Request the evidence packet before you book the audit

We rejected a lot of 500 crimp-top vials in March after a single QC measurement: 38 of 40 closures released below 1.4 N·m on the torque tester, against a drawing callout of 2.0 ± 0.3 N·m. The vials were fine. The caps were fine. The sealing machine was fine. The problem was that the supplier had changed its aluminum cap alloy in January and never told us, and our supplier audit checklist had no line item that would have caught it. That is the whole argument for auditing vial vendors at the quality-system level, not at the “do you have ISO 9001” level.

If you are building a vial supplier supplier audit checklist, the items that predict failure are almost never on the generic template. They are the ones that sit one layer below the certificate: change control on polymer grades, lot traceability back to resin, dimensional capability on the finish, and the supplier’s own incoming inspection of the components it buys. This article covers what to demand before the audit, what to measure during it, and how to score a vendor so the result is a number you can defend to your own QA group.

Half of a vial audit is a desk review. If you send a questionnaire and get back a PDF of certificates, you have learned nothing. Ask for records that only a functioning quality system can produce on short notice, and read them before anyone travels. A vendor with a real system turns these around in three to five business days. A vendor without one sends the certificate again.

What belongs in the pre-audit packet

• Twelve months of certificate of analysis (CoA) data as a spreadsheet, not a scanned PDF. You want to run capability analysis yourself. For a 2 mL vial, ask for neck OD, finish height, ID at the crimp, and septum thickness, each with a stated tolerance and the actual measured value per lot.
• Resin and elastomer lot traceability from the finished vial lot back to the incoming material lot, the supplier of that lot, and the date code. If the vendor cannot walk a specific lot number back to a resin lot in under 30 minutes, the traceability claim is decorative.
• The change control procedure, plus a list of every change to a vial, cap, or septum that affected a customer in the last 24 months. This is the single most predictive document in the packet. A vendor that has no list either has no changes or does not track them, and both answers are problems.
• Incoming inspection plans for the components the vendor buys: glass tubing, aluminum shells, elastomer sheet, and any coatings. What is the sampling plan, what is the AQL, and what happens to a rejected lot?
• Calibration records for the dimensional gauges, torque testers, and leak testers used on your product. Look for the gauge that measures the finish, because that is where the money is.
• Deviation and complaint logs for the last 12 months, with root cause and corrective action. Ask specifically for complaints about closure integrity, particulate, and dimensional nonconformance.

I read the change control list first. If a vendor has changed a septum formulation, a cap alloy, or a glass tubing source in the last two years without a customer notification, the audit’s job is to find out how deep that goes. You can cross-check the material side against published references such as the Restek technical resources on septa and inlet consumables, which give you a baseline for what a formulation change actually does to puncture life and bleed.

The quality-system items that actually predict vial failures

Most audit checklists are built around document control, training records, and management review. Those matter, but they do not predict whether your autosampler will jam at 3 a.m. These four items do.

Change control on materials, not just on drawings

The failure mode is silent substitution. A glass tubing supplier changes its draw process, the ID at the crimp neck shifts by 0.15 mm, and the crimper that was set for the old stock now produces closures that pass a visual check and fail a dye leak test. Ask the vendor to show you the trigger list for a customer notification. It should include any change to resin or elastomer grade, any change to a sub-supplier, any change to a mold or tool, and any change to a sterilization or washing process. If the trigger list only covers drawing revisions, the system does not cover the things that break vials.

Dimensional capability on the finish

Ask for Cp/Cpk on the critical dimensions, not just pass/fail against tolerance. A vendor running at Cpk 0.8 is producing a predictable fraction of out-of-spec parts and catching some of them at final inspection. For crimp vials, the dimensions that matter are the neck OD, the finish height, and the ID at the crimp ring. For screw-thread vials, it is the thread profile and the sealing land. If the vendor cannot produce Cpk data because it only inspects at final, that is your answer.

Cleanliness and particulate control

For LC-MS work, a vial that sheds particles or leachables will show up as ghost peaks and baseline drift long before it shows up as a failed dimensional check. Ask how the vendor controls the wash water, whether vials are rinsed with USP purified water or WFI, and what the particulate specification is. A vendor that cannot state a particulate count per vial has not characterized its own product. This is where the Lab Chromatography reference material on injection and sample prep is useful for calibrating what “clean” means for your method.

Closure integrity testing as a routine, not a type test

Ask whether the vendor runs dye ingress or microbial ingress on a sampling basis, and at what frequency. A type test done once at design validation tells you nothing about lot 47. A vendor running dye ingress at, say, one sample per 10,000 closures with a documented reject action is telling you it understands that closure integrity is a process output, not a design property.

Score the vendor on a weighted matrix, not a pass/fail

A checklist that returns “compliant / non-compliant” hides the difference between a vendor that is 90% there and one that is 40% there. Use a weighted score. The weights below are the ones I use for vial vendors, and they reflect where failures actually come from.

Score each item 0 to 5, multiply by the weight, and total to 500. In practice, a vendor below 350 is a vendor you will be chasing for the life of the contract. A vendor between 350 and 420 is workable with a written improvement plan and a tighter incoming inspection on your side. Above 420, you can move to skip-lot receiving inspection after three clean lots.

Why the weights are what they are

Change control and dimensional capability carry 40% between them because those two items produce the failures you cannot see coming. A vendor with perfect traceability and a beautiful complaint log but no capability data will still ship you a lot that jams the autosampler. Traceability tells you what went wrong after it went wrong. Capability tells you it is not going to.

Two field cases that show the checklist working

Case 1: the alloy change that failed torque

Symptom: intermittent cap loosening on a crimp-top line, roughly 2% of vials failing a manual twist check after sealing.

Measurement: torque testing across 40 closures per lot showed a mean of 1.35 N·m against a 2.0 ± 0.3 N·m spec, with the low tail at 1.1 N·m. The old lots from the same vendor were at 2.05 N·m.

Root cause: the vendor had switched aluminum shell alloy to a softer grade to reduce tool wear. The change was never notified. The crimper setting was correct for the old alloy.

Fix: added a line to the audit checklist requiring a written change notification for any change to cap alloy or shell temper, and added an incoming torque check at 10 closures per lot. The vendor reverted the alloy. The torque came back to 2.0 N·m within one lot.

Case 2: the septum that bled at high temperature

Symptom: a new baseline peak appearing in GC-MS runs after a supplier change, at a retention time that matched a siloxane.

Measurement: running a blank injection with a PTFE membrane filter in line to rule out mobile-phase contamination, then comparing septum bleed across two lots with the same method. The new lot showed a siloxane peak roughly four times the area of the old lot at 280 °C inlet temperature.

Root cause: the vendor had changed the silicone septum formulation to reduce cost. The new formulation had a lower maximum continuous use temperature.

Fix: added a septum formulation lock to the purchase specification and a bleed test at the method’s actual inlet temperature as an incoming check. This is the kind of thing Agilent’s GC columns and accessories documentation is useful for referencing, because it states bleed behavior at defined temperatures rather than a vague “high temperature” claim.

The measurements to take during the audit itself

The desk review tells you what the vendor says. The audit tells you what the vendor can do on the day. Bring your own instruments and take these measurements on a sample the vendor pulls from current stock, not a sample prepared for you.

• Dimensional check on 30 vials from one lot: neck OD, finish height, ID at the crimp, and septum thickness. Compare against the drawing. A spread wider than the stated tolerance means the vendor is sorting, not controlling.
• Torque test on 20 sealed closures using the vendor’s own crimper and your torque tester. If your numbers and theirs disagree by more than 10%, one of the two instruments is out of calibration, and it is worth knowing which.
• Dye ingress on 20 closures after sealing, at the vendor’s standard conditions. A single failure in 20 is a finding, not a fluke.
• Particulate count on a rinse of 10 vials with filtered water, counted by your own method. Compare against the vendor’s stated spec.
• Traceability drill: pick a lot number from the warehouse and ask the vendor to produce the resin lot, the incoming inspection record, and the operator who ran the line, in front of you. Time it. Anything over 30 minutes is a fail.

For reference on what a sample container must actually withstand, the U.S. EPA SW-846 test methods set the container requirements for environmental work, and the Thermo Fisher chromatography and sample handling documentation covers the practical side of what a closure has to do in an autosampler. Neither substitutes for your own measurements, but both give you a defensible baseline for what “conforms” should mean.

Decision rule

Score every candidate vial supplier on the weighted matrix before you sign anything, and set the threshold at 420 of 500 for a vendor you will buy from without a written improvement plan. Below 350, walk away regardless of price, because the cost of a failed lot at 3 a.m. on a validated method is larger than any per-unit saving. Between 350 and 420, buy, but hold the vendor to a 90-day CAPA on the two lowest-scoring items and keep full receiving inspection in place until three consecutive lots pass. The checklist is not a document to file. It is a number you use to decide, and the number should be the same every time you run it. If you need a starting point for the product side of the audit, the Vial Supplier reference material and the HPLC filter documentation give you the specifications to audit against.

References & further reading

Related resources from our lab consumables network:

• Vial Supplier at vialsupplier.com
• Lab Chromatography at labchromatography.com
• Hplc Filter at hplcfilters.com
• Ptfe Membrane Filter at ptfemembranefilter.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