Method for the determination of sub-ppm concentrations of perfluoroalkylsulfonate anions in water

Abstract

The determination of sub-ppm concentrations of aqueous perfluoroalkylsulfonate (PFS†) anions, including perfluorooctylsulfonate (PFOS), has been accomplished with a relatively simple mass spectrometric procedure that does not require extraction of the analytes into an organic solvent or a chromatographic separation prior to injection into the negative-ion electrospray ionization mass spectrometer. Sample pretreatment was minimized and consisted of dilution of the aqueous samples of groundwater, surface water, tap water, and distilled water with acetonitrile, addition of dodecylsulfate (DDS) as an internal standard, and, in some cases, addition of known amounts of perfluorobutylsulfonate (PFBS) or PFOS for standard-addition experiments. The linear-response range for PFOS is 25.0 µg L⁻¹ to 2.5 mg L⁻¹. The lower limit of this range is three orders of magnitude lower than an equally straightforward chromatographic method. The relative errors for standard aqueous solutions containing only 25.0 µg L⁻¹ and 2.5 mg L⁻¹ PFOS are ±14% and ±7%, respectively, with 133 µg L⁻¹ DDS as the internal standard. The detection limit and quantification limit for PFOS in these standards are 5.0 µg L⁻¹ and 25.0 µg L⁻¹, respectively. Six different PFS anions, containing three to eight carbon atoms, were identified and quantified in an aqueous film-forming foam (AFFF) formulation using the method of standard additions. Two alkylsulfate anions and two perfluoroalkylcarboxylate anions were also identified in the AFFF formulation.

---

Introduction

The 3M Company recently announced that it is discontinuing production and distribution of perfluorooctylsulfonate-based products because of the suspected toxicity of C₈F₁₇SO₃⁻ (PFOS) and the persistence of some PFOS-based chemicals in workers, in animals, and in the environment.^1–13 The single-dose LD₅₀ of tetraethylammonium perfluorooctylsulfonate for Wistar rats has long been known to be 190 mg kg⁻¹.³ In a recent 3M study, however, more than one-third of the rat pups whose mothers were fed only 1.6 mg kg⁻¹ per day during pregnancy died within four days of birth.⁴ Although the mechanism(s) of toxicity is not known, the persistence of PFOS in the environment is probably related to the inert nature of C–F bonds in perfluoroalkyl chains.^14–16 For example, PFOS was not metabolized by Pseudomonas sp. strain D2.¹⁷ Laboratory rats were only able to excrete 26% of the ¹⁴C-labeled PFOS that was administered.¹⁸

The PFOS anion, its 4-, 5-, 6-, and 7-carbon homologs, and their derivatives are important components in two widely used commercial formulations, Scotchgard protection and repellant products and LIGHT WATER brand aqueous film-forming (fire-fighting) foams (AFFF's).^4,5,12,13 Both have been in use for many years. AFFF's are especially effective at extinguishing liquid-fuel fires and have been used extensively at military and civilian fire-fighting training facilities as well as at military airfields and municipal airports.^12,19–23 In a recent seminal study, Moody and Field detected and quantified several perfluoroalkylcarboxylate anions at concentrations between 125 µg L⁻¹ and 7.1 mg L⁻¹ in groundwater contaminated with untreated AFFF wastewater at Naval Air Station Fallon, NV, and at Tyndall Air Force Base, Panama City, FL.²⁴ Their study confirmed two earlier reports that tentatively identified “fluorinated surfactants” in groundwater at the Tyndall site.^25,26 More recently, perfluoroalkylcarboxylate anions have been quantified from the groundwater around a fire training area at the former Wurtsmith Air Force Base in Oscoda, MI.¹³ Although perfluoroalkylcarboxylate anions are not listed as components of AFFF's in patents or in MSDS sheets,^27,28 their presence in at least one AFFF formulation will be reported in this paper.

Before this work, there were only two published methods for the identification of perfluoroalkylsulfonate (PFS) anions in water, and neither of them is suitable for the quantitative determination of sub-ppm amounts of individual PFS anions. In the first study, PFS anions were identified, but not quantitatively determined, using fast-atom-bombardment and collision-activated-dissociation mass spectrometry.²⁹ In the second study, two commercially available suites of PFS anions at concentrations between 5 and 50 mg L⁻¹ in aqueous acid were quantitatively determined collectively by extraction by a polymeric reverse-phase resin followed by elution onto a multiphase HPLC column and detection by suppressed conductivity.³⁰ However, each PFS anion was not individually quantified by this method.

While this manuscript was in preparation, Hansen et al. reported a mass-spectral method for determining concentrations of PFOS and other fluorochemical anions in blood serum and whole liver tissue.³¹ Due to the complex nature of these matrixes, their method requires (i) extraction of PFOS (and the other fluorochemical anions) into an organic solvent using an ion-pairing reagent, (ii) evaporation of the solvent and reconstitution in another solvent, (iii) HPLC separation of the anions on a Betasil C₁₈ column, and (iv) negative-ion electrospray tandem mass spectrometry (the latter was necessary because of possible biological interferrents).³¹ This important new method, however complex, has now been used to determine the accumulation of PFOS in marine mammals and the global distribution of PFOS in wildlife.^8–10

A second mass-spectral method for the determination of perfluorinated surfactants in surface water was also reported while this manuscript was in preparation.³² Moody et al. describe two techniques, liquid chromatography-tandem mass spectrometry and ¹⁹F NMR spectroscopy, that required the use of solid-phase extraction to preconcentrate the analytes. Using these techniques, PFS anions were quantified in the low µg L⁻¹ concentration range (the linear response range using liquid chromatography-tandem mass spectrometry is 0.85–208 µg L⁻¹). While useful when analyzing surface water, both of these methods require significant sample preparation and relatively long analysis times.

In this paper, we report a simpler mass-spectral method for the quantitative determination of individual PFS anions, ranging from four to eight carbon atoms, in groundwater and other homogeneous aqueous samples. The method is simpler and less time consuming because these matrixes are significantly less complex than blood serum, whole tissue, and surface water. The linear-response range for PFOS is 25.0 µg L⁻¹ to 2.5 mg L⁻¹. The lower limit is three orders of magnitude lower than the lower limit for the HPLC method discussed above.³⁰ Although the linear-response range we report is for a higher concentration range than that reported by Moody et al.,³² our direct-injection mass spectral method is less complex since it does not require preconcentration steps. The direct-injection method has already been used to determine that C₆F₁₃SO₃⁻ (PFHxS) and PFOS anions are still present in groundwater contaminated with untreated AFFF wastewater at the former Wurtsmith Air Force base, Oscoda, MI, six years after the cessation of fire-training activities.¹³ The detection limit (signal-to-noise ratio greater than three) and quantification limit (signal-to-noise ratio greater than five) for PFOS are 5.0 µg L⁻¹ and 25.0 µg L⁻¹, respectively.

Experimental section

Standards and reagents

Sodium dodecylsulfate (Na(DDS), Sigma, >99%) was dried under vacuum at 25 °C for 24 h and was stored in a helium-filled glovebox. Two perfluoroalkylcarboxylic acids, C₁₁F₂₃CO₂H and C₅F₁₁CO₂H (Aldrich), were used as received. Acetonitrile (Fisher, HPLC grade) was used as received. Distilled water was purified and deionized to 18 MΩ with a Barnstead NanoPure purification system. Potassium perfluorooctylsulfonate, K(PFOS), was synthesized from perfluorooctylsulfonyl fluoride (3M) by adding it to KOH in water. The white crystalline compound K(PFOS) is the major product from this reaction, while perfluoroalkylsulfonates with other chain lengths (C7, C6) are also present in small amounts. Five recrystallizations from hot water were carried out to take advantage of the lower solubility of K(PFOS) in cold water compared to PFS salts with shorter chain lengths. The purity of K(PFOS) was determined to be >99% by negative-ion electrospray ionization mass spectrometry ((−)ES-MS) and by ¹⁹F NMR spectroscopy. A yield of 65% was achieved based on perfluorooctylsulfonyl fluoride. Purified K(PFOS) was dried under vacuum and stored in a helium-filled glovebox. The synthesis of potassium perfluorobutylsulfonate, K(PFBS), from perfluorobutylsulfonyl fluoride (3M) and KOH was the same in every respect except that the higher solubility of this salt in water resulted in a much lower yield of purified material (30% based on perfluorobutylsulfonyl fluoride).

The AFFF formulation FC-203CF LIGHT WATER Brand Aqueous Film Forming Foam^® (3M) was used as received. The MSDS sheet for this formulation lists the components as water (70%), 2-(2-butoxyethoxy)ethanol (20%), two alkylsulfate salts (5%), amphoteric fluoroalkylamide derivative (3%), five perfluoroalkylsulfonate salts (1%), triethanolamine (1%), and methyl-1H-benzotriazole (0.1%).²⁷

A groundwater sample was collected from the former Wurtsmith Air Force base near Oscoda, MI by Cheryl Moody Bartel and Jennifer Field.¹³ The sample collection protocols as well as the analysis of field blanks were addressed in a study to be published elsewhere.¹³ The groundwater was withdrawn in June 1999 from a well (labeled FT3) in an area used for fire-fighting training from 1952 to 1993. These fire-training exercises included the use of AFFF formulations that probably contained PFS anions including PFOS.¹³

Water samples from the Ohio River and Cincinnati tap water were donated by E. T. Urbansky of the US Environmental Protection Agency in high-density polyethylene bottles. Horsetooth Reservoir water (Fort Collins, CO) was collected in December 2001 in high-density polyethylene bottles.

Sample preparation

Due to the sensitivity of (−)ES-MS peak intensities to minor changes in experimental conditions, a strict experimental procedure was followed when preparing and analyzing all samples. A 1∶1 (v∶v) acetonitrile∶water matrix was used to efficiently nebulize the anions present in solution. Matrixes containing only water as the solvent were significantly less efficient. All samples were prepared using gas-tight syringes accurate to ±1% or less of their total volume (10, 50 and 500 µL). The syringes were rinsed with 1∶1 (v∶v) acetonitrile∶water at least five times after each use and conditioned by rinsing three times with each new sample before an aliquot of the sample was withdrawn for analysis.

Standard-addition samples were prepared as follows. A groundwater sample was centrifuged for 15 min at 5,000 rpm to separate any fine particulates that cannot be injected into the (−)ES-MS instrument. The samples were not filtered since this might result in some loss of PFOS by adsorption onto the filter. A series of 200-µL aliquots of the centrifuged groundwater supernatant were treated with appropriate volumes of a 5.0 mg L⁻¹ solution of PFOS in 1∶1 (v∶v) acetonitrile∶water. The series of samples were then diluted with 200 µL of acetonitrile and 25 µL of a 2.7 mg L⁻¹ DDS solution in 1∶1 (v∶v) acetonitrile∶water. The diluted sample was then diluted further with a sufficient amount of 1∶1 (v∶v) acetonitrile∶water so that the final volume was 500 µL. This sample preparation process resulted in a 2.5 fold dilution of the original groundwater sample.

The samples used to generate the calibration curve for the direct-injection method were prepared as above except that deionized water was used instead of the groundwater supernatant. Much larger volumes (e.g., 10 mL) of these samples were prepared, but the dilution factors were the same. In this paper, the direct-injection method is defined as a method that requires only dilution and addition of DDS, the internal standard, prior to analysis. This greatly reduces both the sample preparation time and the number of samples to be analyzed.

All samples were prepared using glass laboratory equipment. Since it was not known whether PFS anions might adhere differently to different surfaces and thus interfere with our analyses, we investigated whether 10 mL samples containing known concentrations of K(PFOS) and K(PFBS) gave different results after being rolled for 20 min and left still for 24 h in 1 L Pyrex and high-density polypropylene bottles. It was found that samples not rolled in the large bottles gave the same mass spectral responses as the samples rolled in either glass or plastic.

The quantitative accuracy of the instruments was checked at the beginning, middle and end of each data set collected using a single standard equimolar solution containing 250 µg L⁻¹ PFOS and 133 µg L⁻¹ DDS in a 1∶1 (v∶v) acetonitrile∶water solution. Three to five replicates of each sample were analyzed and the Q test was used to eliminate spurious data. All experimental values in this paper are reported with ±1 standard deviation. Blanks were run after the replicates of each sample to insure that there was no carry-over between samples.

Instrumentation

Both a Fisons VG Quattro single quadrupole mass spectrometer and a Finnigan LCQ Duo mass spectrometer were used to collect the (−)ES-MS data. Instrumental parameters are compared in Table 1. Samples were introduced into the LCQ Duo by continuous infusion from a syringe and into the Quattro by flow injection from a 10 µL sample loop. Results were similar for both instruments with the exception that the Fisons instrument had slightly lower detection and quantification limits (3 µg L⁻¹ and 5 µg L⁻¹, respectively) although it also had higher relative errors. Standard-addition experiments were conducted using only the Finnigan LCQ Duo instrument. All other experiments were done using both instruments.

Table 1 Comparison of instrumental parameters for the Finnigan LCQ Duo and the Fisons VG Quattro mass spectrometers

Finnigan LCQ Duo		Fisons VG Quattro
a Different names have been given to the same components by the two companies.
Spray voltage^a	−3.5 kV	Capillary voltage	−2.9 kV
Capillary voltage	−39 V	Cone voltage	−30 V
Capillary temperature	200 °C	Source temperature	75 °C
Infusion rate	5 µL min^−1	Flow injection rate	4 µL min^−1

---

Several experiments were specifically performed to compare the two spectrometers. A set of two standards containing low and high PFOS concentrations was prepared with four replicates each. The first standard contained 250 µg L⁻¹ PFOS and 133 µg L⁻¹ DDS and had a ratio of (−)ES-MS intensities, I(499)/I(265), of 3.9 ± 0.9 (Fisons) and 4.8 ± 0.4 (Finnigan). The second standard contained 2.5 mg L⁻¹ PFOS and 133 µg L⁻¹ DDS and had values of I(499)/I(265) of 26 ± 4 (Fisons) and 33 ± 3 (Finnigan). In both cases, the ratios of intensities were the same within the error of each spectrometer. In addition, log-log calibration plots were made using each spectrometer. The slopes of the plots were 1.02 ± 0.04 (Fisons) and 0.93 ± 0.02 (Finnigan).

Fluorine-19 NMR spectra of K(PFOS) and K(PFBS) dissolved in acetonitrile-d₃ were recorded using a Varian Inova-300 spectrometer at 25 °C. The ¹⁹F frequency was 282 MHz and the chemical shift was referenced to an external CFCl₃ standard (δ 0.0). Peaks for PFOS were located at δ −80.3 (triplet (t), 3F), −114.2 (multiplet (m), 4F), −120.2 (m, 2F), −121.1 (m, 4F), −121.9 (m, 2F), −125.3 (t, 2F).

The pH was measured using an Orion meter with a Ross^® pH electrode. The conductivity was measured with a Yellow Springs conductivity probe and meter.

Results and discussion

Internal standard choice

Dodecylsulfate (DDS) was used as an internal standard for all (−)ES-MS samples to allow for the quantification of PFS anions. In general, it is important that an ES-MS internal standard has properties similar to the analyte, so that both species will tend to form ions under the same conditions. Several researchers have used an isotopically labeled form (e.g., a perdeuterated or polydeuterated isotopomer) of their analyte as their internal standard, but that is not convenient for perfluorinated compounds.^33–35 By keeping the internal standard similar in structure and mass, it will have a similar sensitivity coefficient to the analyte. The sensitivity coefficient is a constant that is related to the efficiency of an analyte to form individual molecular ions in the electrospray source.^34,35 The ratio of anion peak intensities is equal to the ratio of sensitivity coefficients when the concentrations of the anions are equal.^34,35 In this work, samples were analyzed containing equal concentrations of PFOS and DDS to compare their sensitivity coefficients. The anions PFOS and DDS have different sensitivity coefficients, which were manifested by two different (−)ES-MS peak intensities for the same concentration of the two ions. However, the ratio of these intensities, I(499)/I(265), was relatively constant (to within experimental error) at 1.4–2.2 over the concentration ranges 5.0 µg L⁻¹ to 25.0 mg L⁻¹ PFOS and 2.7 µg L⁻¹ to 13.3 mg L⁻¹ DDS for equimolar concentrations of the two surfactant anions, which makes DDS an appropriate internal standard for PFOS. The validity of DDS as an internal standard should be evaluated for each new matrix. In addition, the sensitivity coefficients for the two perfluorinated anions PFOS and PFBS were found to be the same to within experimental error over the concentration range 5.0 µg L⁻¹ to 499 µg L⁻¹ PFOS and 3.0 µg L⁻¹ to 299 µg L⁻¹ PFBS (these limiting concentrations correspond to equal numbers of moles of the two anions). Based on these observations, we assume that the sensitivity coefficients of C₅F₁₁SO₃⁻, C₆F₁₃SO₃⁻, and C₇F₁₅SO₃⁻ are also equal to that of PFOS and PFBS over equimolar concentration ranges.

Method of standard additions

Two methods of determining unknown concentrations of PFOS (and therefore unknown concentrations of C₄F₉SO₃⁻, C₅F₁₁SO₃⁻, C₆F₁₃SO₃⁻, and C₇F₁₅SO₃⁻) were investigated. A groundwater sample from the former Wurtsmith Air Force base near Oscoda, MI, that was part of a larger study¹³ will serve to exemplify the first method.

The groundwater sample, labeled FT3, consisted of a clear, colorless supernatant and a small amount of solid particles. The pH and specific conductivity of the supernatant were 6.2 and 736 µS cm⁻¹, respectively. Four replicate aliquots of each of four different standard-addition samples were prepared. Along with four replicate aliquots of FT3 that did not contain added PFOS, the spiked aliquots were examined by (−)ES-MS after each was made to be 133 µg L⁻¹ DDS in 1∶1 (v∶v) acetonitrile∶water.

A plot of I(499)/I(265) vs. the concentration of added PFOS is shown in Fig. 1 (each data point is the average over the four replicates). The mass spectrometer detector was found to be linear over the concentration range from 25 µg L⁻¹ to 2.5 mg L⁻¹ PFOS. The absolute value of the x-axis intercept of a linear least-squares fit to the data, 120 ± 20 µg L⁻¹, represents the experimentally determined concentration of PFOS in the diluted groundwater sample. The concentration of PFOS in undiluted FT3 groundwater is 2.5 times that value, 300 ± 50 µg L⁻¹ (17% relative error). The dilution process is described in the Experimental Section.

Fig. 1 A plot of the mass-spectral intensity ratio I(499)/I(265) vs. concentration of the perfluorooctylsulfonate anion (PFOS, C₈F₁₇SO₃⁻) added to groundwater sample FT3 (collected at Wurtsmith Air Force Base, Oscoda, MI) that was diluted 2.5 times with 1∶1 (v∶v) acetonitrile∶water and spiked with 133 µg L⁻¹ dodecylsulfate. Each data point is the average over the four replicates and the error bars shown are for ±1 standard deviation. The absolute value of the x-axis intercept, 120 ± 20 µg L⁻¹, represents the experimentally determined concentration of PFOS in the diluted groundwater sample. Therefore, the concentration of PFOS in undiluted FT3 groundwater is 2.5 times that value, 300 ± 50 µg L⁻¹.

No peaks were detected in the (−)ES-MS spectra of these samples for other PFS anions with three, four, five, or seven carbon atoms. However, there was a detectable peak corresponding to C₆F₁₃SO₃⁻ (PFHxS, m/z 399). A separate standard-addition experiment was not possible since neither PFHxS nor its sulfonyl fluoride precursor are commercially available. Based on the assumption that the sensitivity coefficient of PFOS and PFHxS are equal (see above), the concentration of PFHxS present was determined by comparing the intensity ratio I(499)/I(265) of the sample with no added PFOS with the intensity ratio I(399)/I(265). The result of this data analysis is that the concentration of PFHxS in undiluted FT3 groundwater is 210 ± 50 µg L⁻¹. Therefore, the PFOS to PFHxS molar ratio is 1.2 ± 0.3.

Note that the method of standard additions takes into account variations in the composition of different groundwater samples (i.e., the method employs matrix matching).³⁶ Furthermore, as noted above, the new method reported herein allows a much simpler sample preparation to be used than required by the method developed by Hansen et al. for blood sera and tissue samples³¹ or that described for the method recently reported by Moody et al.³²

Direct-injection method

In some situations, the standard-addition method may be more time consuming than desired for a quick estimate of unknown concentrations of PFS anions. To explore the possibility of a direct measurement (i.e., after addition of DDS but with no added PFOS), the following experiments were performed. A series of twenty aqueous samples (four replicates of five different concentrations) that were spiked with 133 µg L⁻¹ DDS and that varied in concentration of 25.0 µg L⁻¹ to 2.5 mg L⁻¹ PFOS were prepared. A log-log plot of I(499)/I(265) vs. the PFOS concentration with a linear least-squares slope of 0.93 ± 0.02 and a correlation coefficient of 0.995 is shown in Fig. 2. Each point on this calibration graph has a relative standard error ranging from 6 to 14%. This plot demonstrates that unknown concentrations of PFOS in water can be reliably determined over this concentration range using a linear fit to the data.

Fig. 2 A log-log plot of the mass-spectral intensity ratio I(499)/I(265) vs. concentration of perfluorooctylsulfonate for a series of five standards with five different concentrations (four replicates at each concentration). Each standard was spiked with 133 µg L⁻¹ dodecylsulfate (the solvent was 1∶1 (v∶v) acetonitrile∶water). The correlation coefficient for this calibration graph is 0.995 and the slope is 0.93 ± 0.02. For some points the error bars (±1 standard deviation) are smaller than the point.

The FT3 groundwater sample was analyzed using the direct-injection method by preparing samples (four replicates each) diluted by a factor 2.5 that contained 133 µg L⁻¹ DDS as the only added standard. Using the calibration graph, the concentration of PFOS in the FT3 groundwater sample was determined to be 240 ± 40 µg L⁻¹ (cf. 300 ± 50 µg L⁻¹ PFOS by the standard-addition method). The concentration of PFHxS in the FT3 groundwater sample was calculated by the same method described in the Method of standard additions section for PFHxS. The concentration of PFHxS was determined to be 160 ± 40 µg L⁻¹ (cf. 210 ± 50 µg L⁻¹ PFHxS by the standard-addition method).

Several experiments were performed to explore the effects of the water matrix on the intensity ratio I(499)/I(265). First, three natural water matrixes (Horsetooth Reservoir (Fort Collins, CO), Cincinnati tap water, and Ohio River water) were tested using a modified standard-addition method. Spectra for all three matrixes with no added PFOS or DDS exhibited no peak for either PFOS or DDS. Samples (four replicates each) were made from each matrix to contain 133 µg L⁻¹ DDS and two different concentrations of PFOS (50 µg L⁻¹ and 250 µg L⁻¹). The (−)ES-MS intensity ratios I(499)/I(399) for these standard additions are listed in Table 2. Both Cincinnati tap water and water from the Ohio River had significantly different values of I(499)/I(399) compared to the standards. The intensity ratios of Horsetooth Reservoir water were very close to the expected intensity ratio for the 50 µg L⁻¹ PFOS sample and it had the same intensity ratio for the 250 µg L⁻¹ PFOS sample, indicating that this matrix had the least effect on the ionization efficiency of the standards within the spectrometer. Therefore, the direct-injection method can to be used with the Horsetooth Reservoir water samples. However, the variation of the intensity ratios of Cincinnati tap water and Ohio River water shows that the method of standard additions must be used for quantification. Each new matrix would have to be tested as described here to determine the proper method of quantification.

Table 2 Comparison of the (−)ES-MS intensity ratio I(499)/I(265) of standards with the I(499)/I(265) ratio of natural water samples spiked with 133 µg L⁻¹ DDS and two different concentrations of PFOS^a

	50 µg L^−1 PFOS	250 µg L^−1 PFOS
a The data are reported with ±1 standard deviation.
Standards	1.0 ± 0.1	4.7 ± 0.7
Horsetooth Reservoir water	1.3 ± 0.1	5.4 ± 0.2
Cincinnati tap water	1.9 ± 0.3	9 ± 2
Ohio River water	1.9 ± 0.2	7 ± 1

---

Next, the ionic strength was varied to check for changes in the intensity ratio I(499)/I(265) compared with low ionic strength standards. The (−)ES-MS spectra of a series of four aqueous samples (three replicates each) that were 50 µg L⁻¹ PFOS and 133 µg L⁻¹ DDS and varied in specific conductivity from 1.4 µS cm⁻¹ to 742 µS cm⁻¹ by addition of Na₂SO₄ were recorded (the 1.4 µS cm⁻¹-samples had no added Na₂SO₄). The values of I(499)/I(265) varied randomly from 0.18 to 0.36, indicating that a change in specific conductivity over the range 1.4–742 µS cm⁻¹ resulting from a change in ionic strength will result in concentrations of PFOS determined by the direct-injection method that are reliable to within a factor of two.

Finally, the (−)ES-MS spectra of a series of three aqueous samples (three replicates each) that were 50 µg L⁻¹ PFOS and 133 µg L⁻¹ DDS and varied in pH from 6–8 were recorded. The pH was adjusted with solid Na₂HPO₄/KH₂PO₄ buffer (pHydrion buffer; Micro Essential Lab, Inc.) The values of I(499)/I(265) varied randomly from 0.36 to 0.60, indicating that concentrations of PFOS determined by the direct-injection method are reliable to within a factor of 2 over this pH range.

We did not investigate the use of tandem mass spectrometry since the goal of the study was to develop a simple, fast method for quantification of PFS anions. The use of MS-MS fragmentation experiments could be used when evaluating samples from sources where PFS anions could be confused with other anions present in solution.

Analysis of an AFFF concentrate

One formulation of AFFF, FC-203CF LIGHT WATER Brand Aqueous Film Forming Foam^® (3M), was analyzed by (−)ES-MS for the quantitative determination of PFS anions. Four samples were prepared by diluting the AFFF concentrate with a solution of 1∶1 (v∶v) acetonitrile∶water to a 40,000-fold final dilution. The samples also contained 133 µg L⁻¹ DDS as an internal standard. The (−)ES-MS spectrum of the dilute AFFF is shown in Fig. 3. The characteristic peaks of six PFS anions can be seen at m/z 249 (C₃F₇SO₃⁻), 299 (C₄F₉SO₃⁻), 349 (C₅F₁₁SO₃⁻), 399 (C₆F₁₃SO₃⁻), 449 (C₇F₁₅SO₃⁻), and 499 (C₈F₁₇SO₃⁻). Note that only five PFS anions are listed in the MSDS sheet for this formulation of AFFF.²⁷ Two alkylsulfate salts are listed in the MSDS sheet and their peaks can be seen at m/z 209 (C₈H₁₇OSO₃⁻) and 237 (C₁₀H₂₁OSO₃⁻).²⁷ In a parallel experiment for which DDS was not added, no (−)ES-MS peak was observed at m/z 265. Therefore, since DDS or another species with m/z 265 is not present in this formulation of AFFF, DDS is an appropriate internal standard. Peaks are also present for C₇F₁₅CO₂⁻ (m/z 413) and C₅F₁₁CO₂⁻ (m/z 313) as well as for their C₇F₁₅⁻ and C₅F₁₁⁻ fragments at m/z 369 and 269, respectively. Although (−)ES-MS is a soft ionization technique that does not normally cause fragmentation, control experiments with pure samples of C₁₁F₂₃CO₂H and C₅F₁₁CO₂H proved that decarboxylation fragments do form under the instrumental conditions used in this study. Therefore, perfluoroalkylcarboxylate anions are present in at least one formulation of AFFF even though they are not listed in either the patent or the MSDS sheet.^27,28

Fig. 3 Negative-ion electrospray ionization mass spectrum of FC203-LIGHT WATER aqueous film-forming foam. The commercial concentrate was diluted 40,000 times with 1∶1 (v∶v) acetonitrile∶water and spiked with 133 µg L⁻¹ dodecylsulfate (m/z 265). The vertical scale from m/z 230–520 has been expanded eight times. The peaks at m/z 249, 299, 349, 399, 449, 499 are assigned to the perfluoropropyl-, perfluorobutyl-, perfluoropentyl-, perfluorohexyl-, perfluoroheptyl-, and perfluorooctylsulfonate anions, respectively. The peaks at m/z 209 and 237 are assigned to C₈H₁₇OSO₃⁻ and C₁₀H₂₁OSO₃⁻, respectively. The peaks at m/z 413, 369, 313, and 269 are assigned to C₇F₁₅CO₂⁻, C₇F₁₅⁻, C₅F₁₁CO₂⁻, and C₅F₁₁⁻, respectively.

Concentrations of the PFS anions present in this formulation of AFFF were determined using a modified version of the method of standard additions. Four samples were prepared by adding 250 µg L⁻¹ PFOS, 150 µg L⁻¹ PFBS, and 133 µg L⁻¹ DDS to the 40,000-fold diluted AFFF. Four samples were also prepared by adding only 133 µg L⁻¹ DDS to the dilute AFFF. The concentrations of PFOS and PFBS in the AFFF concentrate were determined, by comparing intensity ratios of the unspiked AFFF samples to the samples containing the known concentrations of the three standards and AFFF, to be 9 ± 1 mg L⁻¹ PFOS and 0.6 ± 0.1 mg L⁻¹ PFBS. The concentrations of the other four PFS anions were determined by comparing the intensity ratios of the unspiked samples and concentrations of both PFOS and PFBS to the intensity ratios of the PFS anion of unknown concentrations. This method gives concentration ranges of 0.4–1.5 mg L⁻¹ for C₃F₇SO₃⁻, 0.3–1.3 mg L⁻¹ for C₅F₁₁SO₃⁻, 0.6–2.7 mg L⁻¹ for C₆F₁₃SO₃⁻, and 0.1–0.4 mg L⁻¹ for C₇F₁₅SO₃⁻.

Conclusions

We have herein presented a method for the quantification of perfluoroalkylsulfonate (PFS) anions in aqueous solution. The method allows for the identification and quantification of individual PFS anions instead of the total (collective) concentration of PFS anions present in solution.³⁰ Other published methods for quantifying these anions require at least one extraction or separation step, making the methods more time consuming and more complex.^31,32 Our method has a limit of detection of 5.0 µg L⁻¹ and a linear calibration range of 25.0 µg L⁻¹ to 2.5 mg L⁻¹ for C₈F₁₇SO₃⁻ (PFOS). Both of these values are within the same order of magnitude as those reported by Hansen et al.³¹ Moody et al. have reported a linear calibration range of 0.85–208 µg L⁻¹ for PFOS using liquid chromatography–tandem mass spectrometry,³² slightly lower than that reported here. They also reported a ¹⁹F NMR method that has a much higher linear calibration range and a higher detection limit.³² Importantly, the two methods reported by Moody et al. gave quantitative results for PFOS in surface water that were significantly different from each other, presumably because of the presence of other surfactants that yield ¹⁹F NMR spectra similar to that of PFS anions. We recommend that the method of standard additions presented in this paper be used to quantify PFOS in aqueous solutions. The direct-injection method can be used when the matrix involved is less complex or when a less precise determination of PFOS concentration is acceptable.

Acknowledgements

This work was supported by a grant from the National Science Foundation (CST–9726143) and, in part, by Electrox, Inc. We thank B. J. Clapsaddle and Dr. J. A. Field for expert advice, Dr. F. E. Behr of 3M Company for samples of C₈F₁₇SO₂F and C₄F₉SO₂F and Dr. E. T. Urbansky of the US Environmental Protection Agency for the samples of Cincinnati tap water and Ohio River water.

References

1. A.-K. Sohlenius, A. M. Eriksson, C. Hogstrom, M. Kimland and J. W. DePierre, Pharmacol. Toxicol., 1993, 72, 90.
2. R. G. Schnellmann, Toxic. in Vitro, 1990, 4, 71.
3. J. N. Meussdoerffer and H. Niderprüm, Chem. Z., 1980, 104, 45.
4. A. Tullo, Chem. Eng. News, 2000, 78(May 29), 12.
5. A. Tullo, Chem. Eng. News, 2000, 78(May 22), 9.
6. D. B. Chan and E. S. K. Chian, Environ. Prog., 1986, 5, 104.
7. Fed. Regist., 2000, 65(202),p. 62319.
8. J. P. Giesy and K. Kannan, Environ. Sci. Technol., 2001, 35, 1339.
9. K. Kannan, J. Koistinen, K. Beckmen, T. Evans, J. F. Gorzelany, K. J. Hansen, P. D. Jones, E. Helle, M. Nyman and J. P. Giesy, Environ. Sci. Technol., 2001, 35, 1593.
10. K. Kannan, J. C. Franson, W. W. Bowerman, K. J. Hansen, P. D. Jones and J. P. Giesy, Environ. Sci. Technol., 2001, 35, 3065.
11. G. W. Olsen, J. M. Burris, J. H. Mandel and L. R. Zobel, J. Occup. Environ. Med., 1999, 41, 799.
12. C. A. Moody and J. A. Field, Environ. Sci. Technol., 2000, 34, 3864.
13. C. A. Moody, G. N. Hebert, S. H. Strauss and J. A. Field, Environ. Sci. Technol. , submitted for publication.
14. B. Smart, in Chemistry of Organic Fluorine Compounds II (ACS Monograph 187), ed. M. Hudlicky and A. E. Pavlath, American Chemical Society, Washington, DC, 1995, p. 979.
15. V. Gloeckner, K. Lunkwitz and D. Prescher, Tenside, Surfactants, Deterg., 1989, 26, 376.
16. E. Kissa, Fluorinated Surfactants: Synthesis, Properties, Applications, Marcel Dekker: New York, 1994.
17. B. D. Key, R. D. Howell and C. S. Criddle, Environ. Sci. Technol., 1998, 32, 2283.
18. J. D. Johnson, S. J. Gibson and R. E. Ober, Fundam. Appl. Toxicol., 1984, 4, 972.
19. R. L. Darwin, R. E. Ottman, E. C. Norman, J. E. Gott and C. P. Hanauska, NFPA Journal, 1995, May/June, 67.
20. Y. Hashimoto, DIC Tech. Rev., 1995, 1, 9.
21. S. Szonyi and A. Cambon, Fire Saf. J., 1990, 16, 353.
22. F. Szonyi, S. Szonyi and A. Cambon, Comun. Jorn. Com. Esp. Deterg., 1991, 22, 297.
23. A. A. Briggs, Fire Technol., 1979, 15, 20.
24. C. A. Moody and J. A. Field, Environ. Sci. Technol., 1999, 33, 2800.
25. M. Henley, H. Mayfield, and T. Shelley, presented at the American Chemical Society Pittsburgh Conference, Atlanta, GA, 1997.
26. A. D. Levine, E. L. Libelo, G. Bugna, T. Shelley, H. Mayfield and T. B. Stauffer, Sci. Total Environ., 1997, 208, 179.
27. MSDS sheet for 3M's FC-203CF LIGHT WATER Brand Aqueous Film Forming Foam.
28. R. Alm and R. M. Stern, US Pat., 5085786, 1992.
29. P. A. Lyon, K. B. Tomer and M. L. Gross, Anal. Chem., 1985, 57, 2984.
30. M. Laikhtman and J. S. Rohrer, J. Chromatogr., A, 1998, 822, 321.
31. K. J. Hansen, L. A. Clemen, M. E. Ellefson and H. O. Johnson, Environ. Sci. Technol., 2001, 35, 766.
32. C. A. Moody, W. C. Kwan, J. W. Martin, D. C. G. Muir and S. A. Mabury, Anal. Chem., 2001, 73, 2200.
33. R. B. Cole, Electrospray Ionization Mass Spectrometry: Fundamentals, Instrumentation, and Applications, John Wiley and Sons, New York, 1997.
34. K. S. Webb, P. B. Baker, N. P. Cassells, J. M. Francis, D. E. Johnston, S. L. Lancaster, P. S. Minty, G. D. Reed and S. A. White, J. Forensic Sci., 1996, 41, 938.
35. P. Kebarle and L. Tang, Anal. Chem., 1993, 65, 972A.
36. D. C. Harris, Quantitative Chemical Analysis, W. H. Freeman and Company, New York, 4th edn., 1995, pp. 139–142.

---

Footnote

† Abbreviations: PFS, perfluoroalkylsulfonate; PFOS, perfluorooctylsulfonate; DDS, dodecylsulfate; PFBS, perfluorobutylsulfonate; PFHxS, perfluorohexylsulfonate; AFFF, aqueous film-forming foam; (−)ES-MS, negative-ion electrospray mass spectrometry

---