A 0.45 µm PTFE filter lot changed hands three times before it reached my bench: polymer supplier, filter manufacturer, contract sterilizer. The certificate of analysis that arrived with it named the filter manufacturer and nothing upstream. When an ISO 17025 traceability audit asked us to show that the membrane resin was the same grade we validated our extractables study against, we had a signed CoA, a lot number, and no way to connect the two. That is the failure this article is about.
Most labs can produce traceability for instruments and reference standards. The chain breaks one link earlier, at the consumable. For lab chromatography, the consumable is where the method lives: the column, the filter, the septum, the vial. If you can trace a calibration standard to NIST but cannot trace the filter that touched your sample to a resin batch, your ISO 17025 traceability for lab chromatography is incomplete, and a competent auditor will find the gap.
The traceability chain for a consumable has six links, and the certificate covers two
ISO/IEC 17025:2017 clause 6.5 requires metrological traceability of measurement results, and clause 6.4 covers the handling of items. Neither clause says “consumable,” but the result depends on it. Here is the chain I use when I map a consumable to a result:
- Link 1: raw material batch. Resin lot, glass lot, polymer grade, with a supplier batch number and a specification.
- Link 2: manufacturing lot. The date, line, and process parameters that turned raw material into product. This is the lot on the label.
- Link 3: release testing. The measurements that qualified the lot: pore size distribution, bubble point, extractables, dimensional check.
- Link 4: calibration of the test equipment. The pore-size measurement is only traceable if the porometer was calibrated against a reference with known uncertainty.
- Link 5: the certificate. The document that transmits links 1 through 4 to you in a form you can file.
- Link 6: your receiving record. The lot you actually used, on the date you used it, tied to the method and the sample.
The certificate typically covers links 2, 3, and part of 4. Links 1 and 6 are the ones labs cannot produce. Link 1 is upstream of the supplier’s own purchasing system, and link 6 is upstream of your own bench discipline.
What “unbroken” actually means at the resin level
For a PTFE membrane filter, the resin grade determines the extractables profile. A change from one PTFE dispersion grade to another, even at the same nominal pore size, can shift the fluorescence baseline in an HPLC-fluorescence method. If your extractables study was run on grade A and the lot you used was grade B, the study does not apply. You cannot know which grade you got unless the chain reaches back to the resin batch. Most filter CoAs stop at “PTFE, 0.45 µm, lot XXXX.” That is a product description, not a traceability record.
The same logic applies to PTFE membrane filters: the sintered support, the housing polymer, and the wetting agent are all inputs, and any of them can move extractables. Ask for the resin supplier and grade in writing. If the answer is “proprietary,” ask for a grade-equivalence statement and a change-notification commitment instead. That is the practical substitute.
The link most labs cannot produce is the raw material batch, not the certificate
I have audited maybe forty chromatography labs against ISO 17025. Every one of them could produce a certificate for the column and the filter. Fewer than half could produce a raw material batch record for the filter, and almost none could produce one for the vial. The certificate is easy because it comes in the box. The raw material batch is hard because it lives in the supplier’s ERP and the supplier has no commercial reason to expose it.
This is the structural problem. The traceability requirement is on the result, but the information you need is held by a party with no accreditation obligation to you. ISO 17025 applies to your lab, not to your filter vendor. The vendor may hold ISO 9001, which requires lot identification but not metrological traceability to a reference. Those are different standards with different evidence.
The three questions that separate a real chain from a paper one
When I qualify a consumable supplier for a regulated method, I ask three questions and score the answers:
- Can you name the raw material batch for this lot? A yes with a batch number is a real chain. A yes without a number is a promise.
- Is the release test traceable? The bubble-point or pore-size measurement should reference a calibrated standard with a stated uncertainty. If the CoA says “0.45 µm nominal” and nothing else, the measurement is a promise, not a result.
- How do you notify me of a change? A written change-control clause with a notice period (30 or 90 days) and a re-qualification trigger is worth more than any single CoA.
Two yeses and a written change clause is a supplier I will put on a regulated method. One yes is a supplier I will use for general work and quarantine from validated runs.
A worked example: tracing a 0.45 µm PTFE filter from resin to injection
Take a method that filters aqueous mobile phase through a 0.45 µm PTFE membrane before an HPLC-UV run at 210 nm. The traceability record needs to connect the membrane lot to the blank injection. Here is the chain I would build:
| Link | Record | Who holds it | What it proves |
|---|---|---|---|
| Resin batch | Supplier batch number, grade, spec | Filter manufacturer | The extractables profile matches the validation study |
| Membrane lot | Lot number, date, line | Filter manufacturer | The lot on the label is the lot tested |
| Release test | Bubble point, pore size, extractables | Filter manufacturer | The lot met specification |
| Test calibration | Porometer calibration, uncertainty | Filter manufacturer | The release test is metrologically traceable |
| Certificate | CoA transmitting the above | Filter manufacturer | You can file it |
| Receiving record | Lot received, date, storage | Your lab | The lot you used is the lot you qualified |
| Use record | Lot used, date, method, analyst | Your lab | The result traces to the consumable |
The two rows your lab owns are the last two, and they are the two most often missing. A receiving record that says “filters, PTFE, qty 100” is not a traceability record. It needs the lot number, the date, and the storage condition. A use record that says “filtered mobile phase” is not a traceability record either. It needs the lot number and the analyst.
Why the calibration row is the one auditors test
If the release test for the membrane lot was a bubble-point measurement, that measurement is only traceable if the porometer was calibrated against a reference with a stated uncertainty. Ask for the calibration certificate number and the uncertainty. If the supplier cannot produce it, the release test is an internal check, not a traceable measurement, and the chain stops at link 4. This is where a lot of otherwise-good suppliers fail. They have the lot record and the CoA, but the test behind the CoA is not traceable.
The same test applies to a HPLC filter used inline: the pressure-drop specification is only meaningful if the flow meter behind it was calibrated. For a broader view of how consumables fit the workflow, the Lab Chromatography resource set covers the bench-level practices that support this.
Two field cases where the chain broke, and the measurement that caught it
Case 1: the septum lot that moved the baseline
Symptom: a GC-FID method for residual solvents started showing a rising baseline after 40 injections on a new box of septa. The old box had run 120 injections before baseline drift. The septa were the same part number, same supplier, same catalog spec.
Measurement: we ran a blank injection program (10 µL air, no sample) across five septa from the new lot and five from the old lot, and measured bleed at 250 °C inlet temperature over 60 minutes. The new lot bled at roughly 2.5 times the rate of the old lot at the same temperature. The difference was consistent across all five septa, so it was a lot effect, not a random failure.
Diagnosis: the septum is a silicone rubber disc bonded to a PTFE face. The silicone formulation changed, or the post-cure changed, and the bleed profile moved. The CoA listed “silicone/PTFE, lot XXXX” and nothing about the formulation. We could not trace the change because the raw material batch was not on the certificate.
Fix: we added a receiving test for septa on regulated methods: a 60-minute bleed check at the method inlet temperature, with a pass criterion of baseline rise under a set threshold. We also asked the supplier for a formulation change-notification clause. The supplier agreed to 90 days’ notice. That clause is now part of the purchasing spec, and it is worth more than the CoA.
Case 2: the vial that failed the extractables study
Symptom: an HPLC-UV method at 214 nm showed a small peak at 4.2 minutes that appeared only in samples stored in a particular lot of autosampler vials. The peak was absent in the same samples stored in a different lot of the same catalog number.
Measurement: we ran a solvent blank (methanol, 24-hour contact) through ten vials from each lot and compared peak area at 4.2 minutes. The suspect lot showed a peak area eight times the qualified lot, with an RSD of 6% across the ten vials. The qualified lot was below the detection limit.
Diagnosis: the vial is borosilicate glass with a silicone/PTFE septum and a polypropylene cap. The peak was from the septum, not the glass. The suspect lot used a different septum bonding process, and the bonding residue extracted into methanol. The certificate said “silicone/PTFE septum,” which is true of both lots. It did not say which bonding process.
Fix: we added a lot-qualification extractables test for vials on this method (methanol contact, 24 hours, UV at 214 nm, peak area threshold) and moved the method to a vial supplier who would name the septum bonding process on the CoA. For methods that use a COD vial for digestion or storage, the same lot-qualification logic applies: the glass and the cap are separate inputs and both need a lot record.
What to demand in writing from a consumable supplier
You will not get a full raw-material chain from every vendor. You can get a defensible substitute if you ask for the right documents. Here is the checklist I use, with the acceptance criteria:
- Certificate of analysis with lot number and release-test results. The lot number on the CoA must match the lot on the box, and the release test must have a numeric result, not a pass/fail.
- Calibration reference for the release test. A certificate number and an uncertainty. If the supplier cannot produce it, the test is an internal check.
- Raw material grade statement. The resin or glass grade, even if the batch number is withheld. A grade-equivalence statement is the fallback.
- Change-notification clause. A written commitment to notify you of formulation, process, or site changes with a stated notice period (30 to 90 days) and a re-qualification trigger.
- Extractables data on request. For a regulated method, a lot-specific extractables profile at the method wavelength and contact time.
- ISO 9001 or ISO 17025 accreditation scope. If the supplier holds ISO 17025 for the release test, ask for the scope. It usually covers only part of the test menu.
Two of these, the change-notification clause and the calibration reference, are the ones that separate a supplier you can put on a validated method from one you cannot. The rest are negotiable.
Where the major suppliers already give you more than you use
Agilent, Thermo Fisher, and Restek publish lot-level documentation and technical notes that many labs never request. Agilent’s GC columns and accessories documentation includes lot traceability and test chromatograms for many parts. Thermo Fisher’s chromatography and sample handling product pages carry CoA lookups by lot. Restek’s technical resources include lot-specific test data and method notes. The gap is usually not the supplier’s capability; it is that the lab never asked, and the purchasing spec never required it.
For methods that cite EPA procedures, the reference chain has its own structure. EPA SW-846 methods name the determinative step and the sample preparation, and the consumable is part of the preparation. If your method cites SW-846 3510 or 3520 for extraction, the filter and the vial are part of the method, and the traceability record needs to reach them.
The decision rule: qualify the link you cannot see, not the certificate you can file
ISO 17025 traceability for a chromatography consumable is not a document problem. It is a supplier-qualification problem. The certificate is the easy part; the raw material batch and the calibration behind the release test are the parts that break, and they break because the lab never wrote them into the purchasing spec. The decision rule is this: for any consumable that touches a validated method, require a lot-specific CoA with numeric release results, a calibration reference for the release test, and a written change-notification clause with a stated notice period. If a supplier cannot meet all three, quarantine that consumable from validated runs and use it only on general work. The certificate you can file is not the link that holds. The link you cannot see is the one that decides whether your result traces to anything.
References & further reading
Related resources from our lab consumables network:
- Lab Chromatography at labchromatography.com
- Hplc Filter at hplcfilters.com
- Ptfe Membrane Filter at ptfemembranefilter.com
- Cod Vial at codvial.com
External references and standards: