Stable Carbon Isotope Analysis of Perfluorooctanoic Acid (PFOA) by Microflow‐High Pressure Liquid Chromatography‐Orbitrap Mass Spectrometry

ABSTRACT

Rationale

As contamination of perfluoroalkyl and polyfluoroalkyl substances (PFAS) in the environment becomes better understood, it is increasingly important to develop forensic tools capable of tracing contamination to emission sources. Isotopic methods have been previously inaccessible for PFAS due to limitations of the instrumentation, but high‐pressure liquid chromatography (HPLC) coupled to Orbitrap mass spectrometry (Orbitrap MS) through electrospray ionization (ESI) allows measurement of the isotopic composition of individual PFAS compounds, such as perfluorooctanoic acid (PFOA).

Methods

Six separate supplier‐procured powders of PFOA were analyzed for their δ ¹³C isotopic composition by microflow‐HPLC‐ESI‐Orbitrap MS.

Results

The method presented replicated, within error, the δ ¹³C values measured using an elemental analyzer‐isotope ratio mass spectrometer (EA‐IRMS) for the six different supplier‐procured PFOA powders. The offset from the EA‐IRMS values for repeated analysis was between 0.2‰ and 1.1‰ and the error was between 0.8‰ and 1.5‰. Our Orbitrap‐MS method requires approximately 0.04% of the material required to make an EA‐IRMS measurement.

Conclusions

We developed a method capable of measuring the carbon isotopic composition of PFOA, using HPLC‐Orbitrap MS, with high precision and accuracy. This will allow future research to expand analytical capabilities for additional isotopologues of PFOA and other PFAS compounds.

1. Introduction

Per‐ and polyfluoroalkyl substances (PFAS) are industrially derived fluorinated organic compounds that persist in the environment, and many have been shown to be detrimental to human and ecosystem health [1, 2]. There are more than 12,000 distinct molecules that are considered PFAS compounds [3]. PFAS are used in many industrial processes, consumer products, and firefighting applications, such as textile manufacturing, in the coating for non‐stick pans and raincoats, and as a component of foams used by urban/suburban firefighting departments and military bases [1]. Their accidental or purposeful release to the environment represents one of the major environmental concerns of the 21st century.

PFAS transport in the environment has become so pervasive that PFAS are essentially ubiquitous in all environmental matrices [2, 4]. Quantification of PFAS in the environment has been a robust field of study for more than twenty years, with studies investigating sites from as close to home as tap water used for drinking [5] to as remote as the middle of the Pacific Ocean [6] or the top of Mount Everest [7].

Techniques to identify the origin and source attenuation of PFAS in the environment are novel and remain in the development stage at present [8, 9, 10]. There are numerous manufacturers of PFAS compounds, and the major compounds, such as perfluorooctanoic acid (PFOA) and perfluorooctane sulfonic acid (PFOS), are produced by many of the manufacturers of PFAS, or they were before halting production of PFOA and PFOS in favor of shorter‐chain and other novel PFAS alternatives [11]. Some replacement compounds are company‐specific [12, 13, 14, 15], but could still come from numerous manufacturing facilities owned by the same company or enter natural systems via alternative pathways. Additionally, compounds like PFOA and PFOS have been shown to form from the degradation of precursors such as fluorotelomer alcohols or fluorinated polymers in the environment [16]. As a result, methods are required to forensically identify the source of PFAS contamination in different settings [17]. This is an initial and crucial step in regulating the environmental contamination of PFAS. Current techniques for the chemical fingerprinting of PFAS include identification of potentially “unique” PFAS compounds and the detection of ratios of different PFAS isomers, as compared to a putative source [8]. Newer technologies such as non‐targeted analysis with high resolution mass spectrometers and newer compound identification software present another potential novel tool for the chemical fingerprinting of PFAS [8].

Initial work performing chemical fingerprinting of PFAS looked at the relative abundance of branched isomers of PFAS compounds such as PFOA and PFOS [8, 18], where the branched isomer is a byproduct of the synthetic process [19]. The presence of branched isomers is an indication of electrochemical fluorination (ECF), which is one of the two primary synthesis pathways of perfluoroalkyl substances such as PFOA and PFOS [20, 21, 22]. The ECF synthesis procedure fully fluorinates an intact carbon chain structure [23]. Thus, there should be no theoretical alteration of the stable carbon isotope composition of the starting material. However, the origin and modification or purification of the starting material is proprietary information, making it difficult to hypothesize potential isotopic fractionations incurred prior to the ECF procedure. The absence of branched isomers in a PFAS sample is potentially indicative of telomerization as the synthesis pathway [18, 21]. Telomerization sequentially stitches CF₂ groups together to form the carbon chain of the PFAS [20]. Telomerization theoretically has a higher potential for carbon isotope fractionation during the synthesis process due to the kinetic isotope effects of sequential formation of the C‐C bonds as well as the likely associated isotopic distillation (Rayleigh distillation) of the starting material as the reaction proceeds. Consequently, telomerization presents possibilities for distinct isotope effects resulting in isotopic fingerprints across the carbon chain. No studies to date have analyzed stable carbon isotope fractionation imparted by telomerization of PFAS. These different pathways present possibilities for isotope fractionation and embedded isotopic fingerprints [24].

Recent studies have demonstrated stable carbon and sulfur isotopic differences in individual PFAS molecules such as PFOA and PFOS [25] that might be exploited for use in source apportionment of contaminated environmental samples. Compound‐specific stable isotopic analysis (CSIA) has been used previously for forensic purposes (e.g., [26, 27]), such as for identifying counterfeit or fake food and wine products (e.g., [28]) or determining whether testosterone in athletes' blood is naturally derived or the result of doping (e.g., [29, 30]). Recently, Dombrowski et al. [25] illustrated stable isotopic differences in identical PFAS compounds from different suppliers. Thus, stable isotope analysis of PFAS compounds is capable of matching isotopic compositions of reference material from different manufacturers to values found in the environment, allowing for forensic determination or calculation of contributions from different sources of PFAS to the contamination in the environment. Isotope analysis has long been used, alongside multivariate statistical models, to separate contributions of multiple sources to a single environmental sample (e.g., [31, 32]), something that is much more difficult for other chemical fingerprinting techniques to do. Isotopic analysis of PFAS could also be used in the future for tracking progress of bioremediation experiments (akin to analysis performed by Casiraghi et al. [33], where compound‐specific isotope analysis (CSIA) of chloroethene was used to track the progress of a bioremediation system).

Orbitrap mass spectrometry (Orbitrap MS) is gaining popularity for isotope analysis (e.g., [34, 35, 36]) and presents a unique opportunity for the isotopic characterization of PFAS molecules. Derivatization of PFAS molecules for isotopic analysis can be challenging [37, 38], particularly when the derivative adds additional atoms of carbon that require correction and increase uncertainty [39]. With Orbitrap MS, molecules can remain in the liquid phase and be introduced intact to the mass spectrometer using high pressure liquid chromatography (HPLC) coupled through electrospray ionization (ESI; [34, 36, 40, 41]). Additionally, numerous PFAS molecules, such as PFOA, do not easily convert quantitatively and completely to simple gases via thermal oxidation, as is required by a traditional continuous flow, magnetic sector isotope ratio mass spectrometer (CF‐IRMS; [42, 43]) and/or they produce chemicals such as hydrofluoric acid in the conversion process to simple gases, rapidly degrading the capabilities of the instrument. Orbitrap MS can measure isotopic compositions on whole molecules or larger fragments of molecules [34, 44, 45], meaning that the production of hydrofluoric acid can be avoided. Orbitrap MS also requires significantly less material than other techniques used for isotope analysis. Our method requires 125 ng of PFOA to make a measurement. This compares with EA‐IRMS requiring 400 to 600 μg of PFOA [25], ¹⁹F‐NMR (nuclear magnetic resonance spectroscopy) necessitating 38 mg of PFBA (perfluorobutanoic acid; [46]), and gas chromatography‐combustion‐isotope ratio infrared spectrometry (GC‐C‐IRIS) needing 1.3 mg of PFOA (0.3 mg pure carbon; [47]) for carbon isotope analysis. Lastly, Orbitrap MS presents a possibility for performing multi‐element CSIA, where isotopic relative abundances can be determined for multiple elements in a single molecule simultaneously (e.g., [35, 48]).

We present results of a novel method developed for determining the stable carbon isotopic composition (δ ¹³C) of PFOA using HPLC‐Orbitrap MS (Figures 1 and 2) and compare our results (Figure 3) to traditional isotopic characterization of PFOA using elemental analyzer‐isotope ratio mass spectrometry (EA‐IRMS; [25]). We additionally show that the method has sufficient precision to separate multiple lots of the PFOA from one supplier from one another. We demonstrate the capability of the method by presenting results of a single isotopologue of a single fragment of PFOA, the C₇ fragment, which is the linear chain without the carboxylic acid group (Figure 6). This allows analysis of the doubly‐substituted ¹³C isotopologue, which is difficult to examine with the molecular ion due to the isobaric interference of the ¹⁸O peak. Additionally, it adds a potential second parameter for testing in a multivariate analysis to perform source apportionment and track degradation of PFOA both in controlled settings and in the environment.

FIGURE 1.

(A) Theoretical mass spectrum of the isotopologues of the molecular ion of PFOA. The inset shows the zoomed in regions of the M + 1 and M + 2 peaks. (B) Theoretical mass spectrum of the isotopologues of the C₇ fragment of PFOA. The inset shows the zoomed in regions of the M + 1 and M + 2 peaks.

FIGURE 2.

(A) Example spectra for PFOA analysis of the molecular ion. The top panel is the “Total Ion Chromatogram” or TIC, and the bottom is the mass spectrum for a single scan during the PFOA peak (scan 2788 at a retention time of 3.19 min). The molecular ion chromatographic peak results in a peak height intensity of approximately 1.1 × 10⁹ absolute abundance. The sample or “standard” is injected at t = 0 and begins to elute roughly 0.5 min after time starts. The chromatographic peak elutes for approximately 7.5 min, based on a flow rate of 4 μL/min and an injection volume of 30 μL. It then is “washed out” by the mobile phases, and the TIC returns to baseline by approximately 11 min. The method flows 100% B (100% methanol with 0.1% formic acid) from t = 0 until t = 16 min, where it then begins the wash cycle, transitioning slowly to 100% A (100% water with 0.1% formic acid) and then back to 100% B, before conditioning with 100% B prior to starting the next injection. In the mass spectrum, the mass spectral peak at 412.9658 m/z is the PFOA base peak (M0), the mass spectral peak at 413.9689 is the PFOA molecule with one ¹³C atom (the 13C peak), and the peak at 414.9706 is the PFOA molecule with one ¹⁸O atom (the 18O peak), with a minor contribution from the PFOA molecule with two ¹³C atoms. The tolerance windows and charges input into IsoX are included. (B) Example spectra for PFOA analysis of the C₇ fragment. The top panel is the “Total Ion Chromatogram” or TIC, and the bottom is the mass spectrum for a single scan during the PFOA peak (scan 613 at a retention time of 2.60 min). The PFOA peak results in a peak with a peak height intensity of approximately 4.0 × 10⁸ absolute abundance. The sample or “standard” is injected at t = 0 and follows the same chromatographic parameters as for the molecular ion analysis in (A). In the mass spectrum, the mass spectral peak at 368.9751 m/z is the C₇ fragment base peak (M0), the mass spectral peak at 369.9783 is the C₇ fragment with one ¹³C atom (the 13C peak), and the mass spectral peak at 370.9827 is the C₇ fragment with two ¹³C atoms (the 13C13C peak). The tolerance windows and charges input into IsoX are included.

FIGURE 3.

δ ¹³C values of PFOA obtained by Orbitrap‐MS as compared to δ ¹³C values obtained by EA‐IRMS in Dombrowski et al. [25]. Blue circles represent the average δ ¹³C value of PFOA for individual suppliers (see Table 1 for the list of suppliers), with error bars on both x‐ and y‐axes representing the standard deviation of the measurement for the EA‐IRMS and Orbitrap‐MS methods, respectively. The black dashed line represents the 1:1 line, where sample data points that fall on that line have the same value as determined by Orbitrap‐MS and EA‐IRMS. The purity of the Fluoryx 241009 PFOA may have impacted the isotope results, given that there is significantly more of the branched PFOA in this lot than the others.

FIGURE 6.

Comparison of δ ¹³C values of the ¹³C ratios of the molecular ion of PFOA and the ¹³C ratios of the C₇F₁₅ fragment. All delta values were calculated using the respective Alfa Aesar PFOA measurements as the “standard” value. The dark blue circles represent the δ ¹³C values of the PFOA molecular ion, not corrected to the VPDB international scale, to allow for comparison with the fragment data. The blue squares represent the δ ¹³C values of singly ¹³C substituted PFOA C₇F₁₅ fragment.

2. Experimental Section

2.1. Materials

PFOA reference materials used are the same as those used by Dombrowski et al. [25]. PFOA reference materials were purchased from five suppliers (Alfa Aesar, BeanTown, Fluoryx, Strem, and SynQuest). All solutions were made with glassware that had been combusted or precleaned with MilliQ water that had been additionally purified using an ion exchange resin to further remove PFAS (referred to as “ion exchange water”) and then LCMS‐grade methanol (Honeywell).

Stock solutions of the PFOA reference materials were made by adding known amounts of powdered PFAS to 10‐mL LCMS‐grade methanol (Honeywell) to produce stock solutions. Each reference material had approximately 10 molar equivalents of ammonium hydroxide (LCMS‐grade, Honeywell) added to the solution to stabilize PFOA, as methyl esterification is known to occur when using methanol [49]. Known aliquots of the stock solutions were taken to prepare 10‐μM working solutions by diluting into 10‐mL of LCMS‐grade methanol (Honeywell).

Solutions for the HPLC were as follows: mobile phase A was 100% water with 0.1% formic acid, mobile phase B was 100% methanol with 0.1% formic acid, weak needle wash was 100% methanol (no additive added), strong needle wash was 100% methanol with 0.1% formic acid, and rear seal wash was 75% isopropanol/25% water with 0.1% formic acid. All solutions for the HPLC were made using LCMS‐grade water (Honeywell), LCMS‐grade methanol (Honeywell), LCMS‐grade isopropanol (Honeywell), and Optima LC/MS‐grade formic acid (Fisher Chemical). The Vanquish Neo solutions were sonicated for 30 min prior to use, as is recommended by the manufacturer (Vanquish Neo manual).

2.2. Instrument Parameters for PFOA Molecular Ion

Method development/analysis was performed using a Thermo Scientific Vanquish Neo HPLC in microflow mode coupled to a Thermo Scientific Orbitrap Exploris 240 Mass Spectrometer through an atmospheric pressure ionization (API) source with a heated electrospray ionization (H‐ESI) needle. The HPLC was equipped with a 100‐μL sample loop. The HPLC injector was plumbed directly into a six‐port valve, bypassing a separation column. The six‐port valve was set up such that it could switch between a direct infusion syringe and the HPLC, either being routed to the Orbitrap or to waste. The autosampler was held at 7.0°C.

The HPLC flow was maintained at 4 μL/min for the entirety of the run. The HPLC gradient elution profile was as follows: 0.0–16.0 min (A: 0%, B: 100%), 16.0–18.0 (A: 0%–100%, B: 100%–0%), 18.0–19.0 (A: 100%, B: 0%), 19.0–21.0 (A: 100%–0%, B: 0%–100%). The HPLC then equilibrated for the next run with 100% B. 30 μL of solution was injected for each analytical run.

The system was set for negative polarity, with the spray voltage at 2200 V and the ion transfer tube at 280°C. The H‐ESI needle was set to atmospheric temperature, which was roughly 35°C for all analyses but fluctuated from day to day. While the source gases were manually tuned prior to each analytical run, the parameters did not change dramatically. As an example, the sheath gas was set at 5 arbitrary units, the aux gas at 2 arbitrary units, and the sweep gas at 0 arbitrary units (set to be off).

For the MS parameters, the quadrupole filter was set for a window of 407–421 m/z, the orbitrap resolution was set at 11250, the RF lens was set at 45%, the AGC (Automatic Gain Control) target was set at 120% (yielding approximately 1.2 × 10⁶ ions in the C‐trap for each scan), the maximum injection time was 1000 ms, and microscans were set at 2. Source fragmentation was turned off to maximize the transmission of the molecular ion through the system to the orbitrap. The internal mass calibration was turned off to maximize the number of ions from PFOA entering the orbitrap. Instead, the orbitrap was calibrated using a one‐point mass calibration in negative ion mode and a separate one‐point mass calibration in positive ion mode each day, prior to each analytical run. The orbitrap was additionally calibrated quarterly using the built‐in Mass & System calibration procedure with the Thermo Scientific Pierce Flex‐Mix Calibration Solution.

2.3. Data Analysis for PFOA Molecular Ion

Mass spectra were processed using Thermo Scientific IsoX Software (Version 2022) and RStudio (2023.09.1 + 494). The .tsv file that was used to process the data in IsoX was downloaded from IsoXL (https://isoorbi.shinyapps.io/IsoXL/) and the parameters, including the tolerance window for each mass and the isotopologues included in the data analysis, are in Figure 2. Parameters, particularly the masses of the individual isotopologues, were adjusted on a day‐by‐day basis, selecting values that captured the variability observed across the entire analytical sequence. Variability in the range of 5–10 mmu (millimass units) was observed from day to day. Isotopologue information was extracted by IsoX from the .raw files produced by the orbitrap for each scan in each file (each individual injection), using the parameters in the .tsv file (Figure 2). That extracted information in the processed IsoX file was then put through RStudio code to extract the base peak (M0), the ¹³C peak, and the ¹⁸O peak for just the period of the analytical run that included the sample injection (not the washout time), flag and remove outliers, calculate ratios (R = ¹³C abundance/M0 abundance and R = ¹⁸O abundance/M0 abundance), and average those ratios over the total number of scans identified for each sample injection (within the analytical window identified). While we have generated results for ¹⁸O (with precision on the order of 0.5‰–1.0‰), they have not been validated against TCEA‐IRMS values, so we have elected to not present them here. RStudio output the averaged ratios for each sample for carbon and oxygen to a Microsoft Excel file. These data were then sorted and used to calculate δ values, by comparing “samples” and “standards” using delta notation [35]. The resulting δ values for carbon were converted to the international scale, VPDB, following the method of Kantnerova et al. [35].

2.4. Instrument Parameters for PFOA C₇F₁₅ Fragment

Instrument setup and HPLC parameters are identical to those in Section 2.2. The API/H‐ESI source was set in negative ion mode, with the spray voltage at 2200 V and the ion transfer tube at 280°C. The H‐ESI needle was set to atmospheric temperature, which was roughly 35°C for all analyses but fluctuated from day to day. While the source gases were manually tuned prior to each analytical run, the parameters did not change dramatically. As an example, the sheath gas was set at 5 arbitrary units, the aux gas at 2 arbitrary units, and the sweep gas at 0 arbitrary units (set to be off).

The orbitrap was set to MS² scan mode, with a precursor m/z value of 412.9664, a precursor charge state of −1, and an isolation width of 5 m/z. Using an HCD collision energy of 12 V and a scan range of 364–376 m/z, the C₇F₁₅ fragment was targeted. The orbitrap resolution was set at 22500, the RF lens was set at 45%, the AGC target was set at 120% (yielding approximately 1.2 × 10⁶ ions in the C‐trap for each scan), the maximum injection time was 1000 ms, and microscans were set at 5. The internal mass calibration was turned off to maximize the number of ions from PFOA entering the orbitrap.

2.5. Data Analysis for PFOA C₇F₁₅ Fragment

Mass spectra were processed in an identical fashion to Section 2.3, with different masses substituted in the .tsv file (Figure 2). The .tsv file that was used to process the data in IsoX was also downloaded from IsoXL (https://isoorbi.shinyapps.io/IsoXL/) and the parameters, including the tolerance window for each mass and the isotopologues included in the data analysis, are in Figure 2. Parameters, particularly the masses of the individual isotopologues, were adjusted on a day‐by‐day basis, selecting values that captured the variability observed across the entire analytical sequence. However, there is no conversion to the international isotopic scale because there is no EA‐IRMS value for the C₇F₁₅ fragment.

3. Results and Discussion

Six separate PFOA powders were measured for δ ¹³C values of PFOA. Each had a distinct isotopic value as determined by EA‐IRMS: Alfa Aesar 10199077 (−28.7‰ ± 0.4‰), BeanTown 50025271 (−27.6‰ ± 0.2‰), Strem 34280400 (−27.8‰ ± 0.1‰), SynQuest 00022778 (−46.1‰ ± 0.7‰), Fluoryx 241009 (−30.5‰ ± 0.5‰), and Fluoryx 0020TC (−26.9‰ ± 0.4‰; Table 1; [25]). Our Orbitrap MS method can determine the values of these PFOA materials to within between 0.2‰ and 1.1‰ of the EA‐IRMS value (Figure 3 and Table 1). This excludes the results of Fluoryx 241009; this PFOA was not 100% pure [25] and the impurities may have affected the resulting isotope values determined by EA‐IRMS. The δ values determined for the Orbitrap‐MS method were determined by designating one supplier‐procured PFOA as the “standard” and the others were treated as “samples.” This was then rotated for the next analytical run (on a separate day), such that all supplier‐procured PFOA “samples” were corrected by all supplier‐procured PFOA “standards” over the course of 6 days of analysis. The reported values are averages across all those individual analyses (Figure 3 and Table 1), which are reported in the Supporting Information.

TABLE 1.

Comparison of the δ ¹³C values determined by EA‐IRMS by Dombrowski et al. [25] and the δ ¹³C values determined by Orbitrap MS by this study. The values determined by Orbitrap MS by this study are determined as the average of 20 individual determinations of the δ ¹³C values of the solution from each supplier. For the Orbitrap MS δ ¹³C values, the different supplier solutions were used to correct the δ value determined by the Orbitrap MS method for each other. The purity of the Fluoryx 241009 PFOA may have impacted the isotope results.

PFOA supplier	Lot number	EA‐IRMS δ 13C	Orbitrap MS δ 13C
Alfa Aesar	10199077	−28.7‰ ± 0.4‰	−29.0‰ ± 1.1‰
BeanTown	50025271	−27.6‰ ± 0.2‰	−27.8‰ ± 1.5‰
Strem	34280400	−27.8‰ ± 0.1‰	−27.4‰ ± 1.0‰
SynQuest	00022778	−46.1‰ ± 0.7‰	−45.0‰ ± 0.8‰
Fluoryx	241009	−30.5‰ ± 0.5‰	−33.2‰ ± 2.2‰
Fluoryx	0020TC	−26.9‰ ± 0.4‰	−27.1‰ ± 1.3‰

The average Orbitrap MS δ ¹³C value for Alfa Aesar PFOA (n = 20) is −29.0‰ ± 1.1‰, an offset of −0.3‰ from the EA‐IRMS value. The individual Alfa Aesar PFOA δ ¹³C values range from −30.9‰ to −27.0‰, a range of 3.9‰. The average Orbitrap MS δ ¹³C value for BeanTown PFOA (n = 20) is −27.8‰ ± 1.5‰, an offset of −0.2‰ from the EA‐IRMS value. The individual BeanTown PFOA δ ¹³C values range from −31.6‰ to −25.3‰, a range of 6.3‰. The average Orbitrap MS δ ¹³C value for Strem PFOA (n = 20) is −27.4‰ ± 1.0‰, an offset of 0.4‰ from the EA‐IRMS value. The individual Strem PFOA δ ¹³C values range from −29.6‰ to −25.3‰, a range of 4.3‰. The average Orbitrap MS δ ¹³C value for SynQuest (n = 20) PFOA is −45.0‰ ± 0.8‰, an offset of 1.1‰ from the EA‐IRMS value. The individual SynQuest PFOA δ ¹³C values range from −46.1‰ to −42.8‰, a range of 3.3‰. The average Orbitrap MS δ ¹³C value for Fluoryx PFOA, lot 241009, (n = 20) is −33.2‰ ± 2.2‰, an offset of −2.7‰ from the EA‐IRMS value. The individual Fluoryx lot 241009 PFOA δ ¹³C values range from −36.3‰ to −30.2‰, a range of 6.1‰. Given that this PFOA was not 100% pure and that there was significantly more branched PFOA in this lot (the branched peak for this lot, as determined by LC–MS/MS, had a peak area three times higher than the other lots, as well as the presence of other PFAS) as compared to the others [25], it is likely that the offset is the result of impurities affecting the EA‐IRMS measurement. The average Orbitrap MS δ ¹³C value for Fluoryx PFOA, lot 0020TC, (n = 20) is −27.1‰ ± 1.3‰, an offset of −0.2‰ from the EA‐IRMS value. The individual Fluoryx lot 0020TC PFOA δ ¹³C values range from −29.7‰ to −24.8‰, a range of 4.9‰. These values were all determined using the same concentration of PFOA, 10 μM.

The precision of the method is sufficient to separate certain PFOA lots from one another. Using ANOVA testing, SynQuest 00022778 and Fluoryx 241009 are statistically distinguishable (α value of 0.05) from all other PFOA lots. Alfa Aesar 10199077, BeanTown 50025271, Strem 34280400, and Fluoryx 0020TC are not statistically distinguishable from one another. However, we anticipate with further isotopologue analysis, those standards will separate in multivariate (multi‐element‐isotope) space. Additionally, see Figure S7 for further statistical testing of the comparison of the δ ¹³C_EA and δ ¹³C_Orbitrap values.

Prior to the analysis of the different lots of PFOA, numerous method parameters were tested, all using a single PFOA lot (Alfa Aesar) with varying conditions in the solution or inside the orbitrap. A single lot of PFOA was used for consistency; however, different reference lots were used for delta calculation. Initial method parameters were determined by direct infusion of PFOA solution using the syringe pump and modifying source and mass spectrometer parameters to yield the highest signal and the lowest scan‐to‐scan error. A PFOA solution of 100‐μM PFOA and 4‐M equivalents of sodium hydroxide was used for this initial parameter determination, based on the published method of Kantnerova et al. [35] and the conditions inside the vial of EPA‐1633STK standard purchased from Wellington Laboratories (for the base concentration and composition). To more closely match environmental conditions, the target concentration of PFOA was lowered to 10 μM. To more closely match the extraction procedure of samples utilized, the base composition was changed from sodium hydroxide to ammonium hydroxide [50].

One of the first parameters decided upon was the resolution. For δ ¹³C analysis, we decided on a resolution of 11250, which is the lowest resolution setting on the Orbitrap Exploris 240. As the only interference for the ¹³C peak is from ¹⁷O, which represents only a fraction of the abundance of ¹³C (natural abundance of ¹⁷O is 0.037% and natural abundance of ¹³C is 1.1%; see also Figure 1B). By decreasing the resolution and absorbing the small ¹⁷O peak, we were able to increase the scan rate, allowing for better counting statistics, without having to increase runtime or injection size. Due to the destructive interference of the ion ringlets, it appears that the ¹⁷O peak has very little impact, if any, on the resulting δ ¹³C values, as is evident from the results in Figure 3.

For initial method parameters, the quadrupole window was set using the molar mass of PFOA and assuming room needed on either side to prevent the quadrupole from causing isotopic fractionation [34]. Even though the observations made in Eiler et al. [34] were made on a GC Orbitrap system, which is known to have more significant quadrupole fractionation, we observed that a window of at least 2 Da was required to not affect the detection of the isotopologues of interest. The AGC target was set at 30% to start based on the recommendation of Thermo Scientific engineers. The ion transfer tube temperature, source voltage, RF lens, maximum injection time, and microscans, along with initial targets for the source gas pressures were all set by infusing PFOA at a constant 4‐μL/min rate from the syringe pump and monitoring the total ion chromatograph (TIC; signal intensity) and relative standard deviation (RSD; signal stability) to find the optimal value that produced the highest TIC and lowest RSD combination. The source voltage, ion transfer tube temperature, and RF lens were all set in order to maximize molecular ions versus forming fragment ions inside the source as much as possible, to allow measurement of the molecular ion. This was tested by examining the relative peak areas of the different fragments as parameter values were changed, and the parameter value that yielded the maximum relative amount of the molecular ion, while also maximizing the TIC and yielding the lowest RSD, was chosen (see Figures S1 [spray voltage], S2 [ion transfer tube temperature], and S3 [RF lens] for plots detailing these tests and Table S1 for additional details of the ions extracted for these tests).

Once initial conditions were set, the AGC target was tested by injecting the same PFOA solution through the system and varying the AGC target, while keeping all other parameters the same. The AGC target is a parameter controlled by the software that regulates the number of ions that can travel into the orbitrap for any given scan to control for and regulate space‐charge effects [34]. As shown by our results and the results of others, the AGC target has a dramatic impact on the resulting isotope value determined. The resulting ¹³C/¹²C ratio produced during these trials leveled off at a value of 120%, corresponding to 1.2 × 10⁶ ions allowed in the orbitrap at a time (see Figures S5 and S6 for further details). It was discovered that the Orbitrap software was throttling back the AGC target at a value above 120%, so we utilized the AGC value of 120% as the highest point after the ¹³C/¹²C ratios leveled off before the throttling began. This was discovered because we attempted to set the AGC higher, and the software started throttling the AGC via a parameter called AGC Target adjust. The software is attempting to prevent more than a certain number of ions of any single mass, even if the total number of ions it can accommodate could be more.

The next parameter that was tested was the relative concentration of base (ammonium hydroxide) that produced consistent isotopic results (Figure 4). It has been previously demonstrated that the pH can affect the efficiency of ionization inside the electrospray ionization source [51]. Relative concentrations of ammonium hydroxide were tested using two different PFOA concentrations: 10‐μM PFOA and 100‐μM PFOA. Separate PFOA solutions, all using the same PFOA lot, were made with varying relative concentrations of ammonium hydroxide. These solutions were then diluted to 100‐μM PFOA and 10‐μM PFOA for analysis. Solutions had relative ammonium hydroxide concentrations ranging from 0 molar equivalents to 20 molar equivalents, totaling 19 different ammonium hydroxide concentrations. After being analyzed without any separate PFOA from another supplier and only looking at the change in the ¹³C/¹²C ratio (Figure 4A,C), they were analyzed against a different supplier PFOA lot (with 10‐μM PFOA and 10 molar equivalents) to allow delta values to be determined (Figure 4B,E) and offsets to be calculated (Figure 4D,F). Based on the results of both sets of analysis, we determined that the optimal relative ammonium hydroxide concentration was 10 molar equivalents. This was determined based on 10 molar equivalents yielding a value closest to the EA‐IRMS value with a relatively stable and reproducible result (Figure 4).

FIGURE 4.

The results of varying ammonium hydroxide relative concentrations on the resulting PFOA δ ¹³C values. Plots in the same column (A–C or D–F) are data that were generated on the same day. Plots in the same row (A and D, B and E, and C and F) are the same type of data (¹³C/¹²C ratio, delta value, or delta value offset). For all plots, the relative ammonium hydroxide concentration, in molar equivalents, is plotted on the x‐axis. Plots A and D have the carbon‐13 ratio determined for each injection plotted on the y‐axis. Plots B and E have the δ ¹³C value calculated using bracketing “standards” (a different lot of PFOA) is plotted on the y‐axis. Plots C and F have the offset of the δ ¹³C value from the EA‐IRMS determined value [25]. Each data point represents the results of three separate injections of the PFOA solution with that ammonium hydroxide concentration. Error bars are the standard deviations across the three replicates.

The last parameter that was tested for potential effects on isotopic measurement was the PFOA concentration. Using the stock solution of PFOA (Alfa Aesar) made with 10 molar equivalents of ammonium hydroxide, different dilutions were made to achieve nine different concentrations of PFOA, ranging from 1‐μM PFOA to 25‐μM PFOA, all with 10 molar equivalents of ammonium hydroxide. Again, these were measured against a PFOA solution of a different lot to allow delta values to be calculated (Figure 5). Based on this analysis, we determined that a 10‐μM PFOA concentration was the best target concentration. The 10‐μM PFOA was chosen because its δ ¹³C value best matched the EA‐IRMS value and because its ¹⁸O/¹⁶O ratio appeared to be stable (where slight concentration variations would not dramatically affect the result).

FIGURE 5.

The results of varying PFOA concentrations on the resulting PFOA δ ¹³C values. Plots in the same column (A–C or D–F) are data that were generated on the same day. Plots in the same row (A and D, B and E, and C and F) are the same type of data. For all plots, the PFOA concentration (μM) is plotted on the x‐axis. Plots A and D have the carbon‐13 ratio determined for each injection plotted on the y‐axis. Plots B and E have the δ ¹³C value calculated using bracketing “standards” (a different lot of PFOA) is plotted on the y‐axis. Plots C and F have the offset of the δ ¹³C value from the EA‐IRMS determined value [25]. Each data point represents the results of four separate injections of the PFOA solution with that ammonium hydroxide concentration. Error bars are the standard deviations across the four replicates.

The final parameters that were determined for the PFOA molecular ion (the AGC target, the ammonium hydroxide concentration, and the PFOA concentration) were those that were used for the analysis of the C₇F₁₅ fragment. In addition, we performed the same test to optimize the HCD collision cell voltage for fragmentation as was performed for the optimization of the spray voltage, ion transfer tube temperature, and RF lens (see Figure S4 and Table S1 for further details). There is no EA‐IRMS determined value for the C₇F₁₅ fragment of PFOA for any of the purchased PFOA, so we were not able to perform quite the same analysis that we did for the molecular ion. Because of this limitation, we selected the Alfa Aesar PFOA to be the “standard” and measured the ¹³C ratio (Figure 6) and the ¹³C¹³C ratio of the C₇F₁₅ fragment (Figure 7) and calculated delta values for both against the Alfa Aesar value. The ¹³C¹³C measurement is not a “true” delta value but it was calculated in the same way. Across the different lots of PFOA, each measurement of the δ ¹³C value of the fragment had more variability than the δ ¹³C of the molecular ion, and even more variability in the ¹³C¹³C measurement. This is likely due to both fewer scans across the peak, yielding poorer counting statistics, and smaller abundances overall, yielding more room for small changes to create larger uncertainties. It is also possibly due to an unwanted fractionation during the fragmentation in the collision cell. Due to these results, we suggest that this method is preliminary and include it to demonstrate the possibilities of isotope analysis with PFAS compounds using Orbitrap MS. The fragment method can also be leveraged in the future to make position‐specific isotope analysis measurements on the carbons in the linear chain of PFOA or other pieces of other PFAS.

FIGURE 7.

Analysis of the ¹³C¹³C ratios of the C₇F₁₅ fragment of PFOA. Values in the figure were calculated in the same way as a delta value (sample minus standard and then divided by standard, multiplied by 1000) and, as in Figure 5, Alfa Aesar PFOA was used as the “standard” value. Once again, because there is not a method to measure the EA‐IRMS value of the C₇F₁₅ fragment, these values are not corrected to an international scale.

It is possible that the separation in δ ¹³C values for the C₇ fragment compared to that of the PFOA molecular ion of Fluoryx 241009, not evident for Fluoryx 0020TC, indicates different synthesis pathways for the two lots (Figure 6). Given the relative isomeric purity of most of the materials [25], it is possible that this position specific difference is indicative of some presence of ECF derived PFOA in the material procured (Fluoryx 241009). It is unknown whether manufacturers synthesize the PFAS materials themselves or purchase pre‐made material and purify it, or both. It is also possible that the difference in isotopic composition is a reflection of purity, where some co‐contaminant is filtered out during MS² Scan Mode, but is still present during Full Scan mode. In that case, the C₇ fragment δ ¹³C value would be more reflective of the “true” value of the PFOA.

Presumably, if there are differences in the PFOA δ ¹³C values of different manufacturers for scientific purposes, there are differences in the δ ¹³C values of different manufacturers for industrial purposes. The next steps for applying this method are to acquire and determine the δ ¹³C values of PFOA from different industrial manufacturers and to expand the testing of PFOA to the environment. Additionally, we plan to test how the isotopic composition of PFAS molecules changes during oxidation or hydrolytic processes that would be industrially or environmentally relevant.

4. Conclusions and Future Perspectives

In conclusion, we present a robust method for the determination of δ ¹³C values of PFOA using HPLC‐Orbitrap MS. This method can be applied to environmental samples, with the intention of being used for forensic purposes and to aid in environmental regulation of PFOA contamination in environmental settings. We are already able to separate out in δ ¹³C isotope space PFOA produced by SynQuest and one lot produced by Fluoryx from that produced by Alfa Aesar, BeanTown, Strem, and the other lot produced by Fluoryx. We plan to broaden this method to increase the number of isotopolocules detected and the molecules tested, building out a suite of isotopic tools to forensically identify PFAS contamination in the environment.

Author Contributions

Paul K. Wojtal: writing – original draft, methodology, validation, visualization, writing – review and editing, formal analysis, data curation, software. Brett Davidheiser‐Kroll: writing – review and editing, software, resources. Chad S. Lane: conceptualization, investigation, funding acquisition, writing – review and editing, project administration, supervision. Ralph N. Mead: conceptualization, investigation, funding acquisition, writing – review and editing, project administration, supervision.

Conflicts of Interest

There are no conflicts of interest to declare.

Peer Review

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer‐review/10.1002/rcm.10139.

Supporting information

Data S1: Supporting Information.

Table S1: Masses of fragment base peaks that were input into the .tsv file for IsoX for the fragmentation minimization tests performed to determine the optimal spray voltage, ion transfer tube temperature, RF lens, and HCD collision cell voltage. The theoretical mass was determined by calculating the exact mass of the base peak (all ¹²C or ¹⁶O atoms) for each fragment. The masses monitored for the spray voltage, ion transfer tube temperature, RF lens, and HCD collision cell voltage were determined by examining how the exact mass measured varied throughout the analysis. Where cells are empty in the table, there was no fragment detected for that fragment for that analysis, so the data were not extracted.

Figure S1: A) Spray voltage effect on the fragmentation of PFOA. Spray voltage was changed at increments of 50 V, with the actual spray voltage being slightly different than what was requested. The spray voltage detailed in the figure is the actual spray voltage, not what was requested. Data are shown as a stacked percentage bar plot, where the ion intensities for each fragment (only the base peak) were used to determine fractional contributions to the total ion intensity of PFOA in the orbitrap at each spray voltage. Almost all (> 99.8%) of the PFOA measured at varying spray voltages with base method parameters was in the form of the C₈F₁₅O₂ ion or the C₇F₁₅ fragment. B) Total ion chromatogram over the course of the analysis, with the plotted changes in spray voltage on the secondary y‐axis.

Figure S2: A) Ion transfer tube temperature effect on the fragmentation of PFOA. Ion transfer tube temperature was changed at increments of 10°C, with the actual ion transfer tube temperature being slightly different than the software setpoint. The temperature detailed in the figure is the actual temperature, not the software setpoint. Data are shown as a stacked percentage bar plot, where the ion intensities for each fragment (only the base peak) were used to determine fractional contributions to the total ion intensity of PFOA in the orbitrap at each spray voltage. Almost all (> 99.8%) of the PFOA measured at varying ion transfer tube temperatures with base method parameters were in the form of the C₈F₁₅O₂ ion or the C₇F₁₅ fragment. B) Total ion chromatogram over the course of the analysis, with the plotted changes in ion transfer tube temperature on the secondary y‐axis.

Figure S3: A) RF lens percentage effect on the fragmentation of PFOA. RF lens percentage was changed in increments of 5%. Data are shown as a stacked percentage bar plot, where the ion intensities for each fragment (only the base peak) were used to determine fractional contributions to the total ion intensity of PFOA in the orbitrap at each RF lens percentage. Almost all (> 99.8%) of the PFOA samples measured at varying ion transfer tube temperatures with base method parameters were in the form of the C₈F₁₅O₂ ion or the C₇F₁₅ fragment. B) Total ion chromatogram over the course of the analysis, with the plotted changes in RF lens percentage on the secondary y‐axis.

Figure S4: A) HCD collision cell voltage effect on the fragmentation of PFOA. HCD collision cell voltage was changed at increments of 1 V. Data are shown as a stacked percentage bar plot, where the ion intensities for each fragment (only the base peak) were used to determine fractional contributions to the total ion intensity of PFOA in the orbitrap at each spray voltage. B) Total ion chromatogram over the course of the analysis, with the plotted changes in HCD collision cell voltage on the secondary y‐axis.

Figure S5: The average total ion current of the orbitrap across the entire PFOA peak as a function of the AGC (automated gain control) target percentage. The AGC target percentage is a relative scale, where 100% corresponds to 1.0x10⁶ ions in the C‐trap per scan. The same solution of PFOA was used for all the analyses that make up the data in this figure, where the only variable being changed is the relative AGC percentage. The number next to each data point is the AGC target setpoint in the method.

Figure S6: The results of varying the AGC target on the resulting PFOA isotopic composition measurement: the ¹³C/¹²C ratio (A) and the δ ¹³C value, normalized to the VPDB scale (B). The AGC target (in relative percentage) is on the x‐axis for both plots. The numbers next to the black circles represent the AGC target that was set in the method.

Figure S7: York regression analysis of data in main text Figure 2. δ¹³C values of PFOA obtained by Orbitrap‐MS as compared to δ¹³C values obtained by EA‐IRMS in Dombrowski et al. (2025). Blue circles represent the average δ¹³C value of PFOA for individual suppliers (see Table 2 in the main text for the list of suppliers), with the circles representing the error associated with both the x‐ and y‐axes from the York. The black line represents the regression line and the shaded area is the 95% confidence interval.

Wojtal P., Davidheiser‐Kroll B., Lane C., and Mead R., “Stable Carbon Isotope Analysis of Perfluorooctanoic Acid (PFOA) by Microflow‐High Pressure Liquid Chromatography‐Orbitrap Mass Spectrometry,” Rapid Communications in Mass Spectrometry 39, no. 24 (2025): e10139, 10.1002/rcm.10139.

Data Availability Statement

Data are made available as a part of the Supporting Information associated with this publication. They are additionally available at the NC PFAS Testing Network website.