henhaobio
Sign in

Environmental Science & TechnologyScientific Communication & EducationPeer-reviewed journal article

Updated Guidance for Communicating PFAS Identification Confidence with Ion Mobility Spectrometry

Anna K. BoatmaniD, Jessie R. ChappeliD, Kaylie I. Kirkwood‐DonelsoniD, Jonathon F. FlemingiD, David M. ReifiD, Emma SchymanskiiD, Julia E. RageriD, Erin Shammel BakeriD

PDFDOIPreprintData
In-depth full-text digest available · Jump to digest ↓

Abstract

Over the last decade, global contamination from per- and polyfluoroalkyl substances (PFAS) has become apparent due to their detection in countless matrices worldwide, from consumer products to human blood to drinking water. As researchers implement non-targeted analyses (NTA) to more fully understand the PFAS present in the environment and human bodies, clear guidance is needed for consistent and objective reporting of the identified molecules. Confidence levels for small molecules analyzed and identified with high-resolution mass spectrometry (HRMS) have existed since 2014; however, unification of currently-used levels and improved guidance for their application is needed due to inconsistencies in reporting and continuing innovations in analytical methods. Here, we (i) investigate current practices for confidence level reporting of PFAS identified with liquid chromatography (LC), gas chromatography (GC), and/or ion mobility spectrometry (IMS) coupled with HRMS and (ii) propose a simple, unified confidence level guidance that incorporates both PFAS-specific attributes and IMS collision cross section (CCS) values. SynopsisUnified and simplified requirements to guide confidence level assignment in non-targeted PFAS identification efforts with ion mobility spectrometry. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=109 SRC="FIGDIR/small/634925v2_ufig1.gif" ALT="Figure 1"> View larger version (23K): org.highwire.dtl.DTLVardef@1b9808dorg.highwire.dtl.DTLVardef@1a6e663org.highwire.dtl.DTLVardef@5f8813org.highwire.dtl.DTLVardef@1dea0d1_HPS_FORMAT_FIGEXP M_FIG For Table of Contents only C_FIG

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (18 May 2025) · CC BY 4.0

An audit of 38 PFAS non-targeted studies from 2023 found confidence levels often missing or misapplied; a five-level, sublevel-free checklist with predefined platform tolerances is proposed for ion mobility workflows.

Non-targeted high-resolution mass spectrometry (HRMS) identifications of per- and polyfluoroalkyl substances (PFAS) are graded with 2014, 2020 (collision cross section, CCS) and 2022 (PFAS-specific) confidence schemes, which the authors say are applied subjectively, skew toward overestimation and give little direction when some criteria are unmet. This work merges them into one sublevel-free checklist for ion mobility spectrometry (IMS) data, with tolerances set in advance to enable future automation.

Key results

  1. ShownConfidence reporting in 2023 PFAS non-targeted analysis (NTA) studies was uneven. More than a quarter gave no confidence levels. Among those that did, the 2014 Schymanski small-molecule scale was cited most often, and Level 2 was the most common assignment.

    A PubMed query (30 Sep 2024) returned 426 publications. The 94 from 2023 were screened, and 38 used HRMS for PFAS. Of the 38, 12 reported no confidence levels and 26 did. Studies reported 0 to 102 PFAS each, and mass tolerances were sometimes undefined or as wide as 20 ppm. LC-Orbitrap and LC-QTOF were the most common platforms. · Materials and Methods; Assessing Current Practices for Confidence…; Figure legends

  2. ShownEarlier guidance (Schymanski et al., Celma et al.) called Level 2b (probable structure from diagnostic evidence) usually rare, but Level 2b/2c identifications were common in PFAS NTA. Levels were also applied inconsistently, e.g. Level 2b for structures with acknowledged possible isomers, which belong at Level 3.

    Of the 10 studies that used Level 2 sublevels, 8 reported Level 2b and/or 2c identifications (5 to 87 PFAS at 2b/c per study). Other inconsistencies included reported mass errors above the authors' own thresholds (examples in Figure S2). · Assessing Current Practices for Confidence…; Figure legends

  3. ClaimedProposed scheme: five levels, no sublevels, with per-platform tolerances for each dimension fixed and justified before analysis. Level 1 needs a same-platform standard match in m/z, isotope envelope, CCS and retention time or index, but not mobility-aligned MS2 if diagnostic evidence such as homologues exists.

    Merges the 2014, 2020 (CCS) and 2022 (PFAS) schemes into one checklist (Figure 3, Table S3). Level 2 requires exactly one structure, from library or predicted values (e.g. homologue CCS/RT trendlines); out-of-tolerance matches fall to Level 3. Defaults are given for an Agilent 6560 IMS-QTOF; local manufacturing knowledge may count as diagnostic evidence. · Updated PFAS LC- or GC-IMS-HRMS Confidence…; Confirmed Structure (Level 1); Probable Structure (Level 2); Discussion and Examples of PFAS Identified with…

  4. ShownIn the authors' in-house LC-IMS-HRMS library of 175 PFAS standards, only 3 pairs could overlap across m/z, retention time and CCS at the Table 1 tolerances (branched isomers excluded). This shows how the CCS dimension separates near-isobaric PFAS.

    Hydro-EVE and 6:2 FTS differ by 6.57 ppm in m/z (426.9651 vs 426.9679) but by 11.1% in CCS (151.2 vs 168.1 Ų), so IMS resolves them even if they coelute; both are Level 1 at ±10 ppm and ±0.5% CCS. 8:2 FTS and 8:2 FTPA (17 ppm apart) coelute and have close CCS values: Level 1 only with diagnostic or MS2 evidence, otherwise Level 3. · Discussion and Examples of PFAS Identified with…; Figure legends

  5. ShownPublic experimental CCS data rarely cover PFAS. Together with cases that fail Level 2a (no CCS or spectral library match) yet carry more evidence than Level 2b/c requires, this led the authors to drop the Level 2 sublevels.

    Of 8,099 unique PubChem CIDs with experimental CCS values, 208 are PFAS under the OECD definition implemented in PubChem (12 Nov 2024). The authors also report PFAS found in spectral but not CCS libraries, and PFAS that match in-house CCS values and homologue trendlines but have no diagnostic fragments. · Addressing Gaps in Existing Confidence Level…

Methods, data and reproducibility

The authors combined a PubMed literature review of PFAS non-targeted analysis (manual review of 2023 studies using HRMS) with their own LC-IMS-HRMS experience to find gaps in existing confidence schemes. From this they wrote a unified five-level checklist, illustrated with examples from the lab's in-house PFAS CCS and retention time library.

  • PubMed records matching a PFAS non-targeted analysis query (searched 30 Sep 2024; earliest from 1997) · n = 426 publications
  • 2023 publications manually reviewed; those using HRMS for PFAS analysed in depth for platform, matrix, PFAS count and confidence scheme (Table S1) · n = 94 reviewed; 38 included
  • Baker lab in-house PFAS LC-IMS-HRMS library of experimentally measured retention time and CCS values from authenticated standards · n = 175 PFAS (as evaluated here) · 10.5281/zenodo.14341320
  • PubChem compounds with experimental CCS values, filtered by the OECD PFAS definition implemented in PubChem (12 Nov 2024) · n = 8,099 unique CIDs; 208 PFAS
Methods & tools
PubMed, PubChem, Agilent 6560 IMS-QTOF, LC-IMS-HRMS
Reproducibility gaps
No code is released; the SMISE web application for automated confidence assignment is still in development. Recommended tolerances are given only for the Agilent 6560 platform. The list of reviewed papers (Table S1) and the tolerance worksheet (Table S4) are in the Supporting Information.
Limitations (6)
  • Recommended tolerances are given only for the Agilent 6560 LC-IMS-HRMS platform. Tolerances for other instruments and workflows are left to future consultation with the IMS PFAS community. · stated by the authors
  • Acceptance tolerances that suit every platform cannot yet be set, because methods and prediction tools differ widely: experimental CCS reproducibility can be ≤0.30%, while machine-learning CCS predictions can err by up to 8%. · stated by the authors
  • The probability approach of Metz et al. for ranking multiple candidates is not yet part of the level system, pending community feedback. · stated by the authors
  • The guidance is scoped to PFAS identified with LC- or GC-IMS-HRMS and will need updating as instruments and predictive tools improve. Automated assignment depends on the SMISE tool, which is not yet released. · stated by the authors
  • The literature audit covers one publication year (38 studies from 2023) found with a single PubMed query, so its trends may not hold for other years or databases. · noted in this summary
  • The checklist is not tested empirically. There is no inter-analyst agreement study, and the audited studies' identifications are not re-graded, so there is no evidence yet that it reduces inconsistency. · noted in this summary

Read this if you report PFAS identifications from non-targeted LC- or GC-IMS-HRMS data and want a single checklist for assigning confidence levels, including how to use CCS values and predefined tolerances.

Citations
26
FWCI
7.78
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISBoatman '25Boatman '23Foster '21Teri '25Kirkwood '21McLane-Svob '26
72% semantic matchPharmacology & Toxicology
Anna K. Boatman, Gregory P. Kudzin, Kylie D. Rock+4 more·bioRxiv·Mar 20, 2025· 8 citations