PFAS in Air, Part 3: OTM-45 Calibrations

In my previous two PFAS blog posts I’ve covered a general summary of the draft OTM-45 method and the chromatography for the LC method. In this blog we’ll talk about how to calibrate your instrument for OTM-45.

Calibration curves are a topic near and dear to my heart, as you may have guessed if you’ve read my blog on TO-15A calibrations. While chromatography and sample prep get a lot of attention, calibration curves seem to be more of a box to check, with the focus on how to meet the method requirements instead of how it can positively or negatively impact results. In fact, this focus on method requirements, especially the use of the coefficient of determination (r²), can lead to worse results in many cases. I covered this briefly in my TO-15A blog, but r² is based on absolute error, so a curve fit chosen to maximize r² may have significant relative error on the low end of the calibration curve. Richard Burrows from Eurofins gave a good presentation on this at NEMC 2020 that goes more deeply into the math.

If I’ve talked about this in a previous blog, why am I bringing it up again? Well, OTM-45 follows the lead of EPA 533 and 537.1 (PFAS in drinking water methods) and drops the use of the coefficient of determination and % relative standard deviation (%RSD) for average response factor curves as metrics for passing calibration curves. The only requirement is that the calculated values of the calibration points are within ±10% of the true value. This is both good and potentially bad news. It ties acceptance of the curve to a metric that actually measures relative accuracy of the curve over the whole span, which is good. The potentially bad part is that ±10% is a very tight window for recovery that labs may struggle with. EPA Methods 533 and 537.1 have limits of ±30%, or ±50% for points at or below the minimum reporting limit. While TO-15A doesn’t drop r² and %RSD requirements, it also includes a recovery limit of ±30%. In the work I’ve done for OTM-45, I was able to get 12 compounds under the 10% limit, 32 between 10-20%, 5 that were between 20-30%, and none above 30%. Just to be clear, that is 37 out of 49 compounds failing to meet the ±10% criteria. See Table 1 for the full results.

Table 1: Calibration accuracy results. Results highlighted in yellow are greater than ±10%, results in red are greater than ±20%.

These results were consistent over several attempts to calibrate, so I began to troubleshoot possible causes. The failures were not consistently happening on a single calibration point, so it didn’t seem to be an issue with any specific standard. The calibration standards were all diluted from a single stock solution I mixed up, so failures couldn’t be associated with single spiking issues. The internal standard areas were consistent, and compounds that shared internal standards were both passing and failing the +/-10% criteria, so it didn’t seem to be an internal standard issue. The passing compounds were all either perfluoroalkylcarboxylic acids (PFCAs) or in a group OTM-45 nebulously describes as “additional targets”, but those are the largest compound groups in the method, so that doesn’t necessarily mean much.

In the end, given that there was no common thread to the failures and that I was meeting the criteria given in other PFAS methods, I accepted the results as is rather than trying to fight to meet the very strict limits laid out in OTM-45. I felt that it was more valuable to make progress on other aspects of the method, such as the extractions, since the calibration results were already acceptable by most other methods. This may be a choice labs running this method will have to make if the +/-10% criteria makes it out of the draft method. If a lab understands their samples well they could limit their compound list and calibration range to their expected results, which may make meeting the calibration criteria easier, but I expect meeting the +/-10% limit for all compounds is something many labs will struggle with.