Last February a QC lab in Ohio rejected an entire incoming lot of 100,000 autosampler vials after a 12-hour sequence produced a creeping baseline at 210 nm that no amount of mobile phase flushing would clear. The vendor’s certificate of analysis said everything was fine. The dimensional report said everything was fine. The problem was that the vendor had no incoming glass-tubing specification on file, had switched tubing suppliers two months earlier, and had no mechanism to notice the change. The lot passed their outgoing inspection because their outgoing inspection only checked what they had always checked. That is the gap an autosampler vial supplier audit checklist is supposed to close, and most checklists I have seen do not close it.

What follows is the version I use. It is built around a single premise: you are not auditing the vials you receive, you are auditing the vendor’s ability to know when something has changed before you do. The physical factory walk-through and the contract clauses are covered elsewhere on this site, so I am not going to repeat them. This is about the quality-system items that predict problems, the evidence you request before anyone sets foot on site, and the arithmetic that turns a pile of observations into a defensible score.

If I only had one hour with a new supplier, I would spend 40 minutes on the change-control register and 20 minutes on everything else. Finished-goods inspection tells you what the vials are today. The change-control register tells you what they will be in six months, and whether anyone will tell you when it happens.

Ask for the register covering the last 24 months and read it for four categories of change:

• Resin and tubing source changes. A new borosilicate tubing draw, a new PTFE film lot, a new silicone supplier. Any of these can shift extractables without moving a single dimensional number.
• Mold and tooling changes. A reworked cavity, a new gate, a polished finish. The bore diameter on a spec sheet is often a promise the mold cannot keep after a tooling change, and the change log is where you find out it happened.
• Process parameter changes. Annealing profile, wash chemistry, drying temperature, the shift from a batch wash to a tunnel wash. These are the changes that quietly move a vial from “clean” to “clean enough for UV but not for LC-MS.”
• Site and line changes. A second production line, a relocated warehouse, a new converter for the septa. The physical factory walk-through covers the site itself. The register covers the fact that the site changed at all.

The tell is not the number of entries. It is whether the vendor can show you a customer notification for each one that could plausibly affect a regulated method. A supplier with 14 entries and 14 documented notifications is healthier than a supplier with 2 entries and no notification trail, because the second supplier is either not looking or not telling.

Which quality-system records predict a vial failure before it ships

Every vial vendor will hand you an ISO 9001 certificate. That certificate tells you they have a documented system. It does not tell you whether the system catches the failure modes that actually bite in an autosampler. I look for six records, and I weight them by how directly they sit upstream of a known failure.

The incoming glass-tubing record is the single highest-value document

This is the control point that separates serious vendors from assemblers. Request the incoming inspection records for the last six tubing lots, and look for at least these measured values per lot:

• Bore diameter measured at both ends and mid-length, with the gauge or air-gauge method named. A tolerance of ±0.05 mm on a nominal 6.0 mm ID is realistic; a claim of ±0.01 mm on drawn tubing is not, and I have never seen it hold across a lot.
• Wall thickness with the measurement positions documented. Wall variation drives crimp seal consistency more than most people realize.
• Chemical durability or hydrolytic resistance per USP <660> or the ISO 719 / ISO 720 equivalents, with the actual titration result, not a pass/fail box.
• Visual defect classification against a written standard, with the AQL level stated. AQL 0.65 on critical defects is the number I insist on before shipment; if the vendor cannot state their AQL, they do not have a sampling plan, they have a habit.

If the incoming record shows only a supplier certificate and a “received OK” stamp, you have found the gap. That vendor is trusting their tubing supplier the same way you are being asked to trust them, with no independent measurement in between. When you are buying through a distributor or a autosampler vial catalog house, ask whether the incoming check is performed at the converting site or inherited from a parent factory. Inherited checks are where the Ohio lab’s creeping baseline came from.

Septum lot traceability down to the film and the compound

A septum is a laminate, and the failure modes live at the interfaces. Request, per lot:

• The PTFE film supplier, thickness (typically 0.05 mm to 0.13 mm for a 1 mm total septum), and whether the film is skived or cast.
• The silicone compound supplier and durometer, usually 45 to 55 Shore A for a standard 9 mm septum.
• The lot number linking the finished septum to both inputs, so a complaint can be traced back to a film lot in under a day.

PTFE is the default face and it is not a default septum. The film is what keeps the silicone from touching your sample; the silicone is what provides the reseal. If the vendor cannot name the film thickness, they cannot tell you why a needle coring complaint started in March.

Extractables data with the method attached

Ask for extractables data and then read the method before you read the numbers. A vendor who ran a 24-hour soak in HPLC-grade water at ambient and reports “low extractables” has told you almost nothing about a 30-minute acetonitrile gradient at 40 °C. What I want to see:

• Solvents that match the intended use: water, methanol, acetonitrile, and at least one acidic and one basic aqueous condition.
• Time and temperature stated, ideally bracketing your method (for example 24 h at 40 °C and 60 min at 80 °C).
• Detection by both UV at 210 and 254 nm and by LC-MS or GC-MS, with the reporting threshold stated in µg/mL.
• The actual chromatograms, not a summary table. I want to see whether there is a peak at the retention time of the analyte that failed in your lab.

For GC-facing work, the same logic applies to a GC vial and to a headspace vial, but the relevant extractables are different and the septa are usually thicker. Do not accept a single extractables report as covering the whole catalog. Each format has its own film, its own crimp geometry, and its own failure surface.

The evidence to request before the audit, and what each item actually proves

Request this pack 10 working days ahead. The speed and completeness of the response is itself a data point; a vendor who needs three weeks to assemble the documents has a document-control problem you will inherit.

The Cpk item deserves a note. A vendor can meet a ±0.05 mm bore tolerance with a process centered at the edge and still ship you a lot where 3% of vials are 0.04 mm off nominal. Cpk of 1.33 on bore diameter and on overall height is the floor I will accept for a vial that has to seat in a fixed-height autosampler tray. Below that, you are buying a distribution, not a specification. For method context on why the tray geometry is unforgiving, the Agilent chromatography consumables and columns documentation is a reasonable reference point for what the instrument expects.

Scoring a vendor objectively instead of arguing about impressions

The reason most supplier audits end in a vague “they seemed fine” is that nobody agreed on weights before the visit. I use a 100-point score with five categories, and I set the weights before I open the evidence pack. The weights reflect how much each category predicts a failure you will actually see in a sequence.

Decision rule: below 70, do not qualify. Between 70 and 85, qualify with a documented incoming-testing uplift on your side (dimensional sampling at AQL 0.65 plus a functional crimp test per lot). Above 85, you can move to certificate-based acceptance for non-critical attributes, but keep the functional crimp test. I have never regretted keeping the crimp test.

Two scoring traps. First, do not let a strong extractables report compensate for weak change control; they measure different things, and the change-control failure is the one that arrives unannounced. Second, score the evidence you were given, not the answers you were told. A vendor who says “we always notify customers” but cannot produce a notification is a 12 out of 25, not a 25.

Two field cases where the audit score predicted the failure

Case 1: the crimp torque drift that a Cpk number would have caught

Symptom. A contract lab running EPA 8260 volatiles on a headspace vial saw a 4% RSD blowout on the internal standard across a 40-vial batch, always in the last third of the tray.

Measurement. They pulled 30 vials from the suspect lot and measured seal compression with a crimp-force gauge. Mean crimp force was 1.8 N·m with a standard deviation of 0.22, and the low-force tail correlated with the late-tray failures. The vendor’s process was centered near the low end of their own 1.5 to 2.0 N·m window with a Cpk around 0.9. The lot met specification. The distribution did not.

Fix. The lab added a per-lot crimp-force check on 10 vials (accept 1.6 to 1.9 N·m, reject if any vial falls below 1.5) and moved the vendor to the 70-to-85 score band with incoming uplift. The blowout stopped. The audit lesson: a Cpk question asked before the visit would have flagged this vendor without a single failed run. The U.S. EPA SW-846 test methods put the container inside the analytical batch, so a seal distribution problem is a data-quality problem, not a handling nuisance.

Case 2: the film thickness change that only the change log explained

Symptom. An environmental lab on a vial format for a semivolatile method started seeing needle coring on roughly 1 in 400 injections, with visible PTFE fragments in the septum puncture.

Measurement. Puncture-count testing on 50 septa from the current lot versus a retained lot showed the current film failing at a median of 180 punctures versus 320 on the retained lot. The film thickness measured 0.05 mm on the current lot against 0.10 mm on the retained lot. The vendor’s change log, which the lab had requested during a re-audit, showed a film supplier switch eight weeks earlier with no customer notification.

Fix. The lab added a film-thickness check (0.10 mm ± 0.02) to its incoming spec and required written notification of any film supplier change as a condition of continued qualification. The audit lesson: the complaint log would have shown nothing, because the vendor had not yet received enough complaints. The change log showed it immediately. That is why change control carries 25 points.

The checklist items that map to a real standard, and the ones that do not

A checklist is only as good as its references. Here is how I map the items to standards so the vendor cannot argue about what “good” means.

• Dimensional and fit: reference the GPI 9 mm neck finish for crimp and the 11 mm snap geometry where relevant, and require the vendor’s drawing to state nominal and tolerance for bore, height, OD, and neck. Compare against the instrument maker’s tray specification. The Thermo Fisher chromatography and sample handling documentation is a useful cross-check for what the autosampler expects.
• Glass chemical durability: USP <660> hydrolytic resistance, or ISO 719 / ISO 720. Require the titration result per tubing lot.
• Extractables: no single harmonized standard, so specify the method yourself. Require multi-solvent, time and temperature bracketing your method, and chromatograms. Reference Restek technical resources for chromatography when you need a defensible extractables protocol to hand the vendor.
• Sampling: AQL 0.65 on critical defects, AQL 1.0 to 1.5 on major, AQL 2.5 to 4.0 on minor, with critical defined as any defect that can affect seal integrity, needle entry, or extractables.
• Septum performance: puncture count on a defined needle gauge (typically 22 gauge, bevel type stated), reseal test, and film thickness with tolerance.

The items that do not map to a standard, and where I therefore insist on a written internal specification with a number in it: crimp-force window, wash-water conductivity and final rinse volume, drying temperature and residence time, and the defect classification itself. A vendor who cannot put a number on their wash process does not control it.

What I do with the score after the audit

The decision rule is simple and I apply it without exception. Qualify at 70 or above with incoming uplift, at 85 or above with certificate-based acceptance plus a retained functional crimp test. Re-audit the change-control and incoming-material categories every 12 months, and every time the register shows a change that touches glass, film, or wash chemistry. Keep the extractables report on file with the method attached, because the next time a baseline creeps at 210 nm you will want to know whether the vial changed or your method did. The score is not a report card. It is the number that tells you how much of the incoming inspection you can stop doing yourself, and it is only worth trusting if the vendor’s change log is worth trusting first.

References & further reading

Related resources from our lab consumables network:

• Autosampler Vial at autosamplerhplc.com
• Vial at vials-filters.com
• Headspace Vial at headspacevialsupplier.com
• Gc Vial at gcvialfactory.com

External references and standards:

• Agilent chromatography consumables and columns
• Thermo Fisher chromatography and sample handling
• U.S. EPA SW-846 test methods for waste and materials
• Restek technical resources for chromatography