A 9 mm crimp VOA vial that mic’d 0.4 mm oversize on the body OD passed incoming inspection for eleven lots before the first autosampler jam. The culprit was not the glass. It was a mold cavity running 12 °C hotter than its neighbors on a 24-cavity tool, which pushed the parison out of the blow station before it fully formed. That is the reality of voa vial dimensional tolerance capability: the number on the drawing is a promise, but the number that ships is a distribution, and the distribution is what your autosampler actually sees.
This article is about the gap between those two numbers. If you already know what a VOA vial is and why 40 mL matters, you are in the right place. What follows is how to read a tolerance callout, how to measure whether a supplier can actually hold it, and which dimensions on a 40 mL screw-thread or crimp vial are genuinely hard to control versus which are easy marketing.
Ask a supplier for a drawing and you will usually get height, OD, ID, thread spec, and a note that says “±0.5 mm.” That single callout is meaningless because it lumps together dimensions with wildly different process capability. Height on a blow-molded or injection-stretched vial is easy to hold. Body OD is not, because it depends on parison temperature, blow pressure, mold cooling uniformity, and the tool’s own wear state.
The relevant number for autosampler compatibility is the maximum body OD over the full 40 mL height, not the nominal. A 27.5 mm nominal body with a ±0.5 mm tolerance means a supplier is allowed to ship 28.0 mm, and some will. Most 12-position autosamplers and many tray-based handlers are built around a nominal 27.5 mm body with roughly 0.3 mm of clearance per side. That is 0.6 mm total. If your supplier’s upper tolerance limit is 0.5 mm above nominal, you have already consumed most of your clearance budget before the part even reaches the tray.
I specify 27.5 mm +0.0/−0.3 mm on the body OD for screw-thread vials, and I reject any supplier who cannot show a Cpk of 1.33 or better on that dimension across at least 30 consecutive parts from each cavity. That is not an unreasonable ask. It is the minimum for a dimension that will physically stop your instrument.
What a real drawing looks like versus a catalog spec
A catalog spec says “27.5 mm.” A production drawing says “27.5 +0.0/−0.3, Cpk ≥ 1.33, measured at 3 axial stations per part, 5 parts per cavity per shift.” The difference is not pedantry. It is the difference between a dimension that is controlled and a dimension that is hoped for.
What each molding process can actually hold on a 40 mL vial
Three processes dominate VOA vial production: injection blow molding (IBM), injection stretch blow molding (ISBM), and straight injection molding of the body and neck in a single shot. Each has a different tolerance floor, and the floor is set by physics, not by how much the supplier cares.
| Process | Typical body OD capability (Cpk 1.33) | Neck ID / thread capability | Wall thickness variation | Where it fails |
|---|---|---|---|---|
| Injection blow molding (IBM) | ±0.25 mm | ±0.15 mm on thread pitch diameter | ±0.15 mm | Parison temperature drift between cavities |
| Injection stretch blow (ISBM, 1-step) | ±0.15 mm | ±0.10 mm | ±0.08 mm | Stretch rod alignment, preform concentricity |
| Straight injection (body + neck, one shot) | ±0.10 mm | ±0.08 mm | ±0.05 mm | Cycle time, mold temperature zoning |
Read that table carefully. A straight-injection vial can hold ±0.10 mm on body OD because the mold cavity is the final dimension. An IBM vial cannot, because the parison is re-formed in a blow station and the blow station is where the variation enters. When a supplier quotes you ±0.1 mm on an IBM vial, ask which dimension. If they say body OD, ask for the Cpk data. If they cannot produce it, they are quoting a mold-spec number, not a process-capability number.
For context on how these tolerances interact with the rest of the consumable stack, the septa and caps you pair with the vial have their own dimensional windows. A septa that is 0.05 mm thick at the low end of its tolerance and a cap that is 0.05 mm shallow at the low end of its tolerance can produce a seal that leaks at 15 psi even though every component passed its own inspection. Tolerance stack-up is the enemy, and it lives at the interfaces.
How to measure capability instead of trusting a certificate of analysis
A CoA tells you what one sample from one lot measured. It does not tell you what the process is capable of. Those are different questions, and the second one is the one that predicts whether lot 47 will jam your autosampler.
Capability is measured with a Cpk calculation on a statistically valid sample. For a 40 mL vial body OD, I want:
- Sample size: at least 30 parts per cavity, drawn consecutively, not cherry-picked
- Measurement method: calibrated digital micrometer or optical comparator, 0.001 mm resolution, 3 axial stations per part (top, mid, bottom of the 40 mL body)
- Calculation: Cpk = min[(USL − mean)/(3σ), (mean − LSL)/(3σ)]
- Acceptance: Cpk ≥ 1.33 on body OD and neck thread pitch diameter; Cpk ≥ 1.67 on the sealing land
The sealing land is where the septa compresses against the vial neck. On a crimp vial, that is the flat annulus the aluminum seal crushes onto. On a screw vial, it is the top face of the thread finish. That surface needs to be flat to within 0.05 mm and perpendicular to the bore axis within 0.1 mm. If it is not, the septa will seal on one side and leak on the other, and no amount of torque will fix it.
The measurement that catches a drifting mold
Run a 30-part capability study every quarter, not every year. Molds wear. A cavity that held Cpk 1.6 at tool birth can drift to Cpk 1.1 after 200,000 shots, and the drift is silent because the mean shifts slowly. Plot the mean and range on an X-bar/R chart per cavity. When one cavity’s mean walks 0.08 mm off the tool average, you have a cooling channel problem or a worn cavity insert, and you have about two lots before it becomes a customer complaint.
The dimensions that fight back: thread pitch diameter, sealing land flatness, and concentricity
Body OD is hard. These three are harder.
Thread pitch diameter on a 9 mm screw finish
The GPI 9 mm screw thread (sometimes called the 9-425 finish) has a nominal pitch diameter around 8.4 mm with a tolerance window of roughly 0.1 mm on a good tool. The problem is that thread pitch diameter is measured with a thread micrometer or a go/no-go gauge, and most suppliers only run go/no-go. A no-go gauge catches parts that are too large. It does not catch parts that are too small, and a too-small pitch diameter means the cap threads strip before the septa seats. I have seen a lot where every vial passed the no-go gauge and 3% of caps spun free at 1.2 N·m of torque instead of the specified 1.5 N·m.
The fix is a pitch-diameter micrometer reading on 10 parts per cavity per lot, with a tolerance of 8.40 ±0.05 mm. That is tighter than the GPI spec, and it is what you need if you are running an autosampler that torques caps mechanically.
Sealing land flatness and perpendicularity
This is the dimension that separates a vial that holds vacuum for 24 hours from one that loses it in 90 minutes. Measure it on an optical flat or a profile projector. The land must be flat within 0.05 mm and perpendicular to the bore within 0.1 mm. On an IBM vial, the land is formed in the blow station, and if the blow station’s insert is not concentric with the parison, the land tilts. You will not see it by eye. You will see it as a leak rate that climbs from 0.02 sccm to 0.4 sccm over a 12-hour headspace run.
Concentricity of neck to body
Total indicated runout (TIR) between the neck bore and the body OD should be under 0.15 mm. Above that, the vial sits cocked in the autosampler tray, the needle enters at an angle, and you get coring of the vial septa on the third or fourth injection. Coring is not a septa problem. It is a concentricity problem wearing a septa costume.
Two field cases where the tolerance stack, not the vial, was the failure
Case 1: the 0.4 mm oversize lot that passed CoA
Symptom: A 12-position autosampler jammed on position 7 of a 40-vial run, three times in two days. The needle bent on the second jam.
Measurement: We pulled 20 vials from the lot and mic’d body OD at three stations. Mean was 27.82 mm, max was 27.94 mm, against a drawing of 27.5 +0.0/−0.3. The lot was out of spec on the high side by 0.14 mm at the max. The supplier’s CoA showed 27.61 mm, within spec, because they measured one part from one cavity.
Root cause: Cavity 14 on the 24-cavity tool was running 12 °C hot because of a partially blocked cooling line. The parison in that cavity was softer and blew out further before the mold closed.
Fix: Supplier pulled the tool, cleaned the cooling circuit, and re-ran a 30-part capability study per cavity. Cavity 14 came back at Cpk 1.41 on body OD. We added a requirement for per-cavity Cpk data on every lot. The jams stopped. The lesson: a single-part CoA is not a capability measurement, and a lot that passes a one-part check can still be 0.4 mm out at the tail of the distribution.
Case 2: the leak that was a sealing land, not a septa
Symptom: A client running EPA Method 8260B volatiles on a headspace GC saw 8% of vials fail the initial pressure decay test at 15 psi. They had already switched septa suppliers twice.
Measurement: We put 10 failed vials on a profile projector and measured sealing land flatness. Six of the ten had land tilt of 0.12 to 0.18 mm, against a 0.10 mm perpendicularity spec. The septa were fine. The vial neck was cocked.
Root cause: The supplier had changed blow-station insert tooling three lots earlier, and the new insert was not concentric with the parison guide. Every vial from that tool run had a land that tilted toward one side.
Fix: Supplier re-centered the insert and added a TIR check on the neck-to-body concentricity at 0.15 mm max. Leak rate dropped to 0.3% within one lot. The client had spent roughly $4,000 on septa they did not need. This is why I always measure the vial before I blame the closure.
For reference on what the method actually requires of the container, the U.S. EPA SW-846 test methods set the analytical requirements, but they do not set the dimensional tolerances. Those are yours to specify, and if you do not, the supplier will specify them for you at whatever the tool happens to hold.
A decision checklist for specifying voa vial dimensional tolerance capability
Use this when you write a purchase specification or audit a supplier. Each line is a number you should be able to point to on a drawing or a capability report.
- Body OD: 27.5 mm +0.0/−0.3 mm, Cpk ≥ 1.33, measured at 3 axial stations, 30 parts per cavity
- Height (crimp): 75.0 ±0.5 mm; height (screw): 75.0 ±0.5 mm
- Neck thread pitch diameter (9 mm screw): 8.40 ±0.05 mm, Cpk ≥ 1.33
- Crimp finish OD: 20.0 ±0.15 mm, Cpk ≥ 1.33
- Sealing land flatness: 0.05 mm max; perpendicularity to bore 0.10 mm max; Cpk ≥ 1.67
- Neck-to-body concentricity (TIR): 0.15 mm max
- Wall thickness: 1.2 ±0.15 mm on the body, measured at 4 quadrants
- Bore ID: 24.0 ±0.2 mm (screw) or 22.0 ±0.2 mm (crimp), whichever matches your finish
- Glass weight: 24.0 ±1.5 g, as a proxy for wall uniformity
- Per-cavity Cpk report: required on every lot, not just at PPAP
- Measurement equipment: calibrated micrometer at 0.001 mm resolution, optical comparator for land and TIR
If a supplier cannot meet the Cpk line on body OD, do not negotiate the tolerance. Negotiate the process. A ±0.3 mm tolerance with Cpk 1.0 will ship you out-of-spec parts roughly 0.3% of the time, which on a 10,000-vial order is 30 bad vials, and 30 bad vials is 30 potential autosampler jams. A ±0.3 mm tolerance with Cpk 1.33 ships out-of-spec parts at about 0.003%, which is effectively zero.
What to ask on a supplier audit
Three questions separate a real capability claim from a catalog number:
- “Show me the per-cavity Cpk on body OD for the last three lots.” If they produce a single aggregate Cpk, ask for the per-cavity breakdown. A tool with one bad cavity can hide behind a good aggregate.
- “What is your measurement uncertainty on the OD micrometer, and how often is it calibrated?” If the uncertainty is more than 20% of the tolerance, the capability number is noise.
- “What is your tool maintenance interval, and what triggers a re-qualification?” A tool that runs 500,000 shots without a capability re-check is a tool that will drift on your order.
For cross-reference on how the instrument side constrains these numbers, the Agilent chromatography consumables and columns documentation and the Thermo Fisher chromatography and sample handling guides both publish autosampler tray clearances that map directly to the body OD window. The Restek technical resources library is also useful for septa compression and torque guidance that pairs with the vial tolerances above.
Where the tolerance budget actually goes, and how to stop paying for it twice
Every interface on a VOA vial consumes tolerance budget. The body OD consumes autosampler clearance. The thread pitch diameter consumes cap engagement. The sealing land consumes septa compression. The neck-to-body concentricity consumes needle entry angle. When you add them up, you get a stack that either fits inside the instrument’s window or does not.
The mistake I see most often is that buyers specify tight tolerances on the easy dimensions (height, weight) and loose ones on the hard dimensions (body OD, land flatness), because the easy ones are what the catalog quotes. Flip that. Height at ±0.5 mm is fine. Body OD at ±0.3 mm with Cpk 1.33 is where the money goes, and it is where the jams come from. If you are sourcing voa vial stock for a high-throughput volatiles lab, the OD and land numbers are the two you should refuse to compromise on. The rest you can negotiate.
One more number worth specifying: torque on the cap. For a 9 mm screw cap with a PTFE-faced silicone septa, the target is 1.5 N·m ±0.2 N·m. Below 1.2 N·m the septa does not compress enough to seal at 15 psi. Above 1.8 N·m the cap threads strip or the septa extrudes. If your vial’s thread pitch diameter is at the low end of its tolerance and your cap is at the high end, you will hit the strip threshold before you hit the seal threshold. That is a tolerance stack failure, and it looks exactly like a “bad cap” complaint until you measure the pitch diameter.
For HPLC work on the same platform, the same logic applies to the hplc vial geometry, where the neck OD and the insert fit matter more than the body OD because the autosampler grips the neck. The dimensional discipline is the same. The dimension that matters changes.
The decision rule
Specify body OD at 27.5 mm +0.0/−0.3 mm and sealing land perpendicularity at 0.10 mm max, then require per-cavity Cpk of 1.33 or better on both, measured on 30 consecutive parts per cavity, on every lot. If a supplier cannot produce that data, they cannot hold the tolerance, regardless of what the drawing says. If they can produce it, you have bought a process, not a promise, and the autosampler will tell you which one you got within the first 200 injections.
References & further reading
Related resources from our lab consumables network:
External references and standards: