A customer sent me a certificate of analysis last March for a lot of 2 mL clear crimp vials that had failed his autosampler three times in a week. The sheet said CV < 1.0% on "critical dimensions." I asked which dimensions, over how many cavities, and whether the CV was calculated on the sample or on the mold. He did not know, and neither did the distributor. That datasheet was technically true and practically useless, which is the most common failure mode in this category. Reading a chromatography vial datasheet specifications page is not a reading exercise. It is an audit of what the supplier actually measured, on what sample size, against what tolerance, and whether the number you are looking at is a guarantee or a typical value dressed up to look like one.
I have been on both sides of this sheet. I have written them for a molder and I have rejected them as a buyer. The difference between a useful datasheet and a decorative one comes down to four questions: is the value tested or nominal, what is the sample size, what is the test method, and what happens when the value falls outside the limit. Most sheets answer none of these. This article walks through each line you will see, how it is actually measured, and how to spot the ones designed to survive a procurement review without telling you anything.
Tested versus nominal: the distinction that decides whether the sheet is worth the paper
A nominal value is a design target. A tested value is a measurement with a method, a sample size, and a pass/fail criterion attached. On most chromatography vial datasheets, roughly half the numbers are nominal and are presented as if they were tested. The 9-425 thread, the 32 x 11.6 mm body, the 6.0 mm opening: these are drawing dimensions. They describe the mold, not the parts in the box. A molder can hold a cavity to within ±0.1 mm on the neck OD and still ship a lot that fails your autosampler, because the dimension that matters (concentricity of the neck to the body, or the depth of the seat) is not on the drawing at all.
The rule I use: if the datasheet shows a dimension without a tolerance band and without a sampling statement, treat it as nominal regardless of how it is worded. A real tested value looks like this: Neck OD 9.42 ± 0.08 mm, n = 30 per cavity, measured by optical comparator at 20x, AQL 0.65. That sentence tells you the target, the tolerance, the sample size, the instrument, and the acceptance criterion. Anything less is marketing.
The three numbers that are almost always nominal
- Body OD and height. These come from the mold drawing. Unless the sheet states a gauge and a sample size, they are design intent. In practice a 32 mm vial will run 31.85 to 32.15 mm across a lot, and if your autosampler tray has a 0.3 mm clearance you will find out the hard way.
- Fill volume. The "1.5 mL" or "2.0 mL" on the label is the nominal capacity to the shoulder, not the usable volume. Usable volume depends on the insert, the septum thickness, and the needle depth. It is a design parameter, not a tested one.
- Glass weight. Occasionally listed as a quality proxy. It is a nominal value derived from the mold and the gob weight, and it varies with furnace drift across a shift. A 2 mL vial weighing 3.1 g nominal will run 2.95 to 3.25 g across a production day.
None of this means the vial is bad. It means the datasheet is describing the design, not the lot. If your application is sensitive to any of these, you need a supplier who will quote a tested value with a sampling plan, or you need to qualify the lot yourself. The chromatography vial range at chromvials.com is one place to start comparing which suppliers publish tested values versus which publish drawings.
How the tested lines are actually measured, and where the method hides
Every tested line on a good datasheet carries an implicit method. When the method is missing, the number is unverifiable. Here is how the common lines are actually generated, and what to ask when the method is absent.
Dimensional values (OD, ID, height, neck)
Measured by optical comparator, vision system, or go/no-go gauge. The method matters because a go/no-go gauge tells you pass or fail, not the distribution. A vision system gives you a mean and a standard deviation. If the sheet quotes a CV without telling you the instrument, you cannot tell whether it is a real process capability number or a single-cavity measurement dressed up. Ask for Cpk, not CV. A Cpk of 1.33 or better on the neck OD across all cavities is a meaningful statement. A CV of 0.8% on an unspecified sample is not.
Concentricity and neck-to-body alignment
This is the line that most often goes unmeasured. It is checked on a vision system by comparing the center of the neck bore to the center of the body OD. A good sheet will quote it as TIR (total indicator reading) ≤ 0.15 mm. A bad sheet omits it entirely. Concentricity is the single best predictor of autosampler gripper failures and needle strikes, and if it is not on the sheet, the supplier is not measuring it in production.
Wall thickness
Measured by ultrasonic gauge at four points around the circumference. Nominal 1.0 mm wall on a 2 mL vial will run 0.85 to 1.15 mm. Thin walls fail under crimping torque and crack at the neck. If the sheet quotes a single wall thickness with no range, it is nominal.
Chemical performance (extractables, hydrolytic resistance)
Hydrolytic resistance is tested per USP <660> and reported as a titration volume in mL of 0.01 M HCl per 10 g of glass. A Type I borosilicate vial should come in under 0.85 mL for the powdered glass test. If your sheet quotes "USP Type I" without the titration number, you have a material claim, not a test result. Extractables by LC-MS or GC-MS are almost never on a standard datasheet; they are quoted on request with a specific method (often a 24-hour soak at 40 °C in the intended solvent, analyzed against a calibration curve). Treat any "low extractables" claim without a method and a detection limit as nominal.
Seal integrity and closure performance
Tested by dye penetration (methylene blue under vacuum) or by pressure decay. A good sheet quotes zero dye ingress at 0.5 bar for 30 s across n = 50. A bad sheet says "leak tested." The difference is whether the test would catch a bad lot or just confirm the process exists.
The sample size problem: why a CV of 0.9% can still mean nothing
A CV is only as good as the sample it was calculated on. I have seen sheets where the CV was computed on five vials pulled from a single cavity at the start of a run. That number tells you the mold was stable for the first ten minutes. It tells you nothing about the other 15 cavities or the last four hours of the shift.
The sample size you want to see is stated explicitly, per cavity, at multiple points across the run. A defensible statement looks like: n = 30 per cavity, 16 cavities, sampled at start, middle, and end of run, CV calculated on the pooled sample. If the sheet quotes a CV without an n, ask. If the supplier cannot answer, the number is decorative.
There is a second trap: the CV may be calculated on the wrong dimension. A sheet can quote an impressive CV on the height while the neck OD, which is what the autosampler actually grips, is uncontrolled. Ask which dimension the CV refers to. If the answer is "overall," press for the specific dimension that matters to your instrument.
Typical values are not guarantees, and the wording tells you which is which
Read the verbs. "Meets," "conforms to," and "tested to" imply a pass/fail criterion. "Typical," "nominal," "approximately," and "designed for" are admissions that the value is not guaranteed on any individual lot. A sheet that says "typical hydrolytic resistance < 0.85 mL" is telling you the process usually lands there, not that every lot will.
This matters most on the lines that drive failure. If the sheet says "typical concentricity < 0.15 mm," you cannot reject a lot at 0.25 mm, because the sheet never promised 0.15. If it says "concentricity ≤ 0.15 mm TIR, tested per lot, n = 30," you can. The wording is the contract. Procurement managers who read only the numbers and not the verbs end up with a datasheet that cannot support a rejection.
A quick test for whether a sheet is hiding its method
Look for these tells. Any one of them is a yellow flag; two or more and I would not qualify the supplier without an audit.
- A CV or tolerance quoted with no sample size and no instrument.
- A material claim (USP Type I, low extractables) with no test method or detection limit.
- "Leak tested" or "quality inspected" with no acceptance criterion.
- Dimensions listed without tolerance bands.
- Concentricity, wall thickness, or closure torque omitted entirely.
- A certificate of analysis that repeats the datasheet verbatim rather than reporting lot-specific measurements.
The last one is the most damning. A CoA should carry lot numbers, test dates, measured values, and the initials of the analyst. If it looks like a photocopy of the datasheet with a lot number stamped on top, the supplier is not testing per lot. They are testing once and shipping many.
Comparison: what a tested line looks like versus a nominal line
| Spec line | Nominal presentation (avoid) | Tested presentation (accept) |
|---|---|---|
| Neck OD | 9.4 mm | 9.42 ± 0.08 mm, n = 30/cavity, vision system, Cpk ≥ 1.33 |
| Concentricity | Not listed | TIR ≤ 0.15 mm, tested per lot, n = 30 |
| Wall thickness | 1.0 mm | 0.95 to 1.10 mm, ultrasonic, 4 points, n = 20 |
| Hydrolytic resistance | USP Type I | ≤ 0.85 mL 0.01 M HCl/10 g, USP <660>, per lot |
| Seal integrity | Leak tested | Zero dye ingress, 0.5 bar, 30 s, n = 50 |
| Closure torque (crimp) | Recommended 1.5 N·m | 1.4 to 1.6 N·m, torque tester, n = 10 per lot |
The right-hand column is what you should be asking for. Not every supplier will provide all of it on a standard sheet, but the ones who can answer the questions are the ones worth qualifying. The HPLC vial suppliers who publish method-attached values tend to be the same ones who will send a lot-specific CoA on request.
Two field cases where the datasheet line was the whole story
Case 1: the concentricity line that was not there
A pharma QC lab running a autosampler started getting intermittent "vial not found" errors on a lot of 2 mL crimp vials. The datasheet was clean: dimensions nominal, CV < 1.0%, USP Type I, leak tested. We pulled 50 vials and measured neck-to-body TIR on a vision system. The mean was 0.21 mm, with a max of 0.34 mm. The autosampler gripper needs the neck centered within about 0.15 mm to seat reliably. The datasheet never quoted concentricity, so the lot passed every criterion the supplier published. The fix was a supplier change to one that quotes TIR ≤ 0.15 mm with a per-lot test. The lesson: the most important line on the sheet is often the one that is missing.
Case 2: the CV that was calculated on the wrong dimension
A contract lab qualified a supplier on a datasheet quoting CV 0.7% on height. Six months later they were seeing needle strikes on a GC autosampler. We measured the neck OD across 100 vials from three lots: range 9.28 to 9.61 mm, a spread of 0.33 mm. The height was indeed tight (CV 0.7%), but the neck OD, which is what the needle guide references, was running nearly four times the tolerance the instrument expects. The supplier had quoted the CV on the dimension they could hold, not the one that mattered. The fix was to rewrite the purchase specification to require a Cpk ≥ 1.33 on neck OD specifically, with the sample size and instrument named. If you are writing that spec, the HPLC septa and closure side of the same drawing needs the same treatment: septum thickness and durometer are tested values, not nominal ones, and they drive coring and seal performance.
What to put in your purchase specification so the datasheet becomes enforceable
A datasheet you cannot enforce is a brochure. The way to make it enforceable is to move the tested lines into your purchase specification, with the method, sample size, and acceptance criterion named. Here is the checklist I use when qualifying a new supplier or renegotiating with an existing one.
- Neck OD: tolerance band, Cpk ≥ 1.33, vision system, n = 30 per cavity.
- Concentricity: TIR ≤ 0.15 mm, tested per lot, n = 30.
- Wall thickness: 0.95 to 1.10 mm, ultrasonic, 4 points, n = 20.
- Hydrolytic resistance: ≤ 0.85 mL 0.01 M HCl/10 g per USP <660>, per lot.
- Seal integrity: zero dye ingress at 0.5 bar for 30 s, n = 50.
- Crimp closure torque: 1.4 to 1.6 N·m, torque tester, n = 10 per lot.
- Screw closure torque: 0.8 to 1.2 N·m for 9-425, tested to a defined seating force.
- CoA: lot-specific measured values, test dates, analyst initials, and the method reference for each line.
For reference standards, cite the ones your supplier should already be testing against: U.S. EPA SW-846 for method-specific sample handling requirements, and the USP chapters that reach the vial (the ones covering glass, closures, and extractables). If a supplier cannot tell you which USP chapter their hydrolytic resistance number comes from, they are not testing it. The technical resources at Restek and the consumables documentation from Aijiren and Thermo Fisher are useful cross-references for what an instrument actually needs from the vial, which is the standard your purchase spec should be written against. The broader chromatography products catalogs often show which suppliers publish method-attached specs and which do not.
The decision rule
Read the datasheet as a contract, not a description. Every line is either tested (with a method, a sample size, and an acceptance criterion) or nominal (a design target). If a line that drives your failure mode is nominal, or if the method is missing, the number cannot support a rejection and the sheet is decorative. Qualify only suppliers who will put the tested lines into a purchase specification with named methods and per-lot CoAs, and treat every "typical" as a number you have agreed not to enforce.
References & further reading
Related resources from our lab consumables network:
- Chromatography Vial at chromvials.com
- Hplc Vial at hplcvialfactory.com
- Chromatography Products at chromatographyforsale.com
- Hplc Septa at hplcsepta.com
External references and standards:
