We pulled 96 vials off an autosampler tray after a 26-hour run and found 11 of them had lost between 2.4 and 6.8 mg. Same lot, same method, same septa. The injection volumes were fine. The peak areas were not: the tail of the run drifted 4.7% RSD on the internal standard, and the lab spent two days blaming the column. The actual cause was an oem vial seal failure on a subset of closures, and it took a gravimetric check to prove it. This article is about how that happens, how to measure it, and how to stop paying for it.
Evaporation loss from a sealed 2 mL vial is not a mystery. It is a mass transfer problem with a small number of paths, and each path has a number attached to it. Once you measure the loss instead of arguing about it, the fix usually costs less than the investigation.
When people say “the seal failed,” they usually mean the needle hole did not close. That is one path. There are at least three others, and in my experience the septum face is the least common culprit on a properly crimped vial.
Path 1: the crimp-to-glass interface
This is the gap between the inside of the aluminium skirt and the neck of the vial. On a 11 mm crimp neck, the seal depends on the aluminium being deformed into the neck bead so the septum is compressed against the land. If the crimp is too loose, the septum is not compressed enough and vapour escapes around the outside of the septum, not through it. If it is too tight, you deform the neck or crush the septum and create a different failure. The window is narrow. I have measured good seals at 0.9 to 1.2 N·m on a manual crimper with a torque-limiting handle, and I have watched a lab run at 1.6 N·m because “tighter felt safer” and lose 3% of a volatile analyte in 12 hours.
Path 2: the septum-to-land contact
The septum must sit flat on the glass land. A land that is chipped, out of flatness, or has a molding witness line will leak regardless of crimp force. The dimensional control window on the neck is tight for a reason. If the land is not flat within a few hundredths of a millimetre, you get a micro-channel and the closure cannot compensate.
Path 3: the septum body itself
Silicone is permeable. Not leaky in the sense of a hole, but permeable in the sense that small molecules diffuse through the polymer matrix. A 1 mm silicone septum on a vial holding dichloromethane will lose measurable mass over 24 hours even with a perfect crimp. This is a material property, not a defect. If your analyte is a small, volatile, non-polar molecule, the septum is part of your evaporation budget whether you like it or not.
Path 4: the puncture channel
This is the one everybody talks about, and it is the one I want to spend the most time on because the re-seal behaviour is where OEM closures differ most from each other. More on that below.
If you are sourcing closures, the vial closures reference set is worth reading alongside this, because the septum durometer and the aluminium temper both feed into every path above.
Re-seal after puncture is a durometer and geometry problem, not a brand problem
A septum does not “heal.” It closes because the elastomer is under compression and the material flows back into the puncture channel. How well it does that depends on three things: the durometer of the silicone, the thickness of the septum, and the geometry of the needle that made the hole.
Standard 11 mm silicone septa run around 45 to 55 Shore A. A softer septum (40 Shore A) closes faster around a small needle but deforms more under crimp and is more prone to extrusion. A harder septum (60 Shore A) holds compression better but leaves a more open channel after a 23-gauge needle. For a 22-gauge needle (0.72 mm OD), a 1 mm thick 50 Shore A silicone septum will typically re-seal to a leak rate below the detection limit of a simple gravimetric test within about 30 seconds. For a 16-gauge needle (1.65 mm OD), the same septum may never fully close, and you will see loss from the first hour.
This is where the puncture count matters. A septum rated for 10 punctures at 22 gauge is not rated for 10 punctures at 16 gauge. The channel area scales with the square of the needle diameter, roughly a 5x difference between 22 and 16 gauge. If your method uses a large-bore needle for viscous samples or a wide-bore autosampler needle for a specific instrument, treat the puncture rating as halved or worse.
I have also seen pre-slit septa sold as a fix for this. They are not. A pre-slit septum is designed for a specific needle geometry and a specific injection port, and on a sample vial it is a controlled leak. Use them where they belong (inlet liners, headspace ports) and not on the sample side unless you have measured the loss and accepted it.
Compression set is why a closure that passed at t=0 fails at t=20 hours
Compression set is the permanent deformation an elastomer takes when held under load. A septum that is compressed at 20% for 24 hours will not return to its original thickness when the load is released. That matters because the crimp is a fixed mechanical constraint: the aluminium does not relax, so as the septum takes a set, the contact stress at the glass land drops.
The practical number: a good silicone septum at 20% compression and 25 °C will show a compression set of 10 to 20% after 24 hours. At 40 °C, that climbs. At 60 °C, a standard silicone septum can take a set of 30% or more, and the contact stress falls below the threshold needed to hold a volatile sample. This is why headspace vials at 80 °C lose more than the same vials at ambient, and it is why a method transferred from a 25 °C autosampler to a 40 °C one can show a sudden drift with no change in the vial lot.
The fix is not a different brand. It is a different material or a different compression target. PTFE-faced silicone septa reduce permeability but the PTFE face does not take compression set the same way, so the silicone underneath still does. If you are running hot, look at the septum specification for compression set at your operating temperature, not at 23 °C. If the datasheet only gives a room-temperature number, ask for the hot number or test it yourself.
How to measure evaporation loss gravimetrically, and what the numbers should be
You do not need a mass spectrometer to find a seal failure. You need a balance that reads to 0.1 mg and a controlled environment. Here is the protocol I use.
Equipment and setup
- Analytical balance, 0.1 mg readability, on a marble slab or a balance table, draught shield closed.
- Same room, same temperature, same humidity for tare and all weighings. Log the conditions. A 1 °C swing in a 2 mL vial headspace changes the mass reading by more than you think.
- Fill each vial with 1.0 mL of the actual sample matrix, not water. Water evaporates at a different rate than your analyte and will give you a false sense of security.
- Crimp with a torque-limiting crimper set to the mid-point of your validated window. Record the setting.
- Weigh at t=0, then at 1, 4, 8, 24 hours. Do not open the vials between weighings.
What the results should look like
On a properly sealed 2 mL crimp vial with a 1 mm silicone septum, holding a 50:50 water:methanol mix at 22 °C, I expect total mass loss of less than 1.5 mg over 24 hours. That is roughly 0.15% of the fill mass. If you see 3 to 5 mg, you have a seal problem or a very volatile matrix. If you see more than 10 mg, something is wrong with the crimp, the septum, or the vial neck.
The shape of the curve matters as much as the endpoint. A well-sealed vial loses mass quickly in the first hour (the headspace saturates) and then flattens. A leaking vial keeps losing at a near-constant rate. Plot the four points. A straight line is a leak. A curve that flattens is normal.
| Observation | Likely cause | Next check |
|---|---|---|
| Loss < 1.5 mg / 24 h, curve flattens | Normal | None |
| Loss 1.5 to 4 mg, curve flattens after 4 h | Marginal crimp or permeable septum | Re-crimp a subset at +0.1 N·m and repeat |
| Loss > 4 mg, near-linear | Crimp too loose or land defect | Inspect neck under 10x, check crimper calibration |
| Loss jumps after puncture | Septum not re-sealing (needle too large, durometer too high) | Switch to 22-gauge or softer septum, repeat |
| Loss only at elevated temperature | Compression set | Check compression set spec at operating temp |
| Loss varies by vial position on tray | Draught or tray temperature gradient | Log tray position, move balance |
If you want an independent check on the method, the analytical testing resources cover the validation side of this, and EPA SW-846 methods give you the holding-time and preservation language that most labs are actually trying to meet when they chase a seal failure.
Field case 1: the 4.7% RSD drift that was a crimper, not a column
Symptom: A 26-hour GC run on a volatile halogenated analyte showed internal standard drift of 4.7% RSD across the tray, with the last 20 vials worst. The column had been replaced. The inlet liner had been replaced. The drift persisted.
Measurement: We pulled 12 vials from the back of the tray and weighed them against 12 from the front. The back vials had lost a mean of 5.1 mg versus 1.2 mg at the front. Then we weighed a fresh set of vials crimped at the lab’s standard setting (which turned out to be 0.7 N·m on a manual crimper that had never been calibrated) and a set crimped at 1.0 N·m. The 0.7 N·m set lost 4.8 mg in 24 hours. The 1.0 N·m set lost 1.1 mg.
Fix: Crimper recalibrated to 1.0 N·m, torque-limiting handle added, and a daily crimp check with a torque gauge. The drift dropped to 0.9% RSD on the next run. The column was never the problem.
Field case 2: the headspace method that failed at 60 °C
Symptom: A headspace method validated at 40 °C was transferred to a 60 °C oven for a different matrix. Recovery of a volatile analyte dropped 8% and the RSD doubled. The vial lot was unchanged.
Measurement: Gravimetric loss test at 60 °C on the same closure showed 9.6 mg lost over 6 hours, versus 2.1 mg at 40 °C. The septum was a standard 1 mm silicone, 50 Shore A. Compression set at 60 °C was measured at 28% after 6 hours, versus 12% at 40 °C. The contact stress at the land had fallen below the threshold to hold the seal.
Fix: Switched to a PTFE-faced silicone septum with a higher-temperature silicone backing and a thicker aluminium crimp (0.25 mm versus 0.20 mm skirt) to hold compression. Loss at 60 °C dropped to 2.8 mg over 6 hours, and recovery came back within 1.5% of the 40 °C value.
Both cases have the same lesson: the closure is a mechanical assembly, and the failure is almost always at an interface, not in the bulk material. If you are sourcing the whole assembly, start with the oem vial specification and work outward to the closure, not the other way around.
A troubleshooting checklist you can run in an afternoon
- Weigh a fresh set at t=0 and t=24. If loss is under 1.5 mg, the seal is not your problem. Stop and look elsewhere.
- Check the crimp torque. Use a torque gauge, not feel. Target 0.9 to 1.2 N·m for 11 mm crimp necks. Log it.
- Inspect the neck land on 10 vials under 10x magnification. Look for chips, witness lines, and out-of-flatness.
- Confirm the needle gauge. If you are above 22 gauge, your puncture rating is lower than the datasheet says.
- Check compression set at operating temperature. If the datasheet only gives 23 °C, ask for the hot number or run the test.
- Log tray position and oven temperature for every vial in the run. A gradient across the tray will look like a seal failure.
- Repeat with a reference closure from a known-good lot. If the reference passes and yours fails, it is the closure. If both fail, it is the method or the environment.
For instrument-side context on how the autosampler handles the vial, the Agilent GC accessories documentation and the Thermo Fisher chromatography sample-handling notes both describe the tray and needle geometry that drive puncture behaviour. Restek also publishes useful holding-time and septum guidance that is worth a read before you change a lot.
What to specify so this does not happen again
If I am writing a purchase specification for a volatile method, I put four numbers on it. Septum durometer: 50 ± 5 Shore A. Septum thickness: 1.0 mm ± 0.05 mm. Crimp torque window: 0.9 to 1.2 N·m. Maximum puncture count at the method needle gauge: stated explicitly, not “multi-puncture”. I also ask for compression set at the operating temperature, not at 23 °C, and I ask for the gravimetric loss figure on the certificate if the supplier has one.
If the supplier cannot give you a compression set number at temperature, that is a signal. It does not mean the product is bad. It means you are buying on price and hoping, and hoping is not a specification.
One more thing on the vial body itself. The neck finish is where the closure meets the glass, and a vial that is dimensionally correct at the body can still be out at the neck. The chromatography vial reference material covers the neck and thread dimensions that matter, and it is worth checking against your incoming inspection before you blame the septum.
The decision rule is simple. If gravimetric loss over your actual run length exceeds 1.5 mg per vial at your operating temperature, you have a seal failure and it will show up as drift, recovery loss, or both. Measure it, fix the crimp or the material, and re-measure. Do not change the column until the balance says you have to.
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: