Perfluoropolyethers. Analytical Method Development for a New Class of Compounds with the Potential to be Long- Lived Environmental Contaminants

Transcription

1 Perfluoropolyethers Analytical Method Development for a New Class of Compounds with the Potential to be Long- Lived Environmental Contaminants By Robert Anthony Di Lorenzo A thesis submitted in conformity with the requirements for the degree of Master of Science Graduate Department of Chemistry University of Toronto Copyright by Robert Anthony Di Lorenzo (2012)

2 Perfluoropolyethers: Analytical Method Development for a New Class of Compounds with the Potential to be Long-Lived Environmental Contaminants Robert Anthony Di Lorenzo Master of Science Graduate Department of Chemistry University of Toronto 2012 i Abstract Perfluoropolyethers (PFPEs) are used in a remarkably large number of industrial applications including thin-film lubricants, greases, heat transfer fluids, cosmetics, and EPA-approved food contact paper coatings and are marketed for their chemical inertness. Although desired industrially, it is also the property of most environmental concern. The lack of literature concerning the environmental impact of these compounds suggests a need to assess and characterize their environmental fate and transport. This work describes efforts to develop methods to characterize, identify and quantify various congeners of PFPEs through chromatographic, mass spectral and nuclear magnetic resonance techniques. The PFPEs exhibited unusual behavior during ionization by ESI, suggesting the possibility of structural lability during analysis. A preliminary assessment of the environmental degradation of a PFPE-phosphate congener is also described, which showed rapid sorption to sewage sludge particulate matter and the possible presence of multiple PFPEs present in the technical product mixture used for analysis. ii

3 ii Acknowledgements First and foremost, I would like to thank my supervisor, Professor Scott Mabury, for allowing me the opportunity to mature in a laboratory and institution with a history of harboring successful chemists, researchers and intellectuals. His passion for research is highly contagious and remarkably encouraging. I would also like to thank Professor Rebecca Jockusch and Dr. Eric Reiner for serving on my committee and providing valuable feedback throughout the course of my research. I would like to express my gratitude for the guidance, both academic and practical, provided by the members of the Mabury group. A special thanks goes out to Dr. Barbara Weiner, Derek Jackson and Keegan Rankin for their numerous conversations and their calm ear while I vented my frustrations. My sincere appreciation goes out to the rest of the Department of Chemistry for keeping me sane through endeavours outside of the laboratory. To my family and friends, thank you for putting up with me over the past couple of years. I know I could be quite irritable at times, but I hope I made up for it with the copious amounts of bacon. iii

4 iii i ii iv v vi vii Table of Contents Abstract... ii Acknowledgements... iii List of Figures... vi List of Tables... vi List of Appendices... vii List of Abbreviations... vii 1 Introduction Introduction to Perfluoroalkyl Compounds Introduction to Perfluoropolyethers PFPE Synthesis PFPE Applications Prior Analysis and Motivation Characterization of PFPE-diol and PFPE-diphosphate F NMR Fingerprint Spectra Average Molecular Weight Determination MS-Infusion by ESI-QqQ and ESI-qTOF Conditions and Sample Preparation Fingerprint Spectra Thoughts and Considerations with Molecular Weight Distributions MALDI-TOF Conditions and Sample Preparation Molecular Weight Distribution HPLC-ESI-QqQ MRM Method Motivation and Approach Analysis of PFPE-diol Conditions and Sample Preparation Evaluation of Method Analysis of PFPE-diphosphate Conditions and Sample Preparation Evaluation of Method Analysis of PFPE-carboxylate iv

5 Conditions and Sample Preparation Evaluation of Method LC-ESI-QqQ Parent Ion Method for PFPE-diphosphate Approach Conditions and Sample Preparation Evaluation of Method Waste Water Treatment Plant Biodegradation Study Motivation Method Experimental Design Extraction Method Results F NMR LC-ESI-QqQ MRM Revisited Polymer Solubility Conclusions and Future Directions Non-quadrupole MS Method Potential Data Dependant Acquisition Analysis Appendix Literature Cited v

6 iv List of Figures Figure 1: Schematic of Industrial Fluorotelomerization Process... 3 Figure 2: Classes of Fluorinated Polymers... 4 Figure 3: General Structure of a) branched and b) linear PFPEs Figure 4: Polymerization Pathway for the Synthesis of the Peroxidic PFPE Figure 5: Reduction of a linear Peroxidic PFPE to form a PFPE capable of functionalization Figure 6: Various forms of PFPEs with their marketed trade names Figure 7: Structures of analytes of interest. p/q ratios typically range from 0.5-3, where n=1 or 2 for the Fomblin HC/P Figure 8: 19F NMR Spectrum of Fluorolink-D Figure 9: 19F NMR Spectrum of Fomblin HC/P Figure 10: ESI-QqQ Infusion Spectrum of a PFPE-diol Figure 11: ESI-QqQ Infusion Spectrum of a PFPE-diphosphate Figure 12: ESI-QqQ Infusion Spectrum of a PFPE-diphosphate, zoomed to show peak spacing Figure 13: MALDI-TOF spectrum of the PFPE-diol showing characteristic peak spacing and a molecular weight distribution higher than that obtained via ESI methods Figure 14: MALDI-TOF spectrum of the PFPE-diphosphate showing characteristic peak spacing and a molecular weight distribution higher than that obtained via ESI methods Figure 15: Sample Chromatograms for the separation of a PFPE-diol Figure 16: Selected chromatograms showing no elution detected for chromatographic separation of the PFPE-diphosphate Figure 17: Structure of perfluoro-2,5-dimethyl-3,6-dioxanoic acid, or PFPEC Figure 18: ESI-QqQ infusion spectrum of perfluoro-2,5-dimethyl-3,6-dioxanoic acid showing weak signal from the parent ion and a strong signal from a possible fragment Figure 19: Parent Ion Scan for Parents of 190m/z of the PFPE-diphosphate Figure 20: Degradation pathway of PAPs to their corresponding PFCAs Figure 21: Proposed degradation pathway of the PFPE-diphosphate to a novel PFCA Figure 22: Schematic of the Purge and Trap experimental design Figure 23: TICs of selected samples and standards of the biodegradation study Figure 24: 100ppm extracted from various media in an attempt to enhance aqueous solubility and prevent sorption to particulate matter v List of Tables Table 1: Calculation of normalized peak area for Fomblin Table 2: ESI-QqQ Infusion Parameters Table 3: ESI Source Parameters for HPLC method for PFPE-diol Table 4: MRM Parameters for HPLC Method for PFPE-diol Table 5: HPLC gradient elution parameters for PFPE-diol Table 6: Solvent and ion-pairing combinations for HPLC separation of PFPE-diphosphate Table 7: Optimized Source parameters for the chromatography of PFPEC Table 8:MRM Parameters for HPLC Method for PFPEC vi

7 Table 9:HPLC gradient elution parameters for PFPEC Table 10: Source parameters for parent-ion scan analysis of PFPE-diphosphate vi List of Appendices 5.1 ESI-qTOF infusion spectrum of the PFPE-diol 5.2 ESI-qTOF infusion spectrum of the PFPE-diphosphate 5.3 Bottle Contents for Waste Water Treatment Plant Biodegradation Study vii List of Abbreviations % Percent amu Atomic Mass Units CFC Chlorofluorocarbon CID Collision Induced Dissociation Da Daltons DDA Data Dependant Acquisition ECF Electrochemical Fluorination EPA United States Environmental Protection Agency ESI Electrospray Ionization ev Electron Volts FDA United States Food and Drug Administration FTICR Fourier Transform Ion Cyclotron Resonance FTOH Fluorotelomer Alcohol GC Gas Chromatography GPC Gel Permeation Chromatography h Hours HPLC High Performance Liquid Chromatography IR Infrared kv Kilovolts LC Liquid Chromatography LTQ Linear Trap Quadrupole m/z Mass to Charge Ratio MALDI Matrix Assisted Laser Desorption Ionization MeOH Methanol mg Milligram MHz Megahertz ml Millilitre mm Millimeter vii

8 mm MRM MS nm NMR PAPs PFCA PFOA PFOS PFPE PFPEC PFPMIE PFSA ppb ppm PTFE QqQ qtof s SIMS TIC TOF UV V μl μm Millimolar Multiple Reaction Monitoring Mass Spectrometry Nanometer Nuclear Magnetic Resonance Polyfluoroalkyl Phosphates Perfluoroalkyl carboxylic acids Perfluorooctanoic acid Perfluorooctane sulfonic acid Perfluoropolyether Perfluoropolyether Carboxylate, 2,5-dimethyl-3,6-dioxanonanoic acid Perfluoroprolymethylisopropylether Perfluoroalkyl sulfonic acids parts per billion parts per million Polytetrafluoroethylene Triple Quadrupole Quadrupole-Time of Flight Seconds Secondary Ion Mass Spectrometry Total Ion Current Time of Flight Ultraviolet Volts Microlitre Micrometer (Micron) viii

9 1 Introduction 1.1 Introduction to Perfluoroalkyl Compounds Per- and polyfluoroalkyl compounds have long been used in a wide variety of commercial applications for their many advantageous properties. Due to an extremely strong C-F bond, compounds containing a perfluoroalkyl moiety, CF3(CF2)x-R, are generally thermally stable and chemically inert. 1 In addition, the fluorinated alkyl chains are both hydrophobic and lipophobic, giving rise to their widespread use in surface treatments as surfactants and stain-repelling coatings. 1 These fluorinated chains are used alone, usually if terminally functionalized, as polymer additives or polymer side chains. Although chemical inertness is desired in industrial and commercial products, it raises an issue from an environmental standpoint: if compounds are chemically inert, they have the potential to be long-lived environmental contaminants. Since one of the first discoveries of organic fluorine being confirmed in human sera by Taves and coworkers in the late 1960s, there have been a suite of studies aimed at assessing routes of exposure to organic fluorine. 2 More recently, with the discovery of the ubiquitous nature of perfluorooctanoic acid (PFOA) and perfluorooctane sulfonic acid (PFOS), researchers continue to assess contamination sources of these specific perfluorinated compounds. 3,4 The levels of PFOA, PFOS and other perfluoroalkyl carboxylic acids (PFCAs) and perfluoroalkyl sulfonic acids (PFSAs) observed worldwide cannot be entirely attributed to the production levels and partitioning properties of these compounds. Due to their potential toxicity, PFOA and PFOS have been phased out of production, yet concentrations of these compounds have not declined as significantly as one would suggest from the reduced direct exposure. 5 It has 1

10 been of recent importance to determine secondary, indirect exposure routes of PFOA and other PFCAs, such as from volatile precursors, and to determine their degradation pathways. 6 There are two major methods by which traditional fluorinated alkyl compounds are made: electrochemical fluorination, ECF, and telomerization. ECF is a process by which an alkylated analogue of the desired fluorinated compound is dissolved in a hydrofluoric acid and subsequently subjected to electrolysis. In this way, all of the hydrogens in the compound are replaced by fluorines. The process is radical in nature such that isomeric rearrangements are very common. 7 The discovery of perfluorinated compound with multiple branched and linear isomers, easily determined by GC and NMR, is characteristic of a compound synthesized by ECF. Telomerization is a process by which the linearity or branched nature of the compound can be controlled. 8 The process used in industry to synthesize fluorotelomer compounds can be seen in Figure 1. A telogen, in this case pentafluoroethyl iodide, is reacted with an olefin or taxogen, in this case, tetrafluoroethylene, to yield an iodo-terminated oligomer. This is then reacted with ethylene, which is inserted between the iodide terminus and the fluoroalkyl chain. These can be then further functionalized to modify the properties to a desired state for a given application. In recent years, it has been discovered by a variety of researchers that fluorotelomer compounds will degrade into PFCAs through routes including atmospheric oxidation, and microbial and bio-degradation. 6,9,10,11 For this reason, companies have begun to phase out the PFCA precursors and have looked to new compounds to perform similar tasks. 12 2

11 Figure 1: Schematic of Industrial Fluorotelomerization Process Fluorinated organics can also take the form of fluorinated polymers. This classification can be broken down into three sub categories as outlined in figure 2. The first classification is Fluoropolymers. These are compounds synthesized by the polymerization of a fluorinated olefin, such that the carbon-only backbone is directly bonded to the fluorine atoms. 13 The primary example of this type of compound is polytetrafluoroethylene (PTFE), more commonly known under the trade name of Teflon. The second class of fluorinated polymer are Side-Chain-Fluorinated Polymers and are characterized by an aliphatic polymer backbone containing branched chains of fluorinated, aliphatic and other various repeating units. 13 The final classification of fluorinated polymers, and the focus of this thesis, are Perfluoropolyethers, PFPEs. 3

12 Figure 2: Classes of Fluorinated Polymers 1.2 Introduction to Perfluoropolyethers Perfluoropolyethers (PFPEs) are currently being produced by a suite of companies including DuPont, Solvay Solexis, and 3M as alternatives to some of the fluorinated polymers described above. 13 The general structure for branched and linear PFPEs can be seen in figure 3. They are polymeric species containing repeating perfluorinated methyl-, and ethyl- or isopropyl-ether units and can have a variety of application specific end-groups. Figure 3: General Structure of a) branched and b) linear PFPEs. X=F, H, CH2OH, alkyl phosphate, alkyl carboxylate etc. 4

13 1.3 PFPE Synthesis Unlike the majority of the compounds mentioned above, PFPEs are not synthesized by either ECF or flurorotelomer based methods. Instead, they are synthesized by a photoinitiated oxidative polymerization method patented by Solvay Solexis. 14 This polymerization process of a linear PFPE is outlined in figure If a linear PFPE is desired, the polymerization is performed with tetrafluoroethylene in the presence of a controlled partial pressure of O2. If a branched PFPE is desired, the reaction is typically performed with hexafluoropropylene in the presence of a controlled partial pressure of O2. The diagram below outlines the synthesis of a linear congener, but the branched mechanism is analogous. The polymerization is initiated by irradiating an allyl acyl fluoride with UV light ( nm) at -40 C to produce an acyl fluoride radical. Since this process is performed in the presence of oxygen, an O2 molecule will then attach to the acyl radical and propagation will begin. Polymeric propagation can occur in three ways. First, the fluorinated olefin can add to either a terminal alkoxy- or peroxy radial and propagate to form a carbon centered radical. Second, oxygen can add to terminal carbon centered radicals and propagate to form a terminal peroxy radical. Finally, two terminal peroxy radicals can couple to result in two alkoxy radicals and the release of O2. Methoxy units are incorporated in the final polymer through the random β-scission of ethoxy radicals to produce methoxy radicals and COF2. The reaction is terminated through the random coupling of alkoxy radicals to produce interior peroxy units. This primary polymer has been termed the peroxidic perfluoropolyether, whose peroxy units must be reduced to produce the final PFPE. 5

14 Figure 4: Polymerization Pathway for the Synthesis of the Peroxidic PFPE. The process to reduce the peroxidic PFPE to produce a form that has the capability of being functionalized with a suite of functional groups is outlined in figure 5. The first step is an overview of the polymerization process described earlier. In the second step, the peroxidic PFPE is treated with methanol and hydrogen iodide at -40 C, such that all peroxy units are cleaved resulting in polymers with acid and methyl-ester termini. Careful control of the relative concentrations of allyl acyl fluoride initiator, oxygen and perfluorinated olefin can lead to a precise tuning of the average molecular weight of the final polymer. Reduced concentration of initiator, relative to that of oxygen and olefin will lead to a higher molecular weight peroxidic PFPE, but will slow the overall synthesis. By reducing the concentration of oxygen relative to the perfluorinated olefin in the reaction chamber, the amount of peroxy units in the peroxidic PFPE will be decreased, thus will increase the molecular weight of the final PFPE upon cleavage of the peroxy linkages. 14 Typical average 6

15 molecular weights for the end product PFPEs synthesized by this method range from approximately 500Da, for products such as heat transfer fluids to Da+ for high performance lubricants with a ratio of perfluoromethylether to perfluoroethylether groups, denoted in the figure as p and q respectively, between Figure 5: Reduction of a linear Peroxidic PFPE to form a PFPE capable of functionalization. 1.4 PFPE Applications PFPEs are generally marketed as highly thermally stable, electrically resistant, nonflammable, non-ozone depleting and chemically inert, but it is their minor variation in internal and, more importantly, termini structure that dictates their specific product application. Unfunctionalized polymers, those with either an F or H at their terminus, are used either as heat transfer fluids replacing CFCs, if they are of low molecular weight, sub-1000 amu, or as high-performance lubricants, if they are of higher molecular weight, usually greater than 5000 amu. PFPEs differ from traditional fluoropolymers in that they tend to be liquids even at high molecular weights due to the extra degree of freedom imparted by the 7

16 oxygen between the fluorinated repeating units. As such, high molecular weight PFPEs are well suited for lubrication under extremely high temperature applications, such as the high-performance automotive and aerospace industries. Alcohol functionalized PFPEs are commonly used as hard disk drive coatings, protecting the magnetic material from friction and surface scratches; as polymer additives and processing aids, just as PFOA was used in the processing of Teflon; and as precursors to additional functionalization. Carboxylate and phosphate terminal PFPEs are both marketed by Solvay Solexis as FDA-approved food grade grease proofing agents under their Solvera brand name, while the same phosphate based PFPE is marketed as a mild exfoliant and texture modifier in cosmetics under its Fomblin HC brand name. All of these PFPEs as well as their common trade names can be seen below in figure 6. Figure 6: Various forms of PFPEs with their marketed trade names. 8

17 1.5 Prior Analysis and Motivation Aside from one study performed by Young, et al., 16 the main focus in the study of PFPEs are their tribological properties, of engineering nature. 17 That is to say, only the performance properties of the polymer, such as viscosity, have been explored after the polymer has been subject to mechanical degradation, for example in the piston of an engine or on the surface of a disk drive. In the study performed by Young et al., 16 a low molecular weight, non-functionalized, branched PFPE, perfluoropolymethylisopropylether (PFPMIE) was analyzed to determine its atmospheric lifetime and global warming potential. The study indicated the compound was inefficiently oxidized by Cl and OH radicals, leading to a lower-limit atmospheric lifetime of roughly 800 years. Combined with its strong absorption of IR radiation in the atmospheric window, it was determined that this PFPE would have a global warming potential on the 100 year time scale of 9000x that of CO2. This strongly indicates a need to assess additional environmental properties of additional PFPEs. The current methods for the analysis of PFPEs are quite different from those employed to analyze fluorotelomer based small molecules and their metabolites. Where LC-ESI-MS/MS and GC-MS methods are used to analyze the fluorotelomers, current analytical methods for PFPEs are either surface-based mass spectral techniques, or NMR. 18,19,20 This primarily stems from the specific information required from analysis of these two sets of compounds. Where fluorotelomer compounds are being analyzed to determine quantitative measurements of concentrations in various media, PFPEs are being analyzed predominantly for qualitative characterization purposes. One needs the method for analysis of PFPEs to be tolerant of a distribution of molecular weights. PFPEs will not be seen at a single m/z ratio in a mass spectrum, but rather at a range of m/zs, hence a time- 9

18 of-flight (TOF) MS analyzer is commonly employed. Matrix Assisted Laser Desorption Ionization (MALDI) and Secondary Ion Mass Spectrometry (SIMS) are commonly coupled sources to the TOF analyzer due to their soft nature and ability to allow high molecular weight compounds to be promoted to the gas phase. One can also gain insight into the molecular weight average of a PFPE by NMR, by determining the ratio of peak areas of fluorines within repeating units and fluorines on terminal, non-repeating units. By NMR, one can also determine whether the polymer is branched or linear in nature through the different chemical shifts of the fluorines in the repeating units. 20 Unfortunately, it is thought to be difficult to determine the structure end-groups of the PFPEs by NMR, since 13 C, 1 H NMR spectra would be highly contested. It has been shown that unfunctionalized PFPEs have high yields in both positive and negative modes of TOF-SIMS analysis. 21 This can be beneficial for the confirmation of the presence of these types of compounds in complex matrices. PFPEs are also readily ionized by MALDI techniques in positive mode, due to their relatively high mass and affinity to form adducts with lithium cations. 19 MALDI minimizes the possibility of source fragmentation and generally produces singly charged ions such that spectra are relatively easy to interpret. A characteristic of PFPEs, as well as all polymers and oligomers, is a distribution of molecular weights in such that peak separations in mass spectra allow the determination of the nature of the repeating units present in the polymer. Hence, in a singly charged non-fragmented spectrum of any PFPE, one would expect to see peak separations corresponding to methyl- and ethyl- or isopropyl-ether repeating units. This will be explored more in the following chapter. Although characteristic peak separations for PFPE type polymers can be obtained using the mass spectral methods described above, it is 10

19 difficult to gain insight into end-group identity, as well. This requires tandem mass spectrometry, which is rarely hyphenated with MALDI and SIMS ionization although becoming more common, or another orthogonal analysis method. Due to the surface nature of the methods presented above, it is extremely challenging to gain a quantitative picture of the amount of polymer present in a sample since one would need to obtain a perfectly uniform distribution of the polymer on the analysis plate, which is difficult, if not, impossible to obtain. Moreover, these methods are not nearly sensitive enough to detect the concentrations that would be found in environmental samples without sample concentration or extended analysis time. PFPEs have been touted as replacements for fluorotelomer compounds since their short ether linkages prevent them from degrading into long chain PFCAs, such as PFOA. But it is clear that PFPEs still have one major property, and selling point, of environmental concern: their chemical inertness. It is clear that the gap in the literature on the environmental impact if a wide variety of PFPEs needs to be filled, and it must begin with developing a method for identifying and quantifying various PFPE analogues. The work herein will focus on the analysis and characterization of two functionalized, linear PFPEs: a PFPE-diol, Fluorolink-D, and a PFPE-diphosphate, Fomblin HC/P2-1000, both shown in figure 7, below. Figure 7: Structures of analytes of interest. p/q ratios typically range from 0.5-3, where n=1 or 2 for the Fomblin HC/P

20 2 Characterization of PFPE-diol and PFPE-diphosphate F NMR All spectra obtained in this section were acquired on a Varian Mercury 400MHz NMR for 64 scans with a relaxation delay of 0.5 s. Approximately 1 mg of the samples were prepared in and locked to approximately 700 μl d4-methanol. The Fomblin HCP2-1000, described below as the PFPE-diphosphate, was acquired from Arch Personal Care Products and the Fluorolink D, described below as the PFPE-diol, was acquired from Alfa Aesar Fingerprint Spectra Although PFPEs are very complex polymers by nature with multiple repeating units and a wide mass distribution, the 19 F NMR spectra collapse down to a few simple peaks that can be easily assigned to the multiple different repeating units present in each PFPE. Each spectrum contains three distinct regions, an interior methylene oxide region from -50ppm to -60ppm, an interior ethylene oxide region from -90 ppm to -95 ppm and a terminal CF region from -75 ppm to -89 ppm. 22 The 19 F NMR spectra for both the Fluorolink D and Fomblin can be seen in figures 8 and 9 respectively. The multiple peaks in each region arise from the different combination of neighbouring repeating units. In the methylene oxide region, the furthest downfield peak corresponds to a -CF2O- unit neighbouring a -CF2CF2Ounit on both sides. The middle peak in the methylene oxide region corresponds to a -CF2Ounit neighbouring a -CF2O- on one side and a -CF2CF2O- on the other. Finally, the furthest upfield peak in the methylene oxide region corresponds to a -CF2O- neighbouring a -CF2O- unit on both sides. In the ethylene oxide region, the downfield peak corresponds to the CF2 12

21 group nearest a CF2O- unit, whereas an upfield peak will correspond to the CF2 group nearest a -CF2CF2O- unit. For example, in the following sequence CF2OCF2CF2OCF2CF2O-, the first bolded CF2 group will be downfield of the second bolded CF2 group. In the terminal region, the downfield peak corresponds to a CF2 neighbouring a -CF2O- in addition to the terminus, and the upfield peak corresponds to a CF2 group neighbouring a -CF2CF2O- unit. Figure 8: 19F NMR Spectrum of Fluorolink-D 13

22 Figure 9: 19F NMR Spectrum of Fomblin HC/P It is worthwhile to note that although the interior repeating unit peaks, i.e. the methylene oxide and ethylene oxide regions, are fairly insensitive to the nature of the termini of the PFPE, the terminal CF2 region is highly sensitive to a change in end group. Comparing the two spectra above, a change in end group from a diol to an alkyl phosphate yielded an unambiguous downfield shift of ~2 ppm of the termini region of the spectra, while the other regions remained mainly unchanged. This shows some potential for end group determination of unknown PFPEs, or a way to quantify changes during a degradation study. Unfortunately, this is the least sensitive area of the spectrum as it only integrates to four total fluorines for each iteration of the polymer. 14

23 2.1.3 Average Molecular Weight Determination Since each region in the NMR spectrum is clearly defined and the terminal region is known to integrate to four fluorines, one can use the terminal region to normalize the peak signal and determine the number average molecular weight of the polymer. 22 First, the largest peak in the spectrum was integrated to 100 units. This is typically one of the peaks in the ethylene oxide region. Each peak in the region is then summed and divided by the number of fluorine atoms to which the region corresponds to give a fluorine normalized peak area, a i. Table 1 below shows a sample calculation for the PFPEdiphosphate, Fomblin using the following equation a i Peak Area region # of Corresponding Fluorines Responding Signal Δ (ppm) Peak Area # of Corresponding Fluorines a i CF 2O- -CF 2CF 2O- CF 2O-Terminus Table 1: Calculation of normalized peak area for Fomblin 15

24 One can then determine the number concentration of each of the internal repeating units by dividing by half the normalized area of the CF2O-Terminus since there are two termini: ai CF O 2 # CF O endgroup ai CF CF O 2 2 # CF CF O a i endgroup a i 7.26 Now that the average number concentration of each repeating unit is known, the molecular weights of each unit can be used to calculate an average mass of the entire polymer. Using this method, the number average molecular weight for the PFPEdiphosphate, Fomblin, is calculated to be ~1800 amu, which is roughly the average molecular weight as reported by Solvay Solexis on their technical data sheets of 2000 amu. In the same way, the PFPE-diol, Fluorolink, has a calculated number average molecular weight of ~2700 amu, which is much higher than the molecular weight as reported by Solvay Solexis on their technical data sheets of 1000 amu. The number average molecular weights calculated herein can differ from those reported due to the fact that other species may be present in the technical products analyzed and are overlapping with the methylene oxide and ethylene oxide regions of the NMR spectra, yielding positively biased calculated number average molecular weights. Also, there is no information as to how the average molecular weights were determined by the producing company. These could have been calculated by GPC, MALDI-TOF spectra, ESI-MS infusion or similarly by NMR before the technical mixture was made. 16

25 2.2 MS-Infusion by ESI-QqQ and ESI-qTOF Conditions and Sample Preparation All ESI-QqQ spectra were acquired on a Waters Quattro Micro Triple Quadrupole Mass Spectrometer in negative mode. ESI-qTOF spectra were obtained on a AB/Sciex QStarXL qtof Mass Spectrometer in negative mode and acquired by Dr. Matthew Forbes, Laboratory Director of the Advanced Instrumentation for Molecular Structure (AIMS), University of Toronto. All samples were prepared in commercially available HPLC grade methanol to a concentration of 10 μg/ml for ESI-QqQ. Infusion was performed at a rate of 50 μl/min for ESI-QqQ. The methanol was used without further purification. The Fomblin HCP2-1000, described below as the PFPE-diphosphate, was acquired from Arch Personal Care Products and the Fluorolink D, described below as the PFPE-diol, was acquired from Alfa Aesar. Optimized conditions for ESI-QqQ infusions for both compounds are listed in the table below. Parameter PFPE-diphosphate PFPE-diol Capillary Voltage (kv) Cone Voltage (V) Extractor Voltage (V) RF Lens Voltage (V) Source Temperature ( C) Desolvation Temp. ( C) Cone Gas Flow (L/hr) OFF OFF Desolvation Gas Flow (L/hr) Table 2: ESI-QqQ Infusion Parameters 17

26 2.2.2 Fingerprint Spectra Although the spectra obtained by MS infusions are quite a bit more complex than those obtained by NMR, one can still elucidate fingerprints, or characteristic details of the spectra, that are indicative of a PFPE. In particular, the peak spacing between iterations of the polymer will be characteristic of PFPEs. Figure 10 below shows the ESI-QqQ infusion spectrum for the PFPE-diol, Fluorolink. The characteristic peak spacings of 116 m/z, 66 m/z, 50 m/z and 16 m/z correspond directly to the CF2CF2O-, -CF2O-, -CF2-, and O repeating units respectively. In this way, without having any high-resolution data of isotope peaks, one can confidently state that this is a predominantly singly charged spectrum. The ESI-qTOF spectrum shows similar information and confirms the singly charged spectrum. It can be seen in the appendix. Figure 10: ESI-QqQ Infusion Spectrum of a PFPE-diol 18

27 It is worthwhile to note that the apparent molecular weight distribution is significantly lower than both the calculated number average molecular weight (2700 amu), shown above, and the reported average molecular weight on the technical data sheets (1000 amu). Reasons for this are discussed in section Figure 11 below shows the ESI-QqQ infusion spectrum for the PFPE-diphosphate, Fomblin. The characteristic peak spacings, in this case, are different from the PFPE-diol spectrum, above. Here we see peak spacings of 58 m/z, 33 m/z, 25 m/z, and 8 m/z corresponding to the CF2CF2O-, -CF2O-, -CF2-, and O repeating units respectively. In this case, the peak spacings are halved which suggests this is a doubly charged region. The complexity of the spectrum is greater for the PFPE-diphosphate due to the additional aliphatic ethylene oxide repeating units nearing the termini of the structure. According to technical the technical data sheets for Fomblin HC/P2-1000, the aliphatic repeating units have n=1 or 2. The ESI-qTOF spectrum shows similar information and confirms the doubly charged spectrum through both the peak spacings and isotope distributions. It, also, can be seen in the appendix. It is worthwhile to note that the molecular weight distribution is lower than both the calculated number average molecular weight (1800 amu), shown above, and the reported molecular weight on the data sheets (2000 amu), even when accounting for the doubly charged nature of the spectrum. Reasons for this are discussed in section

28 Figure 11: ESI-QqQ Infusion Spectrum of a PFPE-diphosphate =22m/z 44amu CH 2 CH 2 O =25m/z 50amu CF 2 =33m/z 66amu CF 2 O =58m/z 116amu CF 2 CF 2 O Figure 12: ESI-QqQ Infusion Spectrum of a PFPE-diphosphate, zoomed to show peak spacing 20

29 2.2.3 Thoughts and Considerations with Molecular Weight Distributions Considering the fact that the molecular weight distributions as identified by MSinfusion are lower than reported on technical data sheets and calculated by NMR, there are two phenomena that could be occurring in the ESI-MS process: mass discrimination and insource fragmentation. Mass discrimination is a phenomenon in which mass analyzers will transmit certain masses more efficiently than others, leading to an inaccurate representation of a mass distribution. For example, depending on its settings, a quadrupole mass analyzer will transmit masses in the 1000m/z range more efficiently than those in the 2000m/z range, possibly leading to a negatively biased perceived molecular weight distribution. Due to the fact that both ESI-QqQ and ESI-qTOF infusions showed identical mass distributions with two different mass analyzers, it leads one to believe that mass discrimination in the mass analyser alone cannot account for the discrepancy in molecular weight distribution. Another source of mass discrimination can occur during the ionization itself. Since compounds that are more surface active are more easily ionized during the ESI process because they reside on the outside of the ESI-droplet, this can also lead to an apparent molecular weight distribution shift. One would expect that surface activity would increase with molecular weight, such that the larger PFPEs would preferentially reside on the outside of ESI droplets and be preferentially ionized. This is contradictory to the negatively biased molecular weight distributions observed during infusion experiments. In a study by G. Carignano et al., it was found that surface activity increased significantly with the increasing chain length of a small PFPE-carboxylate with two to five ether repeating units and one hydrophilic carboxylate head group. 23 With two hydrophilic head groups present, 21

30 as in the PFPE-diphosphate, one would expect this trend to continue into larger molecular weight regimes, but it is still unclear whether or at what point the trend would reverse, such that larger PFPE based surfactants would have diminishing surface activity. ESI is an extremely soft method of ionization, so it is difficult to believe that compounds that are marketed as being thermally stable and chemically inert would fragment in-source. Though uncommon, it is not unheard of. ESI inherently forms an electrochemical cell during the process of ionization, such that in negative mode, the analyte has the possibility of being reduced. Although the alkyl phosphate head groups have the potential to be reduced, this would not significantly affect the molecular weight distribution. It is unlikely that ether linkages or carbon-carbon bonds would reduce yielding cleavage, but the electrochemistry offers another fragmentation pathway. Furthermore, one would expect to see the PFPE-diol as a doubly charged species. The PFPE-diphosphate has the potential to be a quadruple charged species, although it is difficult for molecules of this size to support four negative charges during this mass analysis. These two things are not enough to confirm the speculation of in-source fragmentation, but give strong arguments toward it. 2.3 MALDI-TOF Due to the fact that ESI-MS methods have been giving conflicting information in terms of the molecular weight distributions and the potential for in source fragmentation, it is worthwhile to explore a technique that solely gives singly charged ions and is known to be one of the softest ionization techniques for mass spectrometry. MALDI-TOF has been used in combination with unfunctionalized PFPEs and it has been shown that these PFPEs 22

31 have relatively high ionization efficiencies in the positive mode. 19 This bodes well for the analysis of the functionalized PFPE-diol and PFPE-diphosphate Conditions and Sample Preparation All MALDI-TOF spectra were acquired using a Waters MALDI micro MX in positive reflectron mode. The ionization parameters are as follows: Reflectron Voltage, 5200 V; Source Voltage, V; Pulse Voltage, 1950 V; MCP Detector Voltage, 2350 V; Laser Energy, 250%; and Matrix Suppression, 800 amu. The matrix used for all samples was a 10 mg/ml solution of dithranol in choloroform with lithium trifluoroacetate as the cationization agent. The MALDI plates were first rubbed with graphite, then spotted with 0.7 μl of matrix solution, allowed to dry, then spotted with 0.7 μl of sample solution prepared in methanol Molecular Weight Distribution Mass analysis performed by MALDI-TOF shown in figures 13 and 14 shows molecular weight distributions that are higher those seen during ESI infusion experiments for both PFPEs studied. This can be attributed to either the reduction of source fragmentation, or from the ability of MALDI more efficiently ionize higher molecular weight species. Irrespective, the molecular weight distribution determined by MALDI-TOF of the PFPE-diphosphate (~1250 amu) was not consistent with reported values in the technical data sheet (2000 amu), or was the PFPE-diol determined by MALDI-TOF (~1500 amu) in agreement with the molecular weight calculated by NMR (2700 amu). The characteristic peak spacings of 116m/z and 66m/z corresponding to fluorinated ethylene oxide and fluorinated methylene oxide repeating units present in both compounds 23

32 was clearly identified in both spectra. In this case, the PFPE-diphosphate gave a singly charged spectrum, as one would expect from MALDI. The spectrum was slightly more convoluted than that of the diol due to the extra aliphatic ethylene oxide repeating units near the termini corresponding to a peak shift of 44m/z. Figure 13: MALDI-TOF spectrum of the PFPE-diol showing characteristic peak spacing and a molecular weight distribution higher than that obtained via ESI methods Although this method seems quite promising for direct analysis of PFPEs, one must note that sample to sample reproducibility was not strong. The quality of the spectrum was highly dependent upon how the analyte crystallizes with the matrix. For this reason, 24

33 sensitivity tends to be low and quantification without the use of an internal standard would be challenging. Figure 14: MALDI-TOF spectrum of the PFPE-diphosphate showing characteristic peak spacing and a molecular weight distribution higher than that obtained via ESI methods 2.4 HPLC-ESI-QqQ MRM Method Motivation and Approach In order to be able to identify and quantify various PFPEs in environmental samples, MS-infusion and MALDI-TOF analyses simply were not sufficient. These methods are not sufficiently quantitative without an internal standard, nor are they exploiting the limits of detection capable of a mass spectrometer. For a triple-quad MS, it is most sensitive in 25

34 multiple reaction monitoring (MRM) mode, so this will be used as the primary mode for method development. Coupling the MS to the HPLC will allow for enhanced sensitivity, give an extra degree of identification and make the method easily quantifiable with the use of an appropriate internal standard. Due to the similarity in structure of the PFPE-diol and PFPEdiphosphate to fluorotelomer based surfactants, an HPLC separation method was optimized based on methods previously developed by the Mabury group Analysis of PFPE-diol Conditions and Sample Preparation All HPLC-MS analysis was performed using a Waters Acquity UPLC coupled to a Waters Quattro-Micro Triple Quadrupole Mass Spectrometer in negative ion mode. Separation was performed on a Gemini-NX 3 μm C18-Column (50 x 4.6 mm, Phenomenex). LC gradients and MRM transitions are listed below. All separations were performed using commercially available HPLC grade methanol and water and were buffered with 1 mm ammonium acetate. All chemicals were used without further purification. Samples were prepared in methanol and water to match the initial conditions of the HPLC gradient. Three transitions for the four most abundant masses from the molecular weight distribution were chosen for redundancy and confirmation. The three transitions chosen were M 113 m/z, corresponding to [CF3CO2] - ; M 228 m/z corresponding to [OCF2OCF2OCF2CH2O] - ; and M M-182 m/z corresponding to the loss of CF2OCF2CF2O. These transitions, along with their cone voltages and collision energies can be seen below. 26

35 Parameter Value Capillary Voltage (kv) 3.00 Cone Voltage (V) Transition Dependant Extractor Voltage (V) 3 RF Lens Voltage (V) 0.5 Source Temperature ( C) 120 Desolvation Temp. ( C) 350 Cone Gas Flow (L/hr) Off Desolvation Gas Flow (L/hr) 350 Table 3: ESI Source Parameters for HPLC method for PFPE-diol Transition Dwell (s) Cone Voltage (V) Collision Energy (ev) > > > > > > > > > > > > Table 4: MRM Parameters for HPLC Method for PFPE-diol Time (min) Flow (ml/min) %Methanol %Water Table 5: HPLC gradient elution parameters for PFPE-diol 27

36 Evaluation of Method A sample chromatogram for both the total ion chromatogram (TIC) and selected MRMs are shown in figure 13. MRM transitions not displayed exhibited identical chromatography. The TIC showed fairly sharp and well resolved peaks corresponding to different iterations of the polymer. On observation of the individual MRMs, each showed multiple peaks of their own, even though single peaks were expected. This was true for every transition in the method. The only way that this is believed to occur is through insource fragmentation of the parent PFPEs. That is to say, multiple different chain lengths are being separated via HPLC prior to entering the source, but are showing up at the same MRM transition having the same parent m/z and fragmenting with the same pattern. Furthermore, the third transition chosen for each one of the parent ions (M M-182m/z) corresponds to a loss of CF2O-CF2CF2O. These should be interior repeating units, thus it seems highly unlikely that an intact parent ion will cleave in the middle, lose the interior units and reassemble to make a new, intact polymer. One would not expect this to be a favourable loss pathway for fragmentation, although complex rearrangements and loss of internal fragments are possible. This would be a reasonable fragmentation pathway if the polymer has already cleaved before entering the collision cell such that these interior units are now at the termini, further suggesting that in-source fragmentation is possible for PFPEs. 28

37 Figure 15: Sample Chromatograms for the separation of a PFPE-diol Analysis of PFPE-diphosphate Conditions and Sample Preparation All HPLC-MS analysis was performed using a Waters Acquity UPLC coupled to a Waters Quattro-Micro Triple Quadrupole Mass Spectrometer in negative ion mode. Separation was performed on a Gemini-NX 3 μm C18-Column (50 x 4.6 mm, Phenomenex) 29

38 using various combinations of solvents, ion paring reagents, gradient run times and mobile phase flow rates. The table below describes some of the various combinations used for separation. Solvent Combination Ion-Pairing Agent Mobile Phase ph Methanol/Water 1mM ammonium acetate 4 Methanol/Water 1mM formic acid 2 Methanol/Water 10mM ammonia 10 Acetonitrile/Water 10mM ammonia 10 Acetonitrile/Water 1mM tertbutylammonium sulfate 4 Methanol/Water 1mM ZnCl 2 7 Methanol/Acetonitrile/Water 1mM 1-methylpiperdine 9 Table 6: Solvent and ion-pairing combinations for HPLC separation of PFPE-diphosphate Multiple transitions monitored during HPLC elution were chosen from the most abundant parent ions. The transitions monitored were M 79 m/z corresponding to [PO3]-; M 248 m/z corresponding to [HPO4(CH2CH2O)2CH2CF2]- or [CF2OCF2OCF2CF2O]-; M 113 m/z corresponding to [CF3CO2] - ; M 123m/z corresponding to [PO3CH2CH2]-. M 289 m/z corresponding to [PO3CH2CH2OCF2CF2] Evaluation of Method A sample chromatogram for selected MRMs is shown in figure 14. MRM transitions not displayed exhibited identical chromatography. It is clear that no peaks were observed for any transition. This was true for every solvent combination, gradient elution and ion pairing reagent used. The inability to chromatograph these compounds or the inability to detect their elution can be attributed to a few factors. There are potentially four negative 30

39 charges that must be neutralized or ion-paired to have optimal reverse-phase chromatography. Without neutralizing these charged sites, the chromatography has the potential to be so poor that the peaks are simply bleeding into the background signal. These negative charges have the potential to minimize any hydrophobic interaction with the stationary phase leading to minimal retention and rapid elution. Conversely, if the charged groups are successfully neutralized or ion-paired, the relatively large size of these PFPEs could lead to strong retention on the column. Their inability to be readily solubilized in traditional, non-fluorinated mobile phases would further this retention. Figure 16: Selected chromatograms showing no elution detected for chromatographic separation of the PFPEdiphosphate. 31

40 Although a chromatographic method could not be developed using traditional reversed-phase methods on a C-18 column, there is the potential to chromatograph these PFPE-diphosphate using other column chemistries. A pentafluorophenyl (PFP) type column might be advantageous since separation will rely more on the fluorine-fluorine interaction, rather than relative hydrophobicites of the analytes. Shorter aliphatic columns, such as C-8 or even C-4 might also prove successful due to their minimized interaction with the analyte to allow for elution under less drastic conditions. One could combine the change in column chemistry with a shorter column to further promote analyte elution Analysis of PFPE-carboxylate In order to gain insight into the behavior of the complex PFPEs in the mass spectrometer and during chromatography, a small molecule analogue of a PFPE, perfluoro- 2,5-dimethyl-3,6-dioxanonanoic acid, which can be seen below and will be referred to as PFPEC, was obtained and subjected to MS-Infusion, CID and was chromatographed. Figure 17: Structure of perfluoro-2,5-dimethyl-3,6-dioxanoic acid, or PFPEC Conditions and Sample Preparation All HPLC-MS analysis was performed using a Waters Acquity UPLC coupled to a Waters Quattro-Micro Triple Quadrupole Mass Spectrometer in negative ion mode. Separation was performed on a Gemini-NX 3 μm C18-Column (50 x 4.6 mm, Phenomenex). Solutions for infusion were prepared in 100% HPLC grade methanol and solutions for 32

41 HPLC analysis were prepared to match initial solvent composition of the elution gradient. All chromatography was performed using 1mm ammonium acetate as a buffer in both solvents. Four transitions were monitored for redundancy and confirmation. The transitions chosen were 495 m/z 185 m/z, corresponding to the parent ion fragmenting to [CF3CF2CF2]-; 495 m/z 119 m/z, corresponding to the parent fragmenting to [CF3CF2]-; 185 m/z 185 m/z to monitor any iteration of 185 m/z in the chromatogram; and 185 m/z 119 m/z, corresponding to [CF3CF2CF2]- fragmenting to [CF3CF2CF2]-. Optimized source and MRM parameters can be seen in the tables below Parameter Value Capillary Voltage (kv) 2.78 Cone Voltage (V) Transition Dependant Extractor Voltage (V) 5 RF Lens Voltage (V) 0.8 Source Temperature ( C) 120 Desolvation Temp. ( C) 350 Cone Gas Flow (L/hr) Off Desolvation Gas Flow (L/hr) 350 Table 7: Optimized Source parameters for the chromatography of PFPEC Transition Dwell (s) Cone Voltage (V) Collision Energy (ev) > > > > Table 8:MRM Parameters for HPLC Method for PFPEC 33

42 Time (min) Flow (ml/min) %Methanol %Water Table 9:HPLC gradient elution parameters for PFPEC Evaluation of Method The initial infusion spectrum for PFPEC is shown in figure 16. The parent ion peak (M-H)- at m/z=495 is not nearly the most abundant peak in the spectrum. Rather, the peak at m/z=185 dominates the signal. This was arising from either a major impurity in the PFPEC sample, or it, too, was fragmenting in source. m/z=185 would correspond to [CF3CF2CF2]-, which is located at one terminus of the molecule and is a viable fragment. In order to determine whether this was truly an in-source fragment or simply an impurity, the PFPEC was subjected to chromatography where both the 495 m/z parent and the 185 m/z parent were monitored. 34

43 m/z=185 m/z=495 Figure 18: ESI-QqQ infusion spectrum of perfluoro-2,5-dimethyl-3,6-dioxanoic acid showing weak signal from the parent ion and a strong signal from a possible fragment. Upon chromatographing the compound using a very basic gradient method, both the 495 m/z transitions and the 185 m/z transitions eluted at the same time, confirming that they belong to the same initial molecule and that the peak at 185 m/z in the infusion spectrum is due to source fragmentation. The conditions used for ionization were kept mild, thus this ether must be very labile in ESI. The peak integration for the 185>119 transition was 14x that of the 495>185 transition. This observation further suggests that the large PFPEs analyzed previously fragment in-source yielding a negatively biased molecular weight distribution. 35

44 Overall, PFPEC chromatographed very easily and was extremely sensitive in the MS. It has the potential to be used as an excellent internal standard for later quantifying peak areas of polymeric PFPEs. 2.5 LC-ESI-QqQ Parent Ion Method for PFPE-diphosphate Approach As much as using MRM with triple quad based mass spectrometers allows for the most sensitive and most selective type of analysis, it is not ideal for analyzing polymers. MRM is ideally suited for the analysis of small molecules because they exist at only one mass to charge ratio which can be selected for quite easily. Conversely, polymers exist over a range of mass to charge ratios. When selecting single masses out of an entire distribution, one is not gaining knowledge about the entire polymer, but just a small subset. Furthermore, quantification and even identification can become quite difficult since polymeric distributions commonly change from batch to batch, such that an m/z that has been selected as the most abundant from an in-house standard might be one of the least abundant or possibly not present in an environmental sample due to the shift in molecular weight. In this way, a sample could be determined a non-detect simply because one is not observing the proper mass. The fact that a polymer s mass is not constant from batch to batch makes mass spectral analysis of polymers extremely difficult. If one wants to still observe polymers by MS, a more general approach is needed. TOF and FT based MSs would be excellent for obtaining full scans, but are usually either not capable of tandem MS, or are not readily available at all. Using a parent ion scan in a triple- 36

45 quad based MS, one can gain insight into the full distribution of a polymer, while still taking advantage of its MS/MS capabilities Conditions and Sample Preparation A parent-ion method was employed for the PFPE-diphosphate. Two daughter ions, 190m/z corresponding to [CF2OCH2CH2OPO3H]-, and 248m/z corresponding to [HPO4(CH2CH2O)2CH2CF2]- or [CF2OCF2OCF2CF2O]-, were selected. A scan was performed such that only parents fragmenting to these two ions would show in the final mass spectrum. Samples for infusion were prepared in HPLC grade methanol and samples for HPLC analysis were prepared to match the initial gradient conditions. Samples were prepared to a concentration of 10 μg/ml and infused at a rate of 50 μl/min. Parameter Value Capillary Voltage (kv) 2.5 Cone Voltage (V) 35 Extractor Voltage (V) 5 Collision Energy (ev) 30 Source Temperature ( C) 120 Desolvation Temp. ( C) 350 Cone Gas Flow (L/hr) Off Desolvation Gas Flow (L/hr) 350 Table 10: Source parameters for parent-ion scan analysis of PFPE-diphosphate. 37

46 2.5.3 Evaluation of Method A sample parent-ion scan for PFPE-diphosphate parents of 190m/z can be seen in figure 19 below. The sensitivity has decreased compared to the direct infusion mass spectra described in section 2.2.2, due to the fact that the transmission efficiency in a parent ion scan is decreased as compared to a non-fragmented infusion scan, but the signal surrounding the polymer peaks has dropped down to the baseline, due to the selectivity of the parent-ion scan. The significantly decreased background signal as compared to a direct MS scan leads one to believe that there are a suite of unidentified, proprietary compounds present in the technical mixture (Fomblin HC-P2-1000) used for analysis, and the PFPEphosphate is not the only compound present in the end product. The enhanced selectivity demonstrates the potential for the parent-ion scan to be used in conjunction with LC separations to give more suitable information during polymeric analysis. 38

47 Figure 19: Parent Ion Scan for Parents of 190m/z of the PFPE-diphosphate. Unfortunately, similar chromatographic issues arose as with the MRM-based analysis of the PFPE-diphosphates. Each chromatogram resulted with no peaks under the same solvent and ion-pairing conditions. If chromatographic issues were not a concern, this would be an excellent method to use in combination with a sensitive MRM-based method, where the parent-ion scans could confirm the molecular weight distributions of analytes identified in the MRM-method, or potentially identify PFPEs not detected due to a shift in average molecular weight. 39

48 3 Waste Water Treatment Plant Biodegradation Study 3.1 Motivation Although the HPLC-MS based methods showed great complexity and still need improvements in order to become useful and quantitative in the future, analysis by NMR showed the potential to be straightforward and quantitative. Although relative sensitivity could be an issue for the analysis of trace levels of PFPEs in environmental samples, quantifying changes during a degradation study at higher concentrations should be easily monitored by changes in NMR spectra. It has been recently shown that polyfluoroalkyl phosphates, PAPs, can be microbially hydrolyzed in waste water treatment plant sludge to the corresponding alcohol. 25 A mono-alkyl PAP, monopap, is shown as the first compound in figure 20. A suite of studies have also shown that these fluorotelomer alcohols have the potential, through atmospheric and biological pathways, to further degrade into terminal PFCAs. 26,27,28,29 A simplified version of the degradation pathway is shown in figure 20. Due to the structural similarity of PAPs to the PFPE-diphosphate, one can propose an analogous degradation pathway. If this pathway is followed, it would result in a new PFCA that has never been monitored in any environmental sample. With the widespread use of PFPE-based surfactants, their degradation to a terminal acid could significantly contribute to the unknown portion of organo-fluorine detected in global environmental samples. 40

49 Figure 20: Degradation pathway of PAPs to their corresponding PFCAs Figure 21: Proposed degradation pathway of the PFPE-diphosphate to a novel PFCA 41

50 3.2 Method Experimental Design The experiment was performed using a purge-and trap system employed previously by Lee et al 25 and Dinglasan et al. 26 Here, 500 ml polypropylene bottle are fitted with house drilled caps and septa to fit 100 mg Orbo Amberlight XAD-2 cartridges. Mixed liquor, a combination of sewage sludge and raw waste water, was obtained from the Ashbridges Waste Water Treatment Plant in Toronto, ON. The experiment contained the following bottle types: No Flow Control (n=2) to determine losses due to sorption to bottle surfaces, Purge Control (n=2) to determine losses due to volatilization of the analyte, Sterile Control (n=2) containing autoclaved mixed liquor to normalize for degradation not associated with microbial activity, Viability Control (n=1) to ensure microbial activity by monitoring the known degradation product of a chosen control, Mixed Liquor Only (n=2) to monitor for background levels of the PFPE-diphosphate and Experimental (n=3). The 6:2 monopap was used in the viability control and its primary degradation product, the 6:2 fluorotelomer alcohol, 6:2 FTOH, was monitored by GC. The structures of both compounds can be seen in Figure 20. Mixed liquor was diluted with a phosphate deprived mineral media in order to promote the PFPE-diphosphate as a microbial phosphate source. The mineral media was composed of the following: 1% (v/v) of 19.9 μg/l FeCl2.4H2O, 0.9 μg/l p-aminobenzoate, 0.9 μg/l nicotinic acid, 40 mg/l (NH4)6Mo7O24.4H2O, 50 mg/l H3BO3, 30 mg/l ZnCl2, 4 mg/l CoCl2.6H2O, 10 mg/l (CH3COO)2Cu.H2O, 17 mg/l FeCl2.4H2O. 42

51 A schematic of the experimental setup can be seen in the diagram below and the bottle contents are outlined in the appendix. Upon filling the bottles with their appropriate solvents, spikes and internal standards, they are continuously purged with carbon-filtered in-house air for the duration of the experiment, while volatiles are trapped by the XAD cartridges. Bottles were sampled at t=0, 1 h, 4 h, 6 h, 17 h, and every 24 h following by extracting 1 ml samples via micropipette. XAD cartridges were also exchanged at each time point. Figure 22: Schematic of the Purge and Trap experimental design Extraction Method An ion pairing method developed by Hansen et al. for the quantitative extraction of a number of per- and polyfluorinated acids and surfactants. 30 Due to the fact that these samples would be analyzed by NMR, and not by HPLC-MS, extensive sample clean-up is not 43

52 required. For this reason, a simple freezing and lyophilzation is employed to halt any microbial activity, then the sample is reconstituted in the desired solvent for analysis. The 1mL aqueous extracts were immediately flash frozen at -196 C. Samples were either directly lyophilized or stored at -20 C until lyophylization. Upon lyophilisation, the samples were reconstituted in 1 ml of methanol and vortexed for 1 minute. The NMR sample was then prepared by adding 800 μl of the sample extract, 150 μl of d4-methanol for locking, and 50 μl of 2000 ppm trifluorotoluene in methanol as an internal standard. XAD-cartridges were opened and the packing material was vortexed in 3x4mL portions of ethyl acetate. The fractions were combined, evaporated to dryness under nitrogen and reconstituted in 1mL ethyl acetate for analysis by GC-MS. All NMR spectra obtained in this section were acquired on a Varian Mercury 400MHz NMR for a minimum of 128 scans with an optimized relaxation delay of 5.5 s 3.3 Results F NMR Each NMR spectrum acquired, aside from the no flow control bottles, showed no peaks in the spectrum up to 4096 scans. The XAD extracts also showed no sign of a hydrolysis product. The lack of peaks in the NMR signal can be attributed to a combination of PFPE solubility, immediate sorption to the bottle walls and, more importantly, immediate sorption to particulate matter in the sewage sludge. Sorption to the particulate matter yields PFPE-diphosphates that are unable to extracted by lyophilisation and a simple methanol extract, unable to be hydrolyzed and, thus, unable to be analyzed by liquids NMR in this method. 44

53 3.3.2 LC-ESI-QqQ MRM Revisited Since the PFPE-diol described in earlier sections is proposed to be a degradation product of the PFPE-phosphate, revisiting the previously established LC-ESI-QqQ method for the analysis of the PFPE-diol might lead to some discovery about the degradation of the PFPE-phosphate. The LC-ESI-QqQ method had been shown to be fairly sensitive and highly selective so that a degradation product, although not observable by NMR, might be seen by LC-MS. All the samples were re-extracted using 950 μl of methanol and 50 μl of hexafluoroisopropanol to enhance solubility. Samples were then 2-fold diluted into a final solvent ratio of 50:50 MeOH:H2O to match the initial conditions of the gradient. 50 ppb of PFPEC was added as an internal standard and was monitored for the previously optimized 185>119 transitions. The same LC method and MRM transitions described in section were employed. As seen in figure 20, all samples that were spiked with the PFPE-phosphate showed signals in the chromatogram, even at t=0. This leads one to believe that there is a PFPE-diol impurity within the PFPE-diphosphate technical material. The chromatography of the sample and sterile control is markedly different from that of the 500ppb PFPE-diol standard, thus the molecular weight distribution probably differs. Furthermore, there are two separate groups of peaks that arise from the degradation study samples. This could be either due to two separate groups of PFPE-diols present in the sample, or possibly a PFPE with a different terminus that has the same parent masses and fragments as those chosen for MRMs. This bimodal peak distribution is consistent for all experimental time-points analyzed by LC-MS. 45

54 t=0 t=0 Figure 23: TICs of selected samples and standards of the biodegradation study Polymer Solubility In an effort to further explore the biodegradation potential of the PFPE-diphosphate, one would need to perform the experiment under conditions promoting the aqueous solubility of the PFPE. A mixture of a surfactant and hydrotrope are commonly used to solubilize large organic molecules in aqueous media. Previous work in the Mabury lab has shown that the surfactant dodecylamine hydrochloride and the hydrotrope sodium salicylate in a 10:6:1 ratio of Surfactant:hydrotrope:analyte by mass, was successful at solubilizing a high molecular weight fluorinated side chain polymer in water

55 The same ratio of surfactant and hydrotrope was tested along with the addition of 1% ethanol to further enhance the aqueous solubility of the PFPE-diphosphate. Into a 50mL propylene tube, a 50/50 mixture of mineral media and autoclaved mixed liquor was spiked with 100ppm of PFPE-diphosphate under the conditions of 1% ethanol, the 10:6:1 ratio of hydrotrope:surfactant:analyte and the combination of both. The mixtures were left to shake for 24 hours, were extracted according to the protocol outlined in section and analyzed by NMR. In each case, there still was no signal in the NMR spectrum. This is shown visually in figure 24. It is clear that the surfactant and hydrotrope are not enough to prevent rapid sorption of the PFPE-diphosphate. Using harsher conditions would not be beneficial due to the viability of the microbial system required to perform the degradation experiment. Figure 24: 100ppm extracted from various media in an attempt to enhance aqueous solubility and prevent sorption to particulate matter 47

56 4 Conclusions and Future Directions It is quite clear that PFPEs are a suite of extremely complex molecules; they have the behaviour of a small molecule with the structure of a polymer. For this reason, nontraditional methods must be employed for their proper analysis. High resolution NMR has the potential to elucidate the details of these polymers: determining their average number molecular weight distribution through relative peak intensities, their identity through terminal CF2 chemical shifts, and their concentrations through the use of appropriate internal standards. Sample preparation is simple and quantification is robust. The only drawback is the time required to obtain the sensitivity of a mass spectrometry based techniques although work by Ellis et al has shown that 19 F NMR can be used to analyze solutions of simple perfluorinated acids, such as mono-, di- and trifluoroacetic acid, down to concentrations in the ppb range after sample concentration. 32 A high-throughput, highsensitivity method would be one that has not currently been explored in any capacity. The possibility of in-source fragmentations of PFPEs in ESI is very intriguing. It is speculative that compounds touted as thermally stable and chemically inert cleave in an ESI source, but data presented here suggests just that. The small molecule analogue, PFPEC, shows strong evidence for the source fragmentation through both infusion and LC-MS experiments presented. 4.1 Non-quadrupole MS Method Due to the polymeric nature of PFPEs, they are inherently not suited for quadrupole based instruments. As soon as a scan method is employed with a quadrupole mass analyzer, sensitivity is reduced 100 fold and selectivity is severely diminished. The use of 48

57 MRMs to monitor a polymeric species does not give the full picture of the polymer; choosing 5 peaks out of 50 present in a spectra only tells 10% of the story. In the same way, analyzing 5 peaks out of 50 present in a spectra is only analyzing 10% of the analyte. Furthermore, many commercial quadrupole mass analyzers have a mass cut-off of roughly 2000 m/z. PFPEs are known to be of far greater molecular weight and, thus, will not be seen even if they are extremely abundant in a sample, depending on the charge state. If a mass spectral method is going to be used to analyze PFPEs, it would have to be with a mass analyzer capable of scanning the entire spectrum of peaks without loss of sensitivity. TOF and FT based instruments (Orbitrap, FTICR), are ideally suited. These mass analyzers also tend to be high resolution, thus exact masses can be used to determine end-group chemistry. 4.2 Potential Data Dependant Acquisition Analysis Assuming chromatography can be sorted out for PFPE congeners of interest, a hypothetical method for the ideal mass spectral analysis of PFPEs could be modeled after proteomics based analysis, where Data Dependant Acquisitions (DDAs) are heavily responsible for obtaining information. This method would require the use of an LC coupled to LTQ-Orbitrap, or similar mass spectrometer. In this method, all things eluting from the LC column would be directly analyzed by the Orbitrap in a full scan mode. Since an Orbitrap is an Fourier Transform-based mass spectrometer, it can analyze all masses at once. The software mediated DDA can be set to select for mass ranges that contain peaks with separations of 66 m/z or 116 m/z, corresponding to the now well established fluorinated methylene oxide and fluorinated ethylene oxide repeating units of a linear PFPE. These specific mass ranges can then be sent to the LTQ for identification and quantification by 49

58 MS n. In this way, there is the potential to have a non-biased, high throughput, high sensitivity method while maintaining high resolution molecular weight information. 50

59 5 Appendix 5.1 ESI-qTOF infusion spectrum of the PFPE-diol 51

60 5.2 ESI-qTOF infusion spectrum of the PFPE-diphosphate 52