A 12 x 32 mm vial that mics 11.98 mm at the body and 12.04 mm at the shoulder will pass incoming inspection and still jam a 96-position autosampler tray on the third injection. I have seen that exact spread on a lot of 400,000 vials. The body was inside the print. The shoulder was inside the print. The roundness was not, and nobody had asked for it on the drawing. That is the whole problem with oem vial dimensional tolerance capability: most buyers specify two or three numbers and assume the rest is somebody else’s job.
It is not. A tolerance callout is a contract about a manufacturing process, and every process has a floor below which the number becomes a wish. This article is about that floor. Where injection molding lands, where it drifts, how to measure whether a supplier can actually hold a window, and which dimensions on a vial are genuinely hard to control versus which ones any competent molder hits for free.
There are three ways a glass or plastic vial body gets made at volume, and they do not share a tolerance floor.
- Injection molding (plastic, typically PP or COP/COC): general commercial tolerance per OEM vial drawing is ±0.10 mm on a 12 mm body diameter, with a well-run tool and a validated process reaching ±0.05 mm. Wall thickness ±0.03 mm is achievable but only with cavity pressure control, not just velocity-to-pressure transfer.
- Injection blow molding (plastic): ±0.20 to ±0.30 mm on diameter is realistic. The parison reheat and stretch step adds variability you cannot dial out with clamp force. Do not buy blow-molded vials against a ±0.10 mm print and expect capability.
- Tubular glass conversion (glass): ±0.15 mm on outer diameter is typical from a good converter, ±0.10 mm on a selected line. Glass draw tolerances on the tube itself (usually ±0.10 to ±0.15 mm on OD) propagate straight through, so the converter cannot beat the incoming tube.
If a supplier quotes ±0.03 mm on a 12 mm injection-molded body without cavity pressure sensors and a documented capability study, they are quoting a number they have not measured. Ask for the Cpk. If they cannot produce it, the quote is fiction.
Capability is a number, not a certificate
The only honest answer to “can you hold this tolerance” is a process capability index derived from a real sample, taken over time, from more than one lot.
What the indices mean on a vial dimension
Cp is the spread of the process against the tolerance width. Cpk adds the centering penalty. A process with Cp 1.67 and Cpk 0.9 is beautifully consistent and badly off-center; it will fail. For vial dimensions feeding an autosampler, I want Cpk ≥ 1.33 minimum and ≥ 1.67 on the neck OD and the crimp seat, because those two dimensions are the ones that decide whether the vial works.
Sample size matters. A Cpk from 30 parts is noise. You need at least 125 measurements per dimension, drawn across a minimum of three lots and at least three mold cavities if the tool is multi-cavity, because cavity-to-cavity variation is often larger than shot-to-shot variation. A 32-cavity tool where cavity 19 runs 0.04 mm high on neck OD will pass a single-cavity sample and fail in the field.
The measurement system has to be better than the tolerance
Rule of thumb: gauge repeatability and reproducibility should consume no more than 10% of the tolerance band, and never more than 30%. On a ±0.10 mm diameter tolerance, a 0.02 mm band, your measurement system has to resolve to about 0.002 mm. Calipers do not do this. A 0.001 mm resolution optical comparator or a vision system does. If your supplier’s inspection report is built on digital calipers, the Cpk they hand you is a Cpk of the caliper, not the vial.
For dimensional verification tied to a method, run the check through a lab that reports against a documented procedure; an analytical testing partner used to dimensional and functional checks will flag drift you would miss in-house.
The dimensions that are genuinely hard to control
Most of a vial is easy. Diameter, height, and body wall are routine. Four dimensions are where OEM programs actually fail.
Neck outer diameter and the crimp seat
The neck OD on a 11 mm crimp vial is nominally 11.00 mm, and the crimp seat sits at a defined height below the finish. Both are affected by shrinkage, and shrinkage is affected by mold temperature, hold pressure, and the cooling gradient across the neck, which is the thinnest, most thermally awkward section of the part. A 5 °C swing in the neck insert temperature moves the neck OD by roughly 0.01 to 0.02 mm on PP. That is a fifth of your tolerance band from a chiller fluctuation.
This is why neck OD needs Cpk ≥ 1.67 and a documented mold temperature control loop. If the supplier controls mold temperature with a single thermocouple on the manifold and no per-zone control on the neck insert, the number will drift across a shift.
Concentricity of neck to body
This is the one almost nobody puts on the drawing and the one that causes the most autosampler faults. A neck that is 0.05 mm off-axis to the body means the vial sits tilted in the tray, the needle enters at an angle, and the septum is punctured off-center. Off-center punctures core the septum instead of slicing it, and cored septa shed fragments into the vial. I have pulled silicone fragments out of an HPLC injection port that traced back to a 0.06 mm concentricity error on a lot of vials that passed every dimensional check on the certificate.
Specify concentricity ≤ 0.05 mm TIR (total indicator reading) between neck bore and body OD. Measure it on a mandrel with a dial indicator, not by eye.
Bottom thickness and the internal floor
For autosampler vials with a pulled-point or conical interior, the residual floor thickness at the point controls how far the needle can travel before it bottoms out. Too thin and the needle punches through into the tray; too thick and you lose recoverable volume. On a 300 µL conical insert, floor thickness at the point should be 0.25 to 0.40 mm with a tolerance of ±0.05 mm. This is a hard dimension because it is the last place the melt reaches and the first place to flash or short.
Flatness of the base
A base that is not flat rocks in the tray. Flatness within 0.10 mm across the base diameter is the working number for a 12 mm vial. Warped bases come from uneven ejection and from cooling too fast on the core side. If the supplier ejects before the part is rigid, the base cups and you get a rocker. You will not see it in a stack of vials on a bench; you will see it as intermittent needle crashes on positions 40 through 60 of a tray.
Where the tolerance budget actually goes
Here is the comparison I use when a customer asks what to put on a drawing. It is the practical tolerance floor per dimension per process, based on what I have seen validated, not what a molder will quote on a first pass.
| Dimension | Nominal | Injection molded (PP/COP) | Tubular glass | Criticality |
|---|---|---|---|---|
| Body OD | 12.00 mm | ±0.05 mm (Cpk 1.67) | ±0.10 mm | Medium |
| Neck OD (11 mm finish) | 11.00 mm | ±0.05 mm (Cpk 1.67) | ±0.08 mm | High |
| Neck-to-body concentricity | 0 TIR | ≤0.05 mm TIR | ≤0.08 mm TIR | High |
| Height | 32.00 mm | ±0.10 mm | ±0.15 mm | Low |
| Base flatness | 0 mm | ≤0.10 mm | ≤0.12 mm | Medium |
| Internal floor (conical) | 0.30 mm | ±0.05 mm | n/a (molded insert) | High |
| Wall thickness | 1.00 mm | ±0.03 mm | ±0.05 mm | Low |
Notice the pattern. The dimensions that matter to the instrument are the ones with the tightest floor and the highest measurement cost. The dimensions the certificate usually reports (height, body OD) are the easy ones. That mismatch is why a lot can be “in tolerance” and still not work.
Two field cases where the drawing was the problem
Case 1: the drifting neck that passed every certificate
Symptom: a customer running a chromatography vial program on a 2 mL crimp format started seeing cap-seal failures on roughly 1 in 400 vials, always late in a shift. The crimper would not seat the aluminum seal flush.
Measurement: we pulled 200 vials from three lots and measured neck OD on a vision system at 0.001 mm resolution. Lot A ran Cpk 1.51. Lot C ran Cpk 0.94, mean shifted +0.04 mm. The certificate said “11.00 ±0.10 mm, conforming.” It was conforming. It was also off-center in a band that the crimper could not absorb.
Fix: the neck insert on cavity bank 3 had a cooling channel partially blocked. Mold temperature on that zone was running 12 °C above setpoint. Once the channel was cleaned and per-zone control was added, Cpk came back to 1.62. The drawing changed too: we moved the neck OD tolerance to ±0.05 mm with a Cpk 1.67 requirement and a per-cavity report.
Case 2: the concentricity error that shed silicone into the port
Symptom: an HPLC method developed a pressure spike every 40 to 60 injections. The guard column was catching material. The material turned out to be silicone fragments.
Measurement: septa were being punctured off-center. We measured concentricity on the vial lot and found 0.09 mm TIR against a print that did not specify concentricity at all. The needle was entering at an angle and coring the vial closures rather than slicing them. Puncture count on the septum dropped from a nominal 5 to 2 before visible coring.
Fix: added concentricity ≤0.05 mm TIR to the drawing, switched the septum to a PTFE-faced silicone with a bonded slit, and re-qualified the closure against the vial lot. Pressure spikes stopped. The lesson: a dimensional tolerance you did not write is a tolerance you do not own.
A decision checklist for qualifying a tolerance claim
When a supplier sends you a capability report, run this. It takes an hour and saves a recall.
- Is the sample ≥125 per dimension? Fewer than that and the Cpk is not statistically meaningful.
- Does it span ≥3 lots and ≥3 cavities? Single-lot, single-cavity data hides the variation that bites you.
- Is the gauge R&R ≤10% of tolerance? If they used calipers on a ±0.05 mm band, reject the report.
- Is the Cpk ≥1.33 on general dimensions and ≥1.67 on neck OD and concentricity? Anything below is a process that will drift out of the window within a few lots.
- Is the mean centered? A Cpk of 1.33 with the mean at the nominal edge is a process waiting to fail.
- Is there a control plan? Per-zone mold temperature, cavity pressure, and a documented reaction plan when a dimension drifts.
- Is the measurement method named? Vision system, optical comparator, or CMM with a stated resolution. “Calipers” is not a method.
How to write a vial drawing that a molder can actually hold
The drawing is where capability is won or lost. A few rules I apply.
Specify the critical dimensions and only the critical dimensions at tight tolerance. A drawing with 14 tight callouts forces the molder to over-control everything and drives cost without improving function. Put the tight band on neck OD, concentricity, and internal floor. Leave height and body OD at commercial tolerance.
Use geometric dimensioning and tolerancing, not just linear dimensions. A concentricity callout with a datum on the body OD is enforceable. A note that says “neck must be centered” is not.
State the capability requirement, not just the tolerance. “Neck OD 11.00 ±0.05 mm, Cpk ≥1.67, per-cavity data required” is a different order than “11.00 ±0.05 mm.” The first gets you a validated process. The second gets you a number on a certificate.
Reference the closure interface. The vial and the closure are one dimensional system. If you are qualifying a new vial, qualify it against the closure you will actually use, and check the crimp or screw interface at the same time. The GPI 11 mm finish and the 13-425 screw thread both have defined dimensional envelopes; a vial that meets the envelope on paper and misses it in practice is a vial that will not seal. Suppliers with a documented interface spec, like the closure lines at vialclosures.com, will give you the mating dimensions to design against.
Ask for the mold layout. A 32-cavity tool has 32 slightly different parts. If the supplier cannot tell you which cavity made which vial, they cannot tell you whether cavity 19 is drifting. Per-cavity traceability is the difference between a dimensional tolerance and a dimensional capability.
What the instrument makers assume you already control
Autosampler manufacturers publish vial dimensional envelopes because they have to. Agilent’s chromatography consumables documentation and Thermo Fisher’s sample handling specs both assume a vial body within a defined envelope and a neck that is concentric within a few hundredths of a millimeter. They do not tell you which dimension will fail first, because they assume you have already validated it.
The instrument vendors are right to assume that, and most buyers are wrong to accept it. The envelope is a functional requirement, not a manufacturing tolerance. Your job is to translate the envelope into a drawing with capability numbers and a control plan. Restek’s technical resources are useful here for understanding how vial geometry interacts with injection volume and needle depth; the EPA’s SW-846 methods set the sample-handling context that makes a 0.05 mm concentricity error a data quality problem rather than a nuisance.
The recommendation
Write the drawing around four dimensions: neck OD at ±0.05 mm with Cpk ≥1.67, concentricity at ≤0.05 mm TIR, internal floor at ±0.05 mm, and base flatness at ≤0.10 mm. Require per-cavity capability data from at least 125 measurements across three lots, measured on a system with gauge R&R under 10% of the tolerance band. Leave body OD and height at commercial tolerance and spend the inspection budget on the dimensions that touch the instrument.
If a supplier cannot produce a Cpk for the neck OD and concentricity, they do not have a controlled process, they have a lucky lot. Buy the process, not the lot. The tolerance is only as real as the capability behind it, and the capability is only as real as the measurement you can audit.
References & further reading
Related resources from our lab consumables network:
- Oem Vial at oemsamplevials.com
- Vial Closures at vialclosures.com
- Analytical Testing at labanalyticaltesting.com
- Chromatography Vial at chromvials.com
External references and standards: