Automated targeted and non-targeted LC-Orbitrap MS workflow for analysis of more than 40,000 PFAS compounds

Significance of the topic

Per- and polyfluoroalkyl substances (PFAS) are persistent environmental contaminants with potential human health risks. Their widespread detection in water supplies demands sensitive, reliable analytical workflows. Combining targeted quantitation with non-targeted screening in a single workflow enhances our ability to monitor known PFAS and discover emerging compounds.

Objectives and study overview

This study aimed to demonstrate a fully automated analytical workflow for both targeted and non-targeted analysis of over 40 000 PFAS in drinking water. Key goals included:

• Implementing an automated dispersive liquid-liquid microextraction (DLLME) for sample preparation.
• Quantifying 56 priority PFAS at low part-per-trillion levels using LC-Orbitrap MS.
• Applying a non-targeted screening workflow with advanced spectral libraries and software tools for unknown PFAS annotation.

Methodology and instrumentation

Sample preparation was performed on a Thermo Fisher TriPlus RSH SMART liquid handling platform. DLLME used a small volume of extraction solvent dispersed in 15 mL water samples to concentrate PFAS.

The instrumental setup included:

• Vanquish Flex UHPLC with Acclaim 120 C18 column (0.4 mL/min, 23 min run time).
• Thermo Scientific Orbitrap Exploris 240 mass spectrometer.
• Data acquisition modes: full scan (MS1, 60 000 resolution), selected ion monitoring (SIM, 60 000), and all-ion fragmentation (AIF, 15 000).
• Chromeleon CDS 7.3.2 for targeted data processing.
• Compound Discoverer 3.3 SP3 with predefined PFAS templates for non-targeted annotation.
• Spectral and mass-list resources: EPA and NIST PFAS databases, FluoroMatch fragmentation libraries, Duke University in-silico library, mzCloud, and 2023 NIST HRMS MS/MS libraries.

Main results and discussion

Targeted quantitation achieved limits of quantitation in the low part-per-trillion range for most analytes, with reproducibility (<30% CV) and accuracy (70–130%) confirmed over multiple days. A panel of 56 PFAS was reliably measured in both tap and bottled water from 15 mL sample volumes.

Non-targeted screening detected and annotated over 96% of spiked PFAS at confidence levels 2–4. Level 2 annotations were achieved by matching MS1 and MS2 data against multiple spectral libraries, illustrating the power of resource integration for unknown PFAS discovery.

Visualization tools in Compound Discoverer enabled sample comparison via PCA and orthogonal MS1 plots, highlighting differences in PFAS composition and concentration across water sources.

Benefits and practical applications

The automated DLLME workflow reduces solvent usage, cost-per-sample, and contamination risk. Integrating targeted and non-targeted analyses permits comprehensive PFAS monitoring, supporting regulatory compliance and enabling discovery of novel compounds in environmental and industrial contexts.

Future trends and potential applications

Advancements likely include:

• Expansion of curated PFAS spectral libraries and in-silico prediction tools.
• Machine learning algorithms for higher-confidence annotation of unknowns.
• High-throughput screening in complex matrices such as food and biota.
• Miniaturized, field-deployable microextraction platforms coupled to portable MS.

Conclusion

This study validates a robust, automated LC-Orbitrap MS workflow that unites sensitive targeted quantitation and comprehensive non-targeted screening for PFAS analysis in water. The approach offers high accuracy, reproducibility, and annotation confidence, addressing critical needs in environmental monitoring and research.

Reference

1. Charbonnet JA, McDonough CA, Xiao F et al. Confidence levels in PFAS annotation: Environmental Science & Technology Letters. 2022.
2. Kaufman A, Butcher P, Maden M et al. Simplifying non-targeted analysis of PFAS in complex food matrices. Journal of AOAC International. 2022.