The difference between a guaranteed limit and a typical value is the whole--Aijiren HPLC Vial Factory

A customer sent me a datasheet last March with “CV < 3%” printed on the dimensional table, and when I asked for the sample size and the gauge they used, the answer was 10 vials pulled from a single mold cavity, measured with a hand caliper at 20 °C. That number is not a specification. It is a snapshot of one cavity on one afternoon, and it was being used to qualify a lot of 240,000 vials across three tools. The run that failed had 0.9 mm of needle penetration scatter on a 250 µL insert, and the root cause traced back to exactly the kind of number that looks reassuring on paper and means nothing at the bench. Reading an HPLC vial datasheet specifications sheet is a skill, and most buyers I deal with have never been taught it.

This is not a piece about which vial to buy. It is about the sheet in front of you: which lines are guaranteed by a test method, which are nominal values copied from a drawing, which are marketing, and how to tell the difference before the purchase order goes out. If you already know your GPI thread and your plate height, you are the right reader. If you are still choosing a format, the dimensional selection material on our HPLC vial page covers that separately.

The difference between a guaranteed limit and a typical value is the whole game

Every number on a vial datasheet falls into one of three buckets, and the sheet rarely labels them.

Guaranteed (specification): a limit the manufacturer tests against, with a stated method, sample plan, and acceptance criterion. If the lot fails, it does not ship. These are the only numbers you can put in a method validation document.
Typical (nominal): a value from the design drawing or a historical average. It describes what the part is supposed to be, not what any given lot measured. Useful for fit calculations, useless for qualification.
Marketing: “low extractables,” “high purity,” “MS certified.” No method, no limit, no sample plan. Treat as zero information.

The trap is that all three appear in the same font on the same page. A sheet that says “Residual volume: < 5 µL” could mean anything from a gravimetric test on 30 vials per lot to a number an engineer typed in 2014. You cannot tell by looking. You have to ask, and the sheet should make the answer findable.

There is a fourth category that catches even experienced buyers: the process capability claim. A sheet that says “Cpk > 1.33 on neck OD” is making a statistical statement about a distribution, not a statement about a single part. Capability is only meaningful if you know the specification window it was computed against, the number of subgroups, and the period covered. A Cpk of 1.33 against a ±0.15 mm window is a very different claim from a Cpk of 1.33 against ±0.30 mm. If the sheet gives you the Cpk without the window and the subgroup count, it is a word, not a number.

Which lines on the sheet are actually tested, and how

Here is what a real test looks like for the values that matter most. If the datasheet does not describe something close to this, the number is nominal.

Critical dimensions (neck OD, bore, height, thread)

These are measured with a go/no-go gauge or an optical comparator, not a caliper. A production sheet should state the gauge standard and the sampling frequency. Neck OD on a 9 mm crimp vial is typically held to ±0.15 mm; the bore to ±0.1 mm. If you see “±0.05 mm” on a molded glass part with no method, be suspicious. That is a machining tolerance, not a glass-forming one. For context on how the forming station holds those numbers, the process-control discussion on our site goes deeper, but for datasheet purposes the question is simply: gauge or caliper, and how many per shift.

Height deserves its own line because it is the dimension most often given as a bare nominal. A 12 × 32 mm vial has a design height of 32.0 mm, but the shipping tolerance that governs autosampler gripper function is usually ±0.4 mm, and the crimp seat height is tighter, around ±0.15 mm. If the sheet lists “32 mm” with no tolerance, you are reading the drawing, not the process. Compare that to a sheet that lists “Height 32.0 ± 0.4 mm, optical comparator, 1 per 200.” Only the second one tells you what will actually arrive.

Residual volume

This is the one buyers fixate on and the one most often misreported. The honest test is gravimetric: fill a defined number of vials with water, dispense to a fixed needle depth, weigh the remainder, convert to volume at a stated temperature. A defensible sheet gives you the sample size (I want n ≥ 30 per lot), the needle depth used, and the acceptance limit. A sheet that says “< 5 µL residual” with no method is quoting the insert geometry, not a measurement. The actual residual depends on needle depth, needle bevel, and how the autosampler parks the needle. Change any of those and the number moves by 2 to 4 µL on a 300 µL insert.

I also want the temperature stated. Water density at 20 °C is 0.9982 g/mL; at 25 °C it is 0.9970 g/mL. That 0.12% difference is small, but on a 5 µL residual claim it is the difference between passing and failing at the edge, and it tells you whether the lab that produced the number was paying attention. A sheet that converts gravimetric data to volume without naming a temperature is a sheet where someone skipped a step.

Extractables and leachables

If the sheet claims a low-extractable glass or a washed vial, demand the test method. Real ones reference a solvent, a temperature, a time, and an analytical finish. For example: “Extractables by ICP-MS after 24 h at 60 °C in 5% HNO₃, reported as µg/vial for Na, K, Ca, B, Al.” Or for organics, a GC-MS screen against a known standard. A sheet that says “low extractables” with no solvent and no temperature is telling you nothing. The regulatory frame here is USP <1663> and <1664>, and the elemental impurity limits in USP <232>. If your supplier cannot name the chapter they test against, the claim is not a specification. Our chromatography products pages list which suppliers publish method-referenced extractable data and which do not.

Watch for the difference between a hydrolytic test (Type I borosilicate, autoclaved in water, titrated for alkali release) and an acid-extraction test. They answer different questions. Hydrolytic resistance at 121 °C for 60 minutes tells you about the glass surface stability. It does not tell you what leaches into a 5% nitric acid digest. If your method is trace metals in acid, you need the acid extraction, not the hydrolytic number, and a sheet that offers only the hydrolytic result is answering a question you did not ask.

Seal integrity and torque

Crimp vials are tested by vacuum decay or dye ingress on a sample per lot. Screw vials are tested by capping to a specified torque and checking for leakage under pressure. A real sheet gives you the torque window: for an 8-425 screw thread, 0.5 to 1.0 N·m is typical; for a 9 mm screw, 0.8 to 1.4 N·m. If the sheet gives a single torque number with no window, it is a suggestion, not a spec. The closures themselves are covered at vial closures, but the datasheet line you are reading should tell you the test method and the sample size.

Dye ingress is the test I trust most for crimp vials. The protocol: cap the vial, immerse in a dye solution (usually 0.1% methylene blue), pull a vacuum of about −80 kPa for 2 minutes, release, and inspect for dye in the headspace. A sheet that names this test and gives you the sample size (I want n ≥ 20 per lot) is a sheet from a supplier who understands that a crimp is a mechanical seal, not a cosmetic one.

Septa: the line most often fudged

A septum datasheet should state durometer (Shore A), thickness with tolerance, and a puncture count with the needle gauge and the test condition. “100 punctures” means nothing without “with a 22-gauge needle at 1 µL injection volume, leak tested at 1 bar after every 10 punctures.” I have seen sheets claim 200 punctures where the test was a single needle at 26 gauge with no leak check. That is a different product. Our HPLC septa reference covers the durometer and coating trade-offs; the datasheet question is whether the puncture count came with a needle spec and a leak criterion.

Thickness is the other under-specified line. A silicone/PTFE septum is usually 3.0 ± 0.2 mm for a 9 mm crimp cap. If the sheet gives “3 mm” with no tolerance, the crimp height calculation on your side is off by up to 0.2 mm before you even start. Durometer matters too: 45 to 55 Shore A for a standard silicone septum, softer for high-puncture-count applications, harder for high-pressure. A sheet that omits durometer is telling you the supplier treats the septum as a commodity.

The tell: a sheet that hides its test method

You can spot a weak datasheet in about 90 seconds. Here is the checklist I run.

Line on sheet	Strong sheet says	Weak sheet says
Residual volume	Gravimetric, n=30/lot, needle depth 8 mm, limit < 5 µL	“< 5 µL”
Neck OD	Go/no-go gauge, ±0.15 mm, 1 per 500	“9.0 mm”
Height	Optical comparator, 32.0 ± 0.4 mm, 1 per 200	“32 mm”
Extractables	ICP-MS, 24 h/60 °C/5% HNO₃, µg/vial	“Low extractables”
Puncture count	22 G, 1 bar leak test, 100 punctures	“High puncture resistance”
Torque	0.5–1.0 N·m, dye ingress per lot	“Tight seal”
CV on dimensions	n=30, 3 cavities, optical comparator	“CV < 3%”
Capability	Cpk 1.67 vs ±0.15 mm, 25 subgroups	“Six sigma process”

If more than two rows land in the right-hand column, the sheet is a brochure. Ask for the control plan and the last three lot certificates. A supplier who cannot produce them within a day is not testing.

One more tell: the units. A sheet that mixes millimeters and inches on the same table without labeling, or that gives a dimension to three decimal places on a molded part, was assembled by someone copying from a CAD file. Real molded glass tolerances land at two decimal places in millimeters. Three decimals is a machining spec pasted in.

Case one: the CV that was not a CV

A pharma customer had a method with a 1.8% RSD acceptance criterion on peak area for a 5 µL injection. Runs started drifting to 2.4% RSD on one instrument and not another. The vial datasheet claimed CV < 3% on neck OD. We pulled 30 vials from the failing lot and measured neck OD with a comparator: the actual CV was 1.1%, well within the claim. The problem was not the neck. It was the crimp height, which the sheet listed as a nominal 11.6 mm with no tolerance and no test. The actual spread was 0.35 mm across the lot, and the crimper on that instrument was set to a fixed anvil gap that could not absorb it. The seal was inconsistent, the headspace varied, and the injection volume scattered.

The fix was twofold: we asked the supplier to add crimp height as a gauged dimension with a ±0.15 mm limit and a per-lot certificate, and the customer re-set the crimper to a torque-and-height window rather than a fixed gap. RSD dropped to 1.2%. The lesson is that the number on the sheet was true and irrelevant, because the dimension that actually governed the failure was listed as nominal.

What I would add for anyone reading a similar sheet: the dimension that governs your failure mode is not always the one the supplier thinks is important. The supplier tests neck OD because that is the fit dimension for the autosampler. The customer’s failure was driven by crimp height, which the supplier considered a cap dimension, not a vial dimension. When you spec a vial, you own the failure-mode analysis, and you have to tell the supplier which dimensions to gauge.

Case two: the extractables claim with no solvent

An environmental lab running EPA SW-846 methods (the SW-846 test methods are the reference here) had blank contamination on sodium at 40 to 60 ppb in every run. The vial datasheet said “low extractables glass.” We could not get a method out of the supplier. So we ran our own: 20 vials from the lot, filled with 5% HNO₃, held 24 h at 60 °C, analyzed by ICP-MS. Sodium came back at 0.8 µg/vial, which at a 10 mL fill is 80 ppb. The claim was not false; it was just undefined. The supplier’s “low” referred to a different test at a different temperature.

We switched to a vial with a published method (same ICP-MS conditions, limit < 0.1 µg/vial for Na) and the blanks dropped below 5 ppb. The datasheet line that mattered was not the word “low.” It was the solvent, the temperature, the time, and the analytical finish. Restek’s technical library (Restek technical resources) is a good place to see how method-referenced extractable data is supposed to be presented; compare that format against whatever your supplier sends.

The cost of that lesson was about six weeks of method rework and a re-validation. The cost of the fix was a few cents per vial. That ratio is why I now treat the extractables line as the highest-priority line on any datasheet for a trace-metals method, ahead of residual volume and ahead of dimensions.

Nominal values are not lies, but they are not commitments

I want to be fair to suppliers here. A drawing nominal is a legitimate engineering value. The 11.6 mm crimp height is the design intent. The 32 × 11.6 mm outer envelope is the fit envelope. The GPI 8-425 and GPI 9 mm thread designations are standard references, not measurements. You need these numbers to know the vial will fit your autosampler and your plate. What you cannot do is put a nominal into a validation document and call it a specification.

The practical rule: match the number to the document. If you are writing a method validation or a supplier qualification, only guaranteed-and-tested values go in. If you are doing a fit check, nominal is fine. If you are comparing two suppliers on price, ignore everything in the marketing bucket and compare the control plans. The instrument vendors publish their own consumable specs in a method-referenced format (Agilent’s chromatography consumables documentation and Thermo Fisher’s sample handling references are useful benchmarks), and comparing your supplier’s sheet against those formats tells you quickly whether you are looking at a spec or a sketch.

A decision checklist you can run in ten minutes

Print this and take it to the next supplier review.

Is the test method named? Gauge type, analytical finish, solvent, temperature, time. No method, no credit.
Is the sample size stated? Per lot, per shift, per cavity. A number with no n is a rumor.
Is there a limit or just a value? “< 5 µL” is a limit. “5 µL” is a nominal.
Is the dimension that governs your failure mode tested? If your method is sensitive to crimp height, demand a gauged crimp height with a tolerance.
Can they produce three lot certificates on request? If not, the sheet is aspirational.
Does the extractables claim name a USP chapter? <1663> and <1664> for containers, <232> for elemental limits. If not, it is a word, not a test.
Is the torque window given? A single number is a suggestion. A window with a leak test is a spec.
Does the puncture count come with a needle gauge and a leak criterion? Without both, the number is decorative.
Is the capability claim tied to a window and a subgroup count? Cpk without a window is arithmetic theater.
Are the units consistent and realistic? Two decimal places in millimeters, not three.

What to ask for, and what to accept

When I qualify a vial now, I ask for four documents: the control plan, the last three lot certificates, the extractables report with method, and the dimensional gauge list. I will accept a nominal value on a fit dimension and a guaranteed limit on anything that touches the method. I will not accept a CV without a sample size, a puncture count without a needle spec, or an extractables claim without a solvent. Those three omissions account for most of the datasheet failures I have seen in fifteen years, and they are the easiest to fix at the purchase order stage, before the lot is molded.

I also keep a one-page comparison of my top three suppliers, with the same eight lines from the checklist scored strong or weak. It takes an hour to build and it has saved me from two bad qualifications in the last four years. The sheet is a sales document until you force it to become a test record, and the forcing function is the question you ask before the PO, not the complaint you file after the run fails.

The decision rule is simple: a number you cannot trace to a test method and a sample size is a design intent, not a specification, and it does not belong in a validation file. Read the sheet for what it guarantees, not for what it hopes. If the supplier cannot tell you which is which, that is your answer about the supplier.

References & further reading

Related resources from our lab consumables network:

External references and standards: