A 96-vial sequence on a GC-FID ran 14 hours overnight, and by morning the last 24 injections were showing internal-standard areas 6.2% below the first calibration injection. The column was fine. The inlet liner was new. The autosampler syringe had 400 injections on it, well inside its 1,000-injection service life. The problem was a gc vial seal failure that nobody had measured: the crimp caps had been applied with a hand tool that was delivering roughly 1.1 N·m instead of the 1.7 N·m the cap spec called for, and the septa had taken a permanent set. Every punctured vial was losing 0.8 to 1.4 mg of solvent per hour through the needle track. Over 14 hours, that is enough evaporation loss to shift a volatile analyte’s response by several percent.

This article is about that failure mode specifically. Not coring, not debris in the needle, not contamination peaks. It is about where the leak path lives, how a septum behaves after the needle withdraws, why compression set matters more than people think, and how to actually measure evaporation loss on your own bench instead of trusting a datasheet.

Most people picture a leak as a hole in the septum. It rarely is, or at least not only. On a 9 mm crimp vial, there are four distinct leak paths, and they fail at different rates:

• Through the septum puncture. The needle track itself. A 23-gauge needle leaves a hole roughly 0.34 mm across, and whether it closes depends on the septum elastomer and how long the needle dwelled.
• Between septum and cap. The septum sits in the cap well. If the cap is under-crimped, the septum is not compressed against the glass lip and gas passes around the outside of the septum rather than through it.
• Between cap skirt and vial neck. The crimp deforms the aluminium skirt around the neck bead. Under-crimped, the cap lifts. Over-crimped, the skirt deforms past the bead and the cap never seats again.
• Through the vial body or neck finish. Rare, but real. A neck finish with a chipped lip or a body with a hairline crack from a hot autosampler tray will leak no matter how good the cap is.

The first two are the ones that show up as slow, run-long drift. The third shows up as a vial that is empty at the end of the sequence. The fourth shows up as one bad vial out of 96, which is why single-vial outliers are worth investigating separately from trends.

If you are also fighting needle damage and debris, that is a different article. I will only say here that a cored septum and a leaking septum are not the same failure, and the fixes are different.

Re-seal behaviour after puncture: what the needle track does when you pull out

A septum is not a self-healing membrane. It is an elastomer that has been pierced, and its ability to close the track depends on how much elastic energy is stored in the material around the puncture.

When a needle passes through, it pushes material aside. Silicone septa with a PTFE face on the sample side recover well because the silicone body has high elasticity and the PTFE film acts as a barrier. A 1 mm silicone septum with a 0.1 mm PTFE facing, which is standard on many 9 mm caps, will typically re-seal a 23-gauge track to within 99.5% of its original seal within 30 seconds of needle withdrawal, provided the septum was correctly compressed by the cap. That 0.5% is not zero. Over a 20-hour run, 0.5% of a 1.5 mL headspace exchange per hour is still measurable loss.

Butyl and buna-N septa recover more slowly and take a permanent set sooner. They are fine for short runs and for aqueous samples, but if you are running volatiles on a 24-hour sequence, silicone is not optional.

The variable that catches people is dwell time. If the autosampler needle sits in the vial for 3 seconds, the elastomer relaxes around the needle and the track is wider when the needle withdraws. If it sits for 0.3 seconds, the recovery is better. This is one of those things you can only see if you measure it, which is why the gravimetric test below matters.

A second variable is multiple punctures. A septum rated for 3 punctures is not rated for 30. After the third puncture, the tracks start to merge and the septum stops recovering. I have seen labs run 5 injections from a single vial and wonder why the fifth is 4% low. The septum was never the problem on injection one. It was the problem on injection four.

Compression set is the number your cap torque is actually controlling

Compression set is the permanent deformation a septum retains after it has been compressed and released. It is expressed as a percentage: a septum with 15% compression set, after being compressed to 75% of its original thickness, will recover to 85% of original thickness and stay there. The 15% it lost is the gap that becomes your leak path.

For a 9 mm crimp cap, the septum is compressed from about 3.0 mm to about 1.9 mm when the cap is properly crimped, which is roughly 37% compression. That is the design point. If you crimp to 1.5 mm, you are at 50% compression, which is past the elastomer’s elastic limit and drives compression set up sharply. If you crimp to 2.3 mm, you are at 23% compression, which is not enough to seat the septum against the glass lip.

The practical consequence: cap torque and crimp height are the two variables that control compression set, and both are easy to get wrong with a hand tool. A proper crimper for 9 mm caps should be set to a crimp height of 1.85 to 1.95 mm, measured with a crimp gauge, and should be checked at the start of every shift. The cap spec for most 9 mm aluminium crimp caps calls for a seal that holds at 1.5 bar pressure differential; a hand crimper that is 0.2 mm out of adjustment will fail that test on roughly 1 vial in 8.

Temperature matters too. A septum that sits in a hot autosampler tray at 40°C for 12 hours will take more compression set than the same septum at 20°C. If your method uses a heated tray, either use a septum rated for that temperature (silicone/PTFE is good to about 200°C; butyl is good to about 100°C but takes set faster) or accept that your re-seal performance is worse than the datasheet claims.

Measuring evaporation loss gravimetrically, on your own bench, in an afternoon

You do not need a special instrument to measure this. You need a balance with 0.1 mg readability, a set of vials, and a way to hold them at a constant temperature. Here is the method I use.

The gravimetric loss test, step by step

1. Take 12 vials from the lot you are qualifying. Fill each with 1.0 mL of the solvent your method uses (methanol and dichloromethane are the ones that show loss fastest; water will not show much).
2. Crimp 6 with your production crimper and 6 with a calibrated torque wrench set to the cap spec (typically 1.7 N·m for a 9 mm aluminium cap, but check the cap datasheet).
3. Weigh each vial to 0.1 mg and record the tare. Do this within 60 seconds of crimping.
4. Puncture each vial once with the autosampler needle your method uses, then withdraw. This is the step people skip, and it is the whole point.
5. Place the vials in a tray at your method’s tray temperature (typically 25°C, but use whatever your autosampler runs at) and leave them for the duration of your longest run. If your run is 14 hours, leave them 14 hours.
6. Weigh again at 1, 4, 8, and 14 hours. Record the mass loss at each point.

A properly sealed 9 mm vial with a silicone/PTFE septum, punctured once, should lose less than 0.3 mg per hour at 25°C with methanol. If you are seeing 1.0 mg/hour or more, you have a seal problem. If you are seeing 3.0 mg/hour, you have a cap that is not crimped at all, or a septum that is the wrong material.

Run the test with an unpunctured control set as well. The difference between punctured and unpunctured loss is the re-seal penalty, and that is the number that tells you whether the septum is doing its job after the needle withdraws.

If you want a reference method for the general approach to volatile loss testing, the EPA SW-846 series has several procedures that rely on gravimetric or volatile-loss checks for sample containers, and the logic is the same. You can find them at U.S. EPA SW-846 test methods for waste and materials.

A decision table for seal integrity by vial and cap configuration

Here is how I would choose a configuration for a given run length and solvent class. The numbers are from our own gravimetric testing on 9 mm crimp vials with 1.0 mL methanol fills at 25°C.

Two things to note from that table. First, the difference between standard and high-recovery silicone is small per hour but adds up over a 24-hour run. Second, the under-crimped and over-crimped rows are the ones that show up in real labs, and both are worse than using the wrong septum material.

If you are running an HPLC autosampler alongside your GC, the same logic applies but the numbers are different because the vial is open to the sampler environment rather than sealed with a septum. The HPLC autosampler vial selection criteria are worth a separate read.

Field case 1: the 6.2% drift that was a crimper, not a column

Symptom. A contract lab running EPA 8260 volatiles on a 96-vial sequence saw internal-standard areas drift down 6.2% from the first to the last injection. The drift was smooth and monotonic, which pointed at the instrument, not at individual vials.

Measurement. We ran the gravimetric test above on 12 vials from the same lot, crimped with the lab’s production crimper. Loss was 1.1 to 1.4 mg/hour per vial, well above the 0.3 mg/hour threshold. A crimp height gauge on the production crimper read 2.15 mm, against a spec of 1.85 to 1.95 mm. The crimper had been adjusted six months earlier and never re-checked.

Fix. Re-adjusted the crimper to 1.90 mm, re-ran the gravimetric test (0.25 mg/hour), and re-ran the sequence. Drift dropped to 0.4%, which is within the method’s own acceptance criteria. The column was never the problem. The instrument was never the problem. The crimper was the problem, and nobody had a number for it.

Field case 2: the re-seal failure that only showed up on the fourth injection

Symptom. A pharmaceutical QC lab was running 4 injections per vial to save vial cost. Injections 1 through 3 were fine. Injection 4 was consistently 3 to 5% low on the early-eluting peaks, and the effect was worse on the last vials in the tray.

Measurement. We took 10 vials, filled them with the method’s diluent, and punctured each one 4 times with the autosampler needle, weighing between punctures. Loss after the first puncture was 0.3 mg/hour. After the fourth, it was 1.8 mg/hour. The septum was rated for 3 punctures; the lab was running 4. The fourth puncture was opening a track that the first three had already weakened.

