Many HPLC methods have a line that says “filter through a 0.45 µm syringe filter.” It often does not say which membrane, or how much filtrate to throw away.
That line is treated as housekeeping. It is a sample preparation step, and it can change the result in two directions. The filter can take analyte out, and it can put something in.
Agilent’s own application note on syringe filters puts it plainly. The major concern with filtration is adsorption of the API, “which results in lower assay values or even out-of-specification results.” The second concern is extractables from the filter, which can interfere with the separation, “causing difficult peak integration and even false results” (Agilent 5991-2409EN, p. 2).
The first millilitre loses the most
The clearest pattern in the data is that a dry membrane binds analyte until its sites are filled, and then it stops.
In Agilent’s study, acetaminophen was filtered through four membrane types and the filtrate was collected in fractions (p. 7, Table 4). Compared with a centrifuged sample:
| Membrane | 1st mL | 2nd mL | 3rd mL |
|---|---|---|---|
| Nylon | 97.0% | 99.6% | 99.7% |
| Nylon with glass-fibre prefilter | 93.1% | 99.1% | 99.9% |
| Regenerated cellulose | 99.6% | 99.7% | 99.8% |
| PES | 99.4% | 99.7% | 99.7% |
Their recommendation: discard the first 1 mL when using nylon, and no discard is needed for regenerated cellulose or PES (p. 7).
The same pattern appears for PFAS, at a larger scale. Zheng et al. (2024) found recovery rising with the volume filtered: 80% at 1 mL, 85% at 5 mL and 90% at 10 mL (p. 2).
FDA’s LC-HRMS method for nitrosamines in sartans writes the discard into the method: PVDF filters, first 1 mL discarded (FDA, 2019, p. 3).
A 7% loss on the first millilitre would fail most assay accuracy criteria by itself. And the analyst who filters 1 mL straight into a vial is collecting exactly that millilitre.
The membrane matters, and so does the analyte
The usual starting guidance is about solvent compatibility. Agilent’s version: hydrophilic membranes such as cellulose acetate or PES for aqueous samples, and PTFE or polypropylene for organic samples with aggressive solvents (p. 2).
That decides whether the filter survives the sample. It says less about whether the analyte survives the filter.
Zheng and co-workers tested six syringe filters with three PFAS in water. The best, mixed cellulose ester, recovered 97–101%. The worst, PTFE, recovered 61–80% (p. 1). And the loss depended on the compound: PFOS recovered only 55–68%, while PFOA and PFBA recovered 95–99% (pp. 1–2). Same filter, same water, different functional group, a very different result.
So “PTFE is the safe, inert choice” holds for many organic-solvent samples, and is the wrong choice for PFAS. EPA Method 1633A goes further and bans PTFE-lined caps for PFAS work (EPA 1633A, p. 14). The EU drinking-water guidelines note that drinking water is “commonly” analysed without filtration (EU PFAS technical guidelines, 2024, p. 5).
The filter can add something too
Adsorption lowers results. Extractables raise them, or add peaks that should not be there.
Two examples from the library behind this site:
- The EU reference laboratory for pesticides found that polyester filters leached chlorate and perchlorate, which are themselves analytes in its QuPPe method (EURL-SRM QuPPe-PO, v12.1, p. 4).
- In a 2026 LCGC International interview, M. Vitulli, who tests food-contact materials, described filters certified free of PFAS, bisphenols and phthalates that still added traces of those compounds once the extract was concentrated (Vitulli and Jones, August 2026). It is an interview with no data, so treat it as a practitioner report.
The second point is easy to miss. A filter blank injected directly can look clean. The same blank, evaporated tenfold as the real samples are, may not be. Carry the blank through the whole procedure, including any concentration step.
How to run a filter study
The design in the Agilent note is simple, and it works for any analyte (pp. 2–7):
- Make a reference. Clarify one portion of the sample by centrifugation. Treat its result as 100%.
- Filter in fractions. Collect the 1st, 2nd, 3rd and 5th mL separately, for each membrane you are considering.
- Set a criterion before you look. Agilent used 100 ± 1% relative recovery against the centrifuged sample. Your assay may justify a different number, but set it in advance.
- Test at your real conditions. Your sample solvent, your concentration, and your lowest level. Binding sites fill up, so a filter that is fine at 100% of the label claim can lose a larger fraction at the LOQ.
- Run a filter blank through the full procedure, including evaporation if the method has it.
- Write the result into the method: the membrane, the pore size, the filter diameter, and the discard volume.
Point 6 is where the value is kept. A method that says “0.45 µm nylon, 25 mm, discard the first 1 mL” survives a change of supplier and a new analyst. A method that says “filter” does not.
When to skip the filter
Filtration is there to keep particles out of the column and the instrument. Agilent’s own note describes it as a faster and cheaper alternative to centrifugation, with equal or better particle removal (p. 2). If centrifugation gives clear supernatant and fits your workflow, it removes the filter as a variable. For PFAS in drinking water, skipping filtration is the common practice.
If you do filter, treat the filter like a column: a specified part of the method, tested before routine use.
Sources
- Zhao L, Long W. Syringe Filter Suitability for Sample Preparation in Drug Assays. Agilent application note 5991-2409EN, 2013. A vendor note testing the vendor’s own filters; I use its design and its fraction-by-fraction data.
- Zheng Y-H et al. Using syringe filtration after lab-scale adsorption processes potentially overestimates PFAS adsorption removal efficiency. J Environ Qual 2025; 54(1):20–30. doi:10.1002/jeq2.20640.
- US FDA, CDER. LC-HRMS method for the determination of six nitrosamine impurities in ARB drugs, 2019.
- US EPA. Method 1633A, 2024.
- European Commission. Technical guidelines on PFAS analysis in drinking water, 2024.
- EURL-SRM. QuPPe-PO method, version 12.1.
- Vitulli M, Jones K. Migration vs. extraction: comparing contaminant profiles. LCGC International, 12 August 2026. An interview write-up, no data.
If a recovery problem in your method has survived every other fix, the filter is a cheap thing to test first. Tell me the analyte and the membrane on LinkedIn and I will tell you what I would compare it against. The validation-readiness checklist covers the other sample preparation questions to settle before validation.
Common questions
- Can a syringe filter adsorb my analyte?
- Yes. The membrane and housing can bind analyte, which lowers the result. Agilent names adsorption of the API as the main concern with filtration, because it gives low assay values or out-of-specification results. How much is lost depends on the membrane, the sample solvent and the analyte.
- Should I discard the first mL of filtrate?
- Often, yes. A dry membrane binds the most analyte. In Agilent's acetaminophen data, nylon and glass-fibre nylon filters recovered 97.0% and 93.1% in the first mL and about 100% from the second mL on; regenerated cellulose and PES needed no discard. FDA's nitrosamine method discards the first 1 mL through PVDF. Measure it for your own analyte.
- Which syringe filter membrane should I use for HPLC?
- As a starting point, hydrophilic membranes such as cellulose acetate or PES for aqueous samples, and PTFE or polypropylene for aggressive organic solvents. Then test the actual membrane with your analyte, because binding depends on the analyte as well as the solvent.
- Which syringe filter should I use for PFAS?
- Avoid PTFE. In one study PTFE filters recovered only 61–80% of PFAS while mixed cellulose ester recovered 97–101%, and PFOS was lost the most. The EU drinking-water guidelines note that drinking water is commonly analysed without filtration.
- How do I validate a syringe filter?
- Compare filtered sample with the same sample clarified by centrifugation, which serves as the 100% reference. Collect the filtrate in fractions, for example 1st, 2nd, 3rd and 5th mL, to see whether the first portion is low. Agilent used 100 ± 1% relative recovery as its acceptance criterion. Also run a blank through the filter to check for extractables.
- Can syringe filters add contaminants?
- Yes. Polyester filters have leached chlorate and perchlorate, and filters certified free of PFAS, bisphenols and phthalates still added traces once the extract was concentrated. Carry a blank through the whole procedure, including any evaporation step.