Automated Targeted and Non-targeted LC-Orbitrap MS Workflow for Analysis of More Than 40,000 PFAS Compounds in Environmental Samples

Emerging Environmental Applications for High Resolution Mass Spectrometry
Oral Presentation

Presented by C. Grim
Prepared by T. Astill
ThermoFisher Scientific, 2443 Trent St, VICTORIA, British Columbia, V8r4Z4, Canada


Contact Information: [email protected]; 604-376-6499

ABSTRACT

Per- and polyfluorinated alkyl substances (PFAS) are a class of bioaccumulative, often toxic small molecules featured in environmental targeted methods, including EPA 533 and 537.1. However, these labor-intensive solid phase extraction (SPE) methods require 250 mL of water and only measure a handful of the thousands of known PFAS compounds due to a lack of standard availability and compatibility with SPE sorbents. This work describes an automated dispersive liquid-liquid microextraction (DLLME) method applied to water samples to reduce solvent consumption, cost-per-sample, and sample contamination. These extracts were analyzed using a targeted quantitative and non-targeted screening LC-Orbitrap MS method. Non-targeted results were compared against a 40,000+ PFAS compound library including a new in silico predicted transformation library.
Methods
Samples underwent a low density solvent DLLME process using an advanced customizable autosampler. Data was acquired on an LC-Orbitrap system through two injections: the first using full scan, SIM, and AIF scan modes for targeted quantitation and another injection using full scan DDA MS2 for non-targeted screening. The orbitrap provides high resolution, high mass accuracy data, and the opportunity for retrospective analysis. Data was filtered in Thermo Scientific™ Compound Discoverer™ 3.3 using fragmentation databases, predicted transformations, as well as standard and Kendrick mass defects for CF2. The software searches multiple databases simultaneously, including a new in silico structural library. Additional features like filter-based feature reduction aids in expediting data analysis and imparting a higher confidence in the automated results.

For this method, a panel of 53 PFAS compounds of interest to both North American and European regulatory bodies was quantified alongside untargeted analysis across multiple environmental matrices. Quantitative results show limits of quantitation down to part per trillion levels for most analytes from an initial volume of only 15 mL of sample. The automated DLLME method provides a cleaner, more concentrated sample than direct injection approaches while providing some of the sensitivity gains seen from solid phase extraction. PFAS are present in many common lab supplies such as pipette tips and gloves, meaning removing manual steps also reduces the probability of sample contamination. Methanol, commonly used to dilute or extract samples, frequently contains PFOS, a toxic PFAS of high regulatory concern, at detectable levels.

Untargeted data generated from this experiment includes a list of suspected compounds detected along with their confidence level for each sample matrix. By using an orbitrap mass analyzer, the high mass accuracy reduces the chances of a false formula assignment. With the new in silico library, we expect to see more tentative candidate structures where previously only possible empirical formulas could be determined. Reviewing putative library identifications alongside other factors such as retention time correlated with mass, molecular mass divided by number of carbon atoms, and mass defect divided by number of carbon atoms provides a multifaceted approach to identifying potential PFAS compounds. This overall workflow provides an automated solution to both quantify known PFAS and explore potential unknown PFAS with confidence.