Fix. Two options: switch to a septum rated for 5 punctures (1.3 mm silicone/PTFE), or cap the injections per vial at 3. The lab chose the septum change because it kept the 4-injection workflow and the cost difference was about 8% on the cap. Loss dropped to 0.4 mg/hour after the fourth puncture, and the injection-4 bias disappeared.

This is the case that taught me to always ask “how many punctures per vial” before recommending a septum. The datasheet puncture count is a real number, and it is a hard limit, not a suggestion.

A troubleshooting checklist you can run in 20 minutes

When you suspect gc vial seal failure is behind a drift or a loss, work through this in order. Stop at the first thing that fails.

• Crimp height. Measure 5 caps from the current run with a crimp gauge. Spec is 1.85 to 1.95 mm for 9 mm aluminium. If any are outside, re-adjust the crimper before doing anything else.
• Septum material. Confirm the septum is silicone/PTFE, not butyl or buna-N, if you are running volatiles. Check the temperature rating against the tray temperature.
• Puncture count. Count the injections per vial in the method. If it exceeds the septum’s rated puncture count, that is your answer.
• Gravimetric loss. Run the 12-vial test above with your production crimper. If loss exceeds 0.3 mg/hour at 25°C with methanol, you have a seal problem.
• Vial neck inspection. Look at 20 vials under a loupe. A chipped or uneven neck finish will leak regardless of cap torque. Reject the lot if you see more than 1 in 20.
• Torque check. If you use a torque wrench rather than a crimper, verify it reads 1.7 N·m on a calibration block. Hand tools drift.

If all six pass and you still see loss, the problem is probably not the vial. It is the autosampler tray environment (temperature, draft, or a loose vial in the tray) or the inlet. The Agilent chromatography consumables and columns documentation has good guidance on inlet-side losses that can mimic vial leaks.

What to actually buy, and what to write into the method

For any GC method running volatiles with a sequence longer than 8 hours, I would specify the following in the method document, not just in the purchasing system:

• Vial: 9 mm crimp, clear glass, 2.0 mL, 32 x 11.6 mm, with a neck finish that meets the GPI 9 mm dimensional standard. If you are sourcing, the gc vial spec sheet should state the neck dimensions and the glass expansion coefficient.
• Cap: 9 mm aluminium crimp, magnetic or standard, with a 1.0 mm silicone / 0.1 mm PTFE septum for runs up to 12 hours, or 1.3 mm silicone / 0.1 mm PTFE for longer runs.
• Crimp height: 1.85 to 1.95 mm, verified at the start of each shift with a gauge.
• Punctures per vial: no more than 3 for a standard septum, no more than 5 for a high-recovery septum.
• Gravimetric acceptance: less than 0.3 mg/hour loss per vial at 25°C with methanol, tested on 12 vials from each lot at receipt.

Write the gravimetric test into your receiving procedure. It takes 20 minutes and it catches the crimper drift, the wrong septum lot, and the neck finish problem before they reach a sequence. The cost of a 12-vial test is trivial next to the cost of re-running a 96-vial sequence.

If you are sourcing vials and caps from different suppliers, make sure the cap spec and the vial neck spec are both stated in GPI dimensions. A cap that is 0.1 mm out on the skirt diameter will crimp differently on a vial that is 0.1 mm out on the neck bead, and the combination can fail even when each part passes its own inspection. The sample vial sourcing guides cover the dimensional checks in more detail.

For labs that also handle aqueous or high-boiling samples, the same seal integrity logic applies to the reagent bottle and its closure, though the numbers are different because the bottle is not punctured. The principle is the same: compression set plus a temperature rating plus a measured loss rate.

One more reference worth bookmarking: the Thermo Fisher chromatography and sample handling pages have good application notes on volatile loss from autosampler vials, and Restek technical resources for chromatography has a useful guide on septum selection by analyte class. Neither replaces your own gravimetric test, but both are good starting points for choosing the material.

The decision rule

If a GC sequence shows smooth, monotonic drift in internal-standard response, and the drift correlates with time-in-tray rather than with injection number, treat it as a vial seal problem until a gravimetric test says otherwise. Measure crimp height first, then septum material, then puncture count, then loss rate. Fix the first one that is out of spec, re-run the gravimetric test, and only then look at the instrument. The vial is a sealed vessel with a measured leak rate, and the leak rate is the number that decides whether your sequence holds its calibration from injection one to injection ninety-six.

References & further reading

Related resources from our lab consumables network:

• Gc Vial at gcvialfactory.com
• Reagent Bottle at reagentbottleforsale.com
• Hplc Autosampler Vial at hplcautosamplervials.com
• Sample Vial at samplevialprice.com

External references and standards:

• Agilent chromatography consumables and columns
• Thermo Fisher chromatography and sample handling
• U.S. EPA SW-846 test methods for waste and materials
• Restek technical resources for chromatography