What a molding process can hold on a 12 mm vial body--Sample Vial Manufacturer

A 2 mL clear glass vial lot failed incoming inspection at 11.94 mm body diameter against a 11.90 mm ± 0.10 mm print, and the autosampler gripper on the customer’s 7697A headspace unit dropped three vials in a 96-position tray during the first run. The parts were inside the drawing. The process was not capable. That distinction, between “in tolerance” and “capable,” is what this article is about: what sample vial dimensional tolerance capability actually means on a molding line, how you measure it, and which dimensions on a vial are genuinely hard to hold.

Start with the honest capability numbers by process. Everything else on this page follows from them.

Process	Typical achievable tolerance on Ø12 mm body	Wall thickness control	Neck ID (Ø6.0 mm nominal)	Cosmetic/optical grade
Injection blow molding (IBM), 2-stage	± 0.08 mm	± 0.05 mm	± 0.10 mm	Good
Injection stretch blow molding (ISBM), 1-stage	± 0.06 mm	± 0.04 mm	± 0.08 mm	Very good
Injection molding, 32-cavity, glass-filled PBT or PP	± 0.04 mm	± 0.03 mm	± 0.05 mm	Very good
Injection molding, 64-cavity, same resins	± 0.07 mm	± 0.05 mm	± 0.08 mm	Good
Glass tubing conversion (ND8, ND11 cane)	± 0.15 mm on OD after forming	± 0.08 mm	± 0.12 mm	Excellent (clear, low extractables)

Those are shop-floor numbers from running lines, not catalog claims. The pattern is consistent: cavity count is the single biggest driver of plastic capability, and glass is a different discipline entirely because you inherit the cane tolerance before you form anything. If a supplier quotes ± 0.03 mm on a 64-cavity mold, ask for the SPC data on the worst cavity, not the mean.

I have watched a 32-cavity tool hold a Cpk of 1.51 on body OD across all cavities and a 64-cavity tool on the same resin hold 1.08. Same print. Same resin. The difference was cavity-to-cavity thermal variation in the tool, which is a mold design problem, not a machine problem. That is why sample vial dimensional tolerance capability has to be quoted per cavity count, not per product family.

How capability is actually measured on a vial drawing

Cpk is the number everyone quotes and the number most people misread. On a vial drawing, capability is only meaningful when it is computed per characteristic, per cavity, and over a defined sampling window.

The measurement chain has to be fixed before the number means anything

Body OD on a 12 mm vial is measured with a digital micrometer at a defined axial station, typically 2.0 mm above the base, with a flat anvil, at a controlled 20 ± 2 °C. Change the axial station by 3 mm and you are measuring the shoulder taper, not the body. Change to a blade anvil and you read a different number on the same part. I have seen two labs disagree by 0.09 mm on the same lot because one used a contact micrometer at 2 mm and the other used a vision system at the widest point of the shoulder. Neither was wrong. Neither was measuring the same thing.

Write the measurement method onto the drawing. Nominal, tolerance, gauge, axial station, and temperature. Any supplier that cannot produce a gauge R&R under 10% of tolerance on that method cannot produce a defensible Cpk either.

Why Cpk on the mean hides the cavity that will fail

Take a 64-cavity tool. Pull 30 consecutive shots. Measure one part per cavity per shot. You now have 1,920 data points and 64 distributions. The pooled Cpk might be 1.35. The worst cavity might be 0.72. The pooled number is dominated by the 60 good cavities and tells you nothing about the four that will generate the field failures. This is the same trap described in the sibling article on what Cpk on a vial drawing measures and hides, so I will not repeat it here beyond one rule: always request per-cavity Cpk, and reject any submission that pools cavities.

Ppk versus Cpk on a new tool

A new tool has no history, so you get Ppk (long-term, all variation). Accept Ppk ≥ 1.33 at PPAP. On a running tool, Cpk ≥ 1.33 is the floor and 1.67 is what I specify for any dimension that touches the autosampler. The gap between 1.33 and 1.67 is roughly the difference between a vial that works on a Waters Alliance and a vial that also works on an Agilent 1290 with a robotic plate handler that has 0.3 mm of gripper compliance.

The dimensions that are genuinely hard to control

Not all dimensions on a vial are equal. Four are hard, and the reason each is hard is different.

Neck ID and the 6.0 mm bore

The neck ID sets septum compression and therefore seal integrity. On a 32-cavity injection tool, neck ID is controlled by a core pin that sits in the hottest part of the cavity. Core pin wear of 0.02 mm over 500,000 shots is normal and shows up as a slow drift upward in neck ID, which shows up as a slow drift downward in septum compression force. A neck ID that drifts from 6.00 to 6.06 mm takes a 3.0 mm thick silicone/PTFE septum from 32% compression to 30%. That is inside most prints and outside most seal validation windows. Control neck ID at ± 0.05 mm and monitor core pin wear at every tool PM.

Body OD concentricity to the neck

Concentricity is the dimension nobody puts on the drawing and everybody suffers from. If the neck axis is 0.10 mm off the body axis, the vial tilts in the autosampler nest, the needle enters at an angle, and you core the septum on one side. The mechanical sequence that cores a septum is covered in detail elsewhere on this site; the molding cause is tool misalignment between the neck insert and the body cavity, usually from uneven clamp tonnage on a multi-plate tool. Specify concentricity ≤ 0.05 mm TIR between neck ID and body OD and measure it on a dial indicator with a precision mandrel, not on a vision system that cannot see inside the neck.

Base flatness and the 0.05 mm rock

Glass vials are the worst offenders here. A base that rocks 0.05 mm on a granite plate will sit at an angle in a tray and produce variable needle depth. On a 2 mL glass vial, base flatness is set by the bottom-forming station and by the annealing profile. A supplier that runs its lehr 15 °C too hot will ship vials that look perfect and rock 0.08 mm. Measure base flatness with a dial indicator on a rotating stage, 100% on the first article and AQL 0.65 on production.

Thread form and the GPI 13-425 fit

The GPI 13-425 finish is a 13 mm thread with a 425 (4.25 mm) neck height. The tolerance that matters is not the major diameter; it is the thread root depth and the start position. A cap that engages 1.8 turns instead of 2.5 turns will feel tight and leak. Measure thread engagement with a go/no-go gauge and torque-to-seal with a calibrated torque screwdriver. 0.35 to 0.55 N·m is the working band for a 13-425 PP cap with a PTFE-faced silicone liner. Below 0.30 N·m you get leaks on a 24-hour headspace hold. Above 0.70 N·m you deform the liner and lose the seal on the second use.

Field case: the lot that passed the mean and failed the tray

Symptom: a pharma QC lab running EPA Method 8260 on a headspace autosampler reported intermittent “vial not detected” errors, roughly 3 per 96-vial tray, on a lot that had passed incoming inspection. The incoming report showed body OD mean 11.93 mm, range 11.88 to 11.98, well inside the 11.90 ± 0.10 print. Cpk reported as 1.42.

Measurement that identified it: we pulled 200 vials from the lot and measured body OD at four axial stations, then measured roundness with a two-point micrometer at 0° and 90°. The mean was fine. The roundness spread was 0.14 mm on 6% of parts, and those parts were all from cavities 41 through 48. The pooled Cpk had hidden it because the oval parts were a small fraction of the sample.

Fix: the tool had uneven cooling on the far side of the manifold, and cavities 41 to 48 ran 8 °C hotter than the rest. The supplier added baffling to the cooling circuit and re-ran per-cavity capability. Worst cavity went from Cpk 0.81 to 1.44. We also changed the incoming inspection to measure roundness on 10 parts per cavity, not 10 parts per lot.

Field case: the glass lot that drifted 0.12 mm over six months

Symptom: a contract lab running SW-846 methods on a GC-MS reported increasing needle bend events on a 10 µL syringe, from 1 per 2,000 injections to 1 per 300. No change in method, no change in septa, no change in the autosampler.

Measurement that identified it: we measured neck ID on retained samples from three lots spanning six months. Lot A: 6.02 mm mean. Lot B: 6.07 mm. Lot C: 6.11 mm. The supplier had changed cane tubing source and the forming station had not been re-qualified. A 0.09 mm neck ID increase reduced septum compression enough that the needle entered with 0.04 mm of lateral play, which over thousands of injections bent the syringe.

Fix: added neck ID to the incoming inspection plan at ± 0.05 mm with a plug gauge, and required the supplier to notify on cane source change. The needle bend rate returned to baseline within two lots.

A tolerance stack checklist for incoming inspection

Use this as a decision list, not a data sheet. Every line is a dimension I have seen fail in the field.

Body OD: ± 0.10 mm on 12 mm, measured at 2.0 mm above base, micrometer, 20 °C.
Roundness: ≤ 0.08 mm TIR, measured at 0° and 90°, 10 parts per cavity.
Neck ID: ± 0.05 mm, plug gauge, 100% first article, AQL 0.65 production.
Concentricity neck-to-body: ≤ 0.05 mm TIR, dial indicator on mandrel.
Base flatness: ≤ 0.05 mm rock on granite, 100% first article.
Height: ± 0.15 mm, caliper, 10 per lot.
Thread engagement (GPI 13-425): 2.0 to 3.0 turns to seal, go/no-go gauge.
Torque-to-seal: 0.35 to 0.55 N·m, calibrated screwdriver, 5 per lot.
Per-cavity Cpk: ≥ 1.33 on all above, ≥ 1.67 on neck ID and concentricity.
Gauge R&R: ≤ 10% of tolerance on every characteristic.

If a supplier cannot supply per-cavity data on the first five lines, the rest of the conversation is theater. I would rather buy from a supplier that shows me a worst-cavity Cpk of 1.20 than one that shows a pooled 1.60.

Where OEM and aftermarket diverge on the same drawing

The OEM vial and the aftermarket vial can print the same dimensions and behave differently, because the drawing does not capture the molding window. An OEM tool is qualified at a specific melt temperature, mold temperature, and cycle time, and that window is documented in the PPAP. An aftermarket tool may hit the same nominal dimensions running 20 °C hotter and 15% faster, and the parts will pass dimensional inspection while failing on residual stress. Residual stress shows up as crazing after autoclave, as warpage after two weeks in a hot warehouse, and as base rock after shipping.

Ask for the molding window, not just the drawing. Ask for the residual stress test (polarized light or solvent crazing per ASTM D1939). Ask whether the tool has been re-qualified since the last major repair. If the answer to any of these is “we follow the drawing,” you are buying dimensional compliance and nothing else. For a broader look at how the consumables supply chain is structured, the chromatography products side of the market and the Aijiren manufacturing operation are useful references for how tool qualification is documented at scale.

What to write into the purchase specification

The purchase order is the only document that survives a supplier change. Put the following in it, in this order:

The characteristic list with nominal, tolerance, and measurement method for each dimension.
The sampling plan: per-cavity, per-lot, AQL levels, and the gauge R&R requirement.
The capability requirement: Ppk ≥ 1.33 at PPAP, Cpk ≥ 1.33 in production, ≥ 1.67 on critical dimensions.
The change notification clause: resin, cane source, tool repair, and cavity count changes require re-qualification.
The residual stress and annealing requirement for glass, and the molding window for plastic.
The retention requirement: 12 months of retained samples per lot, so a drift case can be reconstructed.

I have seen a lab avoid a six-figure recall because the supplier had retained samples from the suspect lot and we could measure the neck ID drift in two days instead of two months. Retention is cheap. Reconstruction without it is not.

For the parts themselves, the sample vial catalog and the matched vials with caps line are worth reviewing against the characteristic list above, and the instrument vendor’s own consumable specifications (for example, Agilent’s GC accessories documentation and Thermo Fisher’s chromatography sample handling pages) give you the autosampler-side tolerances the vial has to meet. Restek’s technical resources are also a good cross-check on septum and cap torque guidance.

The decision rule

Buy on worst-cavity capability, not on the drawing and not on the pooled Cpk. If the supplier can show per-cavity Cpk ≥ 1.33 on body OD, neck ID, and concentricity, with a gauge R&R under 10% of tolerance and a documented molding window, the lot will behave. If any of those is missing, the dimensional report is a description of the sample, not a promise about the process, and the 3-in-96 tray failures will find you eventually.

References & further reading

Related resources from our lab consumables network:

External references and standards: