Methods

Standard preparation: Standards were provided by AGE (Luxembourg) and ordered from Wellington Laboratories (Guelph, Ontario).

Sample preparation: Calibration solutions were prepared by dilution of each water sample (ultrapure HPLC-grade water, tap water and bottled mineral water) with methanol at a 50:50 ratio.

Note: it is paramount for all solutions to be prepared and vialed in polypropylene vessels. This is necessary due to the potential adsorption of PFAS compounds to glass vessels, which decreases the sensitivity of analytes over time.⁵

Chromatography: Chromatographic separation was performed using the ExionLC™ AD System, which provides very low carryover and full UHPLC capabilities. The column used was a Phenomenex Luna Omega 3 µm PS C18 100 Å, 100 x 3.0 mm with a Phenomenex Luna Omega 1.6µm PS C18 100 Å, 50 x 2.1 mm delay column. Details of the chromatography are outlined in the supplementary information.³

Mass spectrometry: These experiments were performed using the QTRAP 6500+ System and the SCIEX 7500 System. The systems were operated in negative mode with electrospray ionization (ESI) using the Scheduled MRM™ Algorithm. Data was acquired using Analyst^® Software. Details for the MS conditions are outlined in the supplementary information.³

Data processing: Data was processed using SCIEX OS software. 

Quantitative performance

Due to the lowered limits associated with PFAS analysis, it has become increasingly important to provide higher levels of sensitivity to the environmental industry to ensure that the relevant LLOQs can be achieved with accuracy, precision and robustness. Good chromatography is key to providing good separation and removing any instrument contamination from the analysis (Figure 2). To characterize the sensitivity of the QTRAP 6500+ System, concentration curves were generated in HPLC grade water. Example chromatograms for PFNA across the 10-point PFNA calibration curve highlights the quality of the data (Figure 3). 

Figure 4 highlights some example calibration curves for 3 relevant PFAS compounds, including both % accuracy and %CV (%RSD) data for the calculated concentration at each linearity level. In addition to high accuracy and precision levels, sensitivity is important to ensure that specialized equipment or large injection volumes are unnecessary. The LLOQs for each PFAS monitored in the 3 different water samples are in Table 1, and example chromatograms are shown in Figure 5. 

PFAS stability in glass

A significant issue with PFAS analysis is that long-chain PFAS compounds have low solubility in water samples and tend to stick to the surface of sample containers. This effect increases with chain length, and this is the main reason that polypropylene material is recommended for storage of samples and standards, and that methanol is added during sample preparation. To evaluate this effect, samples were stored in glass vials and the peak areas for all PFAS compounds were analyzed across 34 hours. The example for PFTeDA is shown in Figure 6, and the results across all PFAS are summarized in Table 2. Of the 21 compounds analyzed, 8 were susceptible to this issue in ultrapure HPLC-grade water and 11 were seen to decrease in both tap water and mineral water matrices. To account for this decrease over long analysis times, it is important to either use polypropylene vials during storage and analysis (Figure 7) or utilize deuterated internal standards to mitigate any change in signal intensity. 

Workflow points to consider

Using a delay column

In previous PFAS analyses, contamination has been observed when using an analytical column alone. This contamination appears to originate from the LC system itself, which creates the analytical challenge of separating the contamination from the analyte of interest. To address this challenge, a pre-column or delay column has been used to ensure the observed contamination is well separated from the analyte. The column chosen for this was similar to the analytical column and utilized the same stationary phase.

Reducing injection volume

With past methods that used LC-MS/MS to quantify PFAS compounds in water, it was common to use large injection volumes—possibly even injecting multiple milliliters of sample and performing a trap-and-elute workflow to concentrate this volume and reduce peak broadening. However, this analysis leveraged the high sensitivity of the QTRAP 6500+ System, which enabled the use of a 50 µL injection volume.

Therefore, there is no need for any specialized equipment, such as a CTC autosampler, or for any significant modification to the UHPLC system, other than using a 50–100 µL sample loop. Another benefit of using a reduced injection volume is a reduction in the previously observed issues, such as a high background and contamination, which leads to a more robust and easier-to-implement workflow.  

Reducing blank contamination

Due to the widespread use of PFAS compounds, it is difficult to achieve a clean blank injection because of the prevalence of contamination that usually originates from the analytical instrumentation itself. This issue is mitigated by the use of a delay column, which means that a clean blank was achieved for all 23 analyzed compounds. This is still an issue to consider when performing PFAS analysis, however. 

Recently regulated PFAS compounds

The new EU drinking water directive (released in February 2020) included 2 new PFAS compounds of concern: L-PFUdS and L-PFTrDS. At the time of the directive’s release, the sourcing of these compounds proved to be particularly challenging, and for this reason, they were not included in the bulk of the analysis performed here. However, in late 2020, it became clear that these compounds were now commercially available (Wellington Laboratories, Guelph, Ontario),⁴ and so testing was performed to assess their LLOQ (Table 1). 

Additional analysis performed on the SCIEX 7500 System

Due to sensitivity being a significant challenge when performing PFAS analysis, testing was also performed on the SCIEX 7500 System, which is the most sensitive triple quadrupole MS system in the SCIEX portfolio. All 20 compounds within the 0.1 µg/L regulation were analyzed, with 15 out of 20 analytes providing improved LLOQ values, 3 providing a comparable LLOQ and 2 having a higher LLOQ due to unexpected interferences, which were thought to be caused by contamination within the water used or related to chromatography (data not shown). Figure 8 and Figure 9 show example results from the calibration curve measured for L-PFOS on the SCIEX 7500 System.  

Conclusions

To summarize, an LC-MRM method using a smaller volume direct injection approach has been demonstrated to be sensitive enough to easily meet the current recommended limits set by the European authorities for drinking water. The generated results highlight the accuracy, robustness and precision of the method. Several analytical challenges have been well documented for PFAS analysis in the past, such as blank contamination, compound stability and the need to inject large volumes of samples. This method has successfully addressed these problems, ensuring that it can be much more easily implemented compared with previous methods. It also does not require the use of specialized equipment or extensive modifications to the LC system. In addition, data has been generated using this method on the SCIEX 7500 System to highlight the advantages of increased sensitivity.  

References

1. European Chemicals Agency (ECHA), Perfluoroalkyl chemicals (PFAS). 
2. Council of the European Union, Proposal for a Directive of the European Parliament and of the Council on the quality of water intended for human consumption, February 2020. 
3. Download the supplementary information.
4. Wellington Laboratories, New products: Native Certified Reference Standards for L-PFUdS and L-PFTrDS, November 2020. 
5. US Environmental Protection Agency, Determination of Selected Perfluorinated Alkyl Acids in Drinking Water by Solid Phase Extraction and Liquid Chromatography/Tandem Mass Spectrometry (LC/MS/MS), Method 537 Rev 1, November 2008 (revised October 2015).