The auditor picked a lot of 11 mm septa from the shelf and asked for the calibration certificate behind the durometer reading on the incoming inspection sheet. The reading was 47 Shore A, recorded on 14 March, and the sheet referenced a handheld tester serial number. Nobody could produce a certificate for that tester. The lot passed every functional test we ran, and it still generated a nonconformity, because the traceability chain for a moulded rubber part is not the chain most labs think it is. ISO 17025 traceability for septa fails at the same link almost every time, and it is not the one people check first.
Traceability, in the ISO/IEC 17025:2017 sense (clause 6.5), means an unbroken comparison chain to a national or international reference through calibrations that each carry a stated uncertainty. For an instrument, that chain is familiar. For a consumable, it has to be reconstructed from the material inward, and it looks like this:
- Link 1: the reference material or reference method. For a septum, this is usually a certified reference polymer for hardness (IRMM or NIST reference blocks for Shore A), a certified gas mixture for bleed testing, or a gravimetric standard for extractables.
- Link 2: the calibration of the measuring instrument against that reference, with uncertainty and a stated interval. This is the durometer, the thermogravimetric analyser, the GC used for bleed, the balance.
- Link 3: the measurement of the production lot using that calibrated instrument, on a defined sample plan.
- Link 4: the link from the measured lot to the physical units in the box the lab received. Lot number, date code, mould cavity if the vendor tracks it.
- Link 5: the link from the box to the specific test or instrument where the septum was used.
Most vendors can produce links 1 through 3 on request. Link 4 is where things get thin. Link 5 is where the chain breaks for almost every lab I have audited, and it breaks silently, because nothing in the septum itself carries an identity.
Link 4 is the one that quietly breaks: lot-to-box-to-cavity
A septum is a moulded part. It comes off a multi-cavity tool, gets deflashed, gets washed, gets packed into a bag, and the bag goes into a box. If the vendor’s lot definition is “one day’s production across all cavities,” then a single lot number can span material from two different compound batches and four cavities with different thermal histories. The certificate says the lot met 47 ± 3 Shore A. It does not say which cavity your box came from, and if the vendor does not hold cavity-level data, nobody can tell you.
I have seen this fail in a way that matters. A customer running headspace GC on a vial septa line reported that one box in five gave silicone bleed peaks at m/z 73 and 147 above their internal threshold, while the other four boxes were clean. Same lot number, same certificate. The vendor’s production records showed that lot ran across two presses, and one press had a post-cure oven running 8 °C low for part of the shift. The under-cured cavities bled more. Without cavity or press-level traceability on the box, the customer had no way to quarantine the affected units, and the vendor had no way to know which boxes to recall. The fix was a lot definition change: one lot equals one press, one compound batch, one post-cure cycle.
When you evaluate a septum supplier, ask what defines a lot. If the answer is a date, push back. If the answer is a press and a compound batch with a retained sample per lot, you have link 4.
Link 5 is the link most labs cannot produce, and it is the one auditors ask about
Here is the question that ends the conversation in a 17025 assessment: “Show me which septa were in the inlet when this batch of results was generated, and show me the traceability record for those septa.”
Most labs cannot answer it. The septum is consumed, discarded, and replaced on a schedule that is not tied to a specific analytical run. The lab has a box with a lot number, and a maintenance log that says “replaced inlet septum 12 April.” Nothing connects the two. The chain from the box to the result is broken, and no amount of vendor documentation repairs it, because the break is on the lab side.
This is the link I would spend your effort on. The vendor can only supply links 1 through 4. Link 5 is a lab record-keeping problem, and it is solvable with a small amount of discipline.
What link 5 actually requires
You need to be able to state, for any result, which physical septum lot was in the flow path. That means:
- A septum log that records lot number, install date, install time, instrument ID, and the run sequence or batch ID at installation. One line per change.
- A retained sample per lot, one or two units, bagged and labelled, held for the retention period your quality system requires. This is what lets you re-test if a bleed or coring problem surfaces later.
- A link from the log to the maintenance record, so the maintenance entry references the septum log line. Two fields, not a new system.
That is the whole fix. It takes about 30 seconds per septum change. I have watched labs spend weeks building a LIMS integration for this when a paper log with a lot number column would have satisfied the assessor.
What the vendor’s certificate can and cannot carry
A certificate of analysis for a septum is not a calibration certificate, and treating it as one is a common error. The CoA reports measured values against a specification. It does not, by itself, establish traceability unless it names the reference and the calibrated instrument behind each value. Here is what to demand, line by line:
| Property | Typical spec | What traceability requires | What most CoAs actually say |
|---|---|---|---|
| Hardness | 45 to 50 Shore A | Instrument ID, calibration cert number, reference block ID, uncertainty | “47 Shore A” with no instrument reference |
| Compression set | < 25% after 22 h at 100 °C | Oven calibration, method reference (ASTM D395), sample count | Pass/fail only |
| Bleed (TGA or GC) | < 50 µg/g at 250 °C | Instrument ID, calibration gas or standard, method, detection limit | “Low bleed” or a single number |
| Puncture life | > 100 punctures at 23 G | Needle spec, puncture rate, acceptance criterion, sample size | Not reported at all |
| Extractables | Per USP <661> or client spec | Balance calibration, water grade, gravimetric method | Method cited, no instrument trace |
The right-hand column is the problem. A CoA that reports a number without naming the calibrated instrument behind it is a specification statement, not a traceability document. If your quality system treats it as the latter, you have a gap you can close by asking the vendor for the calibration records behind the lot. A vendor who cannot produce them is telling you something useful.
Field case: the durometer that drifted 4 points over a year
A lab sourcing septa for a GC inlet had a run of inlet leaks that traced to septa seating poorly. The symptom was a retention time shift of 0.04 min on early-eluting peaks and a rising baseline on the FID. The lab pulled the retained samples and re-measured hardness. The current lot read 51 Shore A on their bench durometer, against a spec of 45 to 50. The vendor’s CoA said 47.
The discrepancy was the durometer. The lab’s handheld tester had last been calibrated 14 months earlier and had drifted 4 points high over that period, which the calibration lab confirmed on the next cycle. Every septum the lab had measured in that window was reading high, and the vendor’s units were inside spec all along. The fix was a calibration interval reduction to 6 months for the durometer and a switch to a benchtop tester with a certified reference block check before each session. The takeaway is not “calibrate your durometer.” It is that the lab could not have found this without the retained samples, and the retained samples were only useful because they carried a lot number that matched the CoA.
Field case: the temperature rating that was not in the chain
A different lab was running a high-temperature inlet method at 300 °C with silicone/PTFE septa rated to 250 °C. They knew the rating. What they did not have was a record connecting the rating to the specific application. The septa degraded, the PTFE face delaminated in sheets, and the resulting particulate contaminated the liner and the column inlet. The failure surfaced as a ghost peak at m/z 207 that appeared only on the first injection after a septum change.
The traceability gap here was not a missing certificate. It was a missing link between the consumable’s stated limit and the method’s operating condition. The fix was a septum selection record: for each method, a line stating the inlet temperature, the septum material, the rated maximum, and the margin. Anything running within 20 °C of the rating gets flagged for review. That record is what an assessor wants to see when they ask whether the consumable was fit for the method. It is also the kind of thing that gets skipped because it feels obvious, right up until the delamination.
A decision checklist for evaluating a septum supplier’s traceability
Use this when you are qualifying a new supplier or re-qualifying an existing one. Score each line yes or no. Two or more “no” answers means the chain has a gap you will have to document as a risk.
- Lot definition: one lot equals one compound batch and one press run. Yes / No.
- Retained sample: a sample per lot is held for a stated period and available on request. Yes / No.
- Instrument trace: each CoA value names the calibrated instrument and its calibration certificate. Yes / No.
- Reference standard: the calibration chain for hardness and bleed references a national standard (NIST, IRMM) with uncertainty. Yes / No.
- Cavity or press data: the vendor can tell you which press or cavity produced a given box. Yes / No.
- Change notification: the vendor notifies you before changing compound, cure schedule, or tooling. Yes / No.
- Method reference: test methods cite ASTM, USP, or ISO numbers, not internal codes. Yes / No.
If a supplier cannot answer the first two, the rest do not matter much. Lot definition and retained samples are the foundation, and everything else sits on them.
How to write the traceability statement for your own quality manual
You will eventually need a paragraph in your quality manual that says how septa are handled. Keep it short and make it true. Something like: “Septa are received against a CoA that references calibrated instrumentation. Each lot is assigned a septum log entry on installation, recording lot number, instrument, and date. Retained samples are held for 24 months. Septa are selected against method inlet temperature with a minimum 20 °C margin to the rated maximum.”
That paragraph covers links 4 and 5, which are the ones the lab owns. Links 1 through 3 are the vendor’s, and your incoming inspection is where you verify them. If you are running an HPLC vial or autosampler vial line alongside your GC work, the same logic applies to the septa on those vials, and the same log discipline carries over.
Where the standards actually point, and where they leave a gap
ISO/IEC 17025:2017 clause 6.5 requires traceability for measurement results, and clause 6.6 covers externally provided products and services. A septum is both: it is a product you buy, and its properties feed measurements you report. The standard does not tell you how to trace a moulded rubber part, and the chromatography methods do not either. EPA SW-846 names the consumable only indirectly, and USP <621> does the same. That is why the traceability chain for septa has to be built by the lab and the vendor together, not lifted from a method.
The practical consequence is that you should not wait for a standard to tell you what to record. Record the lot, keep a sample, and know your inlet temperature margin. Those three things cover most of what an assessor will ask. For the instrument side, the calibration records from Agilent, Thermo Fisher, and your own metrology lab are the raw material. For method and material background, the technical notes from Restek are worth reading before you write your selection record.
The decision rule
If you cannot name the septum lot that was in the flow path for a given result, you do not have ISO 17025 traceability for that consumable, regardless of what the certificate says. Fix link 5 first with a septum log and a retained sample per lot, then push the vendor on link 4 by asking what defines a lot and whether cavity or press data exists. Links 1 through 3 are the vendor’s to prove and yours to verify at incoming inspection. The chain is only as strong as the link you can actually produce on the day the assessor asks, and for most labs that link is the one on their own bench, not the one in the vendor’s file.
References & further reading
Related resources from our lab consumables network:
- Septa at septacaps.com
- Vial Septa at vialsepta.com
- Hplc Vial at hplcgc.com
- Autosampler Vial at autosamplerhplc.com
External references and standards:
