J.C.; Martin, J.W.; Ellis, D.A.; Mabury, S.A. Atmospheric chemistry of C2F5CHO: Mechanism of the C2F5C(O)O2 + HO2 reaction. Chemical Physics Letters 2003, 381, 14-21.

(29) Sulbaek Andersen, M.P.; Stenby, C.; Nielsen, C.J.; Hurley, M.D.; Ball, J.C.; Wallington, T.J.; Martin, J.W.; Ellis, D.A.; Mabury, S.A. Atmospheric chemistry of n-CxF2x+1CHO (x=1,3,4): Mechanism of the CxF2x+1C(O)O2 + HO2 reaction. Journal of Physical Chemistry A 2004, 108, 6325-6330.

(30) Hurley, M.D.; Ball, J.C.; Wallington, T.J.; Sulbaek Andersen, M.P.; Nielsen, C.J.; Ellis, D.A.; Martin, J.W.; Mabury, S.A. Atmospheric chemistry of n-CxF2x+1CHO (x = 1,2,3,4): Fate of n-CxF2x+1C(O) radicals. Journal of Physical Chemistry A 2006, 110, 12443-12447.

(31) Sulbaek Andersen, M.P.; Toft, A.; Nielsen, O.J.; Hurley, M.D.; Wallington, T.J.; Chishima, H.; Tonokura, K.; Mabury, S.A.; Martin, J.W.; Ellis, D.A. Atmospheric chemistry of perfluorinated aldehyde hydrates (n-CxF2x+1CH(OH)2, x=1,3,4): Hydration, dehydration, and kinetics and mechanism of Cl atom and OH radical initiated oxidation. Journal of Physical Chemistry A 2006, 110, 9854-9860.

(32) D'eon, J.C.; Hurley, M.D.; Wallington, T.J.; Mabury, S.A. Atmospheric chemistry of N- methyl perfluorobutane sulfonamidoethanol, C4F9SO2N(CH3)CH2CH2OH: Kinetics and mechanism of reaction with OH. Environmental Science and Technology 2006, 40, 1862-1868.

(33) Hurley, M.D.; Sulbaek Andersen, M.P.; Wallington, T.J.; Ellis, D.A.; Martin, J.W.; Mabury, S.A. Atmospheric chemistry of perfluorinated carboxylic acids: Reaction with OH radicals and atmospheric lifetimes. Journal of Physical Chemistry A 2004, 108, 615-620.

(34) Young, C.J.; Mabury, S.A. Atmospheric perfluorinated acid precursors: Chemistry, occurrence and impacts. Reviews of Environmenal Contamination and Toxicology 2009, Submitted.

(35) Howard, P.H.; Meylan, W. 2007. EPA Great Lakes Study for Identification of PBTs to Develop Analytical Methods: Selection of Additional PBTs - Interim Report, EPA Contract No. EP-W-04-019,

60 (36) Waterland, R.L.; Dobbs, K.D. Atmospheric chemistry of linear perfluorinated aldehydes: Dissociation kinetics of CnF2n+1CO radicals. Journal of Physical Chemistry A 2007, 111, 2555- 2562.

(37) Scollard, D.J.; Treacy, J.J.; Sidebottom, H.W.; Balestra-Garcia, C.; Laverdet, G.; LeBras, G.; MacLeod, H.; Teton, S. Rate constants for the reactions of hydroxyl radicals and chlorine atoms with halogenated aldehydes. Journal of Physical Chemistry 1993, 4683-4688.

(38) Wallington, T.J.; Hurley, M.D. A kinetic study of the reaction of chlorine and fluorine atoms with CF3CHO at 295 ± 2 K. International Journal of Chemical Kinetics 1993, 25, 819- 824.

(39) Sulbaek Andersen, M.P.; Nielsen, C.J.; Hurley, M.D.; Ball, J.C.; Wallington, T.J.; Stevens, J.E.; Martin, J.W.; Ellis, D.A.; Mabury, S.A. Atmospheric chemistry of n-CxF2x+1CHO (x=1,3,4): Reaction with Cl atoms, OH radicals and IR spectra of CxF2x+1C(O)O2NO2. Journal of Physical Chemistry A 2004, 108, 5189-5196.

(40) Sulbaek Andersen, M.P.; Hurley, M.D.; Wallington, T.J.; Ball, J.C.; Martin, J.W.; Ellis, D.A.; Mabury, S.A.; Nielsen, C.J. Atmospheric chemistry of C2F5CHO: Reaction with Cl atoms and OH radicals, IR spectrum of C2F5C(O)O2NO2. Chemical Physics Letters 2003, 379, 28-36.

(41) Solignac, G.; Mellouki, A.; Le Bras, G.; Barnes, I.; Benter, T. Reaction of Cl atoms with C6F13CH2OH, C6F13CHO, and C3F7CHO. Journal of Physical Chemistry A 2006, 110, 4450- 4457.

(42) Dobe, S.; Kachatryan, L.A.; Berces, T. Kinetics of reactions of hydroxyl radicals with a series of aliphatic aldehydes. Ber Bunsen-Ges Phys Chem 1989, 93, 847-952.

(43) Sellevåg, S.R.; Kelly, T.; Sidebottom, H.; Nielsen, C.J. A study of the IR and UV-Vis absorption cross-sections, photolysis and OH-initiated oxidation of CF3CHO and CF3CH2CHO. Physical Chemistry Chemical Physics 2004, 6, 1243-1252.

(44) Atkinson, R.; Baulch, D.L.; Cox, R.A.; Crowley, J.N.; Hampson, R.F.; Hynes, R.G.; Jenkin, M.E.; Rossi, M.J.; Troe, J.; Wallington, T.J. Evaluated kinetic and photochemical data for atmospheric chemistry: Volume IV - gas phase reactions of organic halogen species. Atmospheric Chemistry and Physics 2008, 8, 4141-4496.

(45) Solignac, G.; Mellouki, A.; Le Bras, G.; Yujing, M.; Sidebottom, H. The gas phase tropospheric removal of fluoroaldehydes (CxF2X+1CHO, x = 3, 4, 6). Physical Chemistry Chemical Physics 2007, 9, 4200-4210.

(46) Solignac, G.; Mellouki, A.; Le Bras, G.; Barnes, I.; Benter, T. Reaction of Cl atoms with C6F13CH2OH, C6F13CHO and C3F7CHO. Journal of Physical Chemistry A 2006, 110, 4450- 4457.

(47) Kelly, T.; Bossoutrot, V.; Magneron, I.; Wirtz, K.; Treacy, J.J.; Mellouki, A.; Sidebottom, H.; Le Bras, G. A kinetic and mechanistic study of the reactions of OH radicals and Cl atoms with 3,3,3-trifluoropropanol under atmospheric conditions. Journal of Physical Chemistry A 2005, 109, 347-355.

61 (48) Hurley, M.D.; Misner, J.A.; Ball, J.C.; Wallington, T.J.; Ellis, D.A.; Martin, J.W.; Mabury, S.A.; Sulbaek Andersen, M.P. Atmospheric chemistry of CF3CH2CH2OH: Kinetics, mechanisms and products of Cl atom and OH radical initiated oxidation in the presence and absence of NOx. Journal of Physical Chemistry A 2005, 109, 9816-9826.

(49) Hurley, M.D.; Ball, J.C.; Wallington, T.J.; Sulbaek Andersen, M.P.; Ellis, D.A.; Martin, J.W.; Mabury, S.A. Atmospheric chemistry of 4:2 fluorotelomer alcohol: products and mechanism of Cl atom initiated oxidation. Journal of Physical Chemistry A 2004, 108, 5635- 5642.

(50) Chiappero, M.S.; Arguello, G.A.; Hurley, M.D.; Wallington, T.J. Atmospheric chemistry of C8F17CH2CHO: Yield from C8F17CH2CH2OH (8:2 FTOH) oxidation, kinetics and mechanisms of reactions with Cl atoms and OH radicals. Chemical Physics Letters 2008, 461, 198-202.

(51) Papadimitriou, V.C.; Papanastasiou, D.K.; Stefanopoulos, V.G.; Zaras, A.M.; Lazarou, Y.G.; Papgiannakopoulos, P. Kinetic study of the reactions of Cl atoms with CF3CH2CH2OH, CF3CF2CH2OH, CHF2CF2CH2OH, and CF3CHFCF2CH2OH. Journal of Physical Chemistry A 2007, 111, 11608-11617.

(52) Ellis, D.A.; Martin, J.W.; Mabury, S.A.; Hurley, M.D.; Sulbaek Andersen, M.P.; Wallington, T.J. Atmospheric lifetime of fluorotelomer alcohols. Environmental Science and Technology 2003, 37, 3816-3820.

(53) Sulbaek Andersen, M.P.; Nielsen, O.J.; Toft, A.; Nakayama, T.; Matsumi, Y.; Waterland, R.L.; Buck, R.C.; Hurley, M.D.; Wallington, T.J. Atmospheric chemistry of CxF2x+1CH=CH2 (x=1,2,4,6, and 8): Kinetics of gas-phase reactions with Cl atoms, OH radicals, and O3. Journal of Photochemistry and Photobiology A: Chemistry 2005, 176, 124-128.

(54) Butt, C.M.; Young, C.J.; Mabury, S.A.; Hurley, M.D.; Wallington, T.J. Atmospheric chemistry of 4:2 fluorotelomer acrylate (C4F9CH2CH2OC(O)CH=CH2): Kinetics, mechanisms and products of chlorine atom and OH radical initiated oxidation. Journal of Physical Chemistry A 2009, 113, 3155-3161.

(55) Orkin, V.L.; Huie, R.E.; Kurylo, M.J. Rate constants for the reactions of OH with HFC- 245cb (CH3CF2CF3) and some fluoroalkenes (CH2CHCF3, CH2CFCF3, CF2CFCF3, and CF2CF2). Journal of Physical Chemistry A 1997, 101, 9118-9124.

(56) Vesine, E.; Bossoutrot, V.; Mellouki, A.; Le Bras, G.; Wenger, J.; Sidebottom, H. Kinetic and mechanistic study of OH- and Cl-initiated oxidation of two unsaturated HFCs: C4F9CH=CH2 and C6F13CH=CH2. Journal of Physical Chemistry A 2000, 104, 8512-8520.

(57) Singh, H.B.; Thakur, A.N.; Chen, Y.E.; Kanakidou, M. Tetrachloroethylene as an indicator of low Cl atom concentrations in the troposphere. Geophysical Research Letters 1996, 23, 1529-1532.

(58) Mereau, R.; Rayez, M.-T.; Rayez, J.-C.; Caralp, F.; Lesclaux, R. Theoretical study on the atmospheric fate of carbonyl radicals: Kinetics of decomposition reactions. Physical Chemistry Chemical Physics 2001, 3, 4712-4717.

62 (59) Sulbaek Andersen, M.P.; Nielsen, C.J.; Hurley, M.D.; Ball, J.C.; Wallington, T.J.; Ellis, D.A.; Martin, J.W.; Mabury, S.A. Atmospheric chemistry of 4:2 fluorotelomer alcohol (n- C4F9CH2CH2OH): Products and mechanism of Cl atom initiated oxidation in the presence of NOx. Journal of Physical Chemistry A 2005, 109, 1849-1856.

(60) Waterland, R.L.; Hurley, M.D.; Misner, J.A.; Wallington, T.J.; Melo, S.M.L.; Strong, K.; Dumoulin, R.; Castera, L.; Stock, N.L.; Mabury, S.A. Gas phase UV and IR absorption spectra of CF3CH2CH2OH and F(CF2CF2)xCH2CH2OH (x = 2, 3, 4). Journal of Fluorine Chemistry 2005, 126, 1288-1296.

(61) Nakayama, T.; Takahashi, K.; Matsumi, Y.; Toft, A.; Sulbaek Andersen, M.P.; Nielsen, O.J.; Waterland, R.L.; Buck, R.C.; Hurley, M.D.; Wallington, T.J. Atmospheric chemistry of CF3CH=CH2 and C4F9CH=CH2: Products of the gas-phase reactions with Cl atoms and OH radicals. Journal of Physical Chemistry A 2007, 111, 909-915.

(62) Martin, J.W.; Ellis, D.A.; Mabury, S.A.; Hurley, M.D.; Wallington, T.J. Atmospheric chemistry of perfluoroalkanesulfonamides: Kinetic and product studies of the OH and Cl atom initiated oxidiation of N-ethyl perfluorobutanesulfonamide. Environmental Science and Technology 2006, 40, 864-872.

(63) Shoeib, M.; Harner, T.; Ikonomou, M.; Kannan, K. Indoor and outdoor air concentrations and phase partitioning of perfluoroalkyl sulfonamides and polybrominated diphenyl ethers. Environmental Science and Technology 2004, 38, 1313-1320.

(64) Stock, N.L. 2007. Occurrence and fate of perfluoroalkyl contaminants in the abiotic environment. PhD thesis, University of Toronto, Toronto. 252 p.

(65) Antoniotti, P.; Borocci, S.; Giordani, M.; Grandinetti, F. Cl-initiated oxidation of N-ethyl- perfluoroalkanesulfonamides: A theoretical insight into experimentally observed products. Journal of Molecular Structure: THEOCHEM 2008, 857, 57-65.

(66) Woodrow, J.E.; Crosby, D.G.; Mast, T.; Moilanen, K.W.; Seiber, J.N. Rates of transformation of trifluralin and parathion vapors in air. Journal of Agricultural and Food Chemistry 1978, 26, 1312-1316.

(67) OSHA Directorate for Technical Support. Washington, DC, 2000. Anesthetic gases: Guidelines for workplace exposures. Accessed

(68) Al-Ghanem, S.; Battah, A.H.; Salhab, A.S. Monitoring of volatile anesthetics in operating room personnel using GC-MS. Jordan Medical Journal 2008, 42, 13-19.

(69) UNEP. 2000. The Montreal Protocol on Substances that Deplete the Ozone Layer, United Nations Environmental Programme,

(70) Velders, G.J.M.; Madronich, S.; Clerbaux, C.; Derwent, R.; Grutter, M.; Hauglustaine, D.; Incecik, S.; Ko, M.; Libre, J.-M.; Nielsen, O.J.; Stordal, F.; Zhu, T. In IPCC/TEAP Special Report: Safeguarding the Ozone Layer and the Global Climate System; Metz, B., Kuijpers, L., Solomon, S., Andersen, S.O., Davidson, O., Pons, J., de Jager, D., Kestin, T., Manning, M., Meyer, L., Eds.; Cambridge University Press: Cambridge, United Kingdom, 2005.

63 (71) Stemmler, K.; O'Doherty, S.; Buchmann, B.; Reimann, S. Emissions of the refrigerants HFC-134a, HCFC-22, and CFC-12 from road traffic: Results from a tunnel study (Gubrist Tunnel, Switzerland). Environmental Science and Technology 2004, 38, 1998-2004.

(72) Forster, P.; Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R.; Fahey, D.W.; Haywood, J.; Lean, J.; Lowe, D.C.; Myhre, G.; Nganga, J.; Prinn, R.; Raga, G.; Schulz, M.; Van Dorland, R. In Climate Change 2007: The Physical Science Basis; Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L., Eds.; Cambridge University Press: Cambridge, United Kingdom, 2007.

(73) The European Parliament and the Council of the European Union In Directive 2006/40/EC, 2006.

(74) Reimann, S.; Schaub, D.; Stemmler, K.; Folini, D.; Hill, M.; Hofer, P.; Buchmann, B.; Simmonds, P.G.; Greally, B.R.; O'Doherty, S. Halogenated greenhouse gases at the Swiss high alpine site of Jungfraujoch (3580 m asl): Continuous measurements and their use for regional European source allocation. Journal of Geophysical Research 2004, 109, D05307.

(75) Miller, B.R.; Weiss, R.F.; Salameh, P.K.; Tanhua, T.; Greally, B.R.; Mühle, J.; Simmonds, P.G. Medusa: A sample preconcentration and GC/MS detector system for in situ measurements of atmospheric trace halocarbons, hydrocarbons, and sulfur compounds. Anal Chem 2008, 80, 1536-1545.

(76) Hoshino, M.; Kimachi, Y.; Terada, A. Thermogravimetric behaviour of perfluoropolyether. Journal of Applied Polymer Science 1996, 62, 207-215.

(77) Ellis, D.A.; Mabury, S.A.; Martin, J.W.; Muir, D.C.G. Thermolysis of fluoropolymers as a potential source of halogenated organic acids in the environment. Nature 2001, 412, 321-324.

(78) Prevedouros, K.; Cousins, I.T.; Buck, R.C.; Korzeniowski, S.H. Sources, fate and transport of perfluorocarboxylates. Environmental Science and Technology 2006, 40, 32-44.

(79) Russell, M.H.; Berti, W.R.; Szostek, B.; Buck, R.C. Investigation of the biodegradation potential of a fluoroacrylate polymer product in aerobic soils. Environmental Science and Technology 2008, 42, 800-807.

(80) Dinglasan-Panlilio, M.J.A.; Mabury, S.A. Significant residual fluorinated alcohols present in various fluorinated materials. Environmental Science and Technology 2006, 40, 1447-1453.

(81) Jahnke, A.; Huber, S.; Temme, C.; Kylin, H.; Berger, U. Development and application of a simplified sampling method for volatile polyfluorinated alkyl substances in indoor and environmental air. Journal of Chromatography A 2007, 1164, 1-9.

(82) Barber, J.L.; Berger, U.; Chaemfa, C.; Huber, S.; Jahnke, A.; Temme, C.; Jones, K.C. Analysis of per- and polyfluorinated alkyl substances in air samples from Northwest Europe. Journal of Environmental Monitoring 2007, 9, 530-541.

(83) Shoeib, M.; Harner, T.; Lee, S.C.; Lane, D.; Zhu, J. Sorbent-impregnated polyurethane foam disk for passive air sampling of volatile fluorinated chemicals. Anal Chem 2008, 80, 675- 682.

64 (84) Strynar, M.J.; Lindstrom, A.B. Perfluorinated compounds in house dust from Ohio and North Carolina, USA. Environmental Science and Technology 2008, 42, 3751-3756.

(85) United States Environmental Protection Agency. 2010/15 PFOA Stewardship Program, http://www.epa.gov/oppt/pfoa/pubs/pfoastewardship.htm.

(86) Jahnke, A.; Berger, U. Trace analysis of per- and polyfluorinated alkyl substances in various matrices-How do current methods perform? Journal of Chromatography A 2009, 1216, 410-421.

(87) Martin, J.W.; Muir, D.C.G.; Moody, C.A.; Ellis, D.A.; Kwan, W.; Solomon, K.R.; Mabury, S.A. Collection of airborne fluorinated organics and analysis by gas chromatography/chemical ionization mass spectrometry. Analytical Chemistry 2002, 74, 584- 590.

(88) Stock, N.L.; Lau, F.K.; Ellis, D.A.; Martin, J.W.; Muir, D.C.G.; Mabury, S.A. Polyfluorinated telomer alcohols and sulfonamides in the North American troposphere. Environmental Science and Technology 2004, 38, 991-996.

(89) Jahnke, A.; Ahrens, A.; Ebinghaus, R.; Temme, C. Urban versus remote air concentrations of fluorotelomer alcohols and other polyfluorinated alkyl substances in Germany. Environmental Science and Technology 2007, 41, 745-752.

(90) Jahnke, A.; Ahrens, L.; Ebinghaus, R.; Berger, U.; Barber, J.L.; Temme, C. An improved method for the analysis of volatile polyfluorinated alkyl substances in environmental air samples. Analytical Bioanalytical Chemistry 2007, 387, 965-975.

(91) Jahnke, A.; Berger, U.; Ebinghaus, R.; Temme, C. Latitudinal gradient of airborne polyfluorinated alkyl substances in the marine atmosphere between Germany and South Africa (53oN - 33oS). Environmental Science and Technology 2007, 41, 3055-3061.

(92) Shoeib, M.; Harner, T.; Vlahos, P. Perfluorinated chemicals in the Arctic atmosphere. Environmental Science and Technology 2006, 40, 7577-7583.

(93) Oono, S.; Harada, K.H.; Mahmoud, M.A.M.; Inoue, K.; Koizumi, A. Current levels of airborne polyfluorinated telomers in Japan. Chemosphere 2008, 73, 932-937.

(94) Oono, S.; Matsubara, E.; Harada, K.H.; Takagi, S.; Hamada, S.; Asakawa, A.; Inoue, K.; Watanabe, I.; Koizumi, A. Survey of airborne polyfluorinated telomers in Keihan area, Japan. Bulletin of Environmental Contamination and Toxicology 2008, 80, 102-106.

(95) Dreyer, A.; Ebinghaus, R. Polyfluorinated compounds in ambient air from ship- and land- based measurements in northern Germany. Atmospheric Environment 2009, 43, 1527-1535.

(96) Piekarz, A.M.; Primbs, T.; Fields, J.A.; Barofsky, D.F.; Simonich, S. Semivolatile fluorinated organic compounds in Asian and Western U.S. air masses. Environmental Science and Technology 2007, 41, 8248-8255.

(97) Dreyer, A.; Temme, C.; Sturm, R.; Ebinghaus, R. Optimized method avoiding solvent- induced response enhancement in the analysis of volatile and semi-volatile polyfluorinated

65 alkylated compounds using gas chromatography-mass spectrometry. Journal of Chromatography A 2008, 1178, 199-205.

(98) 3M. Fluorochemical use, distribution and release overview, United States Environmental Protection Agency, Public docket AR226-0550.

(99) Shoeib, M.; Harner, T.; Wilford, B.H.; Jones, K.C.; Zhu, J. Perfluorinated sulfonamides in indoor and outdoor air and indoor dust: Occurrence, partitioning, and human exposure. Environmental Science and Technology 2005, 39, 6599-6606.

(100) 3M Company. 2000. Phase-out plan for POSF-based products, US Environmental Protection Agency, Public Docket OPPT-2002-0043-0009.

(101) Boulanger, B.; Peck, A.M.; Schnoor, J.L.; Hornbuckle, K.C. Mass budget of perfluorooctane surfactants in Lake Ontario. Environmental Science and Technology 2005, 39, 74-79.

(102) Kotamarthi, V.R.; Rodriguez, J.M.; Ko, M.K.W.; Tromp, T.K.; Sze, N.D. Trifluoroacetic acid from degradation of HCFCs and HFCs: A three-dimensional modeling study. Journal of Geophysical Research 1998, 103, 5747-5758.

(103) Carr, S.; Treacy, J.J.; Sidebottom, H.W.; Connell, R.K.; Canosa-Mas, C.E.; Wayne, R.P.; Franklin, J. Kinetics and mechanisms for the reaction of hydroxyl radicals with trifluoroacetic acid under atmospheric conditions. Chemical Physics Letters 1994, 227, 39-44.

(104) Thompson, R.S.; Stewart, K.M.; Gillings, E. Trifluoroacetic acid: Accumulation from aqueous solution by the roots of Sunflower (Helianthus annuus), Brixham Environmental Laboratory, Report BL5042/B.

(105) Thompson, R.S.; Stewart, K.M.; Gillings, E. Sodium trifluoroacetate: Toxicity to wheat (Triticum aestivum) in relation to bioaccumulation (by aqueous exposure of the roots), Brixham Environmental Laboratory, Report BL5473/B.

(106) Smyth, D.V.; Thompson, R.S.; Gillings, E. Sodium trifluoroacetate: Toxicity tot he marine alga Skeletonema costatum, Brixham Environmental Laboratory, Report BL4980/B.

(107) Boutonnet, J.C.; Bingham, P.; Calamari, D.; de Rooij, C.; Franklin, J.; Kawano, T.; Libre, J.-M.; McCulloch, A.; Malinverno, G.; Odom, J.M.; Rusch, G.M.; Smythe, K.; Sobolev, I.; Thompson, R.; Tiedje, J.M. Environmental risk assessment of trifluoroacetic acid. Human and Ecological Risk Assessment 1999, 5, 59-124.

(108) Tromp, T.K.; Ko, M.K.W.; Rodriguez, J.M.; Sze, N.D. Potential accumulation of a CFC- replacement degradation product in seasonal wetlands. Nature 1995, 376, 327-330.

(109) Jordan, A.; Frank, H. Trifluoroacetate in the environment. Evidence for sources other than HFC/HCFCs. Environmental Science and Technology 1999, 33, 522-527.

(110) Goss, K.-U. The pKa values of PFOA and other highly fluorinated carboxylic acids. Environmental Science and Technology 2008, 42, 456-458.

66

(111) Burns, D.C.; Ellis, D.A.; Li, X.; McMurdo, C.J.; Webster, E. Experimental pKa determination for perfluorooctanoic acid (PFOA) and the potential impact of pKa concentration dependence on laboratory-measured partitioning phenomena and environmental modeling. Environmental Science and Technology 2008, 42, 9283-9288.

(112) Cheng, J.; Psillakis, E.; Hoffman, M.R.; Colussi, A.J. Acid dissociation versus molecular association of perfluoroalkyl oxoacids: Environmental implications. Journal of Physical Chemistry A 2009, 113, 8152-8156.

(113) McMurdo, C.J.; Ellis, D.A.; Webster, E.; Butler, J.; Christensen, R.D.; Reid, L.K. Aerosol enrichment of the surfactant PFO and mediation of the water-air transport of gaseous PFOA. Environmental Science and Technology 2008, 42, 3969-3974.

(114) Armitage, J.M.; Macleod, M.; Cousins, I.T. Modeling the global fate and transport of perfluorooctanoic acid (PFOA) and perfluorooctanoate (PFO) emitted from direct sources using a multispecies mass balance model. Environmental Science and Technology 2009, 43, 1134- 1140.

(115) Young, C.J.; Furdui, V.I.; Franklin, J.; Koerner, R.M.; Muir, D.C.G.; Mabury, S.A. Perfluorinated acids in Arctic snow: New evidence for atmospheric formation. Environmental Science and Technology 2007, 41, 3455-3461.

(116) Ellis, D.A.; Webster, E. Response to comment on "Aerosol enrichment of the surfactant PFO and mediation of the water-air transport of gaseous PFOA". Environmental Science and Technology 2009, 43, 1234-1235.

(117) De Silva, A.O.; Muir, D.C.G.; Mabury, S.A. Distribution of perfluorocarboxylate isomers in select samples from the North American environment. Environmental Toxicology and Chemistry 2009, 28, 1801-1814.

(118) Yarwood, G.; Kemball-Cook, S.; Keinath, M.; Waterland, R.L.; Korzeniowski, S.H.; Buck, R.C.; Russell, M.H.; Washburn, S.T. High-resolution atmospheric modeling of fluorotelomer alcohols and perfluorocarboxylic acids in the North American troposphere. Environmental Science and Technology 2007, 41, 5756-5762.

(119) Armitage, J.; Cousins, I.T.; Buck, R.C.; Prevedouros, K.; Russell, M.H.; Macleod, M.; Korzeniowski, S.H. Modeling global-scale fate and transport of perfluorooctanoate emitted from direct sources. Environmental Science and Technology 2006, 40, 6969-6975.

(120) Wania, F. Global mass balance analysis of the source of perfluorocarboxylic acids in the Arctic Ocean. Environmental Science and Technology 2007, 41, 4529-4535.

(121) Schenker, U.; Scheringer, M.; Macleod, M.; Martin, J.W.; Cousins, I.T.; Hungerbühler, K. Contribution of volatile precursor substances to the flux of perfluorooctanoate to the Arctic. Environmental Science and Technology 2008, 42, 3710-3716.

(122) Caliebe, C.; Gerwinski, W.; N., T.; Huhnerfuss, H. In Fluoros: Toronto, Canada, 2005; Vol. Available at http://www.chem.utoronto.ca/symposium/fluoros/.

67 (123) Armitage, J.M.; Macleod, M.; Cousins, I.T. Comparative assessment of the global fate and transport pathways of long-chain perfluorocarboxylic acids (PFCAs) and perfluorocarboxylates (PFCs) emitted from direct sources. Environmental Science and Technology 2009, In press.

(124) Li, Y.F.; Macdonald, R.W. Sources and pathways of selected organochlorine pesticides to the Arctic and the effect of pathway divergence on HCH trends in biota: a review. Science of the Total Environment 2005, 342, 87-106.

(125) Butt, C.M.; Muir, D.C.G.; Stirling, I.; Kwan, M.; Mabury, S.A. Rapid response of Arctic ringed seals to changes in perfluoroalkyl production. Environmental Science and Technology 2007, 41, 42-49.

(126) Stock, N.L.; Furdui, V.I.; Muir, D.C.G.; Mabury, S.A. Perfluoroalkyl contaminants in the Canadian Arctic: Evidence of atmospheric transport and local contamination. Environmental Science and Technology 2007, 41, 3529-3536.

(127) Butt, C.M.; Mabury, S.A.; Muir, D.C.G.; Braune, B.M. Prevalence of long-chained perfluorinated carboxylates in seabirds from the Canadian Arctic. Environmental Science and Technology 2007, 41, 3521-3528.

(128) Verreault, J.; Berger, U.; Gabrielsen, G.W. Trends of perfluorinated alkyl substances in herring gull eggs from two coastal colonies in northern Norway: 1983-2003. Environmental Science and Technology 2007, 41, 6671-6677.

(129) Martin, J.W.; Smithwick, M.M.; Braune, B.M.; Hoekstra, P.F.; Muir, D.C.G.; Mabury, S.A. Identification of long-chain perfluorinated acids in biota from the Canadian arctic. Environmental Science and Technology 2004, 38, 373-380.

CHAPTER THREE

Atmospheric Lifetime and Global Warming Potential of a Perfluoropolyether

Cora J. Young, Michael D. Hurley, Timothy J. Wallington and Scott A. Mabury

Published in: Environ. Sci. Technol. 2006 40:2242-2246

Reproduced with permission from Environmental Science and Technology

Copyright ACS 2006

67 68 3.1 Introduction

Recognition of the adverse environmental impact of chlorofluorocarbon (CFC) and Halon release into the atmosphere has led to an international effort to replace these compounds with environmentally acceptable alternatives. Perfluoropolyethers (PFPEs), along with hydrofluoroethers (HFEs), have been used as replacements for CFCs as heat transfer fluids. They are also used for electronic reliability testing. These compounds lack Cl, so they do not contribute to the catalytic destruction of the ozone layer. However, PFPEs and HFEs may be associated with other environmental risks.

Within the Earth’s atmosphere, there are no major components that absorb significant thermal radiation between approximately 750 and 1250 cm-1, creating a window through which heat emitted from the Earth can escape. This region is termed the “atmospheric window”, and anthropogenic compounds in the atmosphere that absorb within this region have the ability to block the escape of terrestrial radiation. Radiative efficiency is a measure of the ability of a

F F F

F3C OC C O C O CF3 m n CF3 F F

Figure 3.1. General structure of PFPMIEs. compound to disrupt the energy balance of the earth, for example, by absorbing IR radiation within the atmospheric window. Fluorinated ethers can contribute to global warming, because both the C-F and C-O bonds absorb in this range. PFPEs and HFEs would be expected to have high radiative efficiencies, as they have multiple C-F and C-O bonds.

Several studies have focused on the atmospheric fate of HFEs, and they have been shown to have lifetimes that vary between days and hundreds of years (1-5). These compounds degrade in the atmosphere through the abstraction of a hydrogen atom by hydroxyl radicals. PFPEs have no abstractable hydrogens, nor do they have any sites for OH addition. Thus, it is likely that their lifetimes will be much greater than those of HFEs. The atmospheric lifetimes of PFPEs are expected to ultimately be limited by photolysis in the upper atmosphere. A long lifetime, in combination with high radiative efficiency, results in a high global warming potential (GWP) for a compound.

69 Perfluoropolymethylisopropyl ethers (PFPMIEs) are sold as mixtures according to their boiling point. The fraction selected here boils at 70°C, has an average molecular weight of 410 and is composed primarily of CF3OCF(CF3)CF2OCF2OCF3 (molecular weight = 386), with smaller amounts of CF3OCF(CF3)CF2OCF2OCF2OCF3 (molecular weight = 452) and longer- chain PFPMIEs (Figure 3.1). Despite the high molecular weight, PFPMIE is volatile and is expected to escape into the atmosphere. Currently, there are no available data concerning the atmospheric fate of PFPEs. To provide such data, the atmospheric chemistry of PFPMIE was investigated. Specifically, the following information was determined using smog chamber/FTIR techniques and by analogy to other long-lived perfluorinated compounds: (i) the kinetics of reactions with chlorine atoms and hydroxyl radicals, (ii) the infrared spectrum, (iii) the atmospheric lifetime, and (iv) the global warming potential.

3.2 Experimental

3.2.1 Chemical preparation

A commercial mixture of Galden HT70, obtained from Solvay Solexis (Thorofare, New Jersey), was purified using fractional distillation to remove any hydrogen-containing impurities. Purity of the distillate was confirmed using 1H NMR, where no evidence of hydrogen impurities was observed. The distillate was still a mixture, because fractional distillation could not isolate a single PFPE. The mixture was composed mainly of CF3OCF(CF3)CF2OCF2OCF3, with smaller amounts of CF3OCF(CF3)CF2OCF2OCF2OCF3 and longer-chain analogues.

3.2.2 Kinetics

Experiments were performed in a 140 L Pyrex reactor interfaced to a Mattson Sirus 100 FTIR spectrometer. The reactor was surrounded by 22 fluorescent blacklamps (GE F15T8-BL), which were used to photochemically initiate the experiments. Chlorine atoms were produced by the photolysis of molecular chlorine:

Cl2 + hv → Cl + Cl

OH radicals were produced by the photolysis of CH3ONO in air:

CH3ONO + hv → CH3O(•) + NO

CH3O(•) + O2 → HO2 + HCHO

70

HO2 + NO → OH + NO2

In relative rate experiments, the following reactions take place:

OH/Cl + reactant → products

OH/Cl + reference → products

It can be shown that:

⎛[X] ⎞ ⎛ k ⎞ ⎛[ref] ⎞ ⎜ t ⎜ x ⎟ ⎜ t ln⎜ ⎟=⎜ ⎟ln⎜ ⎟ [X] kref [ref]t ⎝ t0⎠ ⎝ ⎠ ⎝ 0⎠

where [X]t0, [X]t, [ref]t0, and [ref]t are the concentrations of the compound of interest and reference at times t0 and t, and kX and kref are the rate constants for the reactant and the reference.

Reaction mixtures for k(Cl + PFPMIE) consisted of 88 mTorr of PFPMIE, 1.7 Torr of Cl2, and

2.6 mTorr of CF2ClH in 700 Torr of N2 diluent. Mixtures for k(OH + PFPMIE) consisted of 88-

176 mTorr of PFPMIE, 100-240 mTorr of CH3ONO, and 1.5-2.6 mTorr of C2H2 in 700 Torr total pressure of air diluent. All experiments were performed at 296 ± 1K.CH3ONO was synthesized by the dropwise addition of concentrated sulfuric acid to a saturated solution of

NaNO2. All other reagents were obtained from commercial sources. C2H2 was selected as the reference compound in the OH relative rate experiments because it is the least-reactive compound whose loss can be monitored with high precision in the present system.

Concentrations of reactants and products were monitored by FTIR spectroscopy. IR spectra -1 were derived from 32 coadded interferograms with a spectral resolution of 0.25 cm and an analytical path length of 27.1 m. To check for the unwanted loss of compounds via heterogeneous reactions, reaction mixtures were left to stand in the chamber for 60 min without irradiation; there was no observable (<1%) loss of the reactants.

3.3 Results and Discussion

3.3.1 Kinetics

The kinetics of reaction 1 were measured relative to those of reaction 2:

Cl + PFPMIE → products (1)

71

Cl + CF2ClH → products (2)

UV irradiation of the gas mixture for 4 min led to a 94% consumption of CF2ClH but no -15 3 -1 -1 observable loss (<2%) of PFPMIE. Using k2 = 1.7 × 10 cm molecule s (6), we derive an -17 3 -1 -1 upper limit of k1 < 2 × 10 cm molecule s . The loss of PFPMIE via reaction with Cl atoms is not of atmospheric significance.

The kinetics of reaction 3 were measured relative to those of reaction 4:

OH + PFPMIE → products (3)

OH + C2H2 → products (4)

UV irradiation of the gas mixture led to a loss of C2H2 but no discernible loss of PFPMIE. By analogy to the substantial existing database for fluorinated compounds, if PFPMIE were to be oxidized by reaction with OH radicals, it would be

Table 3.1. Photolysis rate constants and lifetimes calculated between 162 and 210 nm for CHF2-

O-CHF2 (5), used to estimate those of PFPMIE, at different altitudes

Altitude (km) Rate constant (s-1) Lifetime (years)

0 0 N/A 20 0 N/A 40 8.7 × 10-12 3.6 × 103 80 2.0 × 10-10 1.5 × 102 120 4.3 × 10-10 7.3 × 101

expected to result in conversion of PFPMIE into COF2 via an “unzipping” mechanism (7). IR product features attributable to COF2 were sought but not found, and an upper limit of 0.0565 mTorr was established for the formation of this compound. As discussed earlier, CF3OCF

(CF3)CF2OCF2OCF3 is the main component of PFPMIE. Degradation initiated by C-C or C-O bond scission in the CF3OCF(CF3)-or -CF2OCF2OCF3 moieties will give either 2 or 4 molecules of COF2, respectively. Assuming formation of 2 molecules of COF2 we conclude that <0.028 mTorr of PFPMIE was consumed (i.e., <0.032% of the initial PFPMIE concentration of 88.8 -4 mTorr). In this experiment, C2H2 loss was 32.5%. We conclude that k3/k4 < 8.1 × 10 . Using k4

72

-13 3 -1 -1 -16 3 -1 -1 = 8.45 × 10 cm molecule s (8), gives k3 < 6.8 × 10 cm molecule s . Using a global weighted-average OH concentration of 1.0 × 106 molecules cm-3 (9) leads to a lifetime of PFPMIE with respect to reaction with OH radicals of greater than 46 years.

The atmospheric lifetime of a chemical is often approximated by tropospheric degradation via reaction with OH radicals. This study showed that PFPMIE has a minimum lifetime of 46 years with respect to reaction with OH radicals, but it is impossible to determine how much greater. For compounds with long tropospheric lifetimes, including other perfluorinated organic compounds, photolysis in the upper-atmosphere plays a dominant role in atmospheric degradation (10). It is likely that upper-atmosphere photolysis is the main degradation pathway for PFPMIE and that its overall atmospheric lifetime is much greater than 46 years.

3.3.2 Photolysis of PFPMIE

Fluorinated ethers absorb in the vacuum UV (VUV) region with an intensity that decreases with increasing fluorination (5). As discussed by Ravishankara et al. (10), directly measuring the UV spectrum of perfluorinated compounds is difficult, because the cross sections are extremely small. We used the published CHF2-O-CHF2 absorption cross-section data in the region 162-210 nm (5) to provide an upper limit for the rate of PFPMIE photolysis at 40, 80, and 120 km altitudes. Solar flux data was calculated using the procedure described by Minschwaner and coworkers (11), and unity quantum yield of photolysis was assumed. The photolysis rate constant between 162 and 210 nm was calculated using the following equation: 210nm J = ∫φ(λ) σ(λ) I(λ) dλ 162nm which can be approximated as:

210nm J = ∑φσI Δλ 162nm where φ(λ) is the quantum yield, σ(λ) is the absorption cross section and I(λ) is the solar flux at a given wavelength. Photolysis rate constants at different altitudes are shown in Table 3.1. The photolysis of CHF2-O-CHF2 following absorption at 162-210 nm is calculated to occur at a rate of 8.7 × 10-12 s-1 at 40 km, which corresponds to a local lifetime of 3.6 × 103 years. Photolysis should proceed at a faster rate at 80 and 120 km, with rate constants of 2.0 × 10-10 and 4.3 × 10- 10 -1 s and local lifetimes of 160 and 73 years, respectively. CHF2-O-CHF2 does not absorb at

73 wavelengths above 182 nm. At altitudes less than 80 km, the photon flux below 180 nm is feeble, so the photolysis of PFPMIE would not be expected to proceed at an appreciable rate below 80 km, as demonstrated by the long lifetime calculated at 40 km.

Orkin et al. (5) noted that increasing fluorination, even in the α position, dramatically decreased the VUV absorption of ethers in the region 162-210 nm. For example, there was no -22 2 -1 observable absorption (σ < 10 cm molecule ) by CHF2-O-CF3 at 162-210 nm. Absorption by

CHF2-O-CF3 at 162 nm is at least a factor of 5 times weaker than that displayed by CHF2-O-

CHF2. The most common molecules in the PFPMIE sample have three or four ether linkages and would be expected to absorb more strongly than a compound with just one ether linkage. However, the decreased absorption expected from fluorination is expected to be the dominant effect. Hence, it seems reasonable to conclude that the lifetimes presented in Table 3.1 are lower limits for PFPMIE photolysis via VUV radiation at 162-210 nm.

In the calculations presented above, we have only considered absorption by CHF2-O-

CHF2 above 162 nm. For altitudes of 40 km and below, there is no solar flux at wavelengths below 162 nm and neglecting absorption below 162 nm has no impact on the calculated rate of photolysis. However, for altitudes of 80 km and above, there is significant solar flux at wavelengths shorter than 162 nm and absorption below 162 nm must be considered. In particular, at 80 km altitude, Lyman-α radiation at 121.6 nm is available. At this wavelength, a gap in the absorption of oxygen, in combination with enhanced solar flux, results in a high incidence of light. While there are no data available for the absorption cross sections of fluorinated ethers, it is well-established that long-chain perfluoroalkanes absorb strongly at 121.6 nm and, as a result, are photolyzed within a few days at around 80 km (10). It seems reasonable to assume that PFPMIE will also absorb significantly at 121.6 nm and that the lifetime with respect to photolysis at 80 km altitude and above will be on the order of a few days. Nonetheless, when estimating the lifetime of a compound that is persistent in the lower atmosphere, such as PFPMIE, the location as well as reactivity of the compound must be taken into account. The degradation of PFPMIE requires that it be present in the mesosphere before it can undergo photolysis. Only 2 × 10-5 of the atmosphere is found above a 75 km altitude. Consequently, air must cycle through the mesosphere many thousands of times before the entire atmospheric burden would be depleted (10). Although the absolute photolysis rate constant at 80 km is likely to be fast, the time taken for air to cycle through this altitude leads to a long

74 lifetime. Compounds for which the main degradation pathway is upper-atmosphere photolysis have been estimated to have lifetimes of at least 800 years (10). This is the minimum lifetime for PFPMIE.

3.3.3 IR Spectrum and global warming potential of PFPMIE

IR spectra were recorded at 296 K using 0.9-2.4 mTorr of PFPMIE in 700 Torr of air diluent. Typical peak absorbances were in the range 0.05-0.7 and scaled linearly with the PFPMIE concentration. The absolute absorption spectrum is shown in Figure 3.2. The integrated cross section (650-1500 cm-1) of PFPMIE is 5.92 × 10-16 cm2 molecule-1 cm-1. Uncertainties in the cross-section measurement arise from the following sources: sample concentration (2%), sample purity (2%), path length (1.5%), spectrum noise ±(10-20 cm2 molecule-1), and residual baseline offset after subtraction of the background (1.5%). From these individual uncertainties, the total (random) uncertainty in the integrated absorption cross section is ±4%. We prefer to quote a conservative uncertainty of 5%. Hence, the integrated cross section is (5.92 ± 0.30) × 10- 16 cm2 molecule-1 cm-1.

1.2 3.5 -1 )

3.0 -1 1.0

) 2.5 -1 molecule

0.8 2 2.0 cm -18 molecule

2 0.6 1.5 (10 -1 cm ) -1 -17 1.0 0.4 (cm (10 -2 σ 0.5 0.2 W m

0.0 -3 Radiative forcing per unit cross section forcing per unit cross Radiative 10 0.0 800 1000 1200 1400

-1 Wavenumber (cm )

Figure 3.2. IR absorption cross section for PFPMIE (solid line) and n-C6F14 (light dashed line) shown with the radiative forcing per unit cross section of the atmosphere (12) (dotted line).

75 Pinnock et al. (12) have presented a simple method that can be used to estimate instantaneous radiative efficiency from IR absorption spectra. In this method, the region -1 between 0 and 2500 cm is divided up into 250 bands 10 cm-1 wide. The instantaneous radiative efficiency for a 0 → 1 ppbv change in atmospheric concentration of the compound can then be calculated using the expression: 250 Forcing = 10(cm-1) i Fi Σ σ av σ i=1 i 2 -1 i where σ av is the average absorption cross section in band i in units of cm molecule and F σ is the radiative forcing per unit cross section per wavenumber in band i in units of W m-2 (cm-1)-

1(cm2 molecule-1)-1. Using this method, the IR spectrum of PFPMIE, shown in Figure 3.2, and the IR spectrum of CFC-11, reported elsewhere (13), we calculate instantaneous forcings for -2 -1 PFPMIE and CFC-11 of 0.65 and 0.26 W m ppb , respectively. The halocarbon global- warming potential, HGWP, for PFPMIE (relative to CFC-11) can be estimated using the expression:

⎛ IF ⎞⎛τ M ⎞⎛1 exp( −− t τ )/ ⎞ HGWP =⎜ PFPMIE⎟⎜ PFPMIE CFC−11⎟⎜ PFPMIE ⎟ PFPMIE ⎜ ⎟⎜τ ⎟⎜ −− τ ⎟ ⎝ IFCFC−11 ⎠⎝ CFC−11 MPFPMIE ⎠⎝1 exp( t CFC−11)/ ⎠

where IFPFPMIE,IFCFC-11, MPFPMIE, MCFC-11, τPFPMIE, and τCFC-11 are the instantaneous radiative efficiencies, molecular weights, and atmospheric lifetimes of PFPMIE and CFC-11, respectively, and t is the time horizon over which the forcing is integrated. Assuming that

τPFPMIE is 800 years, which was discussed above, and τCFC-11 is 45 years (14), we estimate that the HGWP of PFPMIE (relative to CFC-11) is 1.95 for a 100 year horizon. Relative to CO2, the GWP of CFC-11 on a 100 year time horizon is 4600 (14). Hence, we estimate that relative to

CO2, the GWP of PFPMIE is 9000 for a 100 year time horizon.

The radiative efficiency of PFPMIE, at 0.65 W m-2 ppb-1, is large relative to halogenated compounds found in the atmosphere. To date, the compound with the highest radiative efficiency that has been detected in the atmosphere is SF5CF3, with a radiative efficiency of 0.59

W m-2 ppb-1 (15).

As seen from Table 3.2, the radiative efficiency of PFPMIE is approximately 30% greater than that of the perfluoroalkane with the same number of C-F bonds, n-C6F14. This may reflect the fact that, in addition to the C-F bonds, PFPMIE has three C-O bonds that would be

76 Table 3.2. Lifetimes and GWPs of perfluorohexane and selected fluorinated ethers (14).

Instantaneous GWP (relative to CFC-11) Number radiative of C-F efficiency Lifetime 20 year 100 year 500 year Name Structure bonds (W m-2 ppb-1) (years) horizon horizon horizon CF OCF(CF )CF O PFPMIE 3 3 2 14 0.65 800 1.00 1.84 6.56 CF2OCF3 Perfluoro CF (CF ) CF 14 0.49 3200 0.96 1.96 8.25 hexane 3 2 4 3 HFE- C F OCH 9 0.31 5 0.21 0.08 0.08 7100 4 9 3 HFE- C F OCH CH 9 0.30 0.77 0.03 0.01 0.01 7200 4 9 2 3

HG-01 CHF2OCF2CF2OCHF2 8 0.87 6.2 0.75 0.33 0.28

HG-10 CHF2OCF2OCHF2 6 0.66 12.1 1.19 0.59 0.53

expected to absorb within the atmospheric window. Only a few compounds have been shown to have radiative efficiencies that exceed that of PFPMIE, all of which are HFEs. It is interesting to

note that H-Galden01, (HG-01), CHF2OCF2CF2OCHF2, which has 6 fewer C-F bonds and 1 less ether linkage, nevertheless has a radiative efficiency which is 34% greater than that of PFPMIE (see Table 3.2). Simplistically, it would be expected that the greater the number of C-F and C-O bonds, the greater the absorption, and the greater the radiative efficiency. However, this is not supported by measurements made to date. The integrated cross sections for HG-01, reported by

Cavalli et al. (16) and Myhre et al. (17), (6.04 ( 0.13) × 10-16 and (5.0 ( 0.5) × 10-16 cm molecule- -16 -1 1 are very similar to the value of (5.92 ( 0.30) × 10 cm molecule measured herein for PFPMIE. H-G01 has a larger radiative efficiency value than PFPMIE because a greater fraction of its IR absorption is located at lower frequencies where there is more of a radiative forcing impact (see Figure 3.2). The effect of fluorination on these bonds might provide some explanation of this observation. The vibration frequency of a given bond increases with the bond strength, so stronger bonds absorb at higher wavenumbers. In the case of C-O bonds, α- fluorination increases the strength of the bond, due to the electron withdrawing nature of fluorine (18). Successive fluorine substitution increases this effect. Thus, as an ether becomes more fluorinated, the C-O bond strength should increase and hence absorption wavenumber should increase and the molecule’s absorption may be shifted into a region where the radiative

77 forcing impact is lower (e.g., moving from 1200 to 1250 cm-1, see Figure 3.2). A similar case can be made for C-F bonds. As a carbon atom becomes more fluorinated, in general, each C-F bond becomes stronger (18) and the absorption may move to higher frequency. The range for C- F bonds is given as 1000-1400 cm-1 (19). However, the minimum absorption for aliphatic monofluorinated compounds is 1000 cm-1, while that for polyfluorinated alkanes is 1090 cm-1. A trend was also observed by Good and Francisco (20) of increasing C-O and C-F bond strengths between CH3-O-CH3 and CHF2-OCF3. It is also interesting to note that ethers with a perfluorinated sidechain and a hydrocarbon sidechain (i.e., HFE-7100 and -7200, shown in Table 3.2) have radiative efficiencies that are substantially lower than the H-Galden compounds. Further work is needed to provide a more fundamental understanding of the observed radiative efficiency trends in Table 3.2.

PFPMIE has a long lifetime and a high radiative efficiency leading to a large GWP. PFPMIE has a longer atmospheric lifetime than CFC-11, and hence, the HGWP of PFPMIE increases with the time horizon. As seen from Table 3.2, HFEs have radiative efficiency values which are comparable to (within a factor of 2), but atmospheric lifetimes which are much shorter than, those of PFPMIE. As a result, in general, HFEs have lower GWPs than PFPMIE, and this difference increases with the time horizon. The structure of a molecule affects both its lifetime and radiative efficiency. The results presented herein may assist with the future design of molecules which have shorter lifetimes and lower radiative efficiencies but which still retain useful function.

3.4 Acknowledgments

The authors thank Jessica D’eon, Jamie Donaldson, John Sagebiel, Dan Mathers, and Stan Skonieczny and also Vladimir Orkin for UV cross-section data. Funding was provided by the Natural Sciences and Engineering Research Council of Canada.

78 3.5 Sources Cited

(1) Christensen, L.K.; Sehested, J.; Nielsen, O.J.; Bilde, M.; Wallington, T.J.; Gushin, A.; Molina, L.T.; Molina, M.J. Atmospheric chemistry of HFE-7200 (C4F9OC2H5): Reaction with OH radicals and fate of C4F9OCH2CH2O and C4F9OCHOCH3 radicals. J. Phys. Chem. A 1998, 102, 4839-4845.

(2) Sulbaek Andersen, M.P.; Nielsen, O.J.; Wallington, T.J.; Hurley, M.D.; DeMore, W.B. Atmospheric chemistry of CF3OCF2CF2H and CF3OC(CF3)2H: Reaction with Cl atoms and OH radicals, degradation mechanism, global warming potentials, and empirical relationship between k(OH) and k(Cl) for organic compounds. J. Phys. Chem. A 2005, 109, 3926-3934.

(3) Wallington, T.J.; Hurley, M.D.; Nielson, O.J.; Sulbaek Andersen, M.P. Atmospheric chemistry of CF3CFHCF2OCF3 and CF3CFHCF2OCF2H: Reaction with Cl atoms and OH radicals, degradation mechanism and global warming potentials. J. Phy. Chem. A 2004, 108, 11333-11338.

(4) Oyaro, N.; Sellevag, S.R.; Nielsen, C.J. Study of the OH and Cl-initiated oxidation, IR absorption cross-section, radiative forcing, and global warming potential of four C4- hydrofluoroethers. Environ. Sci. Technol. 2004, 38, 5567-5576.

(5) Orkin, V.L.; Villenave, E.; Huie, R.E.; Kurylo, M.J. Atmospheric lifetimes and global warming potentials of hydrofluoroethers: Reactivity toward OH, UV spectra, and IR absorption cross sections. J. Phys. Chem. A 1999, 103, 9770-9779.

(6) Sokolov, O.; Hurley, M.D.; Wallington, T.J.; Kaiser, E.W.; Platz, J.; Nielsen, O.J.; Berho, F.; Rayez, M.-T.; Lesclaux, R. Kinetics and mechanism of the gas-phase reaction of Cl atoms with benzene. J. Phys. Chem. A 1998, 102, 10671-10675.

(7) Wallington, T.J.; Schneider, W.F.; Sehested, J.; Bilde, M.; Platz, J.; Nielsen, O.J.; Christensen, L.K.; Molina, M.J.; Molina, L.T.; Wooldridge, P.W. Atmospheric chemistry of HFE-7100 (C4F9OCH3): reaction with OH radicals, UV spectra and kinetic data for C4F9OCH2 and C4F9OCH2O2 radicals, and the atmospheric fate of C4F9OCH2O radicals. J. Phys. Chem. A 1997, 101, 82648274.

(8) Sørensen, M.; Kaiser, E.W.; Hurley, M.D.; Wallington, T.J.; Nielsen, O.J. Kinetics of the reaction of OH radicals with acetylene in 25-8000 Torr of air at 296K. Int. J. Chem. Kinet. 2003, 35, 191-197.

(9) Prinn, R.G.; Huang, J.; Weiss, R.F.; Cunnold, D.M.; Fraser, P.J.; Simmonds, P.G.; McCulloch, A.; Salameh, P.; O’Doherty, S.; Wang, R.H.J.; Porter, L.; Miller, B.R. Evidence for substantial variation of atmospheric hydroxyl radicals in the past two decades. Science 2001, 292, 1882-1888.

(10) Ravishankara, A.R.; Solomon, S.; Turnipseed, A.A.; Warren, R.F. Atmospheric lifetimes of long-lived halogenated species. Science 1993, 259, 194-199.

(11) Schneider, W.F.; Wallington, T.J.; Minschwaner, K.; Stahlberg, E. A. Atmospheric chemistry of CF3OH: is photolysis important? Environ. Sci. Technol. 1995, 29, 247-250.

79 (12) Pinnock, S.; Hurley, M.D.; Shine, K.P.; Wallington, T.J.; Smyth, T.J. Radiative forcing of climate by hydrochlorofluorocarbons and hydrofluorocarbons. J. Geophys. Res. 1995, 100, 2322723238.

(13) Ninomiya, Y.; Kawasaki, M.; Gushin, A.; Molina, L.T.; Molina, M.; Wallington, T.J. Atmospheric chemistry of n-C3F7OCH3: Reaction with OH radicals and Cl atoms and atmospheric fate of n-C3F7OCH2O. Environ. Sci. Technol. 2000, 34, 2973-2978.

(14) Houghton, J.T.; Ding, Y.; Griggs, D.J.; Noguer, M.; van der Linden, P.J.; Dai, X.; Maskell, K.; Johnson, C.A. Climate Change 2001: The Scientific Basis. Contribution of Working Group 1 to the Third Assessment Report of the Intergovernmental Panel on Climate Change; Cambridge University Press: New York, 2001.

(15) Nielsen, O.J.; Nicolaisen, F.M.; Bacher, C.; Hurley, M.D.; Wallington, T.J.; Shine, K.P. Infrared spectrum and global warming potential of SF5CF3. Atmos. Environ. 2002, 36, 12371240.

(16) Cavalli, F.; Glasius, M.; Hjorth, J.; Rindone, B.; Jensen, N. R. Atmospheric lifetimes, infrared spectra and degradation products of a series of hydrofluoroethers. Atmos. Environ. 1998, 32, 3767-3773.

(17) Myhre, G.; Nielsen, C.J.; Powell, D.L.; Stordal, F. Infrared absorption cross section, radiative forcing, and GWP of four hydrofluoro(poly)ethers. Atmos. Environ. 1999, 33, 4447- 4458.

(18) Banks, R.E. Organofluorine Chemistry: Principles and Commercial Applications; Plenum: New York, 1994.

(19) Socrates, G. Infrared Characteristic Group Frequencies, 2nd ed.; Wiley: New York, 1994.

(20) Good, D.A.; Francisco, J.S. Structure and vibrational spectra of chlorofluorocarbon substitutes: An experimental and ab initio study of fluorinated ethers CHF2OCF3 (E125), CHF2OCHF2 (E134), and CH3OCF3 (E143A). J. Phys. Chem. A 1998, 102, 18541864.

CHAPTER FOUR

Molecular Structure and Radiative Efficiency of Fluorinated Ethers: A Structure-Activity Relationship

Cora J. Young, Michael D. Hurley, Timothy J. Wallington and Scott A. Mabury

Published in: J. Geophys. Res. 2008 113:D24301

An edited version of this paper was published by AGU.

Copyright 2008 American Geophysical Union

80 81 4.1 Introduction

As a result of the Montreal Protocol and subsequent conventions, ozone-depleting substances, such as chlorofluorocarbons (CFCs) and Halons, are being phased out of production and use (1). This has led to a search for environmentally acceptable alternatives that do not contain chlorine or bromine. Fluorinated ethers are a class of potential CFC replacements and may be released into the atmosphere during use. Prior to the widespread industrial use of such compounds, assessments of their impact on climate change are needed.

Radiative forcing is a metric used to estimate and compare the impacts of climate change processes and is defined as the change in net irradiance (W m-2) at the tropospause caused by a change in an external driver of the climate system. When applied to greenhouse gases the change normally considered is a 1 ppb increase in the concentration of the gas in the troposphere and is termed radiative efficiency with units of W m-2 ppb-1. The overall radiative forcing of a greenhouse gas is determined by multiplying the radiative efficiency of the compound by its atmospheric concentration. Radiative forcing is a useful metric to assess the relative impacts on climate of long-lived greenhouse gases, such as fluorinated ethers (2). Compounds in the atmosphere that absorb within the optically thin spectral region of the atmosphere at approximately 8 – 14 μm (1250 – 700 cm-1), termed the “atmospheric window”, hinder the escape of terrestrial radiation and increase the temperature at the surface of the Earth. Vibrational stretching transitions associated with C-F bonds occur in the atmospheric window and consequently highly fluorinated compounds, particularly fluorinated ethers, tend to have high radiative efficiency values. The compound with the largest radiative efficiency value considered in the latest IPCC report is a fluorinated ether, CHF2OCF2OC2F4OCHF2, at 1.37 W m-2 ppb-1 (2,3).

Radiative efficiency provides a measure of the potential of a compound to affect climate, yet that potential can only be realized if the compound is long-lived in the atmosphere. The overall likelihood of long-lived greenhouse gases to affect climate is assessed through the global warming potential (GWP), which is a function of both radiative efficiency and lifetime. Polyfluorinated compounds, including fluorinated ethers, are relatively unreactive in the troposphere, as the electron-withdrawing nature of fluorine reduces the probability of hydrogen abstraction by the hydroxyl radical. The presence of an ether bond generally increases the reactivity of organic compounds with hydroxyl radicals, yet atmospheric lifetimes are still

82 typically on the order of years (2). In a recent series of papers, Blowers et al. (4-6) have used quantum chemical methods to investigate the influence of molecular structure on atmospheric lifetime (5,6) and to determine whether radiative efficiency of HFEs could be accurately calculated (4). Structure-activity relationship methods to estimate the reactivity towards hydroxyl radicals and hence the atmospheric lifetime of organic compounds are available (7). However, SAR methods to estimate radiative forcing are not available. To assess the impact of fluorinated ethers on climate, it is critical to have a good understanding of the radiative efficiencies of these compounds. Blowers et al. (4) calculated the radiative efficiency of 53 HFEs using results from a moderate-sized quantum chemical calculation and showed that radiative efficiency is not directly proportional to the number of C-F bonds. Previous work from our group has shown that chemical surroundings of a functional group can affect its radiative efficiency (8).

Given that radiative efficiencies of fluorinated ethers are functions of their absorption cross-sections, knowledge of their infrared absorption properties is necessary to understand their radiative properties. However, because of interactions between group frequencies (8,9), the infrared spectra of polyfluorinated compounds are difficult to interpret and the relationship between structure and radiative efficiency is therefore complex. These complex characteristics may be better studied using computational techniques, where molecules can be systematically altered and subtle effects of molecular structure on infrared spectra assessed. The results of Blowers et al. (4) show that infrared absorption cross-sections of HFEs can be determined fairly accurately using density functional theory. Climate sensitivity can then be determined by entering this data into atmospheric models, which can vary in complexity from one-dimensional radiative-convective models to full three-dimensional atmosphere-ocean general circulation models (10). However, rather than having to calculate or measure the input data for each new compound of interest, it would be desirable to have a simple, accurate prescriptive treatment to estimate the radiative efficiency parameters.

Structure-activity relationships (SARs) are available to predict a number of chemical and physical properties from molecular structures. Such properties include octanol-water partitioning coefficient (11), vapour pressure (12) and the rate constant for atmospheric reaction with hydroxyl radicals (7). Using simple arithmetic equations, the user is able to quantitatively determine chemical and physical properties without the time or expense required for experiments or computations. These types of relationships are particularly useful in the

83 development of new compounds where authentic standards may not be available for experimentation. Radiative efficiency is a property that is of particular interest in the development of new compounds, especially in assessing the environmental suitability of replacements for ozone-depleting chemicals. An SAR to estimate radiative efficiency from chemical structure for fluorinated ethers would be valuable for these purposes. The objective of this study was to develop such an SAR for fluorinated ethers.

4.2 Methods

4.2.1 Experimental details

Measurements were performed in a 140 L Pyrex reactor interfaced to a Mattson Sirus 100 FTIR spectrometer. Infrared spectra were derived from 32 coadded interferograms with a spectral resolution of 0.25 cm-1 and an analytical path length of the infrared beam of 27.1 m. Compounds were subjected to repeated freeze/pump/thaw cycling before use. Infrared spectra were recorded at 296 K using 1.54 – 4.26 mTorr of CHF2CF2-O-CH3, CHF2CF2-O-CH2CH3,

CHF2CF2CH2-O-CH3, CF3CF2CF2-O-CH3, CF3CF2CH2-O-CHF2, CHF2CF2CH2-O-CHF2,

CF3CHFCF2-O-CH2CH3, CF3CHFCF2-O-CH2CF3, CF3CF2CF2-O-CHFCF3, CF3CHFCF2-O-

CH2CF2CH2F or CF3CHFCF2-O-CH2CF2CH2F in 700 Torr of air diluent. Vapor pressures were all greater than 0.5 Torr. These compounds will partition into the gas phase following release to the environment. All compounds were obtained from commercial sources and were of purity greater than 99%. Samples were subjected to repeated freeze-pump-thaw prior to measurements before being added to the chamber via a calibrated volume. At least four separate spectra were recorded for each compound, with one spectrum taken while warming from liquid nitrogen to ensure purity. Typical peak absorbances were in the range 0.05 – 0.7 and scaled linearly with the HFE concentration over the ranges explored here.

4.2.2 Computational details

Calculations were performed using GAUSSIAN 03 programs and basis sets (13). One hundred and fifty-four fluorinated ethers with the structures X(CX2)nOCX3 (n = 1 – 4) and

CX3OCX2OCX3 (where X was fluorine or hydrogen) were studied (full list available in supporting information). Ground state geometries and vibrational frequencies of all species were determined using density functional theory with the 6-311+g(d,p) basis set and B3LYP functionals. No correction factors were applied to the resulting frequencies. To calibrate the

84 intensity output of the computations, measured and computed infrared spectral intensities were compared. Infrared spectra for CHF2CF2CH2OCH3, CHF2CF2OCH3, and CF3CF2CF2OCH3 were obtained computationally within the studied set and were also measured experimentally. Infrared spectra were also computed for the remaining set of measured HFEs, which were used exclusively for calibration and were not included in the SAR. Comparing the experimental and computational infrared absorption intensity results for these eleven molecules gave a scaling factor (Icomp / Iexp) which was used to calibrate the spectral intensities determined from the density functional theory calculations. Radiative efficiencies determined from the eleven measured cross-sections were used as a training set to optimize peak width parameters to minimize differences between radiative efficiencies determined from measured spectra and those determined from the SAR. Variability can exist in the Gaussian calculated frequencies and intensities, particularly due to rotational conformers that can exist for HFEs. A study was conducted on the potential impacts of rotational conformers on the SAR. Vibrational frequencies and intensities were determined for three rotational conformers of

CHF2CF2CH2OCH3 and compared to those of the Gaussian-optimized geometry utilized in the SAR. Little difference was observed in the vibrational frequencies, with those of the C-F bond stretches varying by less than 7%, with an average of less than 1.5%. The calculated intensities showed greater variability, with differences averaging approximately 25%. However, the fitting of the modeled data to experimental observations through use of the training set minimizes the impacts of this variability on the utility of the SAR.

4.2.3 Determination of radiative efficiency

Pinnock et al. (14) have presented a method that can be used to estimate radiative forcing from infrared absorption cross-sections. In this method, the region between 0 and 2500 cm-1 is divided up into 250 bands, each 10 cm-1 wide. The cloudy-sky instantaneous radiative forcing for a 0→1 ppbv change in atmospheric concentration of the compound can then be determined from the radiative forcing per unit cross-section using a simple expression. The method of Pinnock et al. (14) was used in the present work to estimate radiative efficiency directly from measured cross-sections. Absorption cross-sections were determined for calculated infrared spectra by calibrating output intensities as described in Section 2.2. Using the measured absorption cross-sections as a training set, peaks corresponding to each stretch were assigned a width of 40 cm-1 and area was determined by multiplying the cross-sectional height by one-half the width. Radiative forcing per unit cross-section values (14) for the mean wavenumber of

85 each absorption were applied to the calculated cross-sectional areas to yield radiative efficiency per functional group (f) values.

4.3 Results and Discussion

4.3.1 Experimental infrared cross-sections

Absolute absorption spectra for the eleven HFEs that were measured in the 140 L Pyrex reactor are available in the Supporting Information (SI). Values for the absorption cross sections were obtained from regression lines derived from Beer’s Law plots. Correlation coefficients for all compounds were greater than 0.99. Integrated cross-sections for 650 – 2000 cm-1 are reported in Table 4.1 and range from 1.36 to 4.44 × 10-16 cm molecule-1. Uncertainties in the cross-section measurement arise from the following sources: sample concentration (2%),

Table 4.1. Measured integrated IR band strengths for hydrofluoroethers (HFEs). Comparison between radiative efficiencies determined from measured cross-sections and by use of the structure-activity relationship (SAR).

Integrated Radiative Efficiency (W m-2 ppb-1) Band Strength -1 % 650-2000 cm Experimental SAR -1 Difference Molecule (cm molecule )

CHF2CF2-O-CH3 1.77E-16 0.260 0.210 -24

CHF2CF2-O-C2H5 1.83E-16 0.275 0.210 -31

CHF2CF2CH2-O-CH3 1.36E-16 0.224 0.154 -45

C3F7-O-CH3 2.86E-16 0.348 0.376 7

CF3CF2CH2-O-CHF2 4.65E-15 0.414 0.424 2

CHF2CF2CH2-O-CHF2 3.61E-15 0.353 0.371 5

CF3CHFCF2-O-C2H5 2.51E-16 0.309 0.334 7

CF3CHFCF2-O-CH2CF3 3.57E-16 0.442 0.439 -1

CF3CF2CF2-O-CHFCF3 4.44E-16 0.576 0.562 -3

CF3CHFCF2-O-CH2CF2CHF2 3.60E-16 0.496 0.488 -2

CF3CHFCF2-O-CH2CF2CF3 2.46E-16 0.485 0.540 10

86

9 9

(a) CHF2CF2-O-CH3 (b) CHF2CF2-O-CH2CH3 6 6

3 3

0 0 (d) CF CF CF -O-CH (c) CHF2CF2CH2-O-CH3 3 2 2 3 6 6

3 3

0 0

) (e) CF3CF2CH2-O-CHF2 (f) CHF2CF2CH2-O-CHF2 ) -1 6 6 -1

3 3 molecule molecule

2 0 0 2 (g) CF CHFCF -O-CH CH (h) CF CHFCF -O-CH CF

cm 3 2 2 3 3 2 2 3 cm 6 6 -18 -18

(10 3 3 (10 σ σ 0 0

(i) CF3CF2CF2-O-CHFCF3 (j) CF3CHFCF2-O-CH2CF2CHF2 6 6

3 3

0 0 (k) CF CHFCF -O-CH CF CF 0 0 0 0 3 2 2 2 3 80 00 20 40 6 1 1 1 Wavenumber (cm-1) 3

0 00 00 00 00 8 10 12 14 -1 Wavenumber (cm )

Figure 4.1. Experimental infrared absorption cross-sections measured for (a) CHF2CF2-O-CH3; (b) CHF2CF2-O-CH3; (c) CHF2CF2CH2-O-CH3; (d) CF3CF2CF2-O-CH3; (e) CF3CF2CH2OCHF2; (f) CHF2CF2CH2OCHF2; (g) CF3CHFCF2-O-CH2CH3; (h) CF3CHFCF2OCH2CF3; (i) CF3CF2CF2OCHFCF3; (j) CF3CHFCF2OCH2CF2CHF2; (k) CF3CHFCF2OCH2CF2CF3.

87 sample purity (2%), path length (1.5%), spectrum noise (± 10-20 cm2 molecule-1), and residual baseline offset after subtraction of background (1.5%). From these individual uncertainties, the total (random) uncertainty in the integrated absorption cross-section is ± 4 %. We prefer to quote conservative uncertainties of ± 5 % (14). Published absorption cross-sections exist for five of the eleven measured fluorinated ethers. Our measured integrated absorption cross- sections for these ethers were all within fourteen percent of the previously published values (15- 17). Quantitative absorption spectra are shown in Figure 4.1 for the six fluorinated ethers for which spectra have not been published previously. Using the method of Pinnock et al. (14), the

Table 4.2. Comparison between calculated vibrational frequencies in this work and calculated and experimental frequencies of Good and Francisco (19).

Frequency (cm-1) This Work Good and Francisco (19) B3LYP/6- B3LYP/6- Stretch 311+g(d, p) 311++g(3df,3pd) Experimental CF asymmetric' 2 1109 1121 (CF3) CF3 symmetric CH3OCF3 1164 1168 1160 (CF3) CF asymmetric'' 2 1277 1278 1264 (CF3) CF asymmetric 2 1055 1054 1088 (CHF2) CF ' asymmetric 2 1113 1115 (CHF ') CHF OCHF 2 2 2 CF symmetric 2 1128 1121 1152 (CHF2) CF ' symmetric 2 1145 1141 (CHF2') CF asymmetric 2 1102 1114 1110 (CHF2) CF symmetric 2 1138 1146 1152 (CHF2) CF2 asymmetric CHF2OCF3 1156 1166 (CF3) CF symmetric 2 1219 1222 1238 (CF3) CF asymmetric' 2 1280 1287 1288 (CF3)

88 measured HFE spectra given in the SI, and the infrared spectrum of CFC-11 reported elsewhere (18), we calculate an instantaneous radiative efficiency for CFC-11 of 0.26 W m-2 ppb-1 and radiative efficiencies for the HFEs ranging between 0.22 and 0.58 W m-2 ppb-1 (Table 4.1).

4.3.2 Validity of computed spectra

Blowers et al. (4) demonstrated that radiative efficiency values for HFEs could be reliably determined from vibrational frequencies predicted by DFT using B3LYP/6-31g* basis sets. They showed this basis set was sufficient to capture the important features of the spectra. The higher basis sets used in the quantum mechanical calculations described in our study ensure results of similar or greater accuracy. As a check of our computational method, calculated vibrational frequencies for C-F stretches in CH3OCF3, CHF2OCHF2 and CHF2OCF3 were compared to those calculated and measured experimentally by Good and Francisco (19). The excellent agreement between the results from the present work and those from Good and Francisco shown in Table 4.2 suggests the absence of significant systematic errors in the present work. In light of the high basis sets used and agreement between measured and calculated spectra shown in Table 4.2, corrections to account for deviations from the harmonic approximation were unnecessary in the present work.

4.3.3 Development and utility of the SAR

For the purposes of the SAR, only C-F bond stretches were considered. Although other vibrational modes exist in the molecules, C-F bond stretches dominate with respect to radiative efficiency. The modeled compounds were selected as the smallest compounds that would give an overview of the chemical environments possible within linear HFEs. Development of an SAR allows extrapolation of these values to larger HFEs. Absorption wavenumbers corresponding to C-F bond stretches in various functional groups were identified in each of the 154 molecules investigated. Average absorption frequencies were determined for each C-F stretch in interior (CHF, CF2) and end (CH2F, CHF2, CF3) functional groups in various environments. The approach outlined by Pinnock et al. (14) was then used to calculate radiative efficiencies for these functional groups (f, as shown in Table 4.3). Assignment of an f value to a given functional group assumes no interactions between vibrational modes. Although this is not strictly true, especially for larger molecules, the SAR described herein provides a simple method to obtain a reasonable estimation of radiative efficiency for HFEs. Radiative efficiency (RE) is estimated according to the following equation:

89 RE = Σf

For example, the radiative efficiency for CF3CHFCF2-O-C2H5 can be determined as follows:

RE = f(CF3 α to CHF) + f(CHF α to CF3 and CF2) + f(CF2 α to O and CHF)

RE = 0.168 + 0.064 + 0.086

RE = 0.318 W m-2 ppb-1

This result is in good agreement with the value of 0.309 W m-2 ppb-1 derived from the experimentally recorded infrared spectrum (see Table 4.1).

The SAR provides radiative efficiency values that are in good agreement with those determined from our measured cross-sections (see Table 4.1) and in reasonable agreement with those reported by others (see Table 4.4). To date, 41 linear HFE cross-sections and corresponding radiative efficiencies have been published (3,15-17,20-25). As seen from Table 4.4, in 37 out of 41 cases the SAR returns values which are within ± 50% of those from experimental studies reported in the literature and in 17 out of 41 cases the SAR values are within ± 25% of those available in the literature. In general, radiative efficiencies were under- predicted using the SAR versus those determined from measurements. Not surprisingly, agreement was better with experimental radiative efficiency values that were determined using the Pinnock et al. (14) method (15-17,21-23) (see Table 4.4). The SAR gives values which are generally lower than those reported in studies which did not use the Pinnock et al. approach (3,20,24,25).

As discussed by Christidis et al. (20), differences in radiative efficiency values reported in the literature are attributed to different infrared spectra and different methods used to convert infrared spectra into radiative efficiency values. In the current study, up to 14% difference was observed between our experimentally measured integrated cross-sections and those published previously. As shown in Table 4.4, Myhre et al. (3) derived radiative efficiency values for HFEs using cross-section data from different sources with the same radiative transfer model and observed differences of up to 32% (for CHF2OCF2OCHF2). Clearly uncertainties in the IR spectra can be a major source of uncertainty in estimates of radiative efficiency.

90 Table 4.3. Radiative efficiencies for structural components of hydrofluoroethers (HFEs) for use in the structure-activity relationship (SAR).

a a Functional f Functional f Group α1 α2 (W m-2 ppb-1) Group α1 α2 (W m-2 ppb-1)

O 0.251 CH2 CH2 0.029

CH2 0.105 CH2 CHF 0.025 CF3 CHF 0.168 CH2 CF2 0.047 CF2 CF2 0.162 CHF CHF 0.062

O (original) 0.137 CHF CF2 0.071

O (corrected) 0.217 CF2 CF2 0.090

Exterior CHF2 CH2 0.077 O O 0.080

CHF 0.089 O CH3 0.028

CF2 0.108 O CH2F 0.045

O 0.082 O CHF2 0.046

CH2 0.030 O CF3 0.018 CH2F CHF 0.030 O CH2 0.041

CF2 0.048 O CHF 0.072

O O 0.098 O CF2 0.040

O CH3 0.101 CH3 CH2 0.008

O CH2F 0.094 CH3 CHF 0.023 Interior O CHF2 0.101 CH3 CF2 0.040

O CF3 0.191 CH2FCH2 0.016

O CH2 0.056 CH2F CHF 0.032 CHF O CHF 0.105 CH2FCF2 0.026

O CF2 0.127 CHF2 CH2 0.027

CH3 CH2 0.045 CHF2 CHF 0.017

CH3 CHF 0.060 CHF2 CF2 0.028 Interior CF2 CH3 CF2 0.064 CF3 CH2 0.045

CH2FCH2 0.029 CF3 CHF 0.035

CH2F CHF 0.054 CF3 CF2 0.061

CH2FCF2 0.078 CH2 CH2 0.006

CHF2 CH2 0.046 CH2 CHF 0.007

CHF2 CHF 0.071 CH2 CF2 0.033

CHF2 CF2 0.069 CHF CHF 0.022

CF3 CH2 0.044 CHF CF2 0.037

CF3 CHF 0.058 CF2 CF2 0.037 CF3 CF2 0.086 a Values were determined through the calculation of infrared spectra for 154 hydrofluoroethers, where the frequency of absorption corresponding to each functional group was identified. These were calibrated to measured spectra in order to determine cross sections. The method of Pinnock et al. (14) was used to determine radiative efficiency. For more details, see Sections 2 and 3.3.

91 Table 4.4. Structure-activity relationship (SAR) predicted versus published radiative efficiencies (REs) determined from experimentally measured cross-sections. Values in brackets include the empirical correction factor for -O-CHF2. Reported RE Calculated RE Percent Structure (W m-2 ppb-1) (W m-2 ppb-1) Difference Type of Forcing Calculation Reference

CF3-O-CH3 0.198 0.251 27 Instantaneous, clear-sky broadband Sihra et al ., 2003

CF3-O-CHF2 0.410 0.388 (0.468) -5 (14) Instantaneous, cloudy-sky narrowband Christidis et al ., 1997

CF3-O-CHF2 0.405 0.388 (0.468) -4 (16) Pinnock et al. method [1995] Heathfield et al ., 1998

CF3-O-CHF2 0.450 0.388 (0.468) -14 (4) Cloudy Orkin et al ., 1999

CF3-O-CHF2 0.424 0.388 (0.468) -8 (10) Instantaneous, clear-sky broadband Sihra et al ., 2002

CHF2-O-CHF2 0.430 0.274 (0.434) -36 (1) Pinnock et al. method [1995] Heathfield et al ., 1998

CHF2-O-CHF2 0.400 0.274 (0.434) -31 (8) Adjusted, cloudy-sky broadband Myhre et al ., 1999

CHF2-O-CHF2 0.440 0.274 (0.434) -38 (-1) Cloudy Orkin et al ., 1999

CF3-O-CH2CH3 0.210 0.251 20 Pinnock et al. method [1995] Oyaro et al ., 2005

CF3-O-CHFCHF2 0.349 0.386 11 Pinnock et al. method [1995] Oyaro et al ., 2005

CF3-O-CHFCF3 0.402 0.437 9 Pinnock et al. method [1995] Oyaro et al ., 2005

CF3-O-CF2CHF2 0.430 0.461 7 Pinnock et al. method [1995] Sulbaek Andersen et al ., 2005

CF3CH2-O-CH3 0.190 0.105 -45 Pinnock et al. method [1995] Oyaro et al ., 2005

CF3CHF-O-CHF2 0.447 0.323 (0.403) -28 (-10) Pinnock et al. method [1995] Oyaro et al ., 2005

CF3CH2-O-CHF2 0.387 0.242 (0.322) -37 (-17) Instantaneous, clear-sky broadband Sihra et al ., 2001

CF3CH2-O-CHF2 0.374 0.242 (0.322) -35 (-14) Pinnock et al. method [1995] Oyaro et al ., 2005

CF3CF2CF2CF2-O-CH3 0.465 0.466 0 Instantaneous, clear-sky broadband Sihra et al ., 2005

CF3CF2CF2-O-CHFCF3 0.563 0.562 0 Pinnock et al. method [1995] Oyaro et al ., 2005

CF3CH2CH2-O-CH3 0.276 0.105 -62 Pinnock et al. method [1995] Oyaro et al ., 2004

CH3CH2-O-CHF2 0.384 0.137 (0.217) -64 (-43) Instantaneous, cloudy-sky narrowband Christidis et al ., 1997

CH3-O-CF2CHF2 0.300 0.210 -30 Pinnock et al. method [1995] Heathfield et al ., 1998

CF3CH2-O-CF2CHF2 0.465 0.315 -32 Pinnock et al. method [1995] Heathfield et al ., 1998

CF3CH2-O-CH2CF3 0.390 0.210 -46 Cloudy Orkin et al ., 1999

CF3CH2-O-CH2CF3 0.362 0.217 -40 Instantaneous, clear-sky broadband Sihra et al ., 2004

CF3CH2-O-CH2CF3 0.334 0.217 -35 Pinnock et al. method [1995] Oyaro et al ., 2004

CF3CHFCF2-O-CF3 0.480 0.585 22 Pinnock et al. method [1995] Wallington et al ., 2004

CF3CHFCF2-O-CH2CH3 0.331 0.334 1 Pinnock et al. method [1995] Oyaro et al ., 2005

CF3CHFCF2-O-CHF2 0.510 0.471 (0.551) -8 (8) Pinnock et al. method [1995] Wallington et al ., 2004

CF3CHFCF2-O-CHF2 0.485 0.471 (0.551) -3 (14) Pinnock et al. method [1995] Oyaro et al ., 2005

CH3CH2-O-CF2CHF2 0.315 0.210 -33 Pinnock et al. method [1995] Heathfield et al ., 1998

CHF2CF2CH2-O-CH3 0.237 0.154 -35 Pinnock et al. method [1995] Oyaro et al ., 2004

CHF2-O-CF2-O-CHF2 0.660 0.372 (0.532) -44 (-19) Adjusted, cloudy-sky broadband Myhre et al ., 1999 Adjusted, cloudy-sky broadband (cross CHF -O-CF -O-CHF 0.870 0.372 (0.532) -57 (-39) Myhre et al ., 1999 2 2 2 section from Cavalli et al., [1998])

CH3-O-C2F4-O-CH3 0.320 0.255 -14 Pinnock et al. method [1995] Sulbaek Andersen et al ., 2004

CHF2-O-C2F4-O-CHF2 0.870 0.529 (0.689) -39 (-21) Adjusted, cloudy-sky broadband Myhre et al ., 1999 Adjusted, cloudy-sky broadband (cross CHF -O-C F -O-CHF 1.010 0.529 (0.689) -48 (-32) Myhre et al ., 1999 2 2 4 2 section from Cavalli et al., [1998])

CH3-(O-C2F4)2-O-CH3 0.610 0.510 -10 Pinnock et al. method [1995] Sulbaek Andersen et al ., 2004

CHF2-O-CF2-O-C2F4-O-CHF2 1.038 0.627 (0.786) -40 (-24) Instantaneous, cloudy-sky narrowband Christidis et al ., 1997

CHF2-O-CF2-O-C2F4-O-CHF2 1.370 0.627 (0.786) -54 (-43) Adjusted, cloudy-sky broadband Myhre et al ., 1999

CHF2-O-CF2-O-C2F4-O-CHF2 1.051 0.627 (0.786) -40 (-25) Instantaneous, clear-sky broadband Sihra et al ., 2006 CH3-(O-C2F4)3-O-CH3 0.830 0.764 -9 Pinnock et al. method [1995] Sulbaek Andersen et al ., 2004

Christidis et al (20) used two different radiation models to determine the radiative efficiencies of three HFEs for clear and cloudy sky conditions with and without stratospheric adjustment. It was shown that an instantaneous, cloudy-sky determination provided the lowest estimation, which was 30-35% lower than the highest estimation determined from an adjusted, clear-sky model (20). An instantaneous determination of radiative efficiency neglects the influence of the equilibration of the stratosphere. Inclusion of the stratospheric adjustment

92 typically increases the radiative efficiency of a compound by approximately 7% (14). In addition, considering cloudy-sky versus clear-sky conditions can decrease the radiative efficiency by roughly 20% (20). The Pinnock et al. (14) method, from which the radiative efficiency values in the SAR are derived, is based on an instantaneous, cloudy-sky narrowband radiation model. Thus, it is not surprising that the SAR gives lower values than reported in studies which allowed for the stratospheric adjustment, such as those determined by Myhre et al. (3).

The under-prediction of some radiative efficiency values by the SAR cannot be attributed entirely to different assumptions in the radiative transfer models (cloudy versus clear sky, with or without stratospheric adjustment). Of the literature values that were not replicated here within ±30%, one-third of those were determined using the Pinnock et al. method (14). It is possible there is a systematic bias due to a common structural feature in the compounds that was poorly predicted. The majority of hydrofluoropolyethers (HFPEs) were under-predicted by the SAR, indicating the possibility that compounds containing multiple ether bonds are not well predicted by the SAR. Although there was poor agreement between published values for some of these compounds (Table 4.4) and a “correct” experimental value could not be determined, it

CHF 60 ) -1 CF ) 2 -1

40 (cm

CH2F -2 W m W

CHF2 -3 20 (10

CF3 Radiance in AtmosphericWindow 0 800 900 1000 1100 1200 1300 1400 -1 Wavenumber (cm ) Figure 4.2. Absorption wavenumber range for each type of C-F stretch in hydrofluoroethers (HFE) functional groups shown with respect to the atmospheric window (…..) (14).

93 appears that the SAR systematically underestimates the radiative efficiency values for certain HFPEs. The three HFPEs for which the SAR gives radiative efficiency values that are within ±15% of the literature values were derived using the Pinnock et al. method and had a common structure, with -O-CH3 terminal groups (21). In contrast, those predicted values for HFPEs that were significantly different from the published values were for those not using the Pinnock et al. method and that had -O-CHF2 terminal groups. Thus, it is likely that under-prediction was not due to the presence of multiple ether bonds, but instead to the nature of the terminal group. In fact, every compound containing a –CHF2 group alpha to an ether oxygen was under-predicted, with almost two-thirds of these under-predicted by more than 30%. This probably reflects the location of absorption of the CF2 symmetric stretch in the -O-CHF2 group, which occurs at 995 to 1106 cm-1 and is coincident with an ozone absorption in the atmospheric window (see Figure 4.2). The radiative forcing per unit infrared cross-section values at 995 - 1106 cm-1 lie in the range (1.10 - 2.61) × 10-3 W m-2 (cm-1)-1 (10-18 cm2 molecule-1)-1, with an average of 1.45 × 10-3 W m-2 (cm-1)-1 (10-18 cm2 molecule-1)-1. From inspection of Figure 4.2 it is clear that a small change in absorption frequency can have a large effect on the radiative efficiency contribution for a number of functional groups. It is possible that the computational approach used here has a slight bias in the vibrational frequencies that it predicts for C-F stretches in the -O-CHF2 -2 -1 group. An empirical correction factor of 0.08 W m ppb per -O-CHF2 was derived to increase the agreement between SAR and experimental values. The value of f for CHF2 alpha to O then becomes 0.221 W m-2 ppb-1. The revised numbers are shown in brackets in Table 4.4. Once this correction is applied, all radiative efficiency values with the exception of one are predicted within ±50% and the majority (27 of 41) are predicted within ±25% of their published values (see Table 4.4).

4.3.4 Radiative efficiency and molecular structure

As seen from Figure 4.2, the atmospheric window becomes less transparent at higher wavenumbers (frequencies) and absorptions at higher frequencies have less potential impact on climate. Within the 154 modeled HFEs, clear trends were observed with respect to the effect of structure on frequency of vibration. The wavenumber ranges for each functional group are shown in Figure 4.2. In general, increasing fluorination of the functional groups (i.e., CHF →

CF2, CH2F → CHF2 → CF3) increased the vibration frequency. On a carbon atom, increasing

94

0.12 ) -1 0.10 ppb -2 0.08

0.06

0.04

0.02 RF per C-F Bondm (W 0.00 CHF CF CH F CHF CF 2 2 2 3

Figure 4.3. Radiative forcing per C-F bond for each functional group, where bars represent the observed range. The empirical correction factor for -O-CHF2 has been included in the CHF2 range. fluorination has been shown to increase C-F bond strengths (26). The vibrational frequency of a bond increases with bond strength, so that vibrations involving stronger bonds absorb at higher wavenumbers. This is consistent with the observed infrared absorption more highly fluorinated functional groups appearing at higher wavenumbers.

Within a given functional group, the lowest vibrational frequency was observed when the functional group was alpha to the ether oxygen. For example, in Figure 3, the effect of substituent groups on the C-F absorption frequency in the CHF group is shown. The frequencies of vibration for C-F bonds adjacent to oxygen were shown to be significantly lower than those alpha to a carbon group using an ANOVA statistical test. To our knowledge, there have not been any studies that examined C-F bond strengths in ether molecules. However, studies have looked at the effects of the oxygen ether on C-H bond strengths in HFEs. It was observed that C-H bonds next to ether bonds were weaker, and this was attributed to the donation of electron density from the oxygen atom (27). Although the behavior of C-H and C-F bonds has been shown to differ in fluorinated molecules (28), this same donation effect may be the cause of the weaker C-F bonds observed alpha to the ether. In Figure 3, a slight decrease in vibrational frequency is also shown for CHF groups adjacent to CHF groups relative to those adjacent to CH2 or CF2 groups. However, this decrease was not statistically significant at the 95% confidence interval using the one-way ANOVA test mentioned above. This similarity in

95 vibrational frequencies is consistent with the generally accepted trend that fluorination at the beta carbon typically does not appreciably affect C-F bonds (28).

Within the atmospheric window, lower frequencies typically correspond to a higher radiative efficiency, due to a decrease in infrared transparency at higher frequencies. Thus, since decreasing fluorination and proximity to the ether oxygen result in lower vibrational frequencies, these factors also tend to cause an increase in radiative efficiency. It can be seen from Table 4.3 that functional groups that are alpha to an oxygen atom have among the highest radiative efficiencies. The CF3 group, when alpha to an ether oxygen, has a radiative efficiency contribution of 0.251 W m-2 ppb-1 which is substantially higher than its contribution of 0.105 - a)

0.3 b) ) -1 CHF2CF2CH2OCH3 CHF CF OCH 2 2 3 ppb -2 0.2

0.1

m (W Efficiency Radiative 0.0 Total RE RE per F

Figure 4.4. (a) Experimental infrared absorption cross-sections for CHF2CF2CH2OCH3 and ….. CHF2CF2OCH3 compared to the atmospheric window ( ) (14); (b) total and radiative efficiency (RE) per C-F bond for each molecule.

96

0.168 W m-2 ppb-1 when it is in other chemical environments. This observation could explain the tendency of HFPEs to have high radiative efficiency values compared to HFEs with the same number of C-F bonds (2). It should be noted that because of absorption by ozone between 1000 and 1100 cm-1 (see Figure 4.2), lower absorption frequencies do not always correspond to higher radiative efficiency. For example, although CHF functional groups have lower absorption frequencies than CF2, the CHF region spans a region of the atmospheric window where transmission is decreased due to absorption by ozone (see Figure 4.2) making its average effective radiative efficiency per C-F bond similar to that of CF2. In addition, the ranges of effective radiative efficiency per C-F bond for CHF and CF2 are large, varying from 0.006 – 0.080 and 0.012 – 0.096 W m-2 ppb-1, respectively (Figure 4.3). This is in contrast to the more consistent effective radiative efficiency per C-F values observed for CH2F, CHF2 and CF3 groups. Depending on the chemical environment, CHF and CF2 can have radiative efficiencies per C-F bond that are higher or lower than for other functional groups. The effect of proximity to the ether group is illustrated by comparing CHF2CF2-O-CH3 and CHF2CF2CH2-O-CH3; molecules with the same number and type of C-F bonds in which the former has fluorination adjacent to the ether oxygen and the latter has a hydrogenated carbon spacer (Figure 4). These two structurally similar compounds have similar experimental absorption cross-sections, which are shown in part (a) of Figure 4. Despite the fact that both compounds contain the same number of fluorine atoms in the same functional groups, CHF2CF2-O-CH3 has a higher integrated band strength (Table 4.1) and a higher radiative efficiency (Figure 4, part (b)) than

CHF2CF2CH2-O-CH3. This observation is consistent with the prediction from the SAR that C-F bonds adjacent to an ether oxygen have lower vibrational frequency and higher radiative efficiency.

4.4 Conclusions

To understand the potential impacts of chemicals intended to replace ozone-depleting substances, it is necessary to recognize their effect on climate. Fluorinated ethers have a wide range of radiative efficiencies that cannot be simply attributed to the number of C-F bonds. Quantification of the influence of molecular structure on radiative efficiency is an important step towards understanding the contribution of HFEs to climate change. The present study shows that chemical environment of the C-F bond in a fluorinated ether affects the vibrational frequency and, hence, the radiative efficiency of functional groups in these compounds. In

97 general, proximity of the C-F bond to an ether oxygen decreases the frequency of vibration and increases the radiative efficiency of a given C-F bond. By inspection of Table 4.3, it is possible to determine the structures of molecules that would have the highest radiative efficiencies. The highest radiative efficiency per functional group is for a CF3 group alpha to O. However, CHF2 groups alpha to O have higher radiative efficiencies than CF3 groups in other chemical environments. Interior groups have radiative efficiencies that are highest when they are next to an ether oxygen. The highest radiative efficiencies for an internal group next to two other internal groups is CF2 alpha to O and CF2. Similarly, the CHF group has the highest radiative efficiency when it is between two ether oxygens. Thus, to design molecules with low radiative efficiency, it is desirable to avoid CF3-O-, CHF2-O-, -O-CF2CF2-O- and -O-CHF-O- groups where possible. In addition, avoidance of fluorination on the carbon alpha to the ether oxygen would minimize radiative efficiency.

Chemical architecture has an influence on the radiative efficiency of fluorinated ethers. However, full determination of potential climate effects also requires knowledge of the atmospheric lifetime. The SAR developed in this study can be used to provide radiative forcing values and can be combined with a hydroxyl radical reactivity SARs (7) to facilitate the design of fluorinated ethers with minimal climate impacts.

4.5 Acknowledgements The authors thank Craig Butt for statistical support and are grateful for the invaluable assistance of Mima Staikova and Jamie Donaldson. Funding was provided by the Natural Science and Engineering Research Council of Canada (NSERC). CJY also appreciates the support of NSERC through a CGS Fellowship.

98 4.6 Sources Cited (1) UNEP. 2000. The Montreal Protocol on Substances that Deplete the Ozone Layer, United Nations Environmental Programme.

(2) Forster, P.; Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R.; Fahey, D.W.; Haywood, J.; Lean, J.; Lowe, D.C.; Myhre, G.; Nganga, J.; Prinn, R.; Raga, G.; Schulz, M.; Van Dorland, R. In Climate Change 2007: The Physical Science Basis; Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L., Eds.; Cambridge University Press: Cambridge, United Kingdom, 2007.

(3) Myhre, G.; Nielsen, C.J.; Powell, D.L.; Stordal, F. Infrared absorption cross section, radiative forcing, and GWP of four hydrofluoro(poly)ethers. Atmospheric Environment 1999, 33, 4447-4458.

(4) Blowers, P.; Moline, D.M.; Tetrault, K.F.; Wheeler, R.R.; Tuchawena, S.L. Prediction of radiative forcing values for hydrofluoroethers using density functional theory methods. Journal of Geophysical Research 2007, 112, D15108.

(5) Blowers, P.; Moline, D.M.; Tetrault, K.F.; Wheeler, R.R.; Tuchawena, S.L. Global warming potentials of hydrofluoroethers. Environmental Science and Technology 2008, 42, 1301-1307.

(6) Blowers, P.; Tetrault, K.F.; Trujillo-Morehead, Y. Global warming potential predictions for hydrofluoroethers with two carbon atoms. Theoretical Chemistry Accounts 2008, 119, 369-381.

(7) Kwok, E.S.C.; Atkinson, R. Estimation of hydroxyl radical reaction rate constants for gas- phase organic compounds using a structure-reactivity relationship--an update. Atmospheric Environment 1995, 29, 1685-1695.

(8) Young, C.J.; Hurley, M.D.; Wallington, T.J.; Mabury, S.A. Atmospheric lifetime and global warming potential of a perfluoropolyether. Environmental Science and Technology 2006, 40, 2242-2246.

(9) Alpert, N.L.; Keiser, W.E.; Szymanski, H.A. IR theory and practice; Plenum Press: New York, 1970.

(10) Houghton, J.T.; Ding, Y.; Griggs, D.J.; Noguer, M.; van der Linden, P.J.; Dai, X.; Maskell, K.; Johnson, C.A., Eds. Climate Change 2001: The Scientific Basis. Contribution of Working Group 1 to the third assessment report of the Intergovernmental Panel on Climate Change; Cambridge University Press: New York, 2001.

(11) Hansch, C.; Leo, A. Substituent constants for correlation analysis in chemistry and biology; John Wiley & Sons: New York, NY, 1979.

(12) Simmons, K.A. A simple structure-based calculator for estimating vapor pressure. Journal of Agricultural and Food Chemistry 1999, 47, 1711-1716.

99 (13) Gaussian 03, Revision B.03. Frisch, M.J.; Trucks, G.W.; Schlegel, H.B.; Scuseria, G.E.; Robb, M.A.; Cheeseman, J.R.; Montgomery, J., J. A.; Vreven, T.; Kudin, K.N.; Burant, J.C.; Millam, J.M.; Iyengar, S.S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G.A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J.E.; Hratchian, H.P.; Cross, J.B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R.E.; Yazyev, O.; Austin, A.J.; Cammi, R.; Pomelli, C.; Ochterski, J.W.; Ayala, P.Y.; Morokuma, K.; Voth, G.A.; Salvador, P.; Dannenberg, J.J.; Zakrzewski, V.G.; Dapprich, S.; Daniels, A.D.; Strain, M.C.; Farkas, O.; Malick, D.K.; Rabuck, A.D.; Raghavachari, K.; Foresman, J.B.; Ortiz, J.V.; Cui, Q.; Baboul, A.G.; Clifford, S.; Cioslowski, J.; Stefanov, B.B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R.L.; Fox, D.J.; Keith, T.; Al-Laham, M.A.; Peng, C.Y.; Nanayakkara, A.; Challacombe, M.; Gill, P.M.W.; Johnson, B.; Chen, W.; Wong, M.W.; Gonzalez, C.; Pople, J.A. Gaussian, Inc., Wallingford, CT, 2004.

(14) Pinnock, S.; Hurley, M.D.; Shine, K.P.; Wallington, T.J.; Smyth, T.J. Radiative forcing of climate by hydrochlorofluorocarbons and hydrofluorocarbons. Journal of Geophysical Research 1995, 100, 23227-23238.

(15) Heathfield, A.E.; Anastasi, C.; McCulloch, A.; Nicolaisen, F.M. Integrated infrared absorption coefficients of several partially fluorinated ether compounds: CF3OCCF2H, CF2HOCF2H, CH3OCF2CF2H, CH3OCF2CFClH, CH3CH2OCF2CF2H, CF3CH2OCF2CF2H and CH2=CHCH2OCF2CF2H. Atmospheric Environment 1998, 32, 2825-2833.

(16) Oyaro, N.; Sellevag, S.R.; Nielsen, C.J. Study of the OH and Cl-initiated oxidation, IR absorption cross-section, radiative forcing, and global warming potential of four C4- hydrofluoroethers. Environmental Science and Technology 2004, 38, 5567-5576.

(17) Oyaro, N.; Sellevag, S.R.; Nielsen, C.J. Atmospheric chemistry of hydrofluoroethers: Reaction of a series of hydrofluoroethers with OH radicals and Cl atoms, atmospheric lifetimes, and global warming potentials. Journal of Physical Chemistry A 2005, 109, 337-346.

(18) Ninomiya, Y.; Kawasaki, M.; Gushin, A.; Molina, L.T.; Molina, M.; Wallington, T.J. Atmospheric chemistry of n-C3F7OCH3: Reaction with OH radicals and Cl atoms and atmospheric fate of n-C3F7OCH2O. Environmental Science and Technology 2000, 34, 2973- 2978.

(19) Good, D.A.; Francisco, J.S. Structure and vibrational spectra of chlorofluorocarbon substitutes: An experimental and ab initio study of fluorinated ethers CHF2OCF3 (E125), CHF2OCHF2 (E134), and CH3OCF3 (E143A). Journal of Physical Chemistry A 1998, 102, 1854-1864.

(20) Christidis, N.; Hurley, M.D.; Pinnock, S.; Shine, K.P.; Wallington, T.J. Radiative forcing of climate change by CFC-11 and possible CFC replacements. Journal of Geophysical Research 1997, 102, 19,597-519,609.

(21) Sulbaek Andersen, M.P.; Hurley, M.D.; Wallington, T.J.; Blandini, F.; Jensen, N.R.; Librando, V.; Hjorth, J.; Marchionni, G.; Avataneo, M.; Visca, M.; Nicolaisen, F.M.; Nielsen, O.J. Atmospheric chemistry of CH3O(CF2CF2)nCH3 (n=1-3): Kinetics and mechanism of

100 oxidation initiated by Cl atoms and OH radicals, IR spectra and global warming potentials. Journal of Physical Chemistry A 2004, 108, 1964-1972.

(22) Wallington, T.J.; Hurley, M.D.; Nielson, O.J.; Sulbaek Andersen, M.P. Atmospheric chemistry of CF3CFHCF2OCF3 and CF3CFHCF2OCF2H: Reaction with Cl atoms and OH radicals, degradation mechanism and global warming potentials. Journal of Physical Chemistry A 2004, 108, 11333-11338.

(23) Sulbaek Andersen, M.P.; Nielsen, O.J.; Wallington, T.J.; Hurley, M.D.; DeMore, W.B. Atmospheric chemistry of CF3OCF2CF2H and CF3OC(CF3)2H: Reaction with Cl atoms and OH radicals, degradation mechanism, global warming potentials, and empirical relationship between k(OH) and k(Cl) for organic compounds. Journal of Physical Chemistry A 2005, 109, 3926- 3934.

(24) Orkin, V.L.; Villenave, E.; Huie, R.E.; Kurylo, M.J. Atmospheric lifetimes and global warming potentials of hydrofluoroethers: Reactivity toward OH, UV spectra, and IR absorption cross sections. Journal of Physical Chemistry A 1999, 103, 9770-9779.

(25) Sihra, K.; Hurley, M.D.; Shine, K.P.; Wallington, T.J. Updated radiative forcing estimates of 65 halocarbons and nonmethane hydrocarbons. Journal of Geophysical Research 2001, 106, 20493-20505.

(26) Smart, B.E. In Molecular Structure and Energetics; Liebman, J.F., Greenberg, A., Eds.; VCH Publishers: Deerfield Beach, FL, 1986; Vol. 3.

(27) Lazarou, Y.G.; Papagiannakopoulos, P. Theoretical investigation of the thermochemistry of hydrofluoroethers. Chemical Physical Letters 1999, 301, 19-28.

(28) Smart, B.E. In Organofluorine chemistry: Principles and commercial applications; Banks, R.E., Smart, B.E., Tatlow, J.C., Eds.; Plenum Press: New York, NY, 1994.

CHAPTER FIVE

Perfluoroalkyl Amines: A New Class of Long-Lived Greenhouse Gases

Cora J. Young, Michael D. Hurley, Timothy J. Wallington and Scott A. Mabury

101 102 5.1 Introduction

Perfluoroalkyl amines (PFAms; CxF2x-1N(CyF2y-1)CzF2z-1) are a class of thermally and chemically stable liquids marketed for use in numerous applications, including electronic testing and as heat transfer agents (1). These compounds are synthesized using electrochemical fluorination (ECF), which yields mixtures of isomer products (2), and are produced as a number of chain-length congeners. Although two PFAm congeners are considered high production volume chemicals (3), no studies on potential environmental impacts have yet been conducted on these compounds.

Perfluorinated compounds tend to be long-lived greenhouse gases (LLGHGs) because they have significant atmospheric lifetimes and are radiatively active. Perfluorinated compounds have no known sinks in the troposphere. As a result, their lifetimes are on the order of hundreds of years and typically dominated by destruction in the upper atmosphere (4,5). Compounds containing C-F bonds can impact climate, because they absorb in the optically thin spectral region of the atmosphere (750 – 1250 cm-1). This leads to a high radiative efficiency (RE), indicating that low atmospheric concentrations can have an effect on climate. Halocarbon compounds contribute 0.337 W m-2, or 13% of the total radiative forcing of LLGHGs (6). This number is assumed to be known with high certainty, but only includes compounds that have been intentionally sought after in the atmosphere. A number of perfluorinated compounds exist, including PFAms, for which atmospheric fate and concentration data is not available. It is likely these chemicals could have an impact on climate.

The objective of this study was to determine the atmospheric lifetime and radiative properties of perfluorotributylamine (PFBAm) and to develop a method for detection of this compound in the atmosphere.

5.2 Methods

5.2.1 Infrared absorption measurements

Measurements were performed in a 140 L Pyrex reactor interfaced to a Mattson Sirus 100 FTIR spectrometer. Infrared spectra were derived from 32 coadded interferograms with a spectral resolution of 0.25 cm-1 and an analytical path length of the infrared beam of 27.1 m. Infrared spectra were recorded at 296 K using 1.54 – 4.26 mTorr of PFBAm. The compound was obtained from a commercial source and was subjected to repeated freeze-pump-thaw before

103 being added to the chamber via a calibrated volume. Four separate spectra were recorded, with one spectrum taken while warming from liquid nitrogen to ensure purity. Peak absorbances were in the range 0.05 – 0.7 and scaled linearly with the PFBAm concentration.

5.2.2 Determination of radiative efficiency

Pinnock et al. (7) have presented a method that can be used to estimate radiative forcing from infrared absorption cross-sections in which the spectral region between 0 and 2500 cm-1 is divided up into 250 bands, each 10 cm-1 wide, assuming a 0→1 ppbv change in atmospheric concentration of the compound. The cloudy-sky instantaneous RE can then be determined from the radiative forcing per unit cross-section using a simple expression. This method was used to estimate RE directly from the measured cross-section of PFBAm.

5.2.3 Physical properties and use of multi-species model

Few physical property measurements for PFAms are available. As a result, we estimated some properties for PFBAm through the SPARC online calculator (8). Calculated properties were verified through comparison with experimentally measured values for other perfluorinated compounds. Physical properties for protonated PFBAm, PFBAmH+, were determined from the properties of PFBAm using the ratio of measured properties for perfluorooctylsulfonic acid (PFOSH) to perfluorooctylsulfanoate (PFOS). Although the pKa of PFBAm is unknown, the pKa of another PFAm congener, perfluorotripentyl amine, has been determined to be less than

0.5. The pKa of PFBAm is likely to be similar, but as the pKa could not be determined explicitly, values from 2 to -2 were used in the model to determine the maximum effect of ionization. The selected model was a high resolution multi-species (HR-MS) model developed by Cahill and Mackay (9). The model includes a separate aerosol phase in each of the air phases. This HR-MS model can describe the fate of inter-converting chemical species, which allows for modeling of processes such as ionization. The model was run with the default environmental settings with all advection processes disabled. Simulations were performed over one day of simulated time, yielding a reasonable first approximation of the importance of ionization, given the speed of acid-base equilibria.

5.2.4 Air sample collection and analysis

Samples were collected using portable air samplers at a rate of ~2 L min-1 for 1 – 4 hours. Samples were collected onto 25 mg of carbotrap in stainless steel tubes that had been

104 conditioned overnight at 380oC under a flow of ultra-high-purity (UHP) nitrogen. Samples were extracted thermally at 350oC under UHP nitrogen for two hours and cryofocused into a sample loop using a custom-built thermal desorber/cryofocuser. A 340oC heating block was applied to the sample loop for 15 min to inject samples via UHP helium flow through a heated transfer line into an Agilent 6890 GC coupled to a 5730 mass spectrometer. Separation was achieved on a GasPro column (60 m × 320 um), beginning at 100oC, increasing at 10oC min-1 to 240oC, followed by a 1oC min-1 increase to 260oC, where the temperature was held for 16 min. The mass spectrometer was operated in electron capture negative ionization mode, screening for m/z 414, 452 and 633. PFAms present analytical challenges due to their insolubility in both aqueous and organic solvents. The only solvents shown to be appropriate were highly fluorinated chemicals, which are not available at high purity. As such, standards were prepared in 1H- perfluoropentane that was obtained from a commercial source and distilled to improve purity. Standards were injected directly into the sample loop and injected as described above. Samples were quantified by external calibration of m/z 633 (M – 2F). While pure standards could be obtained for PFBAm, they were not available for the longer, more highly produced congeners. Methods for the measurement of other congeners are in progress.

5.3 Results and Discussion

5.3.1 Infrared spectrum and radiative efficiency

The absolute absorption spectrum of PFBAm is shown in Figure 5.1. The total integrated cross section is 7.08 × 10-16 cm2 molecule-1 cm-1. Uncertainties in the cross-section measurement arise from the following sources: sample concentration (2%), sample purity (2%), path length (1.5%), spectrum noise ±(10-20 cm2 molecule-1), and residual baseline offset after subtraction of the background (1.5%). From these individual uncertainties, the total (random) uncertainty in the integrated absorption cross section is ±4%. We prefer to quote a conservative uncertainty of 5%, yielding an integrated cross section for PFBAm of (7.08 ± 0.35) × 10-16 cm2 molecule-1 cm-1. Using the method of Pinnock et al. (7) and the measured spectrum, the RE of PFBAm was determined as 0.86 W m-2 ppb-1.

105

12 100 F F F F F F F F

F3C N CF3

F ) F F F F

F -1

10 ) F F 80 -1 F CF3

) F -1 8 molecule 60 2 cm -18 molecule

2 6 (10 -1 ) cm 40 -1 -18

4 (cm (10 -2 σ W m

20 -3 2 (10 Radiative forcing per unit cross section cross unit per forcing Radiative

0 0 600 800 1000 1200 1400 -1 Wavenumber (cm )

Figure 5.1: Infrared absorption cross section for PFBAm (solid line) shown with the radiative forcing per unit cross section of the atmosphere (dotted line) (7).

Table 5.1: Physical-chemical properties for PFBAm, PFBAmH+, PFOSH and PFOS

PFBAm PFBAmH+ PFOSH PFOS Vapour Pressure (Pa) 192 (1) 9.6 x 10-4 9.0 x 10-1 (10) 4.5 x 10-6 (11) Water Solubility (mol m-3) 3.2 x 10-8 (8) 1.6 x 10-6 2.0 x 10-2 (8) 1.0 (12) log KAW 6.4 -0.6 log KOW 0 (8) -5.8 6.09 (8) 0.32 (13)

5.3.2 Physical properties and impact of ionization

Physical properties for PFBAm and its protonated analogue, PFBAmH+, are shown in

Table 5.1. Details for their determination are reported in Appendix B. The high log KAW value was incompatible with the selected model. A sensitivity analysis (details in Appendix B) determined that sensitivity to log KAW was low, so a value of 0 was used to minimize computational time. Model results demonstrated that after emission of PFBAm into air, the

106 majority (> 99.9%) was found in air as PFBAm, with a tiny fraction found as PFBAmH+. PFBAmH+ was also found primarily in the atmosphere, with most in the gas phase. However, a fraction of approximately 11% of PFBAmH+ was also present in the aerosol phase. PFBAmH+ on aerosols could be subject to wet and dry particle deposition. Emission of PFBAmH+ into a water body showed the bulk returning to the atmosphere, but a small amount (6%) present in the sediment. This suggests that sediment could act to sequester PFBAmH+ and, thus, PFBAm. However, model results gathered over fifty days of simulated time and extrapolated to a century timescale indicated this was an insignificant pathway (see Appendix B for details). Thus, ionization to PFBAm+ has a negligible effect on the environmental fate of PFBAm.

5.3.3 Atmospheric lifetime of PFAms

Since PFAms, by analogy to PFBAm, are almost exclusively present in the neutral form, their fate can be determined without the need to include the protonated form. Perfluorinated compounds such as PFAms are unreactive to tropospheric oxidants. As a result, any loss pathways occur in the upper atmosphere. In the stratosphere, reaction with singlet oxygen (O(1D)) can be an important loss mechanism. Although reactions of O(1D) with PFAms have not been examined, reactions of a number of fluorinated compounds have been studied. The primary result of reaction of O(1D) with perfluorinated alkanes is physical quenching of O(1D) to O(3P) (4). In many cases, the yield of O(3P) is unity, within error. In contrast, reactions of 1 3 O( D) with both NH3 and NF3 lead to products other than O( P) (14,15), with NF3 more readily proceeding through reactive channels. Structural requirements for efficient reaction with O(1D) are not well understood. Given known reactivities of amines, it is possible that PFAms could be 1 subject to reaction with O( D). However, it is unlikely that PFAms are more reactive than NF3, given the low bond strength of the N-F bond (276 kJ mol-1 (16)) compared to the N-C and C-F bonds in PFAms. Another potential loss pathway in the upper atmosphere is reaction with mesospheric free electrons, which have been shown to be important in the atmospheric fate of

SF6 (4). Studies have not examined the potential of this fate for amine compounds, but it is chemically feasible as a loss for PFAms. Photolysis by Lyman-α radiation (121.6 nm) in the mesosphere has been demonstrated to be the dominant fate for perfluorinated alkanes and an important fate SF6 (4). It is probable that PFAms are also subject to photolysis at this wavelength. Determining the lifetimes and the relative importance of reaction with O(1D), free electrons and photolysis by Lyman-α radiation is difficult given the limited information. An estimate can be made for the overall lifetime of PFAms by comparison to other perfluorinated

107 compounds (Table 5.2). The compound with the shortest lifetime is NF3, where loss is dominated by stratospheric photolysis, with some contribution from reaction with O(1D) (as discussed above) (5). The lifetimes of the other compounds shown in Table 5.2 depend primarily on reactivity in the mesosphere and, consequently, are much longer. It is unclear whether PFAms would undergo photolytic degradation in the stratosphere. The n → σ* transitions drive photolysis for NF3 and it is possible that a similar mechanism could occur in PFAms. Thus, assuming that PFAms could undergo photolysis in the stratosphere and have some reactivity in the mesosphere, we can quote a conservative lower-limit for the lifetime of PFAms of 500 years.

Table 5.2: Atmospheric lifetimes of some perfluorinated compounds.

Compound Lifetime (years) Ref NF3 550 (5) SF6 3200 (4) C4F10 2600 (6) C5F12 4100 (4) C6F14 3100 (4)

5.3.4 Atmospheric detection of PFBAm

Three air samples were collected in Toronto (43o42΄07.22΄΄N, 79o26΄26.46΄΄W) on October 20-22, 2009 over approximately 200 minutes in the morning, where one sample was collected on each day. A peak corresponding to PFBAm was observed in the extracted sample chromatograms, which was not observed in the extraction of a blank tube (Figure 5.2). Although the retention time was in slight disagreement with the standard (difference of < 1 min), small shifts in retention time are common with this externally controlled GC system. The electron capture-negative ionization spectrum of PFBAm shows the presence of two major ions: m/z 633, corresponding to the loss of two fluorine atoms and m/z 452, corresponding to the loss of one alkyl chain from the nitrogen. Peak shape was well conserved between the standard and the sample in both of the ions (Figure 5.3). It is clear that the observed PFBAm is the sum of multiple peaks, presumably the result of isomers originating from ECF synthesis. From inspection of Figure 5.3, it appears as if there is a difference between the isomer profile of the sample and standard. In fact, a difference was observed between the ion ratios for the sample and standard. The standard yielded a ratio of m/z 633/452 of 1.26 ± 0.04, while the samples had

108 Abundance 3500 3000 2500 2000 1500 1000 500 0 Time--> 14.00 16.00 18.00 20.00 22.00 24.00

Figure 5.2: Chromatogram of GC-MS analysis (m/z 633) of September 28, 2009 sample (–) and blank sorbent tube (…).

a)

b)

Figure 5.3: Comparison between September 28, 2009 sample (–) and authentic standard (…) for a) m/z 633 and b) m/z 452.

109 a ratio of 4.7 ± 0.23. This discrepancy could be due to the differing isomer profiles in the sample. In addition, matrix effects are known to impact some polyfluorinated compounds during GC analysis (17). Further work is underway to determine the source of this inconsistency. Preliminary results suggest that sorption of PFBAm to the carbotrap is not exhaustive. Thus, peak profile differences may be caused by isomer-specific sorption/desorption. The lack of exhaustive extraction also indicates that the levels measured in this study correspond to lower limits of atmospheric concentrations of PFBAm. Using m/z 633 to calibrate the sample, an atmospheric concentration of 0.057 ± 0.019 pptv was determined.

5.4 Environmental Implications

The RE determined above for PFBAm of 0.86 W m-2 ppb-1 is among the highest for long-lived greenhouse gases (Table 5.3). In fact, the Intergovernmental Panel on Climate Change reports only two compounds with higher radiative efficiencies than PFBAm (6), but neither has been detected in the atmosphere. Utilizing the RE of PFBAm and the lifetime determined above for PFAms, a halocarbon global warming potential (HGWP) relative to CFC- 11 can be calculated:

⎛ IF ⎞⎛τ M ⎞⎛1 −exp( −t τ )/ ⎞ HGWP =⎜ PFBAm ⎟⎜ PFBAm CFC−11⎟⎜ PFBAm ⎟ PFBAm ⎜ ⎟⎜τ ⎟⎜ −− τ ⎟ ⎝ IFCFC−11 ⎠⎝ CFC−11 MPFBAm ⎠⎝1 exp( t CFC−11)/ ⎠ where IFPFBAm,IFCFC-11, MPFBAm, MCFC-11, τPFBAm, and τCFC-11 are the instantaneous radiative efficiencies, molecular weights, and atmospheric lifetimes of PFBAm and CFC-11, respectively, and t is the time horizon over which the forcing is integrated. Assuming that τPFBAm is 500 years, and τCFC-11 is 45 years (6), we estimate a HGWP for PFBAm of 1.42 over a 100 year horizon.

The global warming potentials (GWPs) of PFBAm relative to CO2 can be calculated using the GWPs of CFC-11 given in Table 5.3. Over a 100-year time horizon, PFBAm has a GWP of 6740.

Previous to this study, the compound with the highest RE that had been detected in the -2 -1 atmosphere was SF5CF3, with a RE of 0.57 W m ppb and a concentration of 0.12 pptv (as of 1999) (18). The greater RE of PFBAm (0.86 W m-2 ppb-1) causes it to now be the highest-RE compound detected in the atmosphere. Implications for climate depend on the total amount of PFAms released to the atmosphere. PFBAm is not listed as a high production volume (HPV) chemical, but was detected at low levels in the atmosphere. The preliminary determination of

110 Table 5.3: Radiative efficiencies and global warming potentials for selected perfluorinated compounds and PFBAm. From (6) except where indicated.

Global warming potential for given Radiative time horizon efficiency Common name Chemical formula (W m-2 ppb-1) 20 year 100 year 500 year CFC-11 CCl3F 0.25 6730 4750 1620 Nitrogen NF 0.21 12500a 15700a 16600a trifluoride 3 Sulfur SF 0.52 16300 22800 32600 hexafluoride 6 Trifluoromethyl sulfur SF5CF3 0.57 13200 17700 21200 pentafluoride PFC-3-1-10 C4F10 0.33 6330 8860 12500 PFC-4-1-12 C5F12 0.41 6510 9160 13300 PFC-5-1-14 C6F14 0.49 6600 9300 13300 H-Galden 01 CHF2OCF2CF2OCHF2 0.87 5100 1500 460 H-Galden CHF OCF OC F OCHF 1.02b 4430b 1240b 370b 1040x 2 2 2 4 2 PFBAm N(C4F9)3 0.86 5410 6740 6930 a GWPs calculated using method described above with lifetime from Prather and Hsu (5). b Radiative efficiency from Wallington et al. (19); this value incorporated into the GWPs.

the PFBAm atmospheric concentration is 0.057 ± 0.019 pptv, leads to a radiative forcing of 4.9 -5 -2 -5 -2 ± 1.6 × 10 W m . This is comparable to the RF of SF5CF3 of is 6.8 × 10 W m , based on the 1999 levels (18) and is equivalent to approximately 12 % of the RF of HFC-152a, a Kyoto- regulated compound (6). Because the atmospheric concentration of PFBAm is a lower-limit estimate, the corresponding climate impacts are also lower-limits of the potential effects. Two other PFAm congeners, perfluorotripentylamine (PFPAm; N(C5F11)3) and perfluorotrihexylamine (PFHAm; N(C6F13)3) are both considered HPVchemicals (20). Assuming production values are accurate, atmospheric levels of PFPAm and PFHAm are likely to be higher than those of PFBAm. If all of the reported production of these two compounds are eventually released to the atmosphere, this would lead to a global atmospheric concentration of 0.11 – 0.42 parts-per-trillion-by-volume. The radiative efficiencies of PFPAm and PFHAm are likely to be higher than that of PFBAm, as a result of the greater number of C-F bonds. Using the RE of PFBAm as a lower-limit estimate, it is possible to approximate the climate impact of PFAms. Using the determined RE for PFBAm, this would lead to a total addition to radiative forcing of 0.96 – 3.6 x 10-4 W m-2. The total radiative forcing for halocarbons is estimated at

111 0.337 W m-2 (6). However, the bulk of the radiative forcing of halocarbons is dominated by compounds regulated under the Montreal Protocol that are currently in decline. Compounds regulated under the Kyoto protocol account for 0.017 W m-2 of radiative forcing, where the estimated radiative forcing of PFAms is up to 2 % of this value. There is a clear potential for this new class of LLGHGs to impact climate, though more information on other congeners is required.

5.5 Acknowledgements

The authors are grateful to Frank Wania and James Armitage for modeling support. We thank the Chemistry Department machine, glass and electronics shops, along with John Sagebiel, Mike Keith and Trevor VandenBoer for assistance with instrument development. This work was funded by a Natural Science and Engineering Research Council of Canada (NSERC) Discovery Grant.

112 5.6 Sources Cited (1) 3M. 2000. 3M FluorinertTM Electronic Liquid FC-43 Product Information.

(2) Kissa, E. Fluorinated Surfactants and Repellents; Marcel Dekker, Inc.: New York, NY, 2001.

(3) DeMore, W.B.; Sander, S.P.; Golden, D.M.; Hampson, R.F.; Kurylo, M.J.; Howard, C.J.; Ravishankara, A.R.; Kolb, C.E.; Molina, M.J. Chemical kinetics and photochemical data for use in stratospheric modeling, Jet Propulsion Laboratory.

(4) Ravishankara, A.R.; Solomon, S.; Turnipseed, A.A.; Warren, R.F. Atmospheric lifetimes of long-lived halogenated species. Science 1993, 259, 194-199.

(5) Prather, M.J.; Hsu, J. NF3, the greenhouse gas missing from Kyoto. Geophysical Research Letters 2008, 35, L12810.

(6) Forster, P.; Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R.; Fahey, D.W.; Haywood, J.; Lean, J.; Lowe, D.C.; Myhre, G.; Nganga, J.; Prinn, R.; Raga, G.; Schulz, M.; Van Dorland, R. In Climate Change 2007: The Physical Science Basis; Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L., Eds.; Cambridge University Press: Cambridge, United Kingdom, 2007.

(7) Pinnock, S.; Hurley, M.D.; Shine, K.P.; Wallington, T.J.; Smyth, T.J. Radiative forcing of climate by hydrochlorofluorocarbons and hydrofluorocarbons. Journal of Geophysical Research 1995, 100, 23227-23238.

(8) Hilal, S.H.; Carreira, L.A.; Karickhoff, S.W. Prediction of the solubility, activity coefficient, gas/liquid and liquid/liquid distribution coefficients of organic compounds. QSAR Combination Science 2004, 23, 709.

(9) Cahill, T.M.; Mackay, D. A high-resolution model for estimating the environmental fate of multi-species chemicals: Application to malathion and pentachlorophenol. Chemosphere 2003, 53, 571-581.

(10) Hilal, S.H.; Carreira, L.A.; Karickhoff, S.W. Prediction of the vapor pressure, boiling point, heat of vaporization and diffusion coefficient of organic compounds. Quantitative Structure-Activity Relationships 2003, 556, 22.

(11) Wildlife International. Determination of the vapor pressure of PFOS using the spinning rotor gauge method, US Environmental Protection Agency.

(12) Wildlife International. Determination of the water solubility of PFOS by the shake flask method, US Environmental Protection Agency.

(13) Martin, J.W.; Mabury, S.A.; Solomon, K.R.; Muir, D.C.G. Dietary accumulation of perfluorinated acids in juvenile rainbow trout (Oncorhynchus mykiss). Environmental Toxicology and Chemistry 2003, 22, 189-195.

(14) Sanders, N.D.; Butler, J.E.; McDonald, J.R. Product branching ratios in the reaction of 1 O( D2) with NH3. Journal of Chemical Physics 1980, 73, 5381-5383.

113 1 (15) Sorokin, V.I.; Gritsan, N.P.; Chichinin, A.I. Collisions of O( D) with HF, F2, XeF2, NF3, and CF4: Deactivation and reaction. Journal of Chemical Physics 1998, 108, 8995-9003.

(16) Molina, L.T.; Wooldridge, P.W.; Molina, M.J. Atmospheric reactions and ultraviolet and infrared absorptivities of nitrogen trifluoride. Geophysical Research Letters 1995, 22, 1873- 1876.

(17) Jahnke, A.; Ahrens, L.; Ebinghaus, R.; Berger, U.; Barber, J.L.; Temme, C. An improved method for the analysis of volatile polyfluorinated alkyl substances in environmental air samples. Analytical Bioanalytical Chemistry 2007, 387, 965-975.

(18) Sturges, W.T.; Wallington, T.J.; Hurley, M.D.; Shine, K.P.; Sihra, K.; Engel, A.; Oram, D.E.; Penkett, S.A.; Mulvaney, R.; Brenninkmeijer, C.A.M. A potent greenhouse gas identified in the atmosphere: SF5CF3. Science 2000, 289, 611-613.

(19) Wallington, T.J.; Hurley, M.D.; Nielsen, O.J. The radiative efficiency of HCF2OCF2OCF2CF2OCF2H (H-Galden 1040x) revisited. Atmospheric Environment 2009, 43, 4247-4249.

(20) Howard, P.H.; Meylan, W. 2007. EPA Great Lakes Study for Identification of PBTs to Develop Analytical Methods: Selection of Additional PBTs - Interim Report, EPA Contract No. EP-W-04-019.

CHAPTER SIX

Atmospheric Chemistry of Perfluorobutenes (CF3CF=CFCF3 and CF3CF2CF=CF2): Kinetics and Mechanisms of Reactions with OH Radicals and Chlorine Atoms, IR Spectra, Global Warming Potentials, and Oxidation to Perfluorocarboxylic Acids

Cora J. Young, Michael D. Hurley, Timothy J. Wallington and Scott A. Mabury

Published in: Atmos. Env. 2009 43:3717-3724

Reprinted from Atmospheric Environment, 43, Cora J. Young, Michael D. Hurley, Timothy J. Wallington and Scott A. Mabury, Atmospheric Chemistry of Perfluorobutenes (CF3CF=CFCF3 and CF3CF2CF=CF2): Kinetics and Mechanisms of Reactions with OH Radicals and Chlorine Atoms, IR Spectra, Global Warming Potentials, and Oxidation to Perfluorocarboxylic Acids 3717-3724, (2009), with permission from Elsevier

114 115 6.1 Introduction Following ratification of the Montreal Protocol and related agreements, there has been an international effort to replace ozone-depleting chemicals with more environmentally friendly alternatives. Fluorinated compounds are attractive alternatives, as the strength of the C-F bond prevents the formation of free fluorine in the stratosphere and subsequent depletion of ozone. However, C-F bonds absorb in the optically thin spectral region of the atmosphere known as the atmospheric window, leading to a high radiative efficiency of fluorinated compounds. Saturated perfluorocarbons and some hydrofluorocarbons (HFCs) have high radiative efficiencies and are long-lived in the atmosphere, causing them to be classified as long-lived greenhouse gases.

Atmospheric lifetimes of HFCs are typically determined by their reaction with hydroxyl radicals. The presence of an unsaturated moiety in an HFC leads to an increase in rate of reaction with hydroxyl radicals and the additional potential of reaction with ozone. Fluorinated alkenes have been proposed as potential replacements, because of their greatly reduced atmospheric lifetimes relative to saturated HFCs. The atmospheric chemistry of several fluorinated alkenes has already been studied. They have been shown to have lifetimes on the order of days or weeks (1-7) and consequently negligible global warming potentials (1,4-7).

The perfluorobutenes (CF3CF=CFCF3 and CF3CF2CF=CF2) are being considered for industrial applications. Prior to the use of these compounds information on their atmospheric chemistry and environmental impact is needed. Such information is not currently available.

The objective of this study was to examine the atmospheric chemistry of CF3CF=CFCF3 and CF3CF2CF=CF2. Experiments were performed to provide the following information: (i) chlorine atom reaction rates; (ii) hydroxyl radical reaction rates; (iii) products of chlorine atom and hydroxyl radical initiated oxidation; and (iv) infrared spectra.

6.2 Methods

6.2.1 Chemicals

A commercial sample of CF3CF=CFCF3 (97%) was obtained from ABCR Chemicals (Karlsruhe, Germany) and consisted of approximately 70% trans and 30% cis isomers. A sample of CF3CF2CF=CF2 was custom synthesized by ABCR Chemicals (Karlsruhe, Germany).

CH3ONO was synthesized by the drop-wise addition of concentrated sulfuric acid to a saturated solution of NaNO2 in methanol. All other reagents were obtained from commercial sources.

116 Chemicals were subjected to repeated freeze-pump-thaw cycling to remove volatile impurities before use.

6.2.2 Kinetics

Experiments were performed in a 140 liter Pyrex reactor interfaced to a Mattson Sirus 100 FTIR spectrometer. The reactor was surrounded by 22 fluorescent blacklamps (GE F15T8- BL) which were used to photochemically initiate the experiments. Chlorine atoms were produced by photolysis of molecular chlorine:

Cl2 + hv → Cl + Cl

Hydroxyl radicals were produced by photolysis of CH3ONO in air:

CH3ONO + hv → CH3O + NO

CH3O + O2 → HO2 + HCHO

HO2 + NO → OH + NO2

In relative rate experiments the following reactions take place:

OH/Cl + Reactant → products

OH/Cl + Reference → products

It can be shown that:

⎛[X] ⎞ ⎛ k ⎞ ⎛[ref] ⎞ ⎜ t ⎜ x ⎟ ⎜ t ln⎜ ⎟=⎜ ⎟ln⎜ ⎟ [X] kref [ref]t ⎝ t0⎠ ⎝ ⎠⎠⎠ ⎝ 0⎠⎠⎠ where [X]t0, [X]t, [ref]t0, and [ref]t are the concentrations of the compound of interest and reference at times t0 and t, and kX and kref are the rate constants for the reactant and the reference. Mixtures used to determine k(Cl + CF3CF=CFCF3) were composed of 4.1 – 11.5 mTorr CF3CF=CFCF3, 100 mTorr Cl2, and either 4.8 mTorr C2H2 or 25.2 mTorr CH3CH2Cl.

Reaction mixtures used to determine k(Cl + CF3CF2CF=CF2) consisted of 2.8 – 13.2 mTorr

CF3CF2CF=CF2, 100 mTorr Cl2, and either 2.2 – 2.9 mTorr C2H2 or 26.4 mTorr C2H5Cl. To test for potential complications caused by the formation of CF3O radicals, rate constant ratios for CF3CF2CF=CF2 were measured in the absence and in the presence (1.4 – 1.6 mTorr) of NO.

117

NO provides an effective scavenging mechanism for CF3O radicals (8). There was no discernable impact of the presence of NO on the rate constant ratios measured suggesting that complications caused by the presence of CF3O radicals are not significant. Mixtures used to determine k(OH + CF3CF=CFCF3) were made up of 3.7 – 26.0 mTorr CF3CF=CFCF3, 100 mTorr CH3ONO and either 4.3 – 4.7 mTorr C2H2 or 6.9 mTorr C2H4. Reaction mixtures used to determine k(OH + CF3CF2CF=CF2) consisted of 7.9 – 14.8 mTorr CF3CF2CF=CF2, 100 mTorr

CH3ONO and either 2.5 – 4.7 mTorr C2H2 or 3.4 – 4.1 mTorr C2H4. All experiments were performed at 296 ± 1 K in 700 Torr total pressure of air or N2 diluent. Concentrations of reactants and products were monitored by FTIR spectroscopy. Infrared spectra were derived from 32 coadded interferograms with a spectral resolution of 0.25 cm-1 and an analytical path length of 27.7 m. Quoted uncertainties include two standard deviations from the linear least squares regression analysis and uncertainties associated with the IR analyses. Errors associated with the rate constants of the reference compounds contribute an estimated 10% in uncertainty to the determined rate constants.

6.2.3 Products

Products of the chlorine atom- and OH radical- initiated oxidation of CF3CF2CF=CF2 and CF3CF=CFCF3 were monitored using FTIR spectroscopy. Mixtures to study the reaction of chlorine with CF3CF=CFCF3 consisted of 4.9 mTorr CF3CF=CFCF3 and 101 mTorr Cl2. To study the reaction of chlorine with CF3CF2CF=CF2, mixtures of 3.5 – 8.4 mTorr

CF3CF2CF=CF2 and 100 mTorr Cl2 were employed. The reaction of CF3CF2CF=CF2 with hydroxyl radicals was studied using mixtures of 9.7 – 9.9 mTorr CF3CF2CF=CF2, 100 mTorr

CH3ONO, and 50 mTorr NO. Reaction mixtures to study the reaction of CF3CF=CFCF3 with hydroxyl radicals were composed of 4.7 mTorr CF3CF=CFCF3 and 100 mTorr CH3ONO. All product experiments were performed in 700 Torr total pressure of air.

6.3 Results and Discussion

6.3.1 Kinetics of reactions with Cl atoms

The kinetics of reaction (1) were measured relative to reactions (2) and (3)

Cl + CF3CF=CFCF3 → products (1)

Cl + C2H2 → products (2)

118

Cl + C2H5Cl → products (3)

Figure 6.1a shows the loss of CF3CF=CFCF3 versus C2H2 and C2H5Cl following UV irradiation.

The lines through the data in Figure 6.1a are linear least-squares fits, which give k1/k2 = 0.136 ±

0.010 and k1/k3 = 0.952 ± 0.058. Using the apparent second-order rate constant measured in air -11 3 -1 -1 -12 3 -1 -1 at 700 Torr k2 = 5.07 × 10 cm molecule s (9) and k3 = 8.04 × 10 cm molecule s (10), -12 -12 3 -1 -1 we derive values of k1 of (6.89 ± 0.50) × 10 and (7.65 ± 0.47) × 10 cm molecule s . Our ) ) 2.0 t 2.5 t ] ] A 2 B 3

2.0 CF=CF

1.5 2 CF=CFCF CF 3 3 1.5 /[CF /[CF 0 0 t 1.0 t ] ] 3 2 1.0 CF=CF 0.5 2 CF=CFCF

3 0.5 CF 3

Ln ([CF 0.0 0.0 0.0 0.5 1.0 1.5 2.0 2.5 Ln ([CF 0.0 0.5 1.0 1.5 2.0 2.5 Ln ([Reference]t /[Reference]t) Ln ([Reference]t /[Reference]t) 0 0

Figure 6.1: Loss of (a) CF3CF=CFCF3 and (b) CF3CF2CF=CF2 versus C2H2 (▲) and C2H5Cl (●) following exposure to Cl atoms.

final value of k1 is the average of the two determinations, with errors that encompass the -12 3 -1 -1 extremes of the individual measurements: k1 = (7.27 ± 0.88) × 10 cm molecule s . It is important to note that the sample of CF3CF=CFCF3 used was a mixture of 70% trans and 30% cis isomers. Rate constants for the reaction of cis and trans isomers of 2-butene with chlorine atoms have been shown to be indistinguishable within error (11). Rate constant differences have been observed between chlorine atom reactivity of E and Z isomers of CF3CF=CHF (6), but it is not clear how these differences might apply to isomers of CF3CF=CFCF3.

The kinetics of reaction (4) were measured relative to reactions (2) and (3):

Cl + CF3CF2CF=CF2 → products (4)

119

Figure 6.1b shows the loss of CF3CF2CF=CF2 versus C2H2 and C2H5Cl following UV irradiation. The lines through the data in Figure 6.1b are linear least-squares fits, which give -11 3 -1 -1 k4/k2 = 0.322 ± 0.048 and k4/k3 = 2.42 ± 0.31. Using k2 = 5.07 × 10 cm molecule s (9) and -12 3 -1 -1 -11 k3 = 8.04 × 10 cm molecule s (10), we derive values of k4 of (1.63 ± 0.24) × 10 and -11 3 -1 -1 (1.95 ± 0.25) × 10 cm molecule s . Our final value of k4 is the average of the two determinations, with errors that encompass the extremes of the individual measurements: k4 = (1.79 ± 0.41) × 10-11 cm3 molecule-1 s-1.

Chlorine atoms react approximately 2.5 times faster with CF3CF2CF=CF2 than with

CF3CF=CFCF3. The reaction of chlorine atoms with perfluorobutenes proceeds via electrophilic addition to the >C=C< double bond. The lower reactivity of CF3CF=CFCF3 presumably reflects the closer proximity of the electron withdrawing fluorine atom substituents to the >C=C< double bond. To our knowledge, there are no literature data for k1 and k4 to compare with our measurements. Kinetic data are available for the reaction of chlorine atoms -11 3 -1 -1 with perfluoropropene; k(Cl+CF3CF=CF2) = (2.7 ± 0.3) × 10 cm molecule s (3). Consistent with its smaller number of electron withdrawing fluorine substituents, perfluoropropene is more reactive than the perfluorobutenes.

6.3.2 Kinetics of reactions with OH radicals

The kinetics of reaction (5) were measured relative to reactions (6) and (7):

OH + CF3CF=CFCF3 → products (5)

OH + C2H2 → products (6)

OH + C2H4 → products (7)

Figure 6.2a shows the loss of CF3CF=CFCF3 versus C2H2 and C4H4 following UV irradiation.

The lines through the data in Figure 6.2a are linear least-squares fits that give values of k5/k6 = -13 -12 3 0.525 ± 0.080 and k5/k7 = 0.061 ± 0.009. Using k6 = 8.45 × 10 (12) and k7 = 8.52 × 10 cm -1 -1 -13 -13 3 -1 molecule s (13) we derive k5 = (4.44 ± 0.68) × 10 and (5.20 ± 0.77) × 10 cm molecule -1 s , respectively. We cite a final value of k5 that is the average with error limits that include the -13 3 -1 -1 extremes of the individual determinations: k5 = (4.82 ± 1.15) × 10 cm molecule s . The sample of CF3CF=CFCF3 used was a mixture of cis and trans isomers, however the potential impact of these isomers on reactivity is unclear. Studies have shown that 2-butene isomers have equal reactivity, within error (14). Differences in hydroxyl radical reactivity have been

120 observed for E and Z isomers of CF3CF=CHF (6), but it is difficult to relate these differences to isomers of CF3CF=CFCF3. ) 0.25 t 1.0 ) ] t

2 B ] A 3

0.20 0.8 CF=CF 2 CF CF=CFCF

C H 3 3 2 2 0.15 0.6 /[CF /[CF 0 0 t t ] ] 2 3 C H 0.10 2 4 0.4 CF=CF 2

CF=CFCF 0.05 0.2 CF 3 3

Ln ([CF Ln 0.00 0.0 0.0 0.5 1.0 1.5 2.0 2.5 Ln ([CF 0.00.51.01.52.0

Ln ([Reference]t /[Reference]t) Ln ([Reference]t /[Reference]t) 0 0

Figure 6.2: Loss of (a) CF3CF=CFCF3 and (b) CF3CF2CF=CF2 versus C2H2 (▲) and C2H5Cl (●) following exposure to hydroxyl radicals.

The kinetics of reaction (8) were measured relative to reactions (6) and (7):

OH + CF3CF2CF=CF2 → products (8)

OH + C2H2 → products (6)

OH + C2H4 → products (7)

Figure 6.2b shows the loss of CF3CF2CF=CF2 versus C2H2 and C4H4 following UV irradiation.

The lines through the data in Figure 6.2b are linear least-squares fits that give values of k8/k6 =

2.18 ± 0.18 and k8/k7 = 0.24 ± 0.02. Using apparent second order rate constants measured in air -13 -12 3 -1 -1 at 700 Torr, k6 = 8.45 × 10 (12) and k7 = 8.52 × 10 cm molecule s (13) we derive k8 = (1.84 ± 0.15) × 10-12 and (2.04 ± 0.17) × 10-12 cm3 molecule-1 s-1, respectively. We choose to cite a final value of k5 that is the average of the two values together with error limits that -12 3 encompass the extremes of the individual determinations: k8 = (1.94 ± 0.27) × 10 cm molecule-1 s-1.

121

Hydroxyl radicals react approximately four times faster with CF3CF2CF=CF2 than

CF3CF=CFCF3. As with the chlorine atoms, OH radicals react with perfluorobutenes via electrophilic addition to the >C=C< double bond. The closer proximity of the electron withdrawing fluorine atoms to the double bond in CF3CF=CFCF3 presumably contributes to its lower reactivity. The increased steric hindrance of a trifluoromethyl group compared to a fluorine may also play a role in the reduced reactivity of CF3CF=CFCF3. While the kinetics of reactions (5) and (8) have not been studied previously, we can compare our results to the -12 3 -1 -1 measurement (3) of k(OH + CF3CF=CF2) = (2.4 ± 0.3) × 10 cm molecule s . Consistent with its smaller number of electron withdrawing fluorine substituents, perfluoropropene is more reactive than the perfluorobutenes.

6.3.3 Products of Cl atom- and OH radical-initiated oxidation of CF3CF=CFCF3 and

CF3CF2CF=CF2

Figure 6.3 shows IR spectra acquired before (A) and after (B) a 10 second UV irradiation of a mixture of 4.85 mTorr CF3CF=CFCF3 and 100 mTorr Cl2 in 700 Torr of air diluent.

Subtraction of IR features attributable to CF3CF=CFCF3 gives the residual spectrum shown in panel (C). Comparison of panel (C) with a reference spectrum of CF3C(O)F in panel (D) shows the formation of CF3C(O)F as a major product.

Figure 6.4 shows a plot of the observed formation of CF3C(O)F versus the loss of

CF3CF=CFCF3 following irradiation of CF3CF=CFCF3/Cl2 and CF3CF=CFCF3/CH3ONO/NO mixtures in 700 Torr of air diluent. Linear least squares fits to the data give slopes of 196 ±

11% for the CF3CF=CFCF3/Cl2 experiments and 218 ± 20% for the CF3CF=CFCF3/

CH3ONO/NO experiments. The molar yield of CF3C(O)F following chlorine atom or OH radical initiated oxidation of CF3CF=CFCF3 is indistinguishable from 200%. The line through the data in Figure 6.4 has a slope of 2 and is provided to illustrate the data trend. We conclude that the atmospheric oxidation of CF3CF=CFCF3 proceeds via quantitative conversion into

CF3C(O)F.

122

0.6

0.3 A: Before Irradiation

0.0

0.6

0.3 B: After Irradiation

0.0 0.6 Absorbance 0.3 C: B - 0.43*A

0.0 0.6 0.4 D: CF3C(O)F 0.2 0.0 800 1000 1200 1400 1600 1800 2000 -1 Wavenumber (cm )

Figure 6.3: IR spectra acquired before (A) and after (B) UV irradiation of a mixture of 4.85 mTorr CF3CF=CFCF3 and 100 mTorr Cl2 in 700 Torr of air diluent. Panel (C) shows the product spectrum obtained after subtracting features attributable to CF3CF=CFCF3 from panel (B). Panel (D) is a reference spectrum of CF3C(O)F.

6

5

4

3

2 C(O)F Production(mTorr) 3 1 CF

0 0123 Δ CF CF=CF CF (mTorr) 3 2 3

Figure 6.4: Yield of CF3C(O)F following chlorine atom- (●) and hydroxyl radical- (▲) initiated oxidation of CF3CF=CFCF3. The line has a slope of two.

123

Figure 6.5 shows IR spectra acquired before (A) and after (B) a 10 second UV irradiation of a mixture of 7.64 mTorr CF3CF2CF=CF2 and 100 mTorr Cl2 in 700 Torr of air diluent. Subtraction of IR features attributable to CF3CF2CF=CF2 from panel (B) gives the spectrum shown in panel (C). Comparison of panel (C) with the reference spectrum in panel

(D) shows the formation of COF2 as a major product. Subtraction of IR features attributable to

COF2 from panel (C) gives the residual spectrum shown in panel (E). As illustrated by comparison with the reference spectrum for CF3COF in panel (F), the feature centered at approximately 1890 cm-1 in panel (E) is consistent with the carbonyl absorption band expected for the formation of an acyl fluoride product. The product responsible for the carbonyl absorption band shown in panel (E) increased linearly with CF3CF2CF=CF2 consumption.

Figure 6.6A shows a plot of the formation of COF2 versus the loss of CF3CF2CF=CF2 observed following the UV irradiation of CF3CF2CF=CF2/Cl2 and CF3CF2CF=CF2/CH3ONO/NO mixtures in 700 Torr of air diluent. Linear least squares fits give molar COF2 yields of 97 ± 9% and 99 ± 8 % for the chlorine atom and hydroxyl radical initiated oxidation, respectively. Thus,

COF2 is formed with a molar yield which is indistinguishable from 100%. It is expected that

CF3CF2C(O)F is formed as a co-product of COF2. We do not have a reference spectrum for

CF3CF2C(O)F to compare with the residual spectrum in panel E of Figure 6.5. However, as discussed above, the frequency of the absorption band of the unknown product is consistent with that expected for CF3CF2C(O)F. If we assume that the unknown product responsible for the absorption feature at 1890 cm-1 shown in Figure 6.5E is formed in a molar yield of unity then from the integrated absorption in Figure 6.5E over the range 1850-1930 cm-1 and the loss of CF3CF2CF=CF2 (6.69 mTorr) we can calculate an integrated absorption cross section of 2.62 x 10-17 cm molecule-1 for this absorption band. This result is indistinguishable from the integrated absorption cross section of 2.56 x 10-17 cm molecule-1 for the analogous band (1850 – -1 1930 cm ) in our CF3COF reference spectrum shown in Figure 6.5F. The frequency and magnitude of the absorption in Figure 6.5E is consistent with the formation of CF3CF2C(O)F.

Assuming that the absorption in Figure 6.5E is attributable to CF3CF2C(O)F, and the carbonyl absorption band centered at 1890 cm-1 has the same integrated absorption cross section as the analogous feature

124

0.6

0.3 A: Before Irradiation

0.0

0.6

0.3 B: After Irradiation

0.0 0.6

0.3 C: B - 0.47*A

0.0 0.6 Absorbance

0.3 D: COF2

0.0 0.4 E: C - 0.64*D (CF CF C(O)F) 0.2 3 2

0.0

0.6 F: CF3C(O)F

0.3

0.0 800 1000 1200 1400 1600 1800 2000 -1 Wavenumber (cm )

Figure 6.5: IR spectra acquired before (A) and after (B) UV irradiation of a mixture of 7.64 mTorr CF3CF2CF=CF2 and 100 mTorr Cl2 in 700 Torr of air diluent. Panel (C) shows the product spectrum obtained after subtracting features attributable to CF3CF2CF=CF2 from panel (B). Panels (D) and (F) are reference spectra of COF2 and CF3C(O)F. Panel E is the residual spectrum obtained after subtracting features attributable to COF2 from the product spectrum (C).

in CF3COF, then we can calibrate the yields of CF3CF2C(O)F. This approach was used to derive the CF3CF2C(O)F data shown in Figure 6.6. Product yields for CF3CF2C(O)F of 97 ± 9 % and 110 ± 15 % were observed from chlorine atom and hydroxyl radical addition, respectively. The atmospheric oxidation of CF3CF2CF=CF2 leads to the formation of both COF2 and

CF3CF2C(O)F in molar yields indistinguishable from 100%.

125

6 6 A B

5 5

4 4

3 3

Production(mTorr) 2 2 2 C(O)F Production (mTorr) 2 COF CF

1 3 1 CF

0 0 0123456 0123456 Δ CF CF CF=CF (mTorr) Δ CF CF CF=CF (mTorr) 3 2 2 3 2 2

Figure 6.6: Yields of COF2 (a) and CF3CF2C(O)F (b) following chlorine atom-initiated (●) and hydroxyl radical-initiated (▲) oxidation of CF3CF2CF=CF2. The lines have slopes of unity.

6.3.4 Proposed oxidation mechanisms

The reaction of chlorine atoms and hydroxyl radicals with CF3CF=CFCF3 and

CF3CF2CF=CF2 is expected to proceed via addition to the >C=C< double bond leading to a radical which will add O2 to give a peroxy radical. The peroxy radical will react with NO or with another peroxy radical to give an alkoxy radical. The symmetry of CF3CF=CFCF3 leads to a single alkoxy radical (see Figure 6.7). In principle this alkoxy radical could decompose via scission of one of two different C–C bonds giving two sets of products. In practice we observed only one product, CF3C(O)F, showing that decomposition occurs essentially exclusively via scission of the central C–C bond. This preference presumably reflects thermochemical factors.

The observed formation of COF2 and CF3CF2C(O)F from CF3CF2CF=CF2 in molar yields indistinguishable from 100% shows that irrespective of the attacking radical (Cl or OH) or the carbon atom to which it adds (terminal or interior) the fate of the resulting alkoxy radical is decomposition via scission of the terminal C–C bond. This mechanism is illustrated in Figure 6.7b for the radicals generated following addition of OH to the terminal carbon atom.

126

(a) ● OH O OH F CF3 ● F F OH ● +O2/+NO CF3 CF3 F C F F C /-NO2 F C 3 3 F 3 F

O O OH O2 ● HO2 F3C F F CF3 F3C F

(b) ● OH O OH F F F F ● ● OH +O2/+NO F3C F F3C F3C F F F /-NO2 F F F F F F F O F C O OH 3 O2 ● F HO 2 F F F F F F

Figure 6.7: Proposed mechanism for the atmospheric oxidation of (a) CF3CF=CFCF3 and (b) CF3CF2CF=CF2 initiated by hydroxyl radicals. Stable products are given in boxes. In part (b), only addition to the terminal carbon is depicted for simplicity.

6.3.5 Infrared spectra and radiative efficiency of CF3CF=CFCF3 and CF3CF2CF=CF2

IR spectra were recorded at 296 K using 2.2 – 7.5 mTorr of CF3CF=CFCF3 and 1.79 –

9.05 mTorr of CF3CF2CF=CF2 (separately) in 700 Torr of air diluent. The IR features scaled linearly with the perfluorobutene concentration. The absolute absorption spectra are shown in -1 Figure 6.8. The integrated cross sections (650 – 1500 cm ) of CF3CF=CFCF3 and -16 -1 CF3CF2CF=CF2 are 2.88 and 2.14 × 10 cm molecule , respectively. Uncertainties in the cross section measurement arise from following sources: sample concentration (2%), sample purity (2%), path length (1.5%), spectrum noise (± 10-20 cm2 molecule-1), and residual baseline offset after subtraction of background (1.5%). From these individual uncertainties, the total (random) uncertainty in the integrated absorption cross section is ± 4 %. We prefer to quote a conservative uncertainty of ± 5 %. Hence, the integrated cross sections of CF3CF=CFCF3 and -16 -1 CF3CF2CF=CF2 are (2.88 ± 0.14) and (2.14 ± 0.11) × 10 cm molecule , respectively.

127

5 A

4 ) -1

3 molecule 2

2 cm -18 (10

σ 1

0 600 800 1000 1200 1400 1600 1800 2000 Wavenumber (cm-1)

5 B

4 ) -1

3 molecule 2

2 cm -18 (10

σ 1

0 600 800 1000 1200 1400 1600 1800 2000 -1 Wavenumber (cm )

Figure 6.8: Infrared spectra of (a) CF3CF=CFCF3 and (b) CF3CF2CF=CF2.

Pinnock et al. (1995) have presented a simple method that can be used to estimate radiative efficiency from IR absorption spectra. In this method, the region between 0 and 2500 cm-1 is divided up into 250 bands 10 cm-1 wide. The radiative efficiency for a 0→1 ppbv change in atmospheric concentration of the compound can then be calculated using the expression: 250 Radiative efficiency = 10(cm-1) i Fi Σ σ av σ i=1

128

i 2 -1 where σ av is the average absorption cross section in band i in units of cm molecule , and i F σ is the radiative forcing per unit cross section per wavenumber per part-per-billion in band i in units of W m-2 (cm-1)-1 (cm2 molecule -1)-1. Using this method, the IR spectra of

CF3CF=CFCF3 and CF3CF2CF=CF2 shown in Figure 6.8, and the IR spectrum of CFC-11 reported elsewhere (15), we calculate radiative efficiencies for CF3CF=CFCF3, CF3CF2CF=CF2, -2 -1 and CFC-11 of 0.32, 0.29, and 0.26 W m ppb , respectively. Although CF3CF=CFCF3 and

CF3CF2CF=CF2 are isomers the radiative efficiency of CF3CF=CFCF3 is about 10% greater reflecting the fact that chemical structure impacts the radiative efficiency of fluorinated compounds (16).

It should be noted that the model of radiative efficiency utilized here (17) assumes a uniform distribution of the chemical in question. For short-lived compounds such as

CF3CF=CFCF3 and CF3CF2CF=CF2, greater concentrations will be found in the boundary layer, where radiative effects are lower. Radiative efficiencies based on distributions calculated using chemical transport models can be substantially lower for fluoroalkenes (by factors of 23 for

C2F4 and 8 for C3F6) than those based on uniform distributions (1). Thus, our values should be considered upper limits for the radiative efficiencies of CF3CF=CFCF3 and CF3CF2CF=CF2.

6.4 Atmospheric Implications

Assuming an average global concentration of hydroxyl radicals of 1 × 106 molecules -3 cm (18) gives lifetimes for CF3CF=CFCF3 and CF3CF2CF=CF2 with respect to reaction with OH of about twenty-four and six days, respectively. The approximate nature of these lifetimes must be stressed. The concentration of hydroxyl radicals, and hence atmospheric lifetime, varies substantially with season and latitude.

Tropospheric concentrations of chlorine atoms are highly variable and uncertain, levels in the marine boundary layer of up to 1 × 105 molecules cm-3 have been reported (19), with a tropospheric average of <5 – 10 × 102 molecules cm-3 (20). Thus, typical lifetimes for

CF3CF=CFCF3 and CF3CF2CF=CF2 with respect to reaction with chlorine atoms will be on the order of a few years in the free troposphere, and a few weeks under marine boundary layer conditions. Reaction with chlorine atoms is not expected to be a significant fate of

CF3CF=CFCF3 or CF3CF2CF=CF2. Reactions of alkenes with ozone can contribute to degradation of these compounds in the atmosphere. The reaction of ozone with perfluoropropene proceeds with a rate constant of 6.2 × 10-22 cm3 molecule-1 s-1 (1) As

129 discussed for chlorine atom and OH radical reactions we would expect the reactivity of the perfluorobutenes to be lower than perfluoropropene. Hence, the lifetime of the perfluorobutenes with respect to reaction with O3 will probably be greater than 8 years and not significant. We conclude that the atmospheric lifetimes of CF3CF=CFCF3 and CF3CF2CF=CF2 are determined by reaction with hydroxyl radicals and are approximately twenty-four and six days, respectively.

The halocarbon global warming potential for CF3CF2CF=CF2 (relative to CFC-11) can be estimated using the expression:

⎛ RE ⎞⎛τ M ⎞⎛ 1 −exp( −t τ )/ ⎞ HGWP = ⎜ X ⎟⎜ X CFC−11⎟⎜ X ⎟ X ⎜ ⎟⎜τ ⎟⎜ −− τ ⎟ ⎝RE CFC−11⎠⎝ CFC−11 MX ⎠⎝1 exp( t CFC−11)/ ⎠

where REX, RECFC-11, MX, MCFC-11, τX, and τCFC-11 are the radiative efficiencies, molecular weights, and atmospheric lifetimes of the molecule of interest and CFC-11, and t is the time horizon over which the forcing is integrated. Assuming τ(CF3CF=CFCF3) = 24 days,

τ(CF3CF2CF=CF2) = 6 days, and τ(CFC-11)= 45 years (21), we estimate the HGWPs of -3 -4 CF3CF=CFCF3 and CF3CF2CF=CF2 (relative to CFC-11) are 1.3 × 10 and 2.5 × 10 for a 100 year horizon, respectively. Relative to CO2, the GWP of CFC-11 on a 100 year time horizon is

4750 (21). Thus, we estimate that relative to CO2, the GWPs of CF3CF=CFCF3 and

CF3CF2CF=CF2 are approximately 6 and 1, respectively, for a 100 year time horizon. It should be noted that these GWP values are upper limits, due to the likelihood that the radiative efficiency values are over-estimated (see section 3.5). It is clear that neither CF3CF=CFCF3 nor

CF3CF2CF=CF2 will contribute to radiative forcing of climate change.

CF3CF=CFCF3 and CF3CF2CF=CF2 do not contain chlorine, bromine, or iodine and hence they will not have any significant impact on stratospheric ozone. Reaction of

CF3CF=CFCF3 and CF3CF2CF=CF2 with chlorine atoms does not form any stable addition products. The absence of stable chlorinated products as a result of atmospheric oxidation of these compounds suggests negligible impact on stratospheric ozone.

Atmospheric oxidation of CF3CF=CFCF3 gives CF3C(O)F, while oxidation of

CF3CF2CF=CF2 gives COF2 and CF3CF2C(O)F. COF2 is a common product of the atmospheric oxidation of fluorinated compounds and is the dominant product of the degradation of numerous hydrofluorocarbons (22,23), hydrochlorofluorocarbons, (22) and hydrofluoroethers (24,25).

The atmospheric fate of COF2 is hydrolysis to yield CO2 and HF. At the levels expected from

130 atmospheric oxidation of perfluorobutenes (and HFCs in general) the formation of HF is not of any environmental significance. The fate of CF3C(O)F and CF3CF2C(O)F is hydrolysis to form CF3C(O)OH (trifluoroacetic acid, TFA) and CF3CF2C(O)OH (perfluoropropionic acid, PFPrA), respectively. TFA and PFPrA are members of the class of compounds called perfluorocarboxylic acids (PFCA) which have attracted attention because they are persistent and accumulate in biota. However, PFCAs with less than eight carbons such as TFA and PFPrA do not appear to be bioaccumulative (26,27).

TFA is widespread in the environment (28,29) and is produced during the atmospheric degradation of several anthropogenic pollutants (30). TFA has been detected in deep ocean water and appears to be a natural trace component of the oceanic environment (31). It is generally accepted that any additional environmental burden of TFA resulting from the atmospheric degradation of HCFCs and HFCs will not have any significant environmental impact (30). PFPrA is present in the environment and has been detected in rainwater (29), although its source is unclear. Further work is needed to clarify the sources, fate, and environmental impact of PFPrA. As illustrated in the present work, the yields and identity of PFCAs formed in the atmosphere are dictated by the molecular structure of the parent fluorinated organic compounds.

6.5 Acknowledgements

The authors thank Craig Butt for experimental assistance. CJY is grateful to the Natural Science and Engineering Research Council of Canada for a CGS Fellowship.

131 6.6 Sources Cited

(1) Acerboni, G.; Beukes, J.A.; Jensen, N.R.; Hjorth, J.; Myhre, G.; Nielsen, C.J.; Sundet, J.K. Atmospheric degradation and global warming potentials of three perfluoroalkenes. Atmospheric Environment 2001, 35, 4113-4123.

(2) Orkin, V.L.; Huie, R.E.; Kurylo, M.J. Rate constants for the reactions of OH with HFC- 245cb (CH3CF2CF3) and some fluoroalkenes (CH2CHCF3, CH2CFCF3, and CF2CF2). Journal of Physical Chemistry A 1997, 101, 9118-9124.

(3) Mashino, M.; Ninomiya, Y.; Kawasaki, M.; Wallington, T.J.; Hurley, M.D. Atmospheric chemistry of CF3CF=CF2: Kinetics and mechanism of its reactions with OH radicals, Cl atoms, and ozone. Journal of Physical Chemistry A 2000, 104, 7255-7260.

(4) Nielsen, O.J.; Javadi, M.S.; Sulbaek Andersen, M.P.; Hurley, M.D.; Wallington, T.J.; Singh, R. Atmospheric chemistry of CF3CF=CH2: Kinetics and mechanisms of gas-phase reactions with Cl atoms, OH radicals, and O3. Chemical Physics Letters 2007, 439, 18-22.

(5) Søndergaard, R.; Nielsen, O.J.; Hurley, M.D.; Wallington, T.J.; Singh, R. Atmospheric chemistry of trans-CF3CH=CHF: Kinetics of the gas-phase reactions with Cl atoms, OH radicals, and O3. Chemical Physics Letters 2007, 443, 199-204.

(6) Hurley, M.D.; Ball, J.C.; Wallington, T.J. Atmospheric chemistry of the Z and E isomers of CF3CF=CHF; Kinetics, mechanisms, and products of gas-phase reactions with Cl atoms, OH radicals, and O3. Journal of Physical Chemistry A 2007, 111, 9789-9795.

(7) Papadimitriou, V.C.; Talukdar, R.K.; Portmann, R.W.; Ravishankara, A.R.; Burkholder, J.B. CF3CF=CH2 and (Z)-CF3CF=CHF: temperature dependent OH rate coefficients and global warming potentials. Physical Chemistry Chemical Physics 2008, 10, 808-820.

(8) Wallington, T.J.; Schneider, W.F.; Worsnop, D.R.; Nielsen, O.J.; Sehested, J.; DeBruyn, W.; Shorter, J.A. Environmental impact of CFC replacements: HFCs and HCFCs. Environmental Science and Technology 1994, 28, 320A.

(9) Wallington, T.J.; Andino, J.M.; Lorkovic, I.M.; Kaiser, E.W.; Marston, G. Pressure dependence of the reaction of chlorine atoms with ethene and acetylene in air at 295K. Journal of Physical Chemistry 1990, 94, 3644-3648.

2 (10) Wine, P.H.; Semmes, D.H. Kinetics of Cl( PJ) reactions with the chloroethanes CH3CH2Cl, CH3CHCl2, CH2ClCH2Cl, and CH2ClCHCl2. Journal of Physical Chemistry 1983, 87, 3572-3578.

(11) Kaiser, E.W.; Donahue, C.J.; Pala, I.R.; Wallington, T.J.; Hurley, M.D. Kinetics, products and stereochemistry of the reaction of chlorine atoms with cis- and trans-2-butene in 10 - 700 Torr of N2 or N2/O2 diluent at 297 K. Journal of Physical Chemistry A 2007, 111, 1286-1299.

(12) Sørensen, M.; Kaiser, E.W.; Hurley, M.D.; Wallington, T.J.; Nielsen, O.J. Kinetics of the reaction of OH radicals with acetylene in 25-8000 Torr of air at 296 K. International Journal of Chemical Kinetics 2003, 35, 191-197.

132

(13) Calvert, J.G.; Atkinson, R.; Kerr, J.A.; Madronich, S.; Moortgat, G.K.; Wallington, T.J.; Yarwood, G. The Mechanisms of Atmospheric Oxidation of the Alkenes; Oxford University Press: Oxford, 2000.

(14) Atkinson, R. Kinetics and mechanisms of the gas-phase reactions of the hydroxyl radical with organic compounds under atmospheric conditions. Chemical Reviews 1986, 86, 69-201.

(15) Ninomiya, Y.; Kawasaki, M.; Guschin, A.; Molina, L.T.; Molina, M.; Wallington, T.J. Atmospheric chemistry of n-C3F7OCH3: Reaction with OH radicals and Cl atoms and atmospheric fate of n-C3F7OCH2O. Environmental Science and Technology 2000, 34, 2973- 2978.

(16) Young, C.J.; Hurley, M.D.; Wallington, T.J.; Mabury, S.A. Molecular structure and radiative efficiency of fluorinated ethers: A structure-activity relationship. Journal of Geophysical Research 2008, 113, doi:10.1029/2008JD010178.

(17) Pinnock, S.; Hurley, M.D.; Shine, K.P.; Wallington, T.J.; Smyth, T.J. Radiative forcing of climate by hydrochlorofluorocarbons and hydrofluorocarbons. Journal of Geophysical Research 1995, 100, 23227-23238.

(18) Prinn, R.G.; Huang, J.; Weiss, R.F.; Cunnold, D.M.; Fraser, P.J.; Simmonds, P.G.; McCulloch, A.; Salameh, P.; O'Doherty, S.; Wang, R.H.J.; Porter, L.; Miller, B.R. Evidence for substantial variation of atmospheric hydroxyl radicals in the past two decades. Science 2001, 292, 1882-1888.

(19) Spicer, C.W.; Chapman, E.G.; Finlayson-Pitts, B.J.; Plastridge, R.A.; Hubbe, J.M.; Fast, J.D.; Berkowitz, C.M. Unexpectedly high concentrations of molecular chlorine in coastal air. Nature 1998, 394, 353-356.

(20) Singh, H.B.; Thakur, A.N.; Chen, Y.E.; Kanakidou, M. Tetrachloroethylene as an indicator of low Cl atom concentrations in the troposphere. Geophysical Research Letters 1996, 23, 1529-1532.

(21) Forster, P.; Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R.; Fahey, D.W.; Haywood, J.; Lean, J.; Lowe, D.C.; Myhre, G.; Nganga, J.; Prinn, R.; Raga, G.; Schulz, M.; Van Dorland, R. In Climate Change 2007: The Physical Science Basis; Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L., Eds.; Cambridge University Press: Cambridge, United Kingdom, 2007.

(22) Tuazon, E.C.; Atkinson, R. Tropospheric transformation products of a series of hydrofluorocarbons and hydrochlorofluorocarbons. Journal of Atmospheric Chemistry 1993, 17, 179-199.

(23) Taketani, F.; Nakayama, T.; Takahashi, K.; Matsumi, Y.; Hurley, M.D.; Wallington, T.J.; Toft, A.; Sulbaek Andersen, M.P. Atmospheric chemistry of CH3CHF2 (HFC-152a): Kinetics, mechanisms, and products of Cl atom- and OH radical-initiated oxidation in the presence and absence of NOx. Journal of Physical Chemistry A 2005, 109, 9061-9069.

133 (24) Tuazon, E.C. Tropospheric degradation products of novel hydrofluoropolyethers. Environmental Science and Technology 1997, 31, 1817-1821.

(25) Sulbaek Andersen, M.P.; Nielsen, O.J.; Wallington, T.J.; Hurley, M.D.; DeMore, W.B. Atmospheric chemistry of CF3OCF2CF2H and CF3OC(CF3)2H: Reaction with Cl atoms and OH radicals, degradation mechanism, global warming potentials, and empirical relationship between k(OH) and k(Cl) for organic compounds. Journal of Physical Chemistry A 2005, 109, 3926- 3934.

(26) Martin, J.W.; Mabury, S.A.; Solomon, K.R.; Muir, D.C.G. Dietary accumulations of perfluorinated acids in juvenile rainbow trout (Oncorhynchus mykiss). Environmental Toxicology and Chemistry 2003, 22, 189-195.

(27) Martin, J.W.; Mabury, S.A.; Solomon, K.R.; Muir, D.C.G. Bioconcentration and tissue distribution of perfluorinated acids in rainbow trout (Oncorhynchus mykiss). Environmental Toxicology and Chemistry 2003, 22, 196-204.

(28) Scott, B.F.; Macdonald, R.W.; Kannan, K.; Fisk, A.; Witter, A.; Yamashita, N.; Durham, L.; Spencer, C.; Muir, D.C.G. Trifluoroacetate profiles in the Arctic, Atlantic and Pacific Oceans. Environmental Science and Technology 2005, 39, 6555-6560.

(29) Scott, B.F.; Spencer, C.; Mabury, S.A.; Muir, D.C.G. Poly and perfluorinated carboxylates in North American precipitation. Environmental Science and Technology 2006, 40, 7167-7174.

(30) World Meteorological Organization. 2007. Scientific Assessment of Ozone Depletion: 2006.

(31) Frank, H.; Christoph, E.H.; Holm-Hansen, O.; Bullister, J.L. Trifluoroacetate in ocean waters. Environmental Science and Technology 2002, 36, 12-15.

CHAPTER SEVEN

Atmospheric Chemistry of 4:2 Fluorotelomer Iodide (n-C4F9CH2CH2I): Kinetics and Products of Photolysis and Reaction with OH Radicals and Cl Atoms

Cora J. Young, Michael D. Hurley, Timothy J. Wallington and Scott A. Mabury

Published in: J. Phys. Chem. A 2008 112:13542-13548

Reproduced with permission from Journal of Physical Chemistry A

Copyright ACS 2008

134 135 7.1 Introduction

Perfluorocarboxylic acids (PFCAs) are found ubiquitously in water and biota in urban and remote locations such as the Arctic (1,2). Given their use in products associated with industrial society it is not surprising that PFCAs are found in urban areas. However, as these compounds exist mainly as anions in environmental media and are not subject to long-range atmospheric transport, their presence in remote locations is somewhat surprising. Two hypotheses have been put forward to account for the presence of PFCAs in remote regions: (i) oceanic transport of PFCAs followed by transfer to land by marine aerosols (3-5) and (ii) atmospheric oxidation of volatile precursors to PFCAs followed by deposition in precipitation (6-8). Evidence supporting the oceanic hypothesis comes from computer modeling studies 5. Evidence supporting the atmospheric hypothesis comes from field measurements (9,10), laboratory experiments (6,7), and computer models (8). The relative importance of the atmospheric and oceanic pathways is a matter of current debate.

Smog chamber experiments have shown that the atmospheric oxidation of fluorotelomer alcohols (FTOHs) (6,7), fluorosulfamido alcohols (11,12), and fluorinated olefins (13) are likely sources of PFCAs in remote locations. It is possible that other classes of volatile compounds are additional sources of PFCAs. One such class of compounds is fluorotelomer iodides (FTI) which are used in the synthesis of FTOHs, fluorotelomer olefins, and other products. Production of the FTIs and the telomer products is in the range of millions of kg (14,15), but the level of emission of FTIs is unknown. Considering their high use and volatility, it seems reasonable to expect that FTIs will be emitted into the atmosphere, however no measurements have yet been made to quantify such emissions. To our knowledge, FTIs have not been the subject of any environmental studies.

Assessments of whether FTIs can survive long-range transport and degrade to form PFCAs in the environment require data for the atmospheric chemistry of FTIs. There are no reported studies of the atmospheric chemistry of FTIs. The objective of the present study was to improve our understanding of the atmospheric chemistry of FTIs by examining a representative compound, CF3(CF2)3CH2CH2I (4:2 FTI). Smog chamber techniques were used to study: (i) the kinetics of reaction of Cl atoms and OH radicals with 4:2 FTI, (ii) the products of the Cl atom initiated oxidation of 4:2 FTI, (iii) the rate of photolysis, and (iv) the products of 4:2 FTI photolysis.

136 7.2 Experimental

7.2.1 Smog chamber methods

Experiments were performed in a 140-liter Pyrex reactor interfaced to a Mattson Sirus 100 FTIR spectrometer (16). The reactor was surrounded by 22 fluorescent blacklamps (GE F15T8-BL), which have maximum emission at approximately 360 nm and were used to photochemically initiate the experiments. Chlorine atoms were produced by photolysis of molecular chlorine.

Cl2 + hv → Cl + Cl (1)

OH radicals were produced by photolysis of CH3ONO in the presence of NO in air.

CH3ONO + hv → CH3O + NO (2)

CH3O + O2 → HO2 + HCHO (3)

HO2 + NO → OH + NO2 (4)

In the relative rate experiments the following reactions take place.

Cl + Reactant → products (5)

Cl + Reference → products (6)

OH + Reactant → products (7)

OH + Reference → products (8)

Assuming that the reactant and reference compounds are lost solely via reaction with Cl atoms (or OH radicals) and that neither the reactant, nor reference, are reformed in any processes, then it can be shown that ⎛[X] ⎞ ⎛ k ⎞ ⎛[ref] ⎞ ⎜ t ⎜ x ⎟ ⎜ t (9) ln⎜ ⎟=⎜ ⎟ln⎜ ⎟ [X] kref [ref]t ⎝ t0⎠ ⎝ ⎠ ⎝ 0⎠ where [X]t0, [X]t, [ref]t0, and [ref]t are the concentrations of the compound of interest and reference at times t0 and t, and kX and kref are the rate constants for the reactant and the reference. Plots of Ln[X]t0/[X]t) versus Ln([ref]t0/[ref]t) should be linear, pass through the origin, and have a slope of kX/kref.

137

CH3ONO was synthesized by the drop wise addition of concentrated sulfuric acid to a saturated solution of NaNO2 in methanol. The C4F9CH2CHO reference spectrum was obtained using a synthesized standard (7). The C4F9CH2C(O)OOH reference spectrum was derived from the analysis described by Hurley et al (7). Other reagents were obtained from commercial sources. Experiments were conducted in 700 Torr total pressure of N2, or N2/O2 diluent at 296 K. Compounds were monitored using absorption features at the following wavenumbers (cm-1):

C4F9CH2CH2I (740), C4F9CH2CHO (1752), C4F9CHO (1778), C4F9C(O)OOH (1450), COF2

(1944) and CO (2150). To remove volatile impurities C4F9CH2CH2I was subjected to repeated freeze/pump/thaw cycling before use. IR spectra were derived from 32 coadded interferograms with a spectral resolution of 0.25 cm-1 and an analytical path length of 27.1 m. To check for unwanted loss of reactants and reference compounds via heterogeneous reactions, initial reaction mixtures were left to stand in the chamber in the dark for 60 minutes; there was no observable (< 2%) loss of reactants. To check for loss via photolysis a mixture of 8.8 mTorr 4:2 FTI in 700 Torr of air was subjected to 5 minutes of UV irradiation using the output from 22 blacklamps. There was no discernable loss (< 2%) of 4:2 FTI. Photolysis is not a significant loss of 4:2 FTI in experiments using blacklamps (i.e., those described in sections 3.1 and 3.2). Photolysis product studies were carried out in the same setup, except that the 12 blacklamps were replaced with phosphor coated sunlamps (GE-FS40), with maximum emission at approximately 310 nm (see Figure 4 in Taniguichi et al. (17)). Product studies were performed using 17.6 – 26.4 mTorr of 4:2 FTI in 50 Torr oxygen diluent. Unless stated otherwise, quoted uncertainties are 2 standard deviations from least squares regressions.

7.2.2 Offline sample collection and analysis

Offline samples were collected by bubbling approximately 5 L of chamber air through pH 11 aqueous Na2CO3 solution. Sodium carbonate solutions were acidified to pH 4 using HCl and analyzed using a Waters Acquity™ Ultra Performance LC with detection by a Micromass Quattro Micro MS/MS detector. Analytes were separated on a Luna C18 column (2.0 mm x 50 mm x 2.5 µm) maintained at 30ºC. The mobile phase consisted of methanol and water, both containing 10 mM ammonium acetate, with a flow rate 0.2 mL min-1. Separation was achieved using a 7 min gradient starting at 10% methanol and 90% water, holding for 0.5 min, increasing the methanol component over the next 2.5 minutes to 60%, with that composition held for 1.5 min before returning to initial conditions over 0.5 minutes and holding for 2.0 min equilibration period. Analysis was performed in triplicate with 7.5 µL injections using isotopically labeled

138 13 perfluorobutanoic acid ( C4-PFBA, Wellington Laboratories) as an internal standard. PFCAs were analyzed with a cone voltage of 17 V and collision energy of 9 eV and the following transitions were monitored: perfluoropentanoic acid (PFPeA) 263 > 219, perfluorobutanoic acid 13 (PFBA) 213 > 169, C4-PFBA 217 > 172, perfluoropropionic acid (PFPrA) 163 > 119 and trifluoroacetic acid (TFA) 113 > 69.

7.2.3 UV Spectral measurements and photolysis rate calculations

The gas phase UV-visible spectra of ethyl iodide and 4:2 FTI were recorded over the wavelength range 200-400 nm in a 6 cm cell using a commercial dual beam (Lambda 18 Perkin- Elmer) spectrometer with a resolution of 1 nm. Spectra were recorded in 700 Torr of air or nitrogen. Photolysis rates were calculated using the Tropospheric Ultraviolet-Visible (TUV 4.2) package (18) assuming a photolysis quantum yield of unity consistent with other alkyl iodides (19).

7.3 Results and discussion

7.3.1 Kinetics of the Cl + 4:2 FTI reaction

The rate of reaction (10) was measured relative to reactions (11) and (12):

Cl + 4:2 FTI → products (10)

Cl + CH3Cl → products (11)

Cl+ CH3OCHO → products (12)

Reaction mixtures consisted of 1.8 – 4.9 mTorr of 4:2 FTI, 100 – 110mTorr Cl2, and either 15 –

20 mTorr CH3Cl, or 2.8-5.9 mTorr CH3OCHO, in 700 Torr of N2, diluent. The observed loss of 4:2 FTI versus those of the reference compounds is plotted in Figure 7.1. Linear least squares analysis of the data in Figure 7.1 gives k10/k11 = 2.43 ± 0.21 and k10/k12 = 1.01 ± 0.09. Quoted errors include two standard deviations from the regression analysis and uncertainty in the analysis of the IR spectra.

-13 -12 (21) -12 Using k11 = 4.8 x 10 (20) and k12 = 1.3 x 10 gives k10 = (1.17 ± 0.11) x 10 and (1.31 ± 0.12) x 10-12 cm3 molecule-1 s-1. We choose to cite a final value which is the average of the individual determinations together with error limits which encompass the extremes of the -12 3 -1 determinations, hence k10 = (1.24 ± 0.19) x 10 cm molecule

139

1.8 ) t

I] 1.6 2

CH3Cl CH

2 1.4 CH

9 1.2 F 4

/[C 1.0 t0 I] 2 0.8 CH 2 0.6 CH 9

F CH OCHO 4 0.4 3 C 0.2 Ln ([

0.0 0.00.10.20.30.40.50.60.70.8 Ln ([Reference] /[Reference] ) t0 t

Figure 7.1. Decay of 4:2 FTI versus CH3Cl and CH3OCHO in the presence of Cl atoms in 700 Torr of N2 at 295 ± 2 K.

-1 s . While there have been no previous studies of k10, we can compare our result with k(Cl + 4:2 FTOH) = (1.61 ± 0.49) x 10-11 cm3 molecule-1 s-1 (6). Replacement of the alcohol functionality by an iodine atom reduces the reactivity of the molecule towards chlorine atoms by a factor of approximately 13. The presence of an alcohol group generally increases the reactivity of neighbouring C–H bonds; hence the observed qualitative difference between 4:2 FTI and 4:2 FTOH is consistent with expectations.

7.3.2 Kinetics of the OH + 4:2 FTI reaction

The rate of reaction (13) was measured relative to reactions (14) and (15):

OH + 4:2 FTI → products (13)

OH + C2H4 → products (14)

OH + C3H8 → products (15)

Initial reaction mixtures consisted of 7.4 – 20.6 mTorr of C4F9CH2CH2I, 102 mTorr CH3ONO, and either 3.2 – 4.4 mTorr C2H4 or 50 – 68 mTorr C3H8 in 700 Torr total pressure of air diluent. Figure 7.2 shows the loss of 4:2 FTI plotted versus loss of the reference compounds. The uncertainty bars indicate uncertainties in the IR analysis. As a result of interfering absorptions from CH3ONO and its products, the IR analysis of 4:2 FTI has greater uncertainty in the OH

140 than in the Cl rate experiments (compare y-axis error bars in Figures 1 and 2). The lines through the data in Figure 7.2 are linear least squares fits which give k13/k14 = 0.11 ± 0.04 and k13/k15 = 1.26 ± 0.28, uncertainties include two standard deviations from the regressions and -12 -12 uncertainties associated with the IR analysis. Using k14 = 8.52 x 10 (22) and k15 = 1.1 x 10 -13 -12 3 -1 -1 (20) gives k13 = (9.4 ± 3.4) x 10 and k13 = (1.39 ± 0.31) x 10 cm molecule s . Within the admittedly substantial uncertainties, the two determinations are in agreement. We choose to report an average with uncertainties which encompass the extremes of the individual -12 3 -1 -1 determinations, hence, k13 = (1.2 ± 0.6) x 10 cm molecule s .

1.2 ) t I] 2 1.0 C H CH C2H2 3 8 2 CH

9 0.8 F 4 /[C t0

I] 0.6 2

CH C H 2 2 4 0.4 CH 9

F C2H4 4

C 0.2 Ln ([

0.0 0.0 0.5 1.0 1.5 Ln ([Reference] /[Reference] ) t0 t

Figure 7.2. Decay of 4:2 FTI versus C2H4 and C3H8 in the presence of OH radicals in 700 Torr of air diluent at 295 ± 2 K.

The reactivity of 4:2 FTI towards OH radicals can be estimated using the SAR method. Using the approach described by Kwok and Atkinson (23) the reactivity of 4:2 FTI is given as ktotal = k(–CH2–)*F(–CF2–)*F(–CH2I) + k(–CH2–)*F(–CH2–)*F(–I). The database concerning reactions of OH radicals with iodoalkanes was very limited at the time that Kwok and Atkinson developed their SAR factors and consequently the factor F(–CH2I) was not included in their -12 analysis. Recently, Carl and Crowley (24) have reported k(OH+CH3CH2CH2I) = 1.47 x 10 3 -1 -1 cm molecule s . In SAR terms the rate constant k(OH+CH3CH2CH2I) can be expressed as k(–CH3)*F(–CH2–) + k(–CH2–)*F(–CH3)*F(–CH2I) + k(–CH2–)*F(–CH2–)*F(–I). Taking -14 -13 values of k(–CH3) = 1.36 x 10 , k(–CH2–) = 9.34 x 10 , F(–CH2–) = 1.23, F(–CH3) = 1.0, and F(–I) = 0.53 we derive F(–CH2I) = 0.90. Using this factor we can estimate k(OH+ 4:2 FTI) = 9.34 x 10-13 x 0.018 x 0.90 + 9.34 x 10-13 x 1.23 x 0.53 = 0.15 x 10-13 + 6.09 x 10-13 = 6.24 x

141 -13 3 -1 -1 10 cm molecule s with the majority ( 98%) of the reactivity occurring at the –CH2I group. Kwok and Atkinson (23) noted that for the case of haloalkanes there were often significant disagreements (> factor of 2) between rate constants estimated using the SAR method and those measured experimentally. The measured value of k(OH+ 4:2 FTI) = (1.17 ± 0.57) x 10-12 cm3 molecule-1 s-1 is reasonably consistent with expectations based upon SAR calculations.

The value of k(OH + 4:2 FTI) measured in the present work can be used to provide an estimate of the atmospheric lifetime of 4:2 FTI. Using a global weighted-average OH concentration (25) of 1.0 x 106 cm-3 leads to an estimated lifetime of 4:2 FTI with respect to reaction with OH radicals of approximately 10 days. The approximate nature of this lifetime estimate should be stressed; the average daily concentration of OH radicals and temperature vary significantly with both location and season. The quoted lifetime is a global average, but local lifetimes could be significantly shorter or longer. Based on production of FTOHs and related products (15), it seems reasonable to expect environmental release of 6:2, 8:2 and 10:2 FTIs to be the most significant. Studies have shown there is no effect of fluorinated chain length on OH reaction rates for telomer compounds, indicating that lifetimes with respect to OH reaction of higher fluorinated FTIs will be similar to those measured in this study.

7.3.3 Products of Cl + 4:2 FTI reaction

The mechanism of Cl atom initiated oxidation of 4:2 FTI was investigated using mixtures of 4:2 FTI and Cl2 in air. Figure 7.3 shows IR spectra acquired before (A) and after (B) 10 minutes of UV irradiation (blacklamps) of a mixture of 12.2 mTorr 4:2 FTI and 100 mTorr Cl2 in 700 Torr of air. The UV irradiation led to 32% (3.9 mTorr) consumption of 4:2 FTI. Subtracting IR features attributable to 4:2 FTI from Figure 7.3B gives the product spectrum shown in 3C. Comparison with reference spectra of the fluorotelomer peracid,

C4F9CH2C(O)OOH and fluorotelomer aldehyde, C4F9CH2CHO, in panels D and E shows the formation of these compounds. The small IR feature at 1673 cm-1 in panel C does not match any features in our reference library and is attributed to an unknown product. The product -1 feature centered at 1944 cm in panel 3C reflects the formation of COF2. For the experiment shown in Figure 7.3, the yields of C4F9CH2C(O)OOH, C4F9CH2CHO, and COF2 were 1.32 mTorr, 0.44 mTorr , and 0.33 mTorr, respectively. In addition, 0.16 mTorr of

C4F9CH2C(O)OH, 0.17 mTorr of C4F9CHO, and 1.51 mTorr of CO were observed by FTIR spectroscopy. The products

142

0.08

0.06 0.04 A: before irradiation 0.02

0.00

0.06

0.04 B: after irradiation

0.02

0.00

0.06 0.04 C: B - 0.68*A 0.02

0.00 Absorbance 0.30 D: C4F9CH2C(O)OOH 0.15

0.00

0.10

0.05 E: C4F9CH2CHO

0.00

1400 1500 1600 1700 1800 1900 2000 -1 Wavenumber (cm )

Figure 7.3. FTIR spectra of a mixture of 12.2 mTorr 4:2 FTI and 100 mTorr Cl2 in 700 Torr of air before (A) and after (B) 10 minutes UV (blacklamps) irradiation. Panel C is the product spectrum. Reference spectra of the peracid C4F9CH2C(O)OOH and the aldehyde C4F9CH2CHO are given in panels D and E. observed by FTIR spectroscopy account for 59% of the 4:2 FTI loss (evaluated on basis of carbon). By analogy to CH3I (26) it is possible that the chloride C4F9CH2CH2Cl is formed and accounts for some of the 4:2FTI loss. We do not have a reference spectrum for the chloride and are not able to analyze for its formation. Offline analysis by LC-MS-MS revealed the presence of small amounts (of the order of 1% of 4:2 FTI loss) of the perfluorinated carboxylic acids PFPeA (perfluoropentanoic acid), PFBA (perfluorobutanoic acid), PFPrA (perfluoropropanoic acid), and TFA (trifluoroacetic acid) in samples taken after 28 min of irradiation.

7.3.4 UV spectra and photolysis kinetics

Mixtures for UV spectral measurements consisted of 3.0 – 6.0 Torr of ethyl iodide or 2.0 – 5.0 Torr of 4:2 FTI in 700 Torr air. The UV absorption spectra of 4:2 FTI and ethyl iodide

143 recorded in the present work are shown in Figure 7.4 and given in tabular format (each 1 nm) in Appendix C. As shown in Figure 7.4, our absorption cross sections for ethyl iodide are in good agreement with the previous measurements by Roehl et al. (27) and Rattigan et al. (28) A slight red shift is observed for the 4:2 FTI compared to the hydrocarbon analogues. The maximum for 4:2 FTI is 260 nm, while that of ethyl iodide is 257 nm. Red-shifts are common in fluorinated compounds and have been demonstrated up to 26 nm in fluorinated propanones (29), but the separation from the absorptive group and the fluorinated tail in these species presumably decreases the effect. The similarity of the spectra of 4:2 FTI and ethyl iodide is not unreasonable, given the fact that in both molecules the iodine chromophore is attached to a -

CH2CH2X group, where X is H or C4F9. The UV absorption by alkyl iodides involves an n to σ* transition with the promotion of a non bonding electron on the halogen to an antibonding σ orbital of the C-I bond (30). In a saturated molecule such as XCH2CH2I, the

1.2 ) -1 ) 1.2e-18 -1 s 1.0 -2 1.0e-18 0.8

molecule 8.0e-19 2

0.6 photons cm

6.0e-19 15

4.0e-19 0.4

2.0e-19 0.2 Cross Section (cm Cross Section

0.0 0.0 (10 Actinic Flux 250 300 350 400 Wavelength (nm)

Figure 7.4. UV-visible absorption cross sections for 4:2 FTI (gray) and ethyl iodide (dashed) compared to actinic flux at the surface (dotted). Symbols represent literature data from Roehl et al (x) (27) and Rattigan et al (+) (28). nature of X is not expected to have major impact on the n to σ* transition. The quantum yields for photolysis were assumed to be unity, in accordance with short-chain alkyl iodides (30). The effect of fluorination on the quantum yield is unknown. There is a possibility that it could decrease the quantum yield, in which case the photolysis rates presented are upper limits.

144 Photolysis rate constants and lifetimes for 4:2 FTI and ethyl iodide were calculated for the atmosphere over the city of Toronto, ON, Canada (43.72° N, 79.33° W) using the TUV program package and are shown in Table 7.1. Calculated lifetimes of a few days for ethyl iodide are in agreement with previous studies (27,28,30). As indicated in Table 7.1, we calculate a lifetime for 4:2 FTI with respect to photolysis in Toronto of a few days. Photolysis rate constants for 4:2 FTI as a function of latitude are shown in Figure C.1. As seen from Figure 7.5, the absorption by 4:2 FTI at atmospherically relevant wavelengths near 300 nm is weak. Consequently, the relative uncertainties in the absorption cross sections, and hence our estimation of the photolysis lifetimes, are significant. The UV spectra for larger FTIs are expected to be indistinguishable from that for 4:2 FTI and the lifetime estimate for 4:2 FTI photolysis applies to larger FTIs. We conclude from the data in Figure 7.4 and Table 7.1 that the atmospheric lifetime of FTIs with respect to photolysis is comparable to that of ethyl iodide and are approximately a few days in the summer and a few weeks in the winter at mid-latitudes.

Table 7.1. Twenty-four hour averaged photolysis rate constants and lifetimes for ethyl iodide and 4:2 FTIs in Toronto, ON, Canada.

Rate Constant (10-6 s-1) Lifetime (days) Ethyl Iodide 4:2 FTI Ethyl Iodide 4:2 FTI Winter Solstice 1.1 4.5 11 2.6 Spring Equinox 3.6 13 3.2 0.87 Summer Solstice 5.9 21 2.0 0.56 Autumn Equinox 3.5 13 3.3 0.89 Yearly Average 3.5 12.8 3.3 0.90

7.3.5 Photolysis Products

Figure 7.5 shows IR spectra before (A) and after (B) 137 minutes of UV irradiation

(sunlamps) of a mixture of 18.5 mTorr 4:2 FTI in 50 Torr of O2 diluent. The consumption of 4:2 FTI was 17% (3.1 mTorr). Subtracting IR features attributable to 4:2 FTI from Figure 7.5B gives the product spectrum shown in 6C. Comparison with a reference spectrum of the fluorotelomer aldehyde, C4F9CH2CHO, shows the formation of this compound in a yield of 0.89 -1 mTorr. The IR product feature at 1944 cm is COF2 which was formed in a yield of 0.21 mTorr. In addition, CO was observed at a yield of 0.62 mTorr. The observed C4F9CH2CHO accounts for only 29% of the loss of 4:2 FTI. It is expected that photolysis of 4:2 FTI will

145 proceed via elimination of the iodine atom leading to the formation of C4F9CH2CHO as the primary product:

C4F9CH2CH2I + hν → C4F9CH2CH2 + I (16)

C4F9CH2CH2 + O2 + M → C4F9CH2CH2O2 + M (17)

C4F9CH2CH2O2 + C4F9CH2CH2O2 → C4F9CH2CH2O + C4F9CH2CH2O + O2 (18a)

C4F9CH2CH2O2 + C4F9CH2CH2O2 → C4F9CH2CH2OH + C4F9CH2CHO + O2 (18b)

C4F9CH2CH2O + O2 → C4F9CH2CHO + HO2 (19)

There are no data concerning the UV spectrum or photolysis quantum yield of C4F9CH2CHO.

However, by analogy to CF3CH2CHO and C6F13CH2CHO (31,32) it is expected that: (i)

C4F9CH2CHO will undergo photolysis in the chamber leading to the formation of C4F9CHO as a secondary product, and (ii) C4F9CHO will undergo photolysis approximately an order of magnitude more rapidly than C4F9CH2CHO. The fact that the observed yield of C4F9CH2CHO accounts for only 29% of the loss of 4:2 FTI may reflect loss of C4F9CH2CHO (and C4F9CHO) via photolysis. Reaction with iodine containing species formed in the chamber may also contribute to consumption of C4F9CH2CHO.

Photolysis of C4F9CHO will give C4F9 radicals which will add O2 to give C4F9O2 radicals and undergo a series of reactions in which the molecule "unzips" by shedding COF2 units and is degraded to give a CF3O radical. CF3O radicals will combine with CF3O2, C2F5O2,

C3F7O2, and C4F9O2 radicals to give trioxides (CF3O3CxF2x+1) (33). Modest residual product IR features were observed in the region 1050-1300 cm-1 which would be consistent with the formation of such trioxides. It should be noted that trioxides are not formed in the atmosphere because of the extremely low concentration of CxF2x+1O2 radicals and the availability of other reaction partners (NO and CH4) for CF3O radicals. Finally, we note that the fate of the iodine atoms in the chamber system is unclear.

146

0.08

0.06

0.04

0.02 A: before irradiation

0.00

0.06

0.04 B: after irradiation 0.02

0.00

0.06 Absorbance 0.04 C: B - 0.79*A 0.02

0.00

0.10

0.05 D: C4F9CH2CHO 0.00

1400 1500 1600 1700 1800 1900 2000 -1 Wavenumber (cm )

Figure 7.5. FTIR spectra of a mixture of 18.5 mTorr 4:2 FTI in 50 Torr of O2 diluent before (A) and after (B) 137 min of UV (sunlamp) irradiation. Panel C shows the result of subtracting features attributable to 4:2 FTI from panel B. A reference spectrum of C4F9CH2CHO is given in panel D.

7.4 Atmospheric Implications

7.4.1 FTI lifetime

The calculated lifetime of 4:2 FTI with respect to photolysis is substantially shorter than that with respect to reaction with OH. We conclude that the atmospheric fate of 4:2 FTI, and by analogy other fluorotelomer iodides, is likely dominated by photolysis. At mid-latitudes the atmospheric lifetime of fluorotelomer iodides is approximately a few days in the summer and a few weeks in the winter.

147 7.4.2 Oxidation and photolysis products

It is of interest to compare the observed products of the photolysis and the Cl atom initiated oxidation of 4:2 FTI (compare Figures 5C and 3C). The fluorotelomer aldehyde

C4F9CH2CHO and small amounts of COF2 and CO were the only observable products of the photolysis of 4:2 FTI. In contrast, C4F9CH2C(O)OOH, C4F9CH2CHO, C4F9CH2C(O)OH,

C4F9CHO, COF2, and CO were observed following Cl atom initiated oxidation of 4:2 FTI. As discussed in section 3.5, the fluorotelomer aldehyde C4F9CH2CHO is the expected primary product of the photolysis of 4:2 FTI. The formation of C4F9CH2CHO can be explained by a simple mechanism consisting of reactions (16-19). The relatively low amount of C4F9CH2CHO observed can be attributed to loss via photolysis and possibly via reaction with iodine containing species in the chamber.

In the Cl atom initiated oxidation experiments we observed the formation of

C4F9CH2C(O)OOH, C4F9CH2CHO, C4F9CH2C(O)OH, C4F9CHO, COF2, and CO. The simplest qualitative explanation of these products is to propose that as with C4F9CH2CH2OH, reaction with Cl atoms leads to the formation of the fluorotelomer aldehyde C4F9CH2CHO as the sole primary product which is then oxidized further to give secondary products (7,34). However, a quantitative analysis shows that this simple explanation is not entirely adequate. Combining -11 (7) -12 3 -1 k(Cl + C4F9CH2CHO) = 1.84 x 10 with k(Cl + C4F9CH2CH2I) = 1.25 x 10 cm molecule -1 s measured in the present work and assuming that reaction with Cl converts C4F9CH2CH2I into

C4F9CH2CHO in 100% yield we can calculate that for the conditions relevant to the experiment shown in Figure 7.3 with a 32% consumption of C4F9CH2CH2I the yield of C4F9CH2CHO should be 0.60 mTorr. This is somewhat larger than the observed concentration of 0.44 mTorr suggesting the presence of additional C4F9CH2CHO loss processes (perhaps involving iodine containing species). The large yield of the peracid C4F9CH2C(O)OOH and the fact that this species is formed in a yield substantially (factor of approximately 8) greater than that of the acid

C4F9CH2C(O)OH is particularly striking. The large yield of the peracid can not be reconciled with the available data concerning the mechanism of chlorine atom oxidation of C4F9CH2CHO (7,34). This may reflect complexities associated with the presence of reactive iodine atoms or

IO radicals in the chlorine atom initiated oxidation of C4F9CH2CH2I in the smog chamber. The goal of the present work was to improve our understanding of the atmospheric chemistry of

C4F9CH2CH2I. Given that we show photolysis dominates the atmosphere fate of C4F9CH2CH2I,

148 experiments to understand the mechanism of Cl atom initiated oxidation were not pursued further.

7.4.3 Formation of PFCAs

In section 4.1 we concluded that photolysis is the dominant atmospheric loss pathway of FTIs and that the atmospheric lifetime of FTIs is of the order of a few days. The available data for smaller alkyl iodides such as CH3I (35,36), C2H5I (37), and C3H7I (35) and the results from the present work (section 3.5) indicate that photolysis of FTIs occurs via elimination of the iodine atom leading to the formation of the fluorotelomer aldehyde (CxF2x+1CH2CHO):

CxF2x+1CH2CH2I + hν → CxF2x+1CH2CH2 + I (20)

CxF2x+1CH2CH2 + O2 + M → CxF2x+1CH2CH2O2 + M (21)

CxF2x+1CH2CH2O2 + NO → CxF2x+1CH2CH2O + NO2 (22)

CxF2x+1CH2CH2O + O2 → CxF2x+1CH2CHO + HO2 (23)

CxF2x+1CH2CHO is removed from the atmosphere via photolysis and reaction with OH. While there are considerable uncertainties (particularly in the value of the photolysis quantum yield) the available data indicate that the lifetime of CxF2x+1CH2CHO is approximately 4 days and is determined by its reaction with OH radicals (29,31,38). As indicated in the proposed atmospheric degradation mechanism for FTIs in Figure 7.6, the oxidation of CxF2x+1CH2CHO gives the perfluoroaldehyde CxF2x+1CHO. CxF2x+1CHO has a lifetime of approximately a day with respect to photolysis and approximately 20 days with respect to reaction with OH radicals. As discussed elsewhere (6) and indicated in Figure 7.6, there are two mechanisms by which the oxidation of CxF2x+1CHO can lead to the formation of perfluorocarboxylic acids

(PFCAs). First, reaction with OH gives CxF2x+1CO radicals some fraction of which will add O2 to give CxF2x+1C(O)O2 radicals which can then react with HO2 to give PFCAs. Second, photolysis leads directly and reaction with OH leads indirectly to the formation of CxF2x+1 radicals which can add O2 to give the corresponding peroxy radicals which in low NOx environments can react with CH3O2 radicals to give perfluoroalcohols which via the intermediacy of acyl fluorides are converted into PFCAs.

149

1. hv/-I 3. +RO2/-RO/-O2 2. +O2 4. +O2/-HO2 CF3(CF2)3CH2CH2I CF3(CF2)3CH2C(O)H

+OH/-H2O

+O hv/-HCO +HO /-O 2 2 3 CF3(CF2)3CH2C(O)OO• CF3(CF2)3CH2C(O)•

+RO2/-RO/-O2 CF3(CF2)3CH2C(O)OH +HO2/-O2 OR -CO +NO/-NO2 -CO2 CF3(CF2)3CH2C(O)O• CF3(CF2)3CH2• CF3(CF2)3CH2C(O)OOH 1. +O2 2. +RO2/-RO/-O2 3. +O2/-HO2

CF3(CF2)3C(O)H

+OH/-H2O +O 2 hv/-HCO CF3(CF2)3C(O)OO• CF3(CF2)3C(O)•

+HO2/-O3 +RO2/-RO/-O2 OR -CO +NO/-NO2

-CO2 CF3(CF2)3C(O)OH CF3(CF2)3C(O)O• CF3(CF2)3•

+O2

-HF +CH O /-HCHO CF (CF ) C(O)F CF (CF ) OH 3 2 3 2 y 3 2 x CF3(CF2)xOO•

+RO2/-RO/-O2 Hydrolysis +O2 OR +NO/-NO2 -COF2 CF3(CF2)yC(O)OH CF (CF ) • CF (CF ) O• 3 2 y 3 2 x

Figure 7.6. Proposed mechanism for atmospheric oxidation of 4:2 FTI. Shaded compounds were observed in FTIR spectra, compounds in boxes were observed in offline analyses. Subscript “x” represents 0-3 and subscript “y” represents 0-2, where the values will depend on the number of times through the degradation cycle. Thick arrows indicate steps that require a low-NOx environment (see text for details).

The approximately 5-10 day time scale over which FTIs are converted into perfluoroaldehydes during the summer months is sufficient to allow transport over long distances (1700-3400 km at the global average wind speed of approximately 4 m s-1). We show here that the atmospheric degradation of FTIs has the potential to contribute to the observed burden of PFCA pollution in remote locations. To assess the magnitude of this contribution estimates for the flux of FTIs into the atmosphere are required. Further work is needed to provide such estimates.

150 7.5 Acknowledgements

We thank Rachel Chang (University of Toronto) and Malisa Chiappero (Cordoba University) for modeling assistance and Gilles Arsenault of Wellington Laboratories for the donation of PFCA standards. Funding was provided by the Natural Sciences and Engineering Research Council of Canada (NSERC). CJY appreciates the additional support of NSERC through a Canada Graduate Scholarship.

151 7.6 Sources Cited

(1) Martin, J.W.; Smithwick, M.M.; Braune, B.M.; Hoekstra, P.F.; Muir, D.C.G.; Mabury, S.A. Identification of long-chain perfluorinated acids in biota from the Canadian arctic. Environmental Science and Technology 2004, 38, 373-380.

(2) Yamashita, N.; Kannan, K.; Taniyasu, S.; Horii, Y.; Petrick, G.; Gamo, T. A global survey of perfluorinated acids in oceans. Marine Pollution Bulletin 2005, 51, 658-668.

(3) Prevedouros, K.; Cousins, I.T.; Buck, R.C.; Korzeniowski, S.H. Sources, fate and transport of perfluorocarboxylates. Environmental Science and Technology 2006, 40, 32-44.

(4) Armitage, J.; Cousins, I.T.; Buck, R.C.; Prevedouros, K.; Russell, M.H.; Macleod, M.; Korzeniowski, S.H. Modeling global-scale fate and transport of perfluorooctanoate emitted from direct sources. Environmental Science and Technology 2006, 40, 6969-6975.

(5) Wania, F. Global mass balance analysis of the source of perfluorocarboxylic acids in the Arctic Ocean. Environmental Science and Technology 2007, 41, 4529-4535.

(6) Ellis, D.A.; Martin, J.W.; De Silva, A.O.; Mabury, S.A.; Hurley, M.D.; Sulbaek Andersen, M.P.; Wallington, T.J. Degradation of fluorotelomer alcohols: A likely atmospheric source of perfluorinated carboxylic acids. Environmental Science and Technology 2004, 38, 3316-3321.

(7) Hurley, M.D.; Ball, J.C.; Wallington, T.J.; Sulbaek Andersen, M.P.; Ellis, D.A.; Martin, J.W.; Mabury, S.A. Atmospheric chemistry of 4:2 fluorotelomer alcohol: products and mechanism of Cl atom initiated oxidation. Journal of Physical Chemistry A 2004, 108, 5635- 5642.

(8) Wallington, T.J.; Hurley, M.D.; Xia, J.; Wuebbles, D.J.; Sillman, S.; Ito, A.; Penner, J.E.; Ellis, D.A.; Martin, J.W.; Mabury, S.A.; Nielsen, C.J.; Sulbaek Andersen, M.P. Formation of C7F15COOH (PFOA) and other perfluorocarboxylic acids during the atmospheric oxidation of 8:2 fluorotelomer alcohol. Environmental Science and Technology 2006, 40, 924-930.

(9) Butt, C.M.; Muir, D.C.G.; Stirling, I.; Kwan, M.; Mabury, S.A. Rapid response of Arctic ringed seals to changes in perfluoroalkyl production. Environmental Science and Technology 2007, 41, 42-49.

(10) Young, C.J.; Furdui, V.I.; Franklin, J.; Koerner, R.M.; Muir, D.C.G.; Mabury, S.A. Perfluorinated acids in Arctic snow: New evidence for atmospheric formation. Environmental Science and Technology 2007.

(11) Martin, J.W.; Ellis, D.A.; Mabury, S.A.; Hurley, M.D.; Wallington, T.J. Atmospheric chemistry of perfluoroalkanesulfonamides: Kinetic and product studies of the OH and Cl atom initiated oxidiation of N-ethyl perfluorobutanesulfonamide. Environmental Science and Technology 2006, 40, 864-872.

(12) D'eon, J.C.; Hurley, M.D.; Wallington, T.J.; Mabury, S.A. Atmospheric chemistry of N- methyl perfluorobutane sulfonamidoethanol, C4F9SO2N(CH3)CH2CH2OH: Kinetics and mechanism of reaction with OH. Environmental Science and Technology 2006, 40, 1862-1868.

152 (13) Nakayama, T.; Takahashi, K.; Matsumi, Y.; Toft, A.; Sulbaek Andersen, M.P.; Nielsen, O.J.; Waterland, R.L.; Buck, R.C.; Hurley, M.D.; Wallington, T.J. Atmospheric chemistry of CF3CH=CH2 and C4F9CH=CH2: Products of the gas-phase reactions with Cl atoms and OH radicals. Journal of Physical Chemistry A 2007, 111, 909-915.

(14) Howard, P.H.; Meylan, W. 2007. EPA Great Lakes Study for Identification of PBTs to Develop Analytical Methods: Selection of Additional PBTs - Interim Report, EPA Contract No. EP-W-04-019.

(15) DuPont global PFOA strategy - Comprehensive source reduction, Presented to the USEPA OPPT, January 31, 2005.

(16) Wallington, T.J.; Japar, S.M. Fourier transform infrared kinetic studies of the reaction of HONO with HNO3, NO3 and N2O5 at 295 K. Journal of Atmospheric Chemistry 1989, 9, 399- 409.

(17) Taniguchi, N.; Wallington, T.J.; Hurley, M.D.; Guschin, A.G.; Molina, L.T.; Molina, M.J. Atmospheric chemistry of C2F5C(O)CF(CF3)2: Photolysis and reaction with Cl atoms, OH radicals, and ozone. Journal of Physical Chemistry A 2003, 107, 2674-2679.

(18) Madronich, S.; Flocke, S. In Handbook of Environmental Chemistry; Boule, P., Ed.; Springer: Heidelberg, 1998.

(19) Majer, J.R.; Simons, J.P. Adv Photochem 1964, 1, 137.

(20) Atkinson, R.; Baulch, D.L.; Cox, R.A.; Crowley, J.N.; Hampson, R.F.; Hynes, R.G.; Jenkin, M.E.; Rossi, M.J.; Troe, J. Evaluated kinetic and photochemical data for atmospheric chemistry: Volume II – gas phase reactions of organic species. Atmospheric Chemistry and Physics 2006, 6, 3625-4055.

(21) Wallington, T.J.; Hurley, M.D.; Haryanto, A. Kinetics of the gas phase reactions of chlorine atoms with a series of formates. Chemical Physics Letters 2006, 432, 57-61.

(22) Calvert, J.G.; Atkinson, R.; Kerr, J.A.; Madronich, S.; Moortgat, G.K.; Wallington, T.J.; Yarwood, G. The Mechanisms of Atmospheric Oxidation of the Alkenes; Oxford University Press: Oxford, 2000.

(23) Kwok, E.S.C.; Atkinson, R. Estimation of hydroxyl radical reaction rate constants for gas- phse organic compounds using a structure-reactivity relationship--an update. Atmospheric Environment 1995, 29, 1685-1695.

(24) Carl, S.A.; Crowley, J.N. 298 K rate coefficients for the reaction of OH with i-C3H7I, n- C3H7I and C3H8. Atmospheric Chemistry and Physics 2001, 1, 1-7.

(25) Prinn, R.G.; Huang, J.; Weiss, R.F.; Cunnold, D.M.; Fraser, P.J.; Simmonds, P.G.; McCulloch, A.; Salameh, P.; O'Doherty, S.; Wang, R.H.J.; Porter, L.; Miller, B.R. Evidence for substantial variation of atmospheric hydroxyl radicals in the past two decades. Science 2001, 292, 1882-1888.

153

(26) Bilde, M.; Wallington, T.J. Atmospheric chemistry of CH3I: Reaction with atomic chlorine at 1 - 700 Torr total pressure and 295K. Journal of Physical Chemistry 1998, 102, 1550.

(27) Roehl, C.M.; Burkholder, J.B.; Moortgat, G.K.; Ravishankara, A.R.; Crutzen, P.J. Temperature dependence of UV absorption cross sections and atmospheric implications of several alkyl iodides. Journal of Geophysical Research 1997, 102, 12,819-812,829.

(28) Rattigan, O.V.; Shallcross, D.E.; Cox, R.A. UV absorption cross-sections and atmospheric photolysis rates of CF3I, CH3I, C2H5I and CH2ICl. Journal of the Chemical Society, Faraday Transactions 1997, 96, 2839-2846.

(29) Metcalfe, J.; Phillips, D. Photophysical processes in fluorinated acetones. Journal of the Chemical Society, Faraday Transactions 2 1976, 2, 1574-1583.

(30) Calvert, J.G.; Derwent, R.G.; Orlando, J.J.; Tyndall, G.S.; Wallington, T.J. Mechanisms of Atmospheric Oxidation of the Alkanes; Oxford University Press, 2008.

(31) Chiappero, M.S.; Malanca, F.E.; Arguello, G.A.; Wooldridge, S.T.; Hurley, M.D.; Ball, J.C.; Wallington, T.J.; Waterland, R.L.; Buck, R.C. Atmospheric chemistry of perfluoroaldehydes (CxF2x+1CHO) and fluorotelomer aldehydes (CxF2x+1CH2CHO): Quantification of the important role of photolysis. Journal of Physical Chemistry A 2006, 110, 11944-11953.

(32) Solignac, G.; Mellouki, A.; Le Bras, G.; Barnes, I.; Benter, T. Reaction of Cl atoms with C6F13CH2OH, C6F13CHO and C3F7CHO. Journal of Physical Chemistry A 2006, 110, 4450- 4457.

(33) Sehested, J.; Ellermann, T.; Nielsen, O.J.; Wallington, T.J.; Hurley, M.D. UV absorption spectrum, and kinetics and mechanism of the self reaction of CF3CF2O2 radicals in the gas phase at 295 K. International Journal of Chemical Kinetics 1993, 25, 701-717.

(34) Kelly, T.; Bossoutrot, V.; Magneron, I.; Wirtz, K.; Treacy, J.; Mellouki, A.; Sidebottom, H.; Le Bras, G. A kinetic and mechanistic study of the reactions of OH radicals and Cl atoms with 3,3,3-trifluoropropanol under atmospheric conditions. Journal of Physical Chemistry A 2005, 109, 347.

(35) Godwin, F.G.; Paterson, C.; Gory, P.A. Photofragmentation dynamics of n-C3H7I and i- C3H7I at 248 nm Molecular Physics 1987, 61, 827.

(36) Hunter, T.F.; Lunt, S.; Kristjansson, K.S. Photofragmentation of CH3I, CD3I and CF3I. 2 Formation of I( P1/2) as a function of wavelength. Journal of the Chemical Society, Faraday Transactions 2 1983, 79, 303-316.

(37) Shepson, P.B.; Heicklen, J. Photooxidation of ethyl iodide at 22oC. Journal of Physical Chemistry 1981, 85, 2691.

(38) Sellevag, S.R.; Kelly, T.; Sidebottom, H.; Nielsen, C.J. A study of the IR and UV-Vis absorption cross-sections, photolysis and OH-initiated oxidation of CF3CHO and CF3CH2CHO. Physical Chemistry Chemical Physics 2004, 6, 1243-1252.

CHAPTER EIGHT

Overtone-Induced Degradation of Perfluorinated Alcohols in the Atmosphere

Cora J. Young and D.J. Donaldson

Published in: J. Phys. Chem. A 2007 111:13466-13471

Reproduced with permission from Journal of Physical Chemistry A

Copyright ACS 2007

154 155 8.1 Introduction

Perfluorinated carboxylic acids (PFCAs) are ubiquitous in the environment, including in remote regions where local usage is not be expected to be a source (1-3). Evidence suggests that atmospheric degradation of volatile precursors, such as fluorotelomer alcohols (FTOHs) (4,5) and fluorosulfonamides (6,7), is a significant source of PFCA contamination to remote environments (8-11). Thus, a greater understanding of the atmospheric mechanisms by which PFCAs are created is of importance.

Perfluorinated alcohols (PFOHs) are formed from oxidation of perfluorinated radicals, which are products of the atmospheric oxidation of hydrofluorocarbons (12) and FTOHs (5) as well as thermolysis of fluoropolymers (13). PFOHs do not contain easily abstractable hydrogens and are not subject to hydroxyl or ozone attack. In addition, the lifetime due to photolysis of CF3OH has been estimated at one million years for altitudes less than 40 km (14). Given this lack of reactivity of PFOHs in the lower atmosphere, it is possible their persistence in this region may be limited by loss of HF via unimolecular decomposition:

F(CF2)xOH → F(CF2)x-1C(O)F + HF (1)

For CF3OH, this elimination channel is the most energetically favourable decomposition reaction:

CF3OH → COF2 +HF (2) with a calculated reaction barrier of 176 kJ mol-1 (42 kcal mol-1) (15) and a measured activation energy of (158 ± 33) kJ mol-1 (16). This relatively high energy barrier has previously been thought to preclude this reaction from occurring in the atmosphere at a significant rate.

However, CF3OH has been shown to degrade rapidly in smog chamber experiments to form

COF2 and this decomposition is enhanced in the presence of water (17):

CF3OH + H2O → COF2 + HF + H2O (3)

The presence of strong hydrogen bonds in water-containing clusters has been calculated to lower activation energies to reaction in a number of systems, including decarboxylation of dicarboxylic acids (18), dehydration of H2SO4 (19), and HF elimination from CH2FOH (20). This barrier lowering is a consequence of the formation of cyclic transition state structures, which facilitate H-atom transfer processes. It is possible that the addition of water could

156 stabilize the transition state of the decomposition of PFOHs as well, and decrease the barrier to reaction in the gas phase.

Schneider et al. (17) reported quantum chemical calculations that suggest water could act as a catalyst for HF elimination from CF3OH, decreasing the activation barrier for the -1 reaction to 67 kJ mol . Although a significant binding energy of CF3OH to water was calculated, the size of the barrier was still sufficient that the authors discounted the presence of a homogenous reaction in the gas phase and concluded that heterogeneous reactions are the probable source of degradation for CF3OH. Analogous reactions have been proposed to account for loss of HF from longer-chain PFOHs formed from the degradation of FTOHs (5), though this mechanism has not been specifically examined by any experimental or modeling studies.

Although thermal degradation or gas-phase hydrolysis of PFOHs may be energetically unfavourable in the atmosphere, O-H stretching overtone excitations of PFOHs or PFOH-water complexes may be of sufficient energy to overcome the reported barriers to reaction. Reactions induced by overtone absorption have gained attention from atmospheric chemists in recent years, particularly those promoted by excitations of the O-H stretch (21). These absorptions lie in the visible and near-infrared region of the spectrum, where actinic radiation fluxes are high. Although the transitions to overtone-excited levels are typically very weak, they are thus well matched to the most intense portions of the actinic spectrum. The excited vibrational levels accessed by overtone absorption are of sufficient energy to induce reactions with modest activation barriers in the ground electronic state. O-H overtone excitation was proposed (22) and demonstrated (23,24) to be critical in bringing the modeled and measured time dependence of stratospheric hydroxyl concentrations into accord. Overtone-driven dehydration of sulfuric acid (25) has recently been included in models to explain the upper stratospheric Junge layer of aerosols (26).

The objectives of this study were to model the energetics of the unimolecular and water- catalyzed elimination of HF from CF3OH and CF3CF2OH and to assess the importance of overtone-induced decomposition to atmospheric formation of PFCAs. The reaction path for HF elimination from PFOHs is expected to involve considerable motion of the alcoholic H-atom. This suggests that OH vibrational excitation may act to promote the reaction, if the energy thus deposited into the molecule is sufficient to overcome the activation barrier.

157 8.2 Methods

Calculations were performed using GAUSSIAN 03 programs and basis sets (27). Geometries of all species were determined at MP2 level with the 6-311+g(d,p) basis set and using density functional theory (DFT) with the 6-311++g(3df,3pd) basis set and the B3LYP functionals. No significant differences were observed in the optimized geometries calculated using the MP2 versus DFT approaches. Only the results obtained using the higher basis set DFT calculations are presented here. Geometries of bound species were confirmed to be true minima by the absence of any imaginary frequencies within the calculated vibrational spectra. Similarly, geometries of transition states were confirmed by the presence of a sole imaginary frequency in each vibrational spectrum. The motion associated with this frequency was imaged using GaussView. Basis set superposition errors (BSSE) are anticipated to be of low significance for DFT calculations with such a large basis set (28).

In order to determine the equilibrium constant for the formation of alcohol-water complexes, entropy and heat capacity values calculated by Gaussian were used to determine the Gibbs free energies of reaction at different temperatures. From these values, and assuming properties of an ideal gas, the equilibrium constants were determined. The Gaussian output was given for 298 K. Output energies and entropies were corrected for temperature by the following equations:

ΔE(T2) = ΔE(T1) + ΔCpΔT

S(T2) = S(T1) + CplnT2/T1

Unless otherwise stated, all thermodynamic properties are given for a temperature of 273 K, appropriate for the lower troposphere.

Harmonic vibrational frequencies for the O-H stretch on the alcohol were calculated at the B3LYP/6-311++g(3df,3pd) level and scaled by a factor of 0.92 to match the measured frequencies of methanol (MeOH) and trifluoroethanol (TFEtOH) (29). These values were corrected for anharmonicity using the following equation:

2 En = ωe(n + ½) – ωeχe(n + ½)

Values for the anharmonicity constant have been measured, and are similar, for water, nitric acid and water-nitric acid complexes (30). A value of 80 cm-1 was chosen for the PFOHs.

158 Similarly, the anharmonicity constant for the O-H stretch of the PFOH-water complexes was estimated at 95 cm-1 through comparison with water-nitric acid complexes and water dimers (30).

8.3 Results and Discussion

8.3.1 Alcohol-water complexes

The calculated minimum energy structure of the CF3OH•H2O complex is in agreement with that determined by Schneider et al. (17) (Bond lengths and angles are reported in Table

D.1). The structure of the CF3CF2OH•H2O complex is shown in Figure 1 and described in Table -1 D.2. The binding energies determined in the current study are 30 kJ mol for CF3OH•H2O and

O2 O1 H1 H3 H F3 2

F4 C2 C1 F2 F1

F5

Figure 8.1 Optimized structure of CF3CF2OH•H2O. See Table D.2 for bond lengths and angles.

Table 8.1 Calculated relative energies of species (kJ mol-1) at 273 K.

Relative Energies

CF3OH (+ H2O) 0 ‡ [CF3OH] 168.758

CF3OH•H2O -30.406 ‡ [CF3OH•H2O] 59.489

COF2 + HF (+ H2O) 13.999

CF3CF2OH (+ H2O) 0 ‡ [CF3CF2OH] 220.566

CF3CF2OH•H2O -31.041 ‡ [CF3CF2OH•H2O] 58.062 CF COF + HF (+ H O) 3 2 15.363

159

-1 31 kJ mol for CF3CF2OH•H2O (see Table 1). These are quite strong and comparable to that calculated for the nitric acid-water complex, of 31 kJ mol-1 (28) and are in good agreement with that calculated for CF3OH•H2O by Schneider et al. (17) Using calculated heat capacities, energies and entropies, equilibrium constants were determined for the formation of both complex species at temperatures from 263 to 308 K and are displayed in Figure 2. Low, but potentially significant levels of PFOH-water complexes are likely to be present in the atmosphere. The equilibrium constant is greater for the longer-chained alcohol, which is the result of a somewhat larger binding energy and a somewhat lower entropy cost to binding. In a tropical setting with a temperature of 303 K and a relative humidity of 80%, 4.4% and 8.5% of the CF3OH and CF3CF2OH, respectively, are predicted to be in the water-complexed form. In the spring or summer in the Arctic with a temperature of 273 K and a relative humidity of 50%, we estimate that 1.5% and 3.2% of the CF3OH and CF3CF2OH, respectively, would be complexed.

18

16 CF3OH + H2OCF3OH•H2O CF CF OH + H O CF CF OH•H O 14 3 2 2 3 2 2

12

10 eq

K 8

6

4

2

0 260 270 280 290 300 310 320

Temperature (K)

Figure 8.2 Calculated equilibrium constants for the reaction of CF3OH and CF3CF2OH with water to form complexes.

160

8.3.2 Overall features of the reactions

The four reactions considered in this study were as follows:

CF3OH → COF2 + HF (4)

CF3OH•H2O → COF2 + HF + H2O (5)

CF3CF2OH → CF3COF + HF (6)

CF3CF2OH•H2O → CF3COF + HF + H2O (7)

The calculated energies for each species as well as the zero point corrections are shown in Table 2. As expected, the imaginary frequency at the transition state corresponds to alcoholic hydrogen motion between the oxygen atom and an adjacent fluorine (Figure D.1). Optimized geometries of the transition states for each of the four reactions are depicted in Figure 3 and ‡ described in Tables D.1 and D.2. The transition state geometries for [CF3OH] and

Energy Zero-point correction (hartree) (hartree)

CF3OH -413.63567 0.02855 ‡ [CF3OH] -413.56597 0.02308

COF2 -313.13803 0.01396 HF -100.48698 0.00932

CF3OH•H2O -490.11457 0.05274 ‡ [CF3OH•H2O] -490.07739 0.04964

H2O -76.46451 0.02131

CF3CF2OH -651.51032 0.04174 ‡ [CF3CF2OH] -651.42065 0.03503

CF3COF -551.01203 0.02602

CF3CF2OH•H2O -727.98934 0.06481 ‡ [CF3CF2OH•H2O] -727.95267 0.06191

Table 8.2 Calculated energies and zero-point corrections for all species investigated.

161

(a) (b) O F3 F3 H F C 1 C H2 H3 F 1 F O2 F2 2 O1 H1

(c) (d) F 4 F F5 5 O C F4 2 F3 C H F1 2 H C1 2 C1 F O2 F3 1 F2 H1 O1 H3 F2

Figure 8.3 Optimized structures of the calculated transition states for reactions (4) through (7): ‡ ‡ ‡ ‡ (a) [CF3OH] ; (b) [CF3OH•H2O] ; (c) [CF3CF2OH] ; and (d) [CF3CF2OH•H2O] . See Tables D.1 and D.2 for bond lengths and angles.

250

vOH=6 200 v =5 ‡ OH [CF3OH] ) -1 150 vOH=4

vOH=3 100 vOH=4

vOH=2 [CF OH•H O]‡ vOH=3 3 2 50

Energy (kJ mol v =2 vOH=1 OH v =1 0 OH COF2 + HF vOH=0 CF3OH (+ H2O)

vOH=0 CF OH•H O -50 3 2

Figure 8.4 Energetics of reactions (4) and (5) with all energies relative to CF3OH. Also shown are calculated overtone vibrations for CF3OH (black) and CF3OH•H2O (blue).

162

‡ [CF3OH•H2O] are consistent with those determined by Schneider et al (Table D.1) (17). The PFOH-water complex transition states are composed of a cyclic double-hydrogen bond, in which water acts as both donor and acceptor. Energetics of the reactions involving water complexes are expected to be favoured over the uncomplexed reactions due to the increased stability of the six-membered ring formed in the double-hydrogen-bonded structure relative to the four-membered ring transition state present in the unimolecular reaction. The calculated energetics for reactions (4) and (5) are shown in Figure 4, while those of reactions (6) and (7) are shown in Figure 5, with all values presented in Table 1.

The unimolecular decomposition of CF3OH (reaction (4)) was determined to require 169 kJ mol-1 to achieve the transition state. This is in reasonable agreement with calculations by Schneider et al. (17), who determined an energy of 175 kJ mol-1 and with calculations by Fransisco (15), who reported a barrier energy of 189 kJ mol-1. The values also agree well with experimental measurements of the activation energy for decomposition of 158 ± 33 kJ mol-1 made by Huey et al. (16) The water-catalyzed reaction of CF3OH (reaction (5)) required less energy to reach the transition state, with a calculated barrier height of 90 kJ mol-1. Schneider et

250 ‡ [CF3CF2OH] v =6 200 OH

vOH=5 ) -1 150 vOH=4

v =3 100 OH vOH=4 ‡ vOH=2 [CF3CF2OH•H2O] vOH=3 50

Energy (kJ mol Energy vOH=1 vOH=2

0 vOH=1 CF3COF + HF vOH=0 CF3CF2OH (+ H2O)

vOH=0 CF CF OH•H O -50 3 2 2

Figure 8.5 Energetics of reactions (6) and (7) with all energies relative to CF3CF2OH. Also shown are calculated overtone vibrations for CF3CF2OH (black) and CF3CF2OH•H2O (blue).

163 al. (17) also calculated a lowering of the energy barrier for the water-catalyzed reaction, to 67 kJ mol-1, which is slightly lower than the value determined herein. Similarly, unimolecular decomposition of CF3CF2OH (reaction (6)) and the corresponding water-catalyzed reaction (7) are calculated to have transition state energies of 221 and 89 kJ mol-1, respectively. The general lowering of the reaction barrier of the complexed species is consistent with observations that PFOHs degrade faster in smog chamber experiments in the presence of water (17) and confirm the important role of water in the decomposition of PFOHs.

8.3.3 Overtone-induced degradation

There have been no published measurements of the overtone absorption spectra or cross sections for either CF3OH or CF3CF2OH. Measurements have been made for methanol (MeOH), ethanol (EtOH) and trifluoroethanol (TFEtOH) up to the third vibrational overtone of the O-H stretch (29). The observed vibrational frequencies are listed in Table 4. The band centres of the O-H stretching levels of all three alcohols are similar. In fact, those of EtOH and TFEtOH are within the estimated experimental error. The main differences among the alcohols lie in how the intensities vary with vibrational level. Lange et al. (29) demonstrated that electronegative substituents increased the absorption intensity of the OH stretching fundamental. However, alcohols with electronegative substituents also showed larger decreases in absorption strength with subsequent levels of excitation, with the result that there is little variability in intensity at the fourth OH stretch vibrational level. On average, band intensities decrease by approximately an order of magnitude for each excitation level. TFEtOH intensity decreases by a factor of 15 from 1vOH to 2vOH and a factor of 21 from 2vOH to 3vOH. In contrast,

EtOH intensity decreases by a factor of 7.5 from 1vOH to 2vOH and a factor of 16 from 2vOH to

3vOH. Since TFEtOH has a higher fundamental intensity, the intensity of TFEtOH at 2vOH is still higher than that of EtOH and at 3vOH, the intensities are approximately equal.

To estimate the potential for overtone-driven chemistry, we used the overtone absorption frequencies calculated by using the Gaussian harmonic frequencies and estimated anharmonicity parameters, as described above. Values for the resulting O-H vibrational frequencies are shown in Table 4. In order to determine the level of absorption required to induce reaction, O-H stretch vibrational energies were compared to calculated reaction energies in Figures 4 and 5.

164 Table 8.3 Comparison of calculated values (kJ mol-1 at 298 K) for reactions (4) and (5) with literature values.

Theoretical Experimental This work Francisco14 Schneider et al16 Huey et al15 Asher et al37 ‡ [CF3OH] 168.673 188.744 175.979 158 ± 33

CF3OH•H2O -30.209 -41.348 ‡ [CF3OH•H2O] 59.264 71.856

COF2 + HF (+ H2O) 14.146 32.643 15.568 12 ± 6

Reactions (4) and (6) require >169 and >221 kJ mol-1, respectively, to overcome the barrier to

HF loss. For CF3OH and CF3CF2OH, this entails excitation into vOH=5 and 6, respectively.

In contrast, reactions (5) and (7) require >90 and >89 kJ mol-1, respectively, and could therefore be initiated through excitation into the vOH=3 energy level. Alternatively, excitation into the vOH=2 with the addition of thermal energy (24) could also lead to reaction. The first- order photolysis rate constant, J, for the overtone-induced dissociation of PFOH-water complexes can be calculated as follows:

J = ∫φ(λ) σ(λ) I(λ) dλ where I(λ) represents the wavelength dependent flux of available light, σ(λ) gives the quantitiative absorption spectrum of the overtone transition, the quantum yield is given by φ(λ) and the integration is performed over each overtone absorption band. Actinic fluxes for solar radiation were obtained from published sources (31,32) or estimated using the equation for emission from a blackbody. Absorption cross-sections were estimated by taking the ratio of the measured cross section of TFA to that of acetic acid at vOH=3 (33), then scaling measured cross sections of MeOH and EtOH (29,34) to approximate the cross-sections of the CF3OH,

CF3OH•H2O and CF3CF2OH•H2O species. There is insufficient information available about these high vibrational levels to allow accurate estimation of the rate. However, overtone- induced photolysis of trifluoroacetic acid (TFA) at vOH=6 has been calculated to occur with a rate constant between 2.5 × 10-10 and 3.7 × 10-9 s-1, which corresponds to a lifetime between 8 and 127 years (33).

165 The quantum yield for the overtone-induced reactions is not known. For direct bond cleavage-type reactions, such as the dissociation of nitric acid or pernitric acid into NO2 + OH or NO2 + HO2, respectively, the quantum yield is unity for excitation to vibrational levels above the barrier (21). For levels close to the barrier, thermal excitation can aid the dissociation, leading to a temperature-dependent quantum yield (24). The quantum yields for reactions involving molecular rearrangement have not been measured experimentally. Miller and Gerber

Table 8.4 Band centres of PFOHs and PFOH-water complexes.

Band centres (cm-1) Band centres (nm)

1ν OH 2v OH 3v OH 4v OH 5v OH 6v OH 1ν OH 2v OH 3v OH 4v OH 5v OH 6v OH

CH3OH 3681(8) 7199(15) 10541(16) 13706(28) 2717 1389 949 730 a Measured CH3CH2OH 3665(8) 7168(15) 10489(24) 13643(56) 2729 1395 953 732

CF3CH2OH 3657(8) 7148(15) 10466(16) 13620(28) 2734 1399 955 934

CF3OH 3669 6858 10047 13075 15944 18653 2726 1458 995 765 627 536

CF3CF2OH 3646 6811 9977 12983 15828 18514 2743 1468 1002 770 632 540 Calculated CF3OH•H2O 3325 6080 8835 11401 3007 1645 1132 877 CF3CF2OH•H2O 3289 6007 8726 11254 3041 1665 1146 889 aAdapted from Table 1 in Lange et al.(29) Experimental uncertainties are expressed in parentheses and reflect uncertainty in the smallest significant digit(s) of the number presented.

(19) reported results of classical trajectory calculations on a semiempirical potential energy surface for the H2SO4 → H2O + SO3 reaction. Five percent of the trajectories initiated at vOH=6 gave rise to direct dissociation in these calculations, though the water complex was found not to give reaction. Later calculations (35), which included dissociation from vOH=4,5 as well, concluded that under conditions of low collisional deactivation, the quantum yield approaches unity. Another recent calculation (20) finds that reaction can compete with dissociation of the complex for CH2FOH•H2O → HF + CH2O + H2O. Here, a quantum yield for HF elimination was estimated to be 0.015 ± 0.01 using direct dynamics simulations. In the absence of other information, it seems reasonable to assume that the quantum yields for overtone-initiated HF elimination from CF3OH•H2O and CF3CF2OH•H2O are in the range of 0.01 to 0.10. Utilizing a quantum yield of 0.10 and other parameters discussed above, the photolysis rate constants for the complexed species in reactions (5) and (7) are determined to be: 6.1 × 10-8 s-1 and 5.6 × 10-8 s-1, corresponding to lifetimes of 191 and 207 days, respectively. Similarly, the photolysis rate constant for reaction (4) is 1.69 × 10-9 s-1, which corresponds to a lifetime of approximately 19 years. The rate of reaction (6) cannot be estimated due to a lack of measured alcohol cross sections for vOH=6. Overtone-induced photolysis lifetimes for reactions (4) and (6) are expected to be longer than those of reactions (5) and (7) as a result of the decrease in absorption cross

166 section with increasing vibrational levels. Even with an assumed quantum yield for reaction of 0.01 for reaction of the water-complexed species (and 0.1 for the uncomplexed compounds), the former are calculated to be more reactive than the latter. The approximate nature of these lifetimes must be stressed, due to the high degree of uncertainty regarding the quantum yields.

If we assume that the energy barrier calculated for reaction (5) by Schneider et al. (17) is -1 correct (67 kJ mol ), excitation into vOH=2 would be sufficient to induce reaction. Again, the photolysis rate can be estimated for this reaction. The actinic flux was estimated assuming Planck intensity, and a quantum yield of 0.1 was used. Estimated absorption cross sections were estimated using measured MeOH cross-sections at vOH=2, which were scaled with the ratio of TFA to acetic acid cross sections at vOH=3, given that cross sections have not been measured at lower vibrational levels for TFA. A photolysis rate of 3.65 × 10-7 s-1 was calculated, corresponding to a lifetime of approximately 32 days.

8.4 Atmospheric Implications

The calculations performed in this study demonstrate that overtone-induced dissociation could be an important fate for PFOHs when they are complexed with water molecules.

Atmospheric lifetimes for CF3OH•H2O and CF3CF2OH•H2O with respect to this process may be as short as a few months. We stress here that these estimates depend upon a significant quantum yield for the elimination, on the order of 1-10%. Since there is no experimental measurement of this quantity, this remains a significant uncertainty in the result presented here. However, quantum dynamics calculations do indicate an HF elimination yield of ~1% for a water-complexed species, lending some credibility to our assumptions.

These elimination rate constants are only of significance if the complex is present in appreciable amounts in the atmosphere. As discussed above, depending on atmospheric conditions PFOH-water complexes should be present as a few percent of the total PFOH concentration. Making the assumption that overtone absorption cross-sections decrease by approximately an order of magnitude for each successive level implies that PFOH-water complexes, which can react through absorption at vOH=3, would react two or three orders of magnitude faster than PFOHs alone, which require absorption into vOH=5 or 6. The calculated fraction of PFOH-water complex (see above) of a few percent would make the reaction of this species approximately one order of magnitude faster in the atmosphere relative to that of PFOH, assuming similar quantum yields for the complexed and uncomplexed species. Takahashi et al.

167

(20) determined that the formation of water complexes with CH2FOH gives rise to higher reactivity when the complex contained an increasing number of water molecules. It is reasonable to assume that the same trend would hold for PFOHs, making multi-water-PFOH complexes even more reactive than the single water complexes studied here. Further studies would be required to assess the full importance of these larger complexes. Regardless of the complex size, degradation of PFOH-water complexes should be the dominant homogeneous fate for PFOHs in the lower atmosphere.

A previous study examining the fate of CF3OH and CF3OH•H2O reported that the homogeneous gas-phase reaction of CF3OH and water would likely occur with a lifetime on the order of two years (17). This study did not take into account the possibility of additional energy input into the system through overtone absorption. A reaction with the calculated energy barrier of 67 kJ mol-1 (calculated as 89 kJ mol-1 in the current study) would not react with an appreciable rate in the atmosphere. As a result, those authors concluded that heterogeneous processes are likely to be of much greater significance to the fate of this compound.

Heterogeneous reaction rates have been studied for CF3OH by Lovejoy et al. (36). The authors observed that heterogeneous loss on cloud particles was an effective loss process for CF3OH in the troposphere, with a lifetime of approximately 20 s after contact with an average-sized cloud droplet. As a result, the total lifetime of CF3OH in the lower atmosphere depends on the time before contact with a cloud, which is estimated to be two days. In the stratosphere, heterogeneous loss will also be important, but will depend on transport to regions where aerosols are present (36). If this is the case, the lifetime with respect to overtone-induced degradation of PFOH-water complexes into vOH=3 is approximately two orders of magnitude slower than heterogeneous degradation of CF3OH, but this difference will be strongly dependent on meteorological conditions. For instance, in regions that receive a moderate to large amount of precipitation, heterogeneous reaction will be the primary route of degradation of PFOHs. However, in very dry regions, such as the Arctic, there is the potential for overtone-induced degradation to be a modestly competitive loss mechanism. It is also possible that PFOHs could adsorb onto ice in the form of cirrus clouds or on the surface of the Earth and then be subject to heterogeneous photochemistry, though more studies would be necessary to verify this as a degradation mechanism.

One branch of the atmospheric oxidation of FTOHs into PFCAs proceeds through PFOH intermediates. PFOHs are only formed under low-NOx conditions, due to reaction of

168 perfluorinated radicals with alkyl peroxy radicals. This occurs primarily in remote regions, such as the Arctic. The PFOHs are transformed into acid fluorides and finally into PFCAs. The mechanism has not been specifically studied, but was presumed to occur via heterogeneous reaction with water (5,11). Results from the current work suggest that overtone-induced degradation of PFOH-water complexes is potentially a modestly competing mechanism for degradation of PFOHs in the transformation into PFCAs.

The fate of CF3OH is of fundamental interest, but of greater environmental interest is the fate of longer-chained PFOHs. These compounds form acid fluorides through reactions described earlier and ultimately degrade into PFCAs. It is possible that results of this study can be extrapolated to PFOHs with longer perfluorinated chains. Increasing the chain length in water complexes from CF3OH to CF3CF2OH resulted in a slightly lower transition state energy being observed. In addition, the binding energy to form the water complex was somewhat larger for CF3CF2OH, leading to a larger proportion in the complexed form. If these trends were to continue with increasing chain length, it would be expected that longer-chain PFOHs should be more likely to form water complexes. These would then be more likely to degrade into acid fluorides and, thus, PFCAs in the atmosphere. As a result, the overtone-induced degradation of longer-chained PFOHs may play a more important role in their atmospheric fate.

8.5 Acknowledgements

The authors thank Mima Staikova for her invaluable assistance and Scott Mabury for helpful comments. CJY is grateful to the Natural Science and Engineering Research Council of Canada for a CGS Fellowship.

169 8.6 Sources Cited

(1) Martin, J.W.; Smithwick, M.M.; Braune, B.M.; Hoekstra, P.F.; Muir, D.C.G.; Mabury, S.A. Identification of long-chain perfluorinated acids in biota from the Canadian arctic. Environmental Science and Technology 2004, 38, 373-380.

(2) Stock, N.L.; Lau, F.K.; Ellis, D.A.; Martin, J.W.; Muir, D.C.G.; Mabury, S.A. Polyfluorinated telomer alcohols and sulfonamides in the North American troposphere. Environmental Science and Technology 2004, 38, 991-996.

(3) Houde, M.; Martin, J.W.; Letcher, R.J.; Solomon, K.R.; Muir, D.C.G. Biological monitoring of polyfluoroalkyl substances: A review. Environmental Science and Technology 2006, 40, 3463-3473.

(4) Hurley, M.D.; Ball, J.C.; Wallington, T.J.; Sulbaek Andersen, M.P.; Ellis, D.A.; Martin, J.W.; Mabury, S.A. Atmospheric chemistry of fluorinated alcohols: Reaction with Cl atoms and OH radicals and atmospheric lifetimes. Journal of Physical Chemistry A 2004, 108, 1973-1979.

(5) Ellis, D.A.; Martin, J.W.; De Silva, A.O.; Mabury, S.A.; Hurley, M.D.; Sulbaek Andersen, M.P.; Wallington, T.J. Degradation of fluorotelomer alcohols: A likely atmospheric source of perfluorinated carboxylic acids. Environmental Science and Technology 2004, 38, 3316-3321.

(6) Martin, J.W.; Ellis, D.A.; Mabury, S.A.; Hurley, M.D.; Wallington, T.J. Atmospheric chemistry of perfluoroalkanesulfonamides: Kinetic and product studies of the OH and Cl atom initiated oxidiation of N-ethyl perfluorobutanesulfonamide. Environmental Science and Technology 2006, 40, 864-872.

(7) D'eon, J.C.; Hurley, M.D.; Wallington, T.J.; Mabury, S.A. Atmospheric chemistry of N- methyl perfluorobutane sulfonamidoethanol, C4F9SO2N(CH3)CH2CH2OH: Kinetics and mechanism of reaction with OH. Environmental Science and Technology 2006, 40, 1862-1868.

(8) Scott, B.F.; Spencer, C.; Mabury, S.A.; Muir, D.C.G. Poly and perfluorinated carboxylates in North American precipitation. Environmental Science and Technology 2006, 40, 7167-7174.

(9) Butt, C.M.; Muir, D.C.G.; Stirling, I.; Kwan, M.; Mabury, S.A. Rapid response of Arctic ringed seals to changes in perfluoroalkyl production. Environmental Science and Technology 2007, 41, 42-49.

(10) Young, C.J.; Furdui, V.I.; Franklin, J.; Koerner, R.M.; Muir, D.C.G.; Mabury, S.A. Perfluorinated acids in Arctic snow: New evidence for atmospheric formation. Environmental Science and Technology 2007, 41, 3455-3461.

(11) Wallington, T.J.; Hurley, M.D.; Xia, J.; Wuebbles, D.J.; Sillman, S.; Ito, A.; Penner, J.E.; Ellis, D.A.; Martin, J.W.; Mabury, S.A.; Nielsen, C.J.; Sulbaek Andersen, M.P. Formation of C7F15COOH (PFOA) and other perfluorocarboxylic acids during the atmospheric oxidation of 8:2 fluorotelomer alcohol. Environmental Science and Technology 2006, 40, 924-930.

170 (12) Wallington, T.J.; Hurley, M.D.; Ball, J.C.; Kaiser, E.W. Atmospheric chemistry of hydrofluorocarbon 134a: Fate of the alkoxy radical 1,2,2,2-tetrafluoroethoxy. Environmental Science and Technology 1992, 26, 1318-1324.

(13) Ellis, D.A.; Mabury, S.A.; Martin, J.W.; Muir, D.C.G. Thermolysis of fluoropolymers as a potential source of halogenated organic acids in the environment. Nature 2001, 412, 321-324.

(14) Schneider, W.F.; Wallington, T.J.; Minschwaner, K.; Stahlberg, E.A. Atmospheric chemistry of CF3OH: Is Photolysis important? Environmental Science and Technology 1995, 29, 247-250.

(15) Francisco, J.S. Ab initio investigation of the decomposition of trifluoromethanol into carbonyl fluoride and hydrogen fluoride. Chemical Physics Letters 1994, 218, 401-405.

(16) Huey, L.G.; Hanson, D.R.; Lovejoy, E.R. Atmospheric fate of CF3OH 1: Gas phase thermal decomposition. Journal of Geophysical Research 1995, 100, 18,771-718,774.

(17) Schneider, W.F.; Wallington, T.J.; Huie, R.E. Energetics and mechanism of decomposition of CF3OH. Journal of Physical Chemistry 1996, 100, 6097-6103.

(18) Staikova, M.; Oh, M.; Donaldson, D.J. Overtone-induced decarboxylation: A potential sink for atmospheric diacids. Journal of Physical Chemistry A 2005, 109, 597-602.

(19) Miller, Y.; Gerber, R.B. Dynamics of vibrational overtone excitations of H2SO4, H2SO4- H2O: Hydrogen-hopping and photodissociation processes. Journal of the American Chemical Society 2006, 128, 9594-9595.

(20) Takahashi, K.; Kramer, Z.C.; Vaida, V.; Skodje, R.T. Vibrational overtone induced elimination reactions within hydrogen-bonded molecular clusters: the dynamics of water catalyzed reactions in CH2FOH-(H2O)n. Physical Chemistry Chemical Physics 2007, 9, 3864- 3871.

(21) Donaldson, D.J.; Tuck, A.F.; Vaida, V. Atmospheric photochemistry via vibrational overtone absorption. Chemical Reviews 2003, 103, 4717-4729.

(22) Donaldson, D.J.; Tuck, A.F.; Vaida, V. Enhancement of HOx at high solar zenith angles by overtone-induced dissociation of HNO3 and HNO4. Physics and Chemistry of the Earth 2000, 25, 223-227.

(23) Wennberg, P.O.; Salwitch, R.J.; Donaldson, D.J.; Hanisco, T.F.; Lanzendorf, E.J.; Perkins, K.K.; Lloyd, S.A.; Vaida, V.; Gao, R.S.; Hintsa, E.J.; Cohen, R.C.; Swartz, W.H.; Kusterer, T.L.; Anderson, D.E. Twilight observations suggest unknown sources of HOx. Geophysical Research Letters 1999, 26, 1373-1376.

(24) Roehl, C.M.; Nizkorodov, S.A.; Zhang, H.; Blake, G.A.; Wennberg, P.O. Photodissociation of peroxynitric acid in the near-IR. Journal of Physical Chemistry A 2002, 106, 3766-3772.

171 (25) Vaida, V.; Kjaergaard, H.G.; Hintze, P.E.; Donaldson, D.J. Photolysis of sulfuric acid vapor by visible solar radiation. Science 2003, 299, 1566-1568.

(26) Mills, M.J.; Toon, O.B.; Vaida, V.; Hintze, P.E.; Kjaergaard, H.G.; Schofield, D.P.; Robinson, T.W. Photolysis of sulfuric acid vapor by visible light as a source of the polar stratospheric CN layer. Journal of Geophysical Research 2005, 110, D08201.

(27) Gaussian 03, Revision B.03. Frisch, M.J.; Trucks, G.W.; Schlegel, H.B.; Scuseria, G.E.; Robb, M.A.; Cheeseman, J.R.; Montgomery, J., J. A.; Vreven, T.; Kudin, K.N.; Burant, J.C.; Millam, J.M.; Iyengar, S.S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G.A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J.E.; Hratchian, H.P.; Cross, J.B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R.E.; Yazyev, O.; Austin, A.J.; Cammi, R.; Pomelli, C.; Ochterski, J.W.; Ayala, P.Y.; Morokuma, K.; Voth, G.A.; Salvador, P.; Dannenberg, J.J.; Zakrzewski, V.G.; Dapprich, S.; Daniels, A.D.; Strain, M.C.; Farkas, O.; Malick, D.K.; Rabuck, A.D.; Raghavachari, K.; Foresman, J.B.; Ortiz, J.V.; Cui, Q.; Baboul, A.G.; Clifford, S.; Cioslowski, J.; Stefanov, B.B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R.L.; Fox, D.J.; Keith, T.; Al-Laham, M.A.; Peng, C.Y.; Nanayakkara, A.; Challacombe, M.; Gill, P.M.W.; Johnson, B.; Chen, W.; Wong, M.W.; Gonzalez, C.; Pople, J.A. Gaussian, Inc., Wallingford, CT, 2004.

(28) Staikova, M.; Donaldson, D.J. Ab initio investigation of water complexes of some atmospherically important acids: HONO, HNO3 and HO2NO2. Physical Chemistry Chemical Physics 2001, 3, 1999-2006.

(29) Lange, K.R.; Wells, N.P.; Plegge, K.S.; Phillips, J.A. Integrated intensities of O-H stretching bands: Fundamentals and overtones in vapor-phase alcohols and acids. Journal of Physical Chemistry A 2001, 105, 3481-3486.

(30) Kjaergaard, H.G. Calculated OH-stretching vibrational transitions of the water-nitric acid complex. Journal of Physical Chemistry A 2002, 106, 2979-2987.

(31) Neckel, H.; Labs, D. The solar radiation between 3300 and 12500 A. Solar Physics 1984, 90, 205-245.

(32) Finlayson-Pitts, B.J.; Pitts, J.N. Chemistry of the Upper and Lower Atmosphere; Academic Press: San Diego, CA, 2000.

(33) Reynard, L.M.; Donaldson, D.J. Overtone-induced chemistry of trifluoroacetic acid: An experimental and theoretical study. Journal of Physical Chemistry A 2002, 106, 8651-8657.

(34) Phillips, J.A.; Orlando, J.J.; Tyndall, G.S.; Vaida, V. Integrated intensities of OH vibrational overtones in alcohols. Chemical Physics Letters 1998, 296, 377-383.

(35) Miller, Y.; Gerber, R.B.; Vaida, V. Photodissociation yields for vibrationally excited states of sulfuric acid under atmospheric conditions. Geophysical Research Letters 2007, 34, L16820.

172

(36) Lovejoy, E.R.; Huey, L.G.; Hanson, D.R. Atmospheric fate of CF3OH 2: Heterogeneous reaction. Journal of Geophysical Research 1995, 100, 18,775-718,780.

(37) Asher, R. L.; Appelman, E. H.; Tilson, J. L.; Litorja, M.; Berkowitz, J.; Ruscic, B. Journal of Chemical Physics 1997, 106, 9111.

CHAPTER NINE

Perfluorinated acids in Arctic snow: New evidence for atmospheric formation

Cora J. Young, Vasile I. Furdui, James Franklin, Roy M. Koerner, Derek C.G. Muir and Scott A. Mabury

Published in: Environ. Sci. Technol. 2007 41:3455-3461 Featured on the cover of May 15, 2007 edition.

Reproduced with permission from Environmental Science and Technology

Copyright ACS 2007

173 174 9.1 Introduction Widespread contamination of waters and biota with perfluorinated acids (PFAs) has been observed in remote regions, including the High Arctic (1-3). In these areas where local usage would not be expected to contribute significantly to contamination, questions are raised regarding the source of these pollutants.

The low volatility and high water solubility of PFAs in their anionic form implies that they are not susceptible to long range atmospheric transport. Thus, another transport mechanism must be at work. It has been postulated that these compounds could be transported via ocean currents and that with concentrations measured in open ocean water, it is possible to calculate a yearly flux of between 2 and 12 tonnes of perfluorooctanoic acid (PFOA) to the Arctic (4). However, this flux may not contribute significantly to the observed biota contamination, as evidence suggests that transport from the Atlantic into the Arctic Ocean will be at a depth greater than 200m, while atmospherically deposited PFAs may have a greater influence on the biologically productive near surface waters (5). In addition, transport to the Arctic via the ocean is estimated to take on the order of decades (6), leading to a lag between production and observed changes.

The atmospheric oxidation of volatile precursors is another potential source of PFAs to the Arctic. Fluorotelomer alcohols (FTOHs) are volatile and have sufficient persistence in the atmosphere to reach the Arctic (7). They have been observed to undergo atmospheric oxidation to form perfluorocarboxylic acids (PFCAs) under low NOx conditions, such as those found in remote Arctic regions (8,9). Analogously, the source of perfluorooctane sulfonate (PFOS) to the Arctic could be atmospheric oxidation of volatile perfluorooctane sulfamido alcohols (10,11). FTOHs and perfluorooctane sulfamido alcohols are present as residuals in some fluoropolymer products, and these residuals can be released into air (12). Both FTOHs and perfluorooctane sulfamido alcohols have been shown to be present ubiquitously in the atmosphere (13-15), including in the Arctic (3,16). Their presence has also been demonstrated in indoor air, which may be a source to the atmosphere (15).

Industry has responded to the presence of perfluorinated compounds in the environment. In 2001-2002, PFOS and related chemistries, including perfluorosulfamido alcohols containing eight carbons, were voluntarily removed from the market (17). As such, emissions of these potential precursors should be significantly decreased, although not yet zero, due to continued

175 use of products. Additionally, some producers of FTOH-containing polymer products have announced their intention to decrease the residuals present in their products (18).

Temporal analyses of PFAs in ringed seals from the Canadian Arctic show recent trends that are in line with a fast response to changing production patterns of PFAs. PFOS concentrations were observed to increase steadily up until 1998 or 2000, after which concentrations began to decrease (19). Time to transport contaminants via the ocean is on the order of decades, while via the atmosphere is days to weeks. Thus, a short response time points to an atmospheric source.

Using estimated emissions of FTOHs based on air concentrations and a three dimensional model, 0.4 tonnes year-1 of perfluorooctanoic acid was calculated to be deposited at latitudes north of 65˚N via atmospheric oxidation (20). Measuring past and present atmospheric fluxes of PFAs into the Arctic would shed light on whether atmospheric degradation of volatile precursors provides the primary source of PFAs to the Arctic. Measured fluxes would also provide a test of the recent model results.

Ice caps, with their high altitude, should receive contamination solely from the atmosphere. Layers of accumulated snow on ice caps are subject to little change and the temporal record can be reasonably well established (21). Few past studies have looked at persistent organic pollutants in ice caps. The limitations of traditional sampling technology have made it difficult to obtain large enough samples in which to detect the low levels of organic pollutants found in these remote regions (22). Ice caps have the potential to provide long-term temporal trends of atmospheric concentrations, which are important for assessing regulatory effectiveness. Utilizing large volume snow samples and state-of-the-art analytical equipment, it is possible to take advantage of the information stored in ice caps. The objective of this study was to use High Arctic ice caps to determine seasonal cycles, temporal trends and atmospheric fluxes in order to illuminate the source of selected PFAs to the Arctic.

9.2 Experimental

9.2.1 Chemicals

See Appendix E for a full list of chemicals used. Deep ice core water for use as a blank was obtained from the Geological Survey of Canada (Ottawa, ON, Canada) from a core taken on Agassiz Ice Cap in 1977. The ice used was taken from 150 to 200 m depth and is greater

176 than 2000 years old. Deep ice samples were wrapped in polyethylene and stored at the Geological Survey of Canada at -20°C. The outside of the core was removed with stainless steel tools before melting.

9.2.2 Sample collection

Surface samples, representing the end of the melt season to the time of sampling in the following spring were collected in the spring of 2005 and 2006 from locations in the Canadian Arctic: Melville Ice Cap, Melville Island, Northwest Territories (75º 27N, 114º 59W); Agassiz Ice Cap, Ellesmere Island, Nunavut (80º 7N, 73º 1W); and Meighen Ice Cap, Meighen Island, Nunavut (79º 27N, 99º 08W, 2006 only). A map of sampling locations can be found in Appendix E. Depth samples were collected from the Devon Ice Cap, Devon Island, Nunavut (75º 20N, 82º 40W, 1797 m above sea level) in spring 2006 (a depth range-finding study was done in the spring of 2005). To avoid contamination, fluoropolymer products were strictly avoided at all sampling sites. For full sampling details, please see Appendix E.

9.2.3 Sample preparation and analysis

Full method details can be found in the Appendix E. Briefly, samples were concentrated one hundred times using solid phase extraction and eluted with methanol. Samples were analyzed using liquid chromatography with tandem mass spectrometry detection, using an isocratic method (23). Quantification of analytes was done using labelled internal standards, which were available for PFOS, PFOA, perfluorononanoic acid (PFNA) and perfluorodecanoic acid (PFDA). Perfluoroundecanoic acid (PFUnA) quantification was done using labelled PFDA.

9.3 Results and Discussion

9.3.1 QA/QC

For full details on all QA/QC, please see Appendix E. Instrumental contamination was not observed to be a problem. Variability was minimized with the use of labelled internal standards and triplicate injections. Extraction blanks and spike and recoveries were used to validate the extraction method. Recoveries were acceptable and ranged from 83 to 110%. A field blank test indicated that field sampling did not result in contamination of the samples.

177 a) 0 2005

2004 200 2003 2002 2001 Year 400 PFOA 2000 (cm) Depth PFNA 1999 1998

600 1997 1996

0 50 100 150 200 250 -1 Concentration (pg L snow)

0 b) 2005

2004 200 2003 2002 2001 Year 400 2000 Depth (cm) PFDA PFUnA 1999 1998

600 1997 1996

0 10203040 Concentration (pg L-1 snow)

0 c) 2005 2004 200 2003 PFOS 2002 2001 Year 400 2000 Depth (cm) 1999 1998 600 1997 1996 0 20406080 Concentration (pg L-1 snow)

Figure 9.1: Density-corrected concentrations of PFAs on Devon Ice Cap since 1996: (a) PFOA and PFNA; (b) PFDA and PFUnA; and (c) PFOS.

178 9.3.2 Dating arctic snow Density, conductivity and major ion measurements, along with visual inspection of ice layers, are all useful for the determination of age of Arctic snow (24). These properties are all expected to vary over the course of a year and allow seasonal and annual markers to be identified along a profile. Observed concentrations of major ions, densities and conductivities from the Devon Ice Cap are shown in Appendix E. Utilizing this information, along with visual inspection, it is possible to assign dates to the depth of the pit. A pit of depth 6.8 m appears to go back to the year 1996. Little snow is available from 2006, as the majority of snow in the Arctic falls in equal parts in the summer and fall. Thus, the last year for which a flux of PFAs can be determined is 2005.

9.3.3 Concentrations of PFAs in arctic snow

PFAs were observed in all surface and depth samples at pg L-1 concentrations. Observed concentrations were corrected for density by multiplying concentrations measured in meltwater by the observed density of the snow. Concentrations were: 2.6 to 86 pg L-1 for PFOS, 12 to 147 pg L-1 for PFOA; 5.0 to 246 pg L-1 for PFNA;

179 PFOA PFNA 12000 12000

10000 10000 ) 8000 )

-2 8000 -2

6000 6000

Flux (fg cm 4000 Flux (fg cm 4000

2000 2000

0 0 1996 1998 2000 2002 2004 2006 1996 1998 2000 2002 2004 2006 Year Year PFDA PFUnA 1600 1600

1400 1400

1200 1200 ) ) -2 1000 -2 1000 800 800 600 600 Flux (fg cm (fg Flux Flux (fg cm 400 400 200 200 0 0 1996 1998 2000 2002 2004 2006 1996 1998 2000 2002 2004 2006 Year Year PFOS 3000

2500

) 2000 -2

1500

Flux (fg cm 1000

500

0 1996 1998 2000 2002 2004 2006 Year Figure 9.2: PFA fluxes to Devon Ice Cap by year. Mean fluxes with standard error of three replicates indicated by bars. Lines reflect a three-year moving average.

into three sections. An increase was observed from 1996 to 1998 (p < 0.001), followed by a decrease from 1998 to 2001 (p < 0.001). Following 2001, no significant trend was evident in the PFOS concentrations (p = 0.094). This is in general agreement with data produced by Butt

180 et al (19), who observed increasing PFOS levels in Arctic ringed seals up to 1998 or 2000, after which concentrations began to decline.

Surface concentrations were measured at the Devon Ice Cap and additional locations in 2005 and 2006, which represent deposition from 2004 and 2005. Concentrations at the Agassiz and Melville Ice Caps were measured in both years, while those at the Meighen Ice Cap were measured in 2006. Some spatial variation is observed in the measured concentrations and fluxes to the different locations are shown in the Appendix E. Devon Ice Cap shows concentrations that are approximately an order of magnitude higher than those found on other ice caps. There are numerous possible causes of this variability. Devon Island and was the most southerly site, and it is known to receive contamination from both North America and Eurasian sources. More northerly sites, such as Ellesmere Island (Agassiz Ice Cap) receive pollution primarily from Eurasian sources (26). Local atmospheric chemistry could also play a role in spatial variation. In addition, deposition rates are higher in the area of the Devon Ice Cap (27), which are expected to influence observed concentrations.

9.3.4 Effects of annual melting

Since 1985, there has been a significant increase in annual summer melt rates on many Arctic glaciers, including at the Devon Ice Cap. The summers of 2005, 2001 and 1998 were especially warm, leading to more melt than usual (28). Because of this melting, it is likely that migration of some chemicals down the profile occurs. Despite this, the annual chemical signals can still be interpreted with confidence, as shown through analysis of another ice cap from the Canadian Arctic (29). Generally, melt water re-freezes within the annual layer, but in warm summers may percolate into one or more of the layers below (28). Melting temperatures were shown to be followed within a few days by below freezing temperatures. If water freezes sufficiently to form an ice layer, further meltwater will accumulate on the impermeable layer, preventing soluble species from moving down the snowpack. Thus, it is likely that the numerous ice layers observed within the Devon Ice Cap profile have preserved the major ion and PFA record.

Examination of the data reveals an anomalously high concentration of PFOS in 1998. This high level stands out due to the subsequent 500% decrease of PFOS levels. It is possible that melting in the warm summer of 2001 could have caused migration of PFOS, which would have collected on the thick ice layer created in 1998. Thus, the concentration that is represented

181 for 1998 may be artificially inflated by PFOS deposited in 1999 through 2001. This could explain why a drop-off in PFOS concentration is seen in 1999 before the phase-out of these chemistries, which began in 2000 and was presumably completed by the end of 2002. However, no evidence of migration is observed for any of the other PFA species. This suggests that any migration experienced by PFOS would likely have been minor.

9.3.5 Fluxes of PFAs to the Arctic

In order to compare our results with those of the modelling study by Wallington et al (20), it is necessary to estimate deposition to the entire Arctic. Fluxes calculated to each of the ice caps were multiplied by the area of the Arctic to yield a flux of these compounds to the area north of 65°N. These fluxes are estimates only and may not be representative of actual deposition in this region, because of wide variations in precipitation rates. Flux means and ranges are shown in Table 1 for individual compounds. The fluxes, as calculated from different High Arctic ice caps, were somewhat variable, which is expected given the variability of surface concentrations discussed above. Fluxes at Meighen, Melville and Agassiz Ice Caps were all reasonably similar, but those at the Devon Ice Cap were higher. PFOS had the lowest flux of all the PFAs measured, with 31 kg year-1 in 2004 and 33 kg year-1 in 2005. PFOA and PFNA showed the highest fluxes, averaging 271 and 295 kg year-1, respectively, in 2004 and 2005. The total PFCA flux was 313 kg year-1 for 2004 and 651 kg year-1 for 2005. These fluxes agree with those determined through modelling of FTOH degradation. The modelling study by Wallington et al (20) determined a flux of PFOA of 400 kg year-1 to the Arctic from the atmospheric oxidation of 8:2 FTOH. The molar yield of PFNA was also determined to be similar to that of PFOA. The average PFOA flux to the Arctic determined herein for the year 2005 was 271 kg year-1, while that for PFNA was 295 kg year-1, which are within a factor of two to those calculated in the modelling study. These fluxes would be the sum total of all atmospheric precursor sources, which may include minor contributions from fluorosulfamido alcohols (PFOA only) and longer-chain FTOH species. However, the 8:2 FTOH is expected to be the dominant atmospheric precursor to PFOA and PFNA. The agreement of observation with the FTOH modelling study also suggests that assumptions made in the modelling study were reasonable.

182 9.3.6 Sources of PFAs to the Arctic

Given that Arctic ice caps receive contamination solely from the atmosphere, the fluxes of PFOA to the Arctic described above are a result of atmospherically derived contamination. This contamination could come from the atmospheric oxidation of gas-phase volatile precursors, followed by wet or dry deposition onto the ice cap. Alternately, it could come from marine aerosols, moving contamination from oceans. The presence of sodium indicates that sea salt aerosols are deposited onto the ice cap (30). Marine aerosols are known to be coated with a layer of organics that can contain enriched concentrations of organic pollutants. Thus, it is possible that these aerosols could transport PFAs from marine environments to the ice cap. These aerosols are likely to be formed from the sea surface microlayer from production by bubble bursting (31). PFOA, PFNA and PFOS have been shown to be present in remote ocean waters at pg L-1 levels (32,33). For example, one open ocean location (North Atlantic) has been observed to contain 8.6 – 36 pg L-1 PFOS, 15 – 36 pg L-1 PFNA and 160 – 338 pg L-1 PFOA. It is important to note that PFOA concentrations are approximately ten times higher than those of

Table 9.1. Mean fluxes of PFAs to the Arctic for 2004 and 2005. Note that PFUnA concentrations were not measured for 2004 samples.

Flux (kg year-1) Agassiz Ice Devon Ice Meighen Melville Ice Mean Cap Cap Ice Cap Cap Analyte 2004 2005 2004 2005 2005 2004 2005 2004 2005 PFOS 11 18 58 48 23 23 44 31 33 PFOA 104 114 239 587 217 156 165 167 271 PFNA 79 74 132 860 175 94 73 102 295 PFDA 31 20 60 84 32 43 16 45 38 PFUnA 41 62 56 26 46 Total 214 250 431 1593 480 293 280 313 651 PFCA

183 PFNA. Concentrations of these compounds may be present at higher levels in the sea surface microlayer, and likewise in marine aerosols. Kaiser et al (34) calculated an enrichment factor of up to 3 in foam produced from bubbling through a dilute PFOA solution. However, this may bean underestimate, as traditional hydrophobic persistent organic pollutants, such as chlorinated hydrocarbons, can be enriched in the sea surface microlayer by up to a factor of 500 (35). Concentrations of PFOA are similar in the ocean and on the ice cap. Sodium concentrations,which are indicative of ocean contamination, are lower on the ice cap by more than 104. This corresponds to a PFOA enrichment of greater than 104 on the ice cap. Even if we assume a microlayer enrichment of 103, this can not account for the enrichment observed on the ice cap. Sodium to PFA ratios measured on the Devon Ice Cap also vary with time (see Appendix E). This suggests the source of PFAs is unrelated to the ocean. To date, PFDA and PFUnA have not been reported in any PFA ocean monitoring studies, presumably because their concentrations are lower than detectable limits. The presence of PFDA and PFUnA on High Arctic ice caps is indicative of atmospheric oxidation. These compounds do not have significant commercial production (4), so their presence in any media would likely be as a result of atmospheric oxidation. Observed ice cap concentrations of PFDA and PFUnA are similar. The mean ratio of PFDA to PFUnA concentrations is 0.9 ± 0.8. Similarly, the ratio of PFOA and PFNA concentrations is 1.5 ± 0.8. Overall, the concentrations of PFOA and PFNA are approximately an order of magnitude higher than those of PFDA and PFUnA. If the PFCAs in the High Arctic are due to oxidation, the source of PFOA and PFNA would primarily be 8:2

300 30

250 25 snow) snow) -1 -1

200 20

150 15

100 10 m = 1.02 m = 0.99 50 r ² = 0.40 5 r ² = 0.42 PFNA concentration (pg L (pg PFNA concentration 0 PFUnA concentation (pg L 0 0 50 100 150 200 250 300 0 5 10 15 20 25 30 -1 PFOA concentration (pg L snow) PFDA concentration (pg L-1 snow)

Figure 9.3: Correlations between PFCA concentrations on Devon Ice Cap.

184 FTOH, while the source of PFDA and PFUnA would primarily be 10:2 FTOH. Although the 10:2 FTOH can also degrade to form PFNA and PFOA, it is expected to do so in lower quantities than it produces PFDA and PFUnA (9). As discussed above, concentrations of PFOA are approximately an order of magnitude higher than those of PFNA in open ocean waters. Thus, it is unlikely that these could account for the similar PFOA and PFNA concentrations observed on the ice cap.

If all PFAs were coming from the same source, their concentrations would be expected to vary together through time. A graph of PFOA versus PFNA concentrations shows a positive correlation with a slope of 1.02, suggesting these two compounds come from the same source, likely the 8:2 FTOH (Figure 3). A similar exercise with PFDA and PFUnA concentrations also shows a positive correlation with a slope of 0.99, again indicating the same source, probably the 10:2 FTOH. PFNA and PFDA are also positively correlated (see Appendix E). This indicates the sources of these two sets of compounds are related. Dinglasan-Panlilio and Mabury (12) have demonstrated that a significant source of FTOH emissions may be coming from residuals in fluorinated polymer products, which were shown to contain both the 8:2 and 10:2 FTOHs. Although different products contained different amounts of these two FTOHs, the ratio of the residuals is consistent through time, suggesting a relationship between the release of the 8:2 and 10:2 FTOHs. Plots of PFNA, PFDA and PFUnA versus PFOS concentrations show no correlation, indicating that the sources of these compounds are unrelated. A positive correlation is observed between PFOA and PFOS, although the linear fit is lower than observed between PFCAs (see Appendix E). The eight-carbon perfluorosulfamido alcohols that were produced up to the year 2000 may degrade in the atmosphere to form PFOA and shorter-chain PFCAs only, in addition to PFOS (10,11). This could lead to another minor source of PFOA that is related to the source of PFOS.

Changing concentrations of PFAs through the past few years can also give evidence as to contamination source. The observed annual decrease of PFOS is likely in response to the removal of perfluorooctane sulfonyl fluoride (PFOSF) from the market. PFOSF was used to make PFOS and other related chemicals. PFOSF production increased from 1985 through to 2000. In 2001 and 2002, production continued at a decreased level in response to a phase-out beginning in January, 2001 (5). It is also possible that the slight, and sometimes insignificant, increase in PFCAs in 2004 and 2005 may reflect a replacement of the phased-out PFOSF chemistry with telomer-based chemistries. This fast response to changes in production is

185 indicative of a significant atmospheric source. Transport times via the ocean would be on the order of decades, as demonstrated by Li et al for transport of β-hexachlorocyclohexane (6), and consequently could not reflect the abrupt production changes.

Detection of the fluorotelomer acids, also OH mediated atmospheric products of FTOH oxidation, would substantially strengthen the overall connection between volatile precursors and fluorinated acids in the Arctic. For this study, the analytical challenges exceeded our current capabilities, given the expected low concentrations, their significantly higher detection limits and their lability. We are working on this problem in a current study.

PFDA and PFUnA are not produced in large quantities commercially, nor are they detected in ocean waters. As a result, the presence of PFDA and PFUnA on ice caps is an indicator of contamination by atmospheric oxidation. Ratios of PFOA:PFNA and PFDA:PFUnA, as well as the correlation between the PFCAs through time is consistent with the source of these acids being FTOHs. If the source of these compounds was oceanic in nature, a correlation between PFCAs and PFOS would be expected. Overall, there is no correlation between PFCA and PFOS concentrations. This fact points to atmospheric oxidation, whereby PFOS and a small amount of PFOA would be derived from fluorosulfamidoalcohols, while the majority of PFCAs would be derived from FTOHs. The consistency of observed concentrations with FTOH modelling supports this theory. The fast response of ice cap concentrations to the PFOS production changes also suggests an atmospheric source. It is also interesting that temporal trends in snow are similar to those observed in Arctic ringed seals (19), which supports the idea that biota are exposed to contaminants derived primarily from the atmosphere. Melting snow may inject PFAs directly into the upper water column, from which they could be taken up into the food web. This contradicts the hypothesis of Armitage et al (36), which proposes that Arctic contamination is dominated by direct sources transported through the ocean. Evidence obtained from Arctic ice caps is generally in agreement with the precursor alcohol atmospheric reaction and transport hypothesis in order to explain PFA contamination in the Arctic.

186 9.4 Acknowledgements

The authors thank James Zheng, Christian Zdanowicz and Jocelyne Bourgeois of the Geological Survey of Canada and Xiaowa Wang of Environment Canada for sample collection and coordination. The provision of labelled standards by Gilles Arsenault and Wellington Laboratories is appreciated. Thanks also to Craig Butt, Jon Abbatt and Jamie Donaldson of the University of Toronto. The Ontario Ministry of the Environment and Eric Reiner are thanked for use of instrumentation. Special thanks to Dan Walsh of Environment Canada for sampling design and field support. Funding was provided by NSERC, NSTP and Solvay Solexis. CJY appreciates the support of NSERC through a Canada Graduate Scholarship. Work would not have been possible without the support of the Polar Continental Shelf Project (PCSP). This work represents PCSP publication number 027-06.

187 9.5 Sources Cited

(1) Giesy, J.P.; Kannan, K. Distribution of perfluorooctane sulfonate in wildlife. Environmental Science and Technology 2001, 35, 1339-1342.

(2) Martin, J.W.; Smithwick, M.M.; Braune, B.M.; Hoekstra, P.F.; Muir, D.C.G.; Mabury, S.A. Identification of long-chain perfluorinated acids in biota from the Canadian arctic. Environmental Science and Technology 2004, 38, 373-380.

(3) Stock, N.L.; Furdui, V.I.; Muir, D.C.G.; Mabury, S.A. Perfluoroalkyl contaminants in the Canadian Arctic: Evidence of atmospheric transport and local contamination. Environmental Science and Technology 2007, 41, 3529-3536.

(4) Prevedouros, K.; Cousins, I.T.; Buck, R.C.; Korzeniowski, S.H. Sources, fate and transport of perfluorocarboxylates. Environmental Science and Technology 2006, 40, 32-44.

(5) Smithwick, M.M.; Norstrom, R.J.; Mabury, S.A.; Solomon, K.R.; Evans, T.J.; Stirling, I.; Taylor, M.K.; Muir, D.C.G. Temporal trends of perfluoroalkyl contaminants in polar bears (Ursus maritimus) from two locations in the North Amiercan Arctic, 1972-2002. Environmental Science and Technology 2006, 40, 1139-1143.

(6) Li, Y.F.; Macdonald, R.W. Sources and pathways of selected organochlorine pesticides to the Arctic and the effect of pathway divergence on HCH trends in biota: a review. Science of the Total Environment 2005, 342, 87-106.

(7) Ellis, D.A.; Martin, J.W.; Mabury, S.A.; Hurley, M.D.; Sulbaek Andersen, M.P.; Wallington, T.J. Atmospheric lifetime of fluorotelomer alcohols. Environmental Science and Technology 2003, 37, 3816-3820.

(8) Hurley, M.D.; Ball, J.C.; Wallington, T.J.; Sulbaek Andersen, M.P.; Ellis, D.A.; Martin, J.W.; Mabury, S.A. Atmospheric chemistry of fluorinated alcohols: Reaction with Cl atoms and OH radicals and atmospheric lifetimes. Journal of Physical Chemistry A 2004, 108, 1973-1979.

(9) Ellis, D.A.; Martin, J.W.; De Silva, A.O.; Mabury, S.A.; Hurley, M.D.; Sulbaek Andersen, M.P.; Wallington, T.J. Degradation of fluorotelomer alcohols: A likely atmospheric source of perfluorinated carboxylic acids. Environmental Science and Technology 2004, 38, 3316-3321.

(10) D'eon, J.C.; Hurley, M.D.; Wallington, T.J.; Mabury, S.A. Atmospheric chemistry of N- methyl perfluorobutane sulfonamidoethanol, C4F9SO2N(CH3)CH2CH2OH: Kinetics and mechanism of reaction with OH. Environmental Science and Technology 2006, 40, 1862-1868.

(11) Martin, J.W.; Ellis, D.A.; Mabury, S.A.; Hurley, M.D.; Wallington, T.J. Atmospheric chemistry of perfluoroalkanesulfonamides: Kinetic and product studies of the OH and Cl atom initiated oxidiation of N-ethyl perfluorobutanesulfonamide. Environmental Science and Technology 2006, 40, 864-872.

(12) Dinglasan-Panlilio, M.J.A.; Mabury, S.A. Significant residual fluorinated alcohols present in various fluorinated materials. Environmental Science and Technology 2006, 40, 1447-1453.

188 (13) Martin, J.W.; Muir, D.C.G.; Moody, C.A.; Ellis, D.A.; Kwan, W.; Solomon, K.R.; Mabury, S.A. Collection of airborne fluorinated organics and analysis by gas chromatography/chemical ionization mass spectrometry. Analytical Chemistry 2002, 74, 584- 590.

(14) Stock, N.L.; Lau, F.K.; Ellis, D.A.; Martin, J.W.; Muir, D.C.G.; Mabury, S.A. Polyfluorinated telomer alcohols and sulfonamides in the North American troposphere. Environmental Science and Technology 2004, 38, 991-996.

(15) Shoeib, M.; Harner, T.; Wilford, B.H.; Jones, K.C.; Zhu, J. Perfluorinated sulfonamides in indoor and outdoor air and indoor dust: Occurrence, partitioning, and human exposure. Environmental Science and Technology 2005, 39, 6599-6606.

(16) Shoeib, M.; Harner, T.; Vlahos, P. Perfluorinated chemicals in the Arctic atmosphere. Environmental Science and Technology 2006, 40, 7577-7583.

(17) 3M. 2000. Phase-out plan for POSF-based products, 3M, Specialty Materials Markets Group.

(18) United States Environmental Protection Agency. 2010/15 PFOA Stewardship Program.

(19) Butt, C.M.; Muir, D.C.G.; Stirling, I.; Kwan, M.; Mabury, S.A. Rapid response of Arctic ringed seals to changes in perfluoroalkyl production. Environmental Science and Technology 2007, 41, 42-49.

(20) Wallington, T.J.; Hurley, M.D.; Xia, J.; Wuebbles, D.J.; Sillman, S.; Ito, A.; Penner, J.E.; Ellis, D.A.; Martin, J.W.; Mabury, S.A.; Nielsen, C.J.; Sulbaek Andersen, M.P. Formation of C7F15COOH (PFOA) and other perfluorocarboxylic acids during the atmospheric oxidation of 8:2 fluorotelomer alcohol. Environmental Science and Technology 2006, 40, 924-930.

(21) Peters, A.J.; Gregor, D.J.; Teixeira, C.F.; Jones, N.P.; Spencer, C. The recent depositional trend of polycylic aromatic hydrocarbons and elemental carbon to the Agassiz Ice Cap, Ellesmere Island, Canada. Science of the Total Environment 1995, 160/161, 167-179.

(22) Gregor, D.; Peters, A.J.; Teixeira, C.; Jones, N.; Spencer, C. The historical residue trend of PCBs in the Agassiz Ice Cap, Ellesmere Island, Canada. Science of the Total Environment 1995, 161/161, 117-126.

(23) Furdui, V.I.; Crozier, P.W.; Reiner, E.J.; Mabury, S.A. Optimized trace level analysis of polyfluorinated carboxylic and sulfonic acids. Organohalogen Compounds 2006, 211-214.

(24) Koerner, R.M.; Fisher, D.A.; Goto-Azuma, K. A 100 year record of ion chemistry from Agassiz Ice Cap Northern Ellesmere Island NWT, Canada. Atmospheric Environment 1999, 33, 347-357.

(25) Scott, B.F.; Spencer, C.; Mabury, S.A.; Muir, D.C.G. Poly and perfluorinated carboxylates in North American precipitation. Environmental Science and Technology 2006, 40, 7167-7174.

189 (26) Goto-Azuma, K.; Koerner, R.M. Ice core studies of anthropogenic sulfate and nitrate trends in the Arctic. Journal of Geophysical Research 2001, 106, 4959-4969.

(27) Brown, R.D.; Brasnett, B.; Robinson, D. Gridded North American monthly snow depth and snow water equivalent for GCM evaluation. Atmosphere-Ocean 2003, 41, 1-14.

(28) Koerner, R.M. Mass balance of glaciers in the Queen Elizabeth Islands, Nunavut, Canada. Annals of Glaciology 2005, 42, 417-423.

(29) Grumet, N.S.; Wake, C.P.; Zielinski, G.A.; Fisher, D.; Koerner, R.; Jacobs, J.D. Preservation of glaciochemical time-series in snow and ice from the Penny Ice Cap, Baffin Island. Geophysical Research Letters 1998, 25, 357-360.

(30) Legrand, M.; Mayewski, P. Glaciochemistry of polar ice cores: A review. Review of Geophysics 1997, 35, 219-243.

(31) Marty, J.C.; Saliot, A.; Buat-Menard, P.; Chesselet, R.; Hunter, K.A. Relationship between the lipid compositions of marine aerosols, the sea surface microlayer, and subsurface water. Journal of Geophysical Research 1979, 84, 5707-5716.

(32) Yamashita, N.; Kannan, K.; Taniyasu, S.; Horii, Y.; Okazawa, T.; Petrick, G.; Gamo, T. Analysis of perfluorinated acids at parts-per-quadrillion levels in seawater using liquid chromatography-tandem mass spectrometry. Environmental Science and Technology 2004, 38, 5522-5528.

(33) Yamashita, N.; Kannan, K.; Taniyasu, S.; Horii, Y.; Petrick, G.; Gamo, T. A global survey of perfluorinated acids in oceans. Marine Pollution Bulletin 2005, 51, 658-668.

(34) Kaiser, M.A.; Barton, C.A.; Botelho, M.; Buck, R.C.; Buxton, L.W.; Gannon, J.; Kao, C.- P.C.; Larsen, B.S.; Russell, M.H.; Wang, N.; Waterland, R.L. Understanding the transport of anthropogenic fluorinated compounds in the environment. Organohalogen Compounds 2006, 68, 675-678.

(35) Wurl, O.; Obbard, J.P. A review of pollutants in the sea-surface microlayer (SML): a unique habitat for marine organisms. Marine Pollution Bulletin 2004, 48, 1016-1030.

(36) Armitage, J.; Cousins, I.T.; Buck, R.C.; Prevedouros, K.; Russell, M.H.; Macleod, M.; Korzeniowski, S.H. Modeling global-scale fate and transport of perfluorooctanoate emitted from direct sources. Environmental Science and Technology 2006, 40, 6969-6975.

CHAPTER TEN

Conclusions and Future Directions

190 191 10.1 Conclusions

The overall objective of this work was to examine the atmospheric chemistry of poly- and perfluorinated chemicals to better understand their contributions to climate change and the global dissemination of perfluorinated acids. The chemistry behind these processes was probed with laboratory and modelling studies, with monitoring studies undertaken to examine relevance to the real environment.

In this thesis, two new classes of long-lived greenhouse gases (LLGHGs) were discovered: perfluoropolyethers (PFPEs) and perfluoroalkyl amines (PFAms). Within each of these classes, a number of commercially produced compounds exist. To date, only one compound, perfluorotributyl amine (PFBAm), has been identified in the atmosphere in either class. Although the climate impact of PFBAm itself is not significant, the overall impact of the classes of PFPEs and PFAms cannot be assessed, because atmospheric concentrations have not been sought. Two other significant classes of LLGHGs that have been well-studied in the lab are hydrofluoroethers (HFEs) and hydrofluoropolyethers (HFPEs) (1). As far as we are aware, no studies have looked for these compounds in the atmosphere, despite their widespread commercial use. Levels of individual PFPEs, PFAms, HFEs and HFPEs are expected to be low, which may explain the lack of measurement to date. However, despite the low levels and impacts of individual fluorinated LLGHGs, the sum total of these potent greenhouse gases could be significant. The current assessment of climate impact of LLGHGs is assumed to be well understood, but only includes the impacts of gases that have been measured in the atmosphere (1). The large number of known LLGHGs that are commercially produced in high quantities, but have unknown atmospheric levels are not included in this appraisal. In order to fully appreciate the climate effects of halogenated LLGHGs, it is necessary to include the potential impacts of these as-yet-unmeasured compounds.

Gas-phase atmospheric chemistry of a number of volatile precursors to perfluorinated acids (PFAs) has been well characterized, including three new precursors evaluated in this thesis. Although this chemistry has been examined in the lab, it is difficult to assess how the observed mechanisms and kinetics apply to the real atmosphere. Many of the mechanisms rely on low-NOx conditions, or more specifically, the ratio of NO to HO2 or RO2 radicals, for the formation of PFAs. Since, these ratios are constantly changing and not well characterized on a global scale, it is difficult to evaluate the overall impact of the known chemistry. Further, the 192 physical properties of precursor compounds, particularly fluorotelomer compounds, are poorly understood. As a result, it is difficult to predict the prevalence of partitioning, wet and dry deposition and the overall fate of these compounds. The specifically designed monitoring study in this work indicates that atmospheric degradation of volatile precursors is a source of PFAs to the Canadian High Arctic. However, more monitoring studies designed to elucidate sources of PFAs are required to fully appreciate the contribution of volatile precursors to global contamination of PFAs.

We characterized the properties of two new LLGHGs: a PFPE and a PFAm. These each likely represent an example of a class of chemicals that can act as LLGHGs, with other chemicals yet to be examined. One of the new LLGHGs, PFBAm, was observed in the atmosphere at low concentrations. From the observed concentration of PFBAm, it is possible that the cumulative effect of PFAms could be significant.

It is commonly acknowledged that increasing the number of C-F bonds in a compound increases the radiative efficiency (RE) of that compound. However, inspection of REs of fluorinated compounds clearly indicates that not all C-F bonds are of equal radiative potency. The structure-activity relationship we observed will allow for the intelligent design of polyfluorinated ethers with low REs for commercial use, potentially reducing climate impact.

Through the introduction of climate directives, new compounds, such as fluorinated olefins, are being suggested as alternatives to traditional coolants. Our study of two perfluorinated olefins demonstrated they were not LLGHGs due to their short atmospheric lifetimes. However, the exclusive products of atmospheric oxidation were the persistent perfluorocarboxylic acids (PFCAs) in yields of 100 or 200%. Through this previously- unobserved mechanism, it was concluded that any olefin that is fully fluorinated on one side of the olefin moiety will form PFCAs in a yield of unity. Although the low impact of these compounds on climate is certainly desired, it is important to consider all potential environmental effects of large-scale use.

Long-chained and bioaccumulative PFCAs can also be formed from volatile precursors, through a number of known mechanisms. A fluorotelomer iodide was observed to form PFCAs through two of these mechanisms. The reactions observed should extend to all chain-length congeners of fluorotelomer iodides, of which eight are considered high-production volume chemicals (2). Within the perfluorinated radical mechanism, one step that is not well 193 understood is the dehydrofluorination of perfluorinated alcohols. We observed that overtone- induced photolysis could play a role in the degradation of these alcohols and may play a role in the formation of PFCAs in the atmosphere.

Although a number of modelling studies have examined the relative impacts of volatile precursor formation versus direct transportation through ocean currents to the Arctic, there have been few empirical studies designed to specifically examine this question. We attempted to address this by collecting snow from a High Arctic ice cap, which receives contamination solely from the atmosphere. The trends and fluxes determined from the measurements indicated that atmospheric input from volatile precursors is an important source of PFA contamination to the Arctic.

10.2 Future Directions

Stemming from this work are a number of avenues for future work in this area. In particular, detection of both LLGHGs and perfluorinated acid (PFA) precursors in the atmosphere is critical to fully appreciating the full impact of these compounds on the environment.

10.2.1 Perfluoroalkyl amines

This work examined the atmospheric fate, radiative efficiency and atmospheric concentration of one perfluoroalklyl amine (PFAm). A number of other congeners exist, for which production is known to be higher (2). Measurements of radiative efficiencies of these compounds and their atmospheric concentrations are a logical next step in determining the total impact of PFAms on climate. Unfortunately, standards for these compounds are currently lacking. Perfluorotripentyl amine is available only as an industrial mixture of ~85 % purity and standards of perfluorotrihexyl amine are not available. Purification or synthesis of these compounds is necessary for these experiments to be conducted. The chemistry of these high- production volume chemicals is important for understanding the impacts of this class of compounds.

10.2.2 Atmospheric chemistry of new long-lived greenhouse gases

In this work, two novel LLGHGs were described. Examination of commercially available compounds and high-production volume lists on the basis of chemical structure can 194 suggest potential new LLGHGs. For example, any perfluorinated chemical that may be released to the atmosphere is likely to have a significant atmospheric lifetime and radiative efficiency and be considered a LLGHG. One such compound is perfluoro(2-butyltetrahydrofuran). This is produced as a byproduct of electrochemical production of perfluorooctanoic acid or perfluorosulfonyl fluoride and is marketed as a coolant product (3). Identifying and examining the properties of potential LLGHGs is critical to fully understanding the impact of halogenated compounds on climate.

10.2.3 Atmospheric monitoring of fluorinated ethers

The atmospheric fate and radiative properties have been determined for sixteen fluorinated ethers (1), including the perfluoropolyether discussed herein. All of these compounds are highly radiatively active and one represents the compound with the highest radiative efficiency yet measured (4). Some fluorinated ethers are also present on a high- production volume list (2). To our knowledge, atmospheric concentrations of these compounds have never been sought. It is possible that methods have not been developed, because these compounds are less volatile than LLGHGs that are currently measured, presenting difficulties in applying existing methods. Preliminary studies using our thermal desorber/cryofocuser coupled to GC-MS with electron capture negative ionization detection system suggests these compounds can be detected in the atmosphere using this method. Work is in progress to determine atmospheric concentrations of some fluorinated ethers.

10.2.4 Atmospheric monitoring of volatile perfluorinated acid precursors

Current methods used to monitor volatile PFA precursors are based on the methods developed for semi-volatile chemicals (5). The typical method consists of polyurethane foam (PUF) and XAD sorbents, sampled using high-volume samplers over 3 – 4 days, followed by solvent extraction and GC-MS analysis. Since many PFA precursors are volatile, often recoveries for these compounds are very low (6), making accurate measurements difficult. Current methods are suitable for 6:2 through 12:2 fluorotelomer alcohols (FTOHs), 8:2 to 12:2 fluorotelomer olefins (FTOs), fluorotelomer acrylates and perfluorosulfamido compounds. Short-chain FTOHs and FTOs are unsatisfactorily captured, as presumably are fluorotelomer iodides. In order to gain an appreciation for the full range of PFA precursors in the atmosphere, new methods are required. The thermal desorber/cryofocuser method may also be suitable for these compounds. Sampling with portable samplers instead of high-volume would allow greater 195 spatial sampling. Removing the need for solvent extraction would also allow shorter collection times and the possibility of determining temporal trends over short time-courses.

10.3 Sources Cited

(1) Forster, P.; Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R.; Fahey, D.W.; Haywood, J.; Lean, J.; Lowe, D.C.; Myhre, G.; Nganga, J.; Prinn, R.; Raga, G.; Schulz, M.; Van Dorland, R. In Climate Change 2007: The Physical Science Basis; Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L., Eds.; Cambridge University Press: Cambridge, United Kingdom, 2007.

(2) Howard, P.H.; Meylan, W. 2007. EPA Great Lakes Study for Identification of PBTs to Develop Analytical Methods: Selection of Additional PBTs - Interim Report, EPA Contract No. EP-W-04-019,

(3) Banks, R.E., Ed. Organofluorine chemistry: principles and commercial applications; Plenum: New York, 1994.

(4) Wallington, T.J.; Hurley, M.D.; Nielsen, O.J. The radiative efficiency of HCF2OCF2OCF2CF2OCF2H (H-Galden 1040x) revisited. Atmospheric Environment 2009, 43, 4247-4249.

(5) Martin, J.W.; Muir, D.C.G.; Moody, C.A.; Ellis, D.A.; Kwan, W.; Solomon, K.R.; Mabury, S.A. Collection of airborne fluorinated organics and analysis by gas chromatography/chemical ionization mass spectrometry. Analytical Chemistry 2002, 74, 584-590.

(6) Barber, J.L.; Berger, U.; Chaemfa, C.; Huber, S.; Jahnke, A.; Temme, C.; Jones, K.C. Analysis of per- and polyfluorinated alkyl substances in air samples from Northwest Europe. Journal of Environmental Monitoring 2007, 9, 530-541.

APPENDIX A

SUPPORTING INFORMATION FOR CHAPTER FOUR:

Molecular Structure and Radiative Efficiency of Fluorinated Ethers: A Structure-Activity Relationship

196 197

All compounds modeled using density functional theory are listed below:

CF3-O-CF3 CH2FCH2CH2-O-CH2F CH2F-O-CF3 CH2FCH2CHF-O-CH3 CH2F-O-CH2F CH2FCHFCH2-O-CH3 CH2F-O-CH3 CH3CF2CH2-O-CH3 CH3-O-CF3 CH3CH2CF2-O-CH3 CHF2-O-CF3 CH3CH2CH2-O-CHF2 CHF2-O-CH2F CH3CH2CHF-O-CH2F CHF2-O-CH3 CH3CHFCH2-O-CH2F CHF2-O-CHF2 CH3CHFCHF-O-CH3 CH2FCH2-O-CH2F CHF2CH2CH2-O-CH3 CH2FCHF-O-CH3 CF3CH2CH2-O-CH3 CH3CF2-O-CH3 CH2FCF2CH2-O-CH3 CH3CH2-O-CHF2 CH2FCH2CF2-O-CH3 CH3CHF-O-CH2F CH2FCH2CH2-O-CHF2 CHF2CH2-O-CH3 CH2FCH2CHF-O-CH2F CF3CH2-O-CH3 CH2FCHFCH2-O-CH2F CH2FCF2-O-CH3 CH2FCHFCHF-O-CH3 CH2FCH2-O-CHF2 CH3CF2CH2-O-CH2F CH2FCHF-O-CH2F CH3CF2CHF-O-CH3 CH3CF2-O-CH2F CH3CH2CF2-O-CH2F CH3CH2-O-CF3 CH3CH2CH2-O-CF3 CH3CHF-O-CHF2 CH3CH2CHF-O-CHF2 CHF2CH2-O-CH2F CH3CHFCF2-O-CH3 CHF2CHF-O-CH3 CH3CHFCH2-O-CHF2 CF3CH2-O-CH2F CH3CHFCHF-O-CH2F CF3CHF-O-CH3 CHF2CH2CH2-O-CH2F CH2FCF2-O-CH2F CHF2CH2CHF-O-CH3 CH2FCH2-O-CF3 CHF2CHFCH2-O-CH3 CH2FCHF-O-CHF2 CF3CH2CH2-O-CH2F CH3CF2-O-CHF2 CF3CH2CHF-O-CH3 CH3CHF-O-CF3 CF3CHFCH2-O-CH3 CHF2CF2-O-CH3 CH2FCF2CH2-O-CH2F CHF2CH2-O-CHF2 CH2FCF2CHF-O-CH3 CHF2CHF-O-CH2F CH2FCH2CF2-O-CH2F CF3CF2-O-CH3 CH2FCH2CH2-O-CF3 CF3CH2-O-CHF2 CH2FCH2CHF-O-CHF2 CF3CHF-O-CH2F CH2FCHFCF2-O-CH3 CH2FCF2-O-CHF2 CH2FCHFCH2-O-CHF2 CH2FCHF-O-CF3 CH2FCHFCHF-O-CH2F CH3CF2-O-CF3 CH3CF2CF2-O-CH3 CHF2CF2-O-CH2F CH3CF2CH2-O-CHF2 CHF2CH2-O-CF3 CH3CF2CHF-O-CH2F CHF2CHF-O-CHF2 CH3CH2CF2-O-CHF2 198

CH3CH2CHF-O-CF3 CH2FCHFCHFCHF-O-CH3 CH3CHFCF2-O-CH2F CH3CF2CF2CF2-O-CH3 CH3CHFCH2-O-CF3 CH3CF2CF2CH2-O-CH3 CH3CHFCHF-O-CHF2 CH3CF2CF2CHF-O-CH3 CHF2CF2CH2-O-CH3 CH3CF2CHFCF2-O-CH3 CHF2CH2CF2-O-CH3 CH3CF2CHFCH2-O-CH3 CHF2CH2CH2-O-CHF2 CH3CF2CHFCHF-O-CH3 CHF2CH2CHF-O-CH2F CH3CH2CF2CF2-O-CH3 CHF2CHFCH2-O-CH2F CH3CH2CF2CH2-O-CH3 CHF2CHFCHF-O-CH3 CH3CH2CF2CHF-O-CH3 CF3CF2CF2-O-CH2F CH3CH2CHFCF2-O-CH3 CF3CF2CF2-O-CH3 CH3CH2CHFCH2-O-CH3 CF3CF2CH2-O-CH2F CH3CH2CHFCHF-O-CH3 CF3CF2CH2-O-CH3 CH3CHFCF2CF2-O-CH3 CF3CF2CHF-O-CH2F CH3CHFCF2CH2-O-CH3 CF3CF2CHF-O-CH3 CH3CHFCF2CHF-O-CH3 CF3CHFCF2-O-CH2F CH3CHFCHFCF2-O-CH3 CF3CHFCF2-O-CH3 CH3CHFCHFCH2-O-CH3 CF3CHFCHF-O-CH2F CH3CHFCHFCHF-O-CH3 CH2FCF2CF2-O-CH2F CH2FCF2CF2-O-CH3 CHF2CF2CF2-O-CH2F CHF2CF2CF2-O-CH3 CHF2CF2CHF-O-CH2F CHF2CF2CHF-O-CH3 CHF2CHFCF2-O-CH2F CHF2CHFCF2-O-CH3 CHF2CHFCHF-O-CH3 CH2F-O-CF2-O-CF3 CH2F-O-CF2-O-CH2F CH2F-O-CF2-O-CHF2 CH2F-O-CHF-O-CF3 CH2F-O-CHF-O-CH2F CH2F-O-CHF-O-CHF2 CH3-O-CF2-O-CF3 CH3-O-CF2-O-CH2F CH3-O-CF2-O-CH3 CH3-O-CF2-O-CHF2 CH3-O-CHF-O-CF3 CH3-O-CHF-O-CH2F CH3-O-CHF-O-CH3 CH3-O-CHF-O-CHF2 CF3CF2-O-CH2F CF3CF2-O-CH3 CH2FCF2CF2CF2-O-CH3 CH2FCF2CHFCF2-O-CH3 CH2FCH2CF2CH2-O-CH3 CH2FCH2CHFCH2-O-CH3 CH2FCHFCF2CHF-O-CH3 199

)

1

- 1150 m

c 1100 (

r

e 1050

b

m 1000

u

n 950 e

v 900 a 850

W F 3 C F 2 F 2 H C C F H 2 F α C H 3 H C 2 C p 2 G F u C F H 2 ro ro H C G up C H 2 α 1 C O O ge ra ve A Figure A.1: Average absorption wavenumbers for C-F bond absorption wavenumbers for the CHF group in different chemical environments. The α1 groups are indicated by the coloured bars.

APPENDIX B

SUPPORING INFORMATION FOR CHAPTER FIVE:

Perfluorinated Amines: A New Class of Long-Lived Greenhouse Gases

200 201 B.1 Determination of physical properties

Limited information regarding physical properties relevant to the environment is available for PFBAm. The vapour pressure has been measured as 192 Pa and the measured water solubility is less than 5 parts-per-million (1). The octanol-water partitioning coefficient

(KOW) and water solubility were determined using SPARC (2). The air-water partitioning (KAW) coefficient was calculated from the literature value for vapour pressure and the estimated water solubility of PFBAm. The log KAW value of 6.4 determined for PFBAm is unusually high, even for a fluorinated compound. The next highest reported log KAW for a fluorinated compound is 2.8 for hexafluoroethane (3). The vapour pressure of this compound is approximately four orders of magnitude higher than that of PFBAm, but the water solubility is also higher, by almost six orders of magnitude. It appears that the driving force for the unusually high KAW value for PFBAm is the low water solubility. Since the value determined here is from a model that may not be designed to predict properties for highly fluorinated molecules, it is necessary to assess whether this value is realistic. Few measured values of water solubility exist for perfluorinated compounds, presumably due to the inherent difficulty of measuring such low water solubilities. However, solubilities of perfluoroalkanes with five to eight carbons have been measured and recommended values have been published by Horvath and Getzen (4). A relationship can be derived from their data between carbon number (n) and water solubility sat (Cw ):

sat Cw = -1.00n + 2.56

From this, it is possible to extrapolate a water solubility for the linear perfluoroalkane with the same number of carbon atoms as PFBAm (n = 12, perfluorododecane) of 3.6 x 10-10 mol m-3. This value is almost two orders of magnitude smaller than the value calculated by SPARC for PFBAm. This suggests that a water solubility of 3.2 x 10-8 mol m-3 is not unreasonable for

PFBAm and that the unusually high KAW value is legitimate.

B.2 Model Limitations and Sensitivity

In the HR-MS model, input of a log KAW greater than 2 leads to computational difficulties. Increasing log KAW values causes the number of iterations per hour of simulated time to increase exponentially. As a result, the highest log KAW input that allowed a reasonable number of iterations was 2. Because the determined log KAW value of 6.4 could not be used, a

202 sensitivity analysis was performed to determine the effect of using a log KAW value lower than the true value on the fate of PFBAm. Log KAW values from -1 to 2 were entered and sensitivity determined according to the following equation:

Δ result result S = Δ input input

The sensitivity of PFBAm environmental distribution to log KAW value was shown to be less -4 than 10 . Since log KAW was shown to have such little effect, simulations were run with log

KAW values of 0 to minimize computational time.

B.3 Effects of ionization on lifetime of PFBAm

The potential for accumulation of PFBAmH+ in sediment as a loss process for PFBAm was assessed through an emission of PFBAm into air over 10 days, followed by 40 days without any emission. As a conservative estimate, the pKa of PFBAmH+ was assumed to be 2. Accumulation of an extremely small fraction of PFBAmH+ in the sediment was observed for the duration of the simulation. A quadratic regression was fit to the 40-day trend following the cessation of emissions and extrapolated to estimate the effects over longer time periods (Table X.1). Even after one thousand years (assuming no other loss pathways), only 8.4 × 10-12 % of the emitted PFBAm would be found in the sediment.

Table B.1: Fraction of PFBAmH+ (PFBAmH+sed/(ΣPFBAmtot + PFBAmH+tot))accumulated in sediment after the cessation of PFBAm emissions (assuming no other loss pathways).

Time since cessation Fraction of of PFBAm PFBAmH+ emissions (years) accumulated in sediment 1 8.73 × 10-13 2 1.71 × 10-12 5 4.23 × 10-12 10 8.44 × 10-12 50 4.20 × 10-11 100 8.41 × 10-11 1000 8.40 × 10-10 Ionization of PFBAm would only alter the environmental fate of PFBAm if it was present in appreciable quantities and had a significantly different fate. The HR-MS model has

203 shown that PFBAmH+ is only present in extremely small quantities, even when assuming a high pKa value of 2. In addition, PFBAmH+ is present primarily in the air compartment, with small amounts in the water and sediment. Given that the fraction of PFBAmH+ in the water and sediment is small and the fraction of PFBAm in the form of PFBAmH+ is also small, ionization to PFBAmH+ does not have an appreciable effect on the environmental fate of PFBAm.

B.4 Sources Cited (1) 3M. 2000. 3M FluorinertTM Electronic Liquid FC-43 Product Information,

(2) Hilal, S.H.; Carreira, L.A.; Karickhoff, S.W. Prediction of the solubility, activity coefficient, gas/liquid and liquid/liquid distribution coefficients of organic compounds. QSAR Combination Science 2004, 23, 709.

(3) Mackay, D.; Shiu, W.Y.; Ma, K.-C.; Lee, S.C. Handbook of physical-chemical properties and environmental fate for organic chemicals, 2nd Ed.; Taylor & Francis Group: Boca Raton, FL, 2006.

(4) Horvath, A.L.; Getzen, F.W. IUPAC-NIST solubility data series 68. Halogenated aliphatic hydrocarbon compounds C3-C14 with water. Journal of Physical Chemical Reference Data 1999, 28, 649-777.

APPENDIX C

SUPPLEMENTARY INFORMATION FOR CHAPTER SEVEN:

Atmospheric Chemistry of 4:2 Fluorotelomer Iodide (n-C4F9CH2CH2I): Kinetics and Products of Photolysis and Reaction with OH Radicals and Cl Atoms

204 205 Table C.1: Measured UV absorption cross sections for ethyl iodide and 4:2 FTI.

UV Absorption Cross Section UV Absorption Cross Section (10-19 cm2 molecule-1) (10-19 cm2 molecule-1)

Wavelength (nm) Ethyl Iodide 4:2 FTI Wavelength (nm) Ethyl Iodide 4:2 FTI 200 131.27 0.74 260 12.25 11.69 201 102.89 0.74 261 11.62 11.55 202 53.89 0.74 262 11.51 11.68 203 7.83 0.74 263 11.01 11.34 204 0.22 0.74 264 10.80 11.24 205 1.11 1.19 265 10.19 10.70 206 1.11 1.19 266 9.86 10.58 207 1.11 1.19 267 9.30 10.22 208 1.11 1.19 268 8.78 9.80 209 1.11 1.19 269 8.17 9.34 210 0.39 0.23 270 7.83 8.83 211 0.39 0.23 271 7.14 8.53 212 0.39 0.23 272 6.72 7.99 213 0.39 0.23 273 6.27 7.80 214 0.39 0.23 274 5.85 7.07 215 0.26 0.22 275 5.04 6.50 216 0.26 0.22 276 4.93 6.22 217 0.26 0.22 277 4.25 5.40 218 0.26 0.22 278 3.80 4.98 219 0.26 0.22 279 3.40 4.69 220 0.77 0.64 280 3.27 4.54 221 0.42 0.39 281 2.67 3.94 222 0.40 0.19 282 2.67 3.63 223 0.82 0.33 283 2.47 3.45 224 0.66 0.49 284 2.25 3.22 225 0.74 0.40 285 1.86 2.54 226 0.69 0.57 286 1.51 2.08 227 1.14 1.12 287 1.46 2.16 228 0.73 0.42 288 1.40 2.07 229 0.78 0.35 289 1.13 1.71 230 1.10 0.83 290 0.88 1.37 231 1.69 1.45 291 0.74 1.07 232 1.56 1.18 292 0.82 1.44 233 1.96 1.59 293 0.84 1.41 234 2.04 1.50 294 0.65 1.03 235 2.10 1.38 295 0.43 0.96 236 2.54 1.87 296 0.48 0.78 237 2.97 2.32 297 0.47 0.64 238 3.39 2.57 298 0.39 0.62 239 3.61 2.76 299 0.34 0.59 240 3.99 3.04 300 0.18 0.44 241 5.00 3.97 301 0.25 0.32 242 5.23 4.26 302 0.20 0.31 243 5.74 4.63 303 0.29 0.50 244 6.46 5.07 304 0.48 0.15 245 6.87 5.45 305 0.11 0.23 246 7.33 5.96 306 0.11 0.23 247 8.30 6.61 307 0.11 0.23 248 8.74 7.06 308 0.11 0.23 249 9.10 7.40 309 0.11 0.23 250 9.91 8.27 310 0.05 0.13 251 10.36 8.80 311 0.05 0.13 252 11.00 9.47 312 0.05 0.13 253 11.31 9.93 313 0.05 0.13 254 11.65 10.52 314 0.05 0.13 255 11.58 10.39 315 0.11 0.21 256 12.06 11.01 316 0.11 0.21 257 12.08 11.04 317 0.11 0.21 258 12.16 11.48 318 0.11 0.21 259 12.14 11.48 319 0.11 0.21

206 Figure C.1: 24-hour average photolysis rate constants for 4:2 FTI by latitude for winter solstice (Dec 22, 2006), spring equinox (Mar 21, 2007), summer solstice (June 21, 2007) and fall equinox (Sept 23, 2007).

80

60

40

20 Dec 22 Mar 21 0 Jun 21 Sept 23 Latitude -20

-40

-60

-80

0.0 5.0e-6 1.0e-5 1.5e-5 2.0e-5 2.5e-5

24 Hour Averaged Rate Constant

APPENDIX D

SUPPORTING INFORMATION FOR CHAPTER EIGHT:

Overtone-induced degradation of perfluorinated alcohols

207 208

Table D.1: Optimized geometries of ground state and transition state structures for CF3OH and CF3OH•H2O.

‡ ‡ CF3OH [CF3OH] CF3OH•H2O[CF3OH•H2O]

r(C-O1) 1.34526 1.25481 1.32682 1.23247

r(C-F1) 1.34732 1.79229 1.35711 1.71334

r(C-F2) 1.32541 1.29852 1.33069 1.32525

r(C-F3) 1.34732 1.29852 1.35735 1.32597

r(O1-H1) 0.96447 1.18934 0.98472 1.38284

r(O2-H2) 0.96250 1.07475

r(O2-H3) 0.96249 0.96516

r(O2-H1) 1.74084 1.08237

r(F1-H1) 2.44880 1.23647

r(F1-H2) 3.95620 1.33696

∠(O1-C-F1) 112.29 88.33 112.81 107.61

∠(O1-C-F2) 108.58 122.07 109.81 120.13

∠(O1-C-F3) 112.29 122.07 112.81 119.89

∠(C-O1-H1) 109.60 83.89 110.12 112.94

∠(O1-H1-F1) 65.92 125.61

∠(H2-O2-H3) 105.96 110.25

∠(O1-H1-O2) 178.88 153.91

∠(H1-O2-H2) 114.93 90.03

∠(F1-H2-O2) 58.28 153.22 ∠(C-F1-H2) 70.86 102.11

209 Table D.2: Optimized geometries of ground state and transition state structures for CF3CF2OH and CF3CF2OH•H2O.

‡ ‡ CF3CF2OH [CF3CF2OH] CF3CF2OH•H2O[CF3CF2OH•H2O]

r(C1-O1) 1.34432 1.25832 1.32812 1.23689

r(C1-F1) 1.35669 1.77861 1.36674 1.72144

r(C1-F2) 1.35669 1.30933 1.36675 1.33640

r(C1-C2) 1.55582 1.55544 1.55397 1.55194

r(C2-F3) 1.33239 1.32695 1.33384 1.34216

r(C2-F4) 1.33265 1.34148 1.33384 1.32864

r(C2-F5) 1.33265 1.32406 1.33493 1.33419

r(O1-H1) 0.96590 1.20405 0.98695 1.38291

r(O2-H2) 0.96255 1.06332

r(O2-H3) 0.96255 0.96560

r(O2-H1) 1.73335 1.08259

r(F1-H1) 2.44902 1.22451

r(F1-H2) 3.52936 1.36483

∠(O1-C1-F1) 112.04 88.78 112.53 107.97

∠(O1-C1-F2) 112.04 120.05 112.54 118.40

∠(O1-C1-C2) 109.33 123.03 110.30 120.09

∠(F1-C1-C2) 108.58 100.88 107.96 97.03

∠(F2-C1-C2) 108.58 110.92 107.96 109.68

∠(C1-C2-F3) 109.76 111.62 110.40 108.13

∠(C1-C2-F4) 110.29 107.08 110.40 112.92

∠(C1-C2-F5) 110.29 110.73 110.23 110.26

∠(C1-O1-H1) 109.69 83.15 110.39 112.68

∠(O1-H1-F1) 66.14 125.26

∠(H2-O2-H3) 105.98 110.01

∠(O1-H1-O2) 179.66 153.80

∠(H1-O2-H2) 114.84 90.94

∠(F1-H2-O2) 83.47 152.57 ∠(C1-F1-H2) 88.71 101.49

210 Figure D.1: Vector diagrams for the vibrational motion associated with transition states: (a) ‡ ‡ ‡ ‡ [CF3OH] ; (b) [CF3OH•H2O] ; (c) [CF3CF2OH] ; and (d) [CF3CF2OH•H2O] .

ab

cd

APPENDIX E

SUPPORTING INFORMATION FOR CHAPTER NINE:

Perfluorinated acids in arctic snow: New evidence for atmospheric formation

211 212 E.1 Experimental

E.1.1 Chemicals

Potassium perfluorooctane sulfonate PFOS (86.4%) was provided by the 3M Company (St. Paul, MN, USA). Perfluorooctanoic acid (PFOA, 96%), perfluorononanoic acid (PFNA, 97%), perfluorodecanoic acid (PFDA, 98%) and perfluoroundecanoic acid (PFUnA, 95%) were 13 purchased from Sigma-Aldrich (Oakville, ON, Canada). Perfluoro-n-[1,2,3,4- C4]octanoic 13 13 acid, perfluoro-n-[1,2,3,4,5- C5]nonanoic acid, perfluoro-n-[1,2- C2]decanoic acid, sodium 13 13 perfluoro-1-[1,2,3,4- C4]octanesulfonate ( C4-PFOS), were provided by Wellington Laboratories (Guelph, ON, Canada). Methanol (OmniSolv grade, 99.9%) was purchased from EMD Chemicals. HPLC grade water was obtained from Caledon Laboratories (Georgetown, ON, Canada).

E.1.2 Sample Collection

A snow pit was created at or near the highest point on the icecap and 2.2 km upwind from the nearest temporary research site. Duplicate samples were taken horizontally at 25 cm (from surface to 3 m depth) or 20 cm (from 3 m depth to bottom) intervals to a depth of up to 6.8 m using a stainless steel corer of 8.1 cm diameter. Volumes of 0.5 – 1.0 L water equivalents were collected from each depth. Surface samples from all sites consisted of dry snow that had not experienced any melt episodes, which was mixed and placed into bottles using a stainless steel scoop. For all sample collection, clean techniques were used. Products containing fluoropolymer coatings were strictly avoided at the sampling sites. Prior to sampling, the surface layer of the snowpit wall was removed using a stainless steel scraper. Samples were stored in new, unopened polypropylene bottles and were kept frozen or at 4˚C until analysis. Field blanks were taken from an unopened bottle of HPLC grade water, transported in polypropylene bottles and opened for 10 minutes at the sampling location. Densities were measured by collecting a sample of known volume at 10 cm intervals and determining the mass.

E.1.3 Conductivity and Ion Analysis

Samples were taken on the Devon Ice Cap at 10 cm depth intervals using a stainless steel corer 45.30 mm in diameter and 50.65 mm in length to determine conductivity and ion concentrations. Conductivity was measured using an Orion model 135 conductivity meter

213

2- - + (Orion Research, Beverly, MA, USA). Major ions (SO4 , Cl and Na ) were determined by NLET (National Laboratory for Environmental Testing, Environment Canada, National Water Research Institute, Burlington, ON, Canada).

E.1.4 Sample Preparation

Sub-samples of approximately 100 mL were exactly weighed and were concentrated using Oasis HLB solid phase extraction cartridges (3 cc, 60 mg) (Waters, Milford, MA, USA) on a vacuum manifold. Cartridges were conditioned using three column volumes of methanol. Samples were passed through at a rate of approximately 5 mL min-1. Sorbed compounds were eluted using 1 mL of methanol. Method blanks and spike and recoveries were prepared using deep ice core water. Samples were extracted in triplicate, where two of the three sub-samples were taken from one collection replicate and the third sub-sample from the other sample collection replicate. Samples were filtered using Whatman Mini-UniPrep® syringeless polypropylene filters and prepared as 50% methanol and 50% water solutions for high-volume injection into the LC-MS-MS. These mixtures facilitate focusing of the analytes on the head of the column (1). Samples were diluted by a factor of 2.2, with 225 μL of each methanol extract added to 25 μL internal standard mixture in methanol and 250 μL deep ice core water.

E.1.5 Analytical Methods and Data Treatment

Analysis was done using an Agilent 1100 LC, with polytetrafluoroethylene (PTFE) parts removed where possible, with detection by an ABS/Sciex 4000QTrap™ MS/MS detector. Triplicate injections of 100 μL were made and separated on a Genesis C18 column (2.1 mm i.d. x 50 mm, 4μm; Chromatographic Specialties, Brockville, ON, Canada) using a 4 minute isocratic run of 80% methanol and 20% water, both containing 10 mM ammonium acetate (see Supporting Figure 1). Analytes were quantified using isotopically labelled internal standards. PFUnA was quantified only for samples taken in 2006 using isotopically labelled PFDA. Monitored transitions for analytes of interest were the same as those used by Furdui et al (1), adding a second transition for PFOS, 499>80 and 503>80 for native and labelled PFOS, respectively. Although subject to interferences in biological samples (2), this transition gave increased sensitivity with no interference observed for the ice samples analyzed herein. The limits of quantification (LOQ) were defined as the lowest analyte concentration required to give approximately 10 to 1 signal to noise ratio. LOQs for PFOS, PFOA, PFNA, PFDA and PFUnA

214 were 0.5, 1.3, 1.2, 0.8 and 0.6 ng L-1, respectively. These LOQs correspond to sample concentrations of 5, 13, 12, 8 and 6 pg L-1. Values that were above the level of the instrument blank, but below the limit of quantification, were assigned a value of one-half LOQ. Statistical analysis was done using SPSS for Windows 2001 (Chicago, IL, USA).

E.2 QA/QC

E.2.1 Instrumental Contamination and Variability

An isocratic chromatography method was adopted, because it allows for quick analysis of low level PFAs on a system without major modifications. The analytes, particularly PFOA and PFNA, leach at constant level from the fluorinated components of the system and maintain a uniform background (1). Blank injections containing deep ice core water and methanol did not show any detectable levels and thus, did not contribute substantial levels of contamination to the samples (see Supporting Figure 1). At levels close to the limits of the instrument, instrument variability may become a factor in accurate quantification. Variability was minimized by the use of isotopically labelled internal standards, which were available for all analytes except PFUnA. Average variability for replicate injections was acceptable for PFOS (13%), PFOA (7%) and PFNA (12%), but was higher for PFDA (29%) and PFUnA (32%). In order to reduce the impact of this variability, samples were each analyzed in triplicate. No statistical difference was observed between sub-samples taken from the two collection replicates (paired t-test, p > 0.05).

E.2.2 Extraction Quality Control

Any processing of samples may be a source of contamination. A number of procedural blanks were extracted along with samples to determine the level of contamination. These extraction blanks were prepared using deep ice core water, which was observed to have lower contamination than HPLC grade water. However, our analysis indicates that there is still a small background amount of contamination in this water, which was likely acquired following core collection. Extraction blanks were prepared using 25, 50 and 100 mL of deep ice core water. For PFDA, levels were below the LOQ. For those with quantifiable contamination (PFOS, PFOA, PFNA, PFUnA), contamination scaled linearly with blank volume, indicating the majority of the contamination was likely from the water itself, and not a result of the extraction procedure. Extrapolation of these lines gives the contamination incurred from the

215 procedure, which for all analytes was below the LOQ. Sample values were not blank- corrected because of the demonstrated cleanliness of the procedure.

The method efficiency was determined using spike and recoveries (n = 5). Mean recoveries for PFOA, PFNA, PFDA, PFUnA and PFOS were 98±5%, 110±11%, 101±2%, 83±19% and 110±19%, respectively. Sample RSDs for three replicate extractions were variable depending on the analyte. PFOA, PFNA and PFDA had RSDs of 12%, 15% and 17%, respectively. RSDs for PFOS and PFUnA were higher at 26% and 32%, respectively, which could be due to the fact that measured concentrations were closer to the LOQ.

E.2.3 Field Blanks

Due to the low levels of PFAs observed on the ice cap and the ubiquitous nature of these compounds, it is impossible to obtain a true field blank. Instead, concentrations in HPLC water taken to the field were compared with those of HPLC water that had been kept in the original bottle for the duration of the sampling trip. Concentrations in both sample sets for PFNA, PFDA and PFUnA were below the LOQ, eliminating the possibility that any significant contamination was incurred during the sampling. PFOS concentrations were above the LOQ, but were determined to be statistically equivalent (t-test, p = 0.20). This indicates that PFOS contamination was present in the HPLC grade water and was not introduced by the sampling method. Concentrations of PFOA were significantly higher in the water left in the lab than the field blanks (p < 0.001), which is likely due to the presence of a PTFE liner in the packaging of the HPLC grade water. It is difficult to assess whether any PFOA contamination was incurred during sampling and storage, or if the entirety of the field blank levels (mean concentration of 25.5 pg L-1, n = 3) is due to inherent contamination in the HPLC grade water. Our laboratory has observed HPLC grade water to contain detectable levels of PFOA. Thus, any contamination that may have occurred during sampling must have been minor and was significantly less than that incurred by the same material left in its original packaging in the laboratory.

216 Table E.1: QA/QC details for analysis. Number of Replicates PFOA PFNA PFDA PFUnA PFOS LOQ (ng L-1) N/A 1.3 1.2 0.8 0.6 0.5 Recovery (%) 5 98 110 101 83 110 Recovery RSD (%) 5 5 11 2 19 19

Table E.2: Mean (n=3) density-corrected concentrations of PFAs in pg L-1. Numbers in italics represent samples where one or more measurements were below the LOQ. Note that PFUnA was not measured in 2005 samples. Location Sample PFOA PFNA PFDA PFUnA PFOS Devon Depth-0 cm 13.9 5.0 1.5 1.1 4.2 Devon Depth-25 cm 11.9 5.1 1.4 1.0 3.8 Devon Depth-50 cm 147.0 245.5 21.8 16.0 5.4 Devon Depth-75 cm 41.8 48.4 5.0 6.1 7.2 Devon Depth-100 cm 22.1 8.5 1.6 2.4 4.0 Devon Depth-125 cm 118.2 143.0 19.2 9.1 5.6 Devon Depth-150 cm 75.5 234.8 11.9 24.0 5.5 Devon Depth-175 cm 21.5 26.6 2.6 4.1 5.4 Devon Depth-200 cm 27.4 17.5 3.1 3.2 6.8 Devon Depth-225 cm 56.2 82.1 10.5 12.3 2.6 Devon Depth-250 cm 19.2 8.8 3.2 3.7 3.8 Devon Depth-275 cm 18.5 17.5 1.7 6.1 2.7 Devon Depth-300 cm 72.8 122.7 13.2 19.8 3.1 Devon Depth-320 cm 17.8 18.2 1.1 6.0 3.6 Devon Depth-340 cm 60.5 89.5 7.9 26.2 5.5 Devon Depth-360 cm 62.7 49.2 8.0 19.2 10.2 Devon Depth-380 cm 42.1 38.6 2.7 7.9 5.8 Devon Depth-400 cm 52.7 42.2 7.0 17.0 14.9 Devon Depth-420 cm 59.6 81.1 11.5 22.7 9.6 Devon Depth-440 cm 65.1 47.4 7.7 5.4 15.1 Devon Depth-460 cm 74.2 70.1 11.1 8.1 21.6 Devon Depth-480 cm 66.6 66.7 10.2 19.4 19.3 Devon Depth-500 cm 124.2 45.6 7.9 1.5 86.0 Devon Depth-520 cm 93.0 67.0 8.5 11.8 18.5 Devon Depth-540 cm 95.6 41.2 7.9 6.2 34.4 Devon Depth-560 cm 89.8 44.9 3.5 3.2 20.3 Devon Depth-580 cm 116.8 104.3 16.2 27.3 24.8 Devon Depth-600 cm 95.3 59.2 8.2 16.3 17.1 Devon Depth-620 cm 95.1 30.6 2.6 3.0 19.9 Devon Depth-640 cm 46.0 16.4 2.6 1.8 9.8 Devon Depth-660 cm 92.8 43.2 7.8 8.3 27.4 Devon Depth-680 cm 56.8 24.1 4.6 5.1 13.8 Agassiz Surface 2005 13.1 10.0 3.9 N/A 1.4

217

Devon Surface 2005 16.6 9.1 4.2 N/A 4.0 Melville Surface 2005 16.3 9.8 4.5 N/A 2.4 Agassiz Surface 2006 53.7 9.4 2.6 5.1 2.3 Meighen Surface 2006 15.1 12.1 2.2 3.9 1.6 Melville Surface 2006 38.6 7.6 1.6 2.8 4.6

85°

80°

75°

70°

65°

60° -60° -130° -120° -70° -110° -100° -90° -80°

Figure E.1: Sampling site locations: Melville Ice Cap (red), Meighen Ice Cap (yellow), Agassiz Ice Cap (green) and Devon Ice Cap (blue).

218 a) 2500

PFOA 2000 PFNA

PFDA 1500 PFUnA PFOS

1000

Instrument Response 500

0 01234

Time (min)

b) 2500

PFOA 2000 PFNA PFDA

1500 PFUnA PFOS

1000

Instrument Response 500

0 01234 Time (min)

Figure E.2: LC-MS-MS chromatograms for analysis of PFAs: (a) Instrument blank and (b) 2.5 ppt standard.

219 a) b)

0 0 2005

2004

200 2003 200 2002 + - 2001 Ca

Cl Year 2+ 2- 2000 Mg 400 SO 400

Depth (cm) Depth 4 + 1999 (cm) Depth Na + 1998 K

600 1997 600 1996

0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.02 0.04 0.06 0.08 Concentration (ug cm-3) Concentration (ug cm-3) c) d) 0 0

200 200

Depth (cm) 400

Depth (cm) 400

600 600

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.2 0.3 0.4 0.5 0.6 0.7 -3 Conductivity (μS cm-1) Density (g cm )

Figure E.3: Information for dating on the Devon Ice Cap: (a) Major anions; (b) major cations; (c) conductivity and (d) density.

220

80

85o ng PFA m-2

o 80 40 Meighen Ice Cap Agassiz Ice Cap 75o

Melville o Devon Ice Cap 70 Ice Cap 0 65o PFOA PFNA PFDA 60o PFUnA PFOS -60o -130o -120o -70o -110o -80o -100o -90o

Figure E.4: Fluxes of PFAs to ice caps in the High Arctic for 2005.

221

PFOA PFNA 9000 14000

8000 m = -174.7 12000 7000 r ² = 0.08 m = 530.6 10000 ) ) r ² = 0.36 -2 6000 -2 8000 5000 6000 4000 Flux (fg cm Flux (fg cm Flux (fg 3000 4000

2000 2000

1000 0 1994 1996 1998 2000 2002 2004 2006 1994 1996 1998 2000 2002 2004 2006 Year Year

PFDA PFUnA 1200 1600 m = 16.8 1400 1000 m = 26.7 r ² = 0.03 r ² = 0.17 1200 ) 800 ) -2 -2 1000 m c 600 800

600

Flux (fg Flux 400 Flux (fg cm 400 200 200

0 0 1994 1996 1998 2000 2002 2004 2006 1994 1996 1998 2000 2002 2004 2006 Year Year PFOS PFOS-Divided 3000 3000

2500 2500 m = -174.5 m = 879.3 m = -769.0 r ² = 0.43 r ² = 0.94 2000 2000 r ² = 0.84 ) ) -2 -2 1500 cm 1500 fg fg 1000 1000 m = 41.6 r ² = 0.20 Flux (fg cm Flux ( Flux 500 500

0 0

1994 1996 1998 2000 2002 2004 2006 1994 1996 1998 2000 2002 2004 2006 Year Year

Figure E.5: Linear regressions for PFAs between 1996 and 2005.

222 a) 0 2005

2004 200 2003 2002 2001 Year 400 2000

(cm) Depth 1999 1998

600 1997 1996

0 20406080 10-4 Concentration ratio of Na+ to PFOA b) 350

) -3 300 m = -769 r ² = 0.016 250

200 150

100

50 PFOA Concentration (fg cm 0 0.00 0.01 0.02 0.03 0.04 0.05 0.06 + -3 Na Concentration (μg cm )

Figure E.6: (a) concentration ratio of sodium to PFOA with depth on the Devon Ice Cap and (b) correlation between PFOA and sodium concentrations on Devon Ice Cap.

223

300 100

250 snow) m = 9.34 snow) 80 -1 r ² = 0.73 -1 200 m = -0.02 60 r ² = 0.004 150 40 100

20 50

PFNA concentration (pg L 0 PFNA concentration (pg L 0 0 5 10 15 20 25 0 50 100 150 200 250 300 PFDA concentration (pg L-1 snow) PFOS concentration (pg L-1 snow)

100

snow) 80 -1

60 m = 0.24 r ² = 0.30

40

20

PFOA concentration (pg L 0 0 20406080100120140160 PFOS concentration (pg L-1 snow)

Figure E.7: Correlations between PFA concentrations on Devon Ice Cap.

E.3 Sources Cited (1) Furdui, V. I.; Crozier, P. W.; Reiner, E. J.; Mabury, S. A. Organohalogen Compounds 2006, 211-214.

(2) Hansen, K. J.; Clemen, L. A.; Ellefson, M. E.; Johnson, H. O. Environmental Science and Technology 2001, 35, 766-770.