Where the vapor actually leaves a sealed vial

A 96-position autosampler run on a 2 mL crimp vial lost 3.8% of its acetonitrile volume over 18 hours in a 4 °C tray, and the internal standard peak area drifted 11% low by the last injection. The analyst blamed the column. The column was fine. I weighed the vials before and after the run and the mass loss pointed straight at the seal. That is the kind of aijiren seal failure, evaporation and sample loss problem that never shows up as an error code. It shows up as drift, and drift gets misdiagnosed for weeks.

This article is about the seal as a mechanical and diffusive barrier, not as a part number. I’ll cover where vapor actually escapes, what happens to a septum after the needle leaves, how compression set accumulates over a run, and how to put a number on the loss with a balance instead of a guess. If you already know what the product is, skip ahead. Nothing here is introductory.

Most people picture a leak as a hole. In practice a punctured septum is a slow diffusive path, not a pinhole, and the dominant loss mechanism changes with temperature and time. There are four paths I check, in this order.

Path 1: the needle track through the septum

A 23-gauge needle leaves a track roughly 0.64 mm across. Silicone and silicone/PTFE septa are elastomers; they close because the material is under radial compression from the crimp or the screw cap. When the needle withdraws, the track closes to somewhere between 60% and 95% of its original seal depending on formulation, dwell time, and how many times that same spot was punched. A septum that re-seals to 90% still leaks. It just leaks slower than one that re-seals to 60%. Over 200 injections in a single run, a track that never fully closes is a continuous diffusion channel.

Path 2: the septum-to-glass interface

This is the one that gets missed. The seal is not the septum alone. It is the compression of the septum against the flat land of the vial finish. If the crimp is under-torqued, or the vial finish has a chip or a mold seam at the land, vapor escapes sideways under the septum skirt. You cannot see it and a visual inspection will not catch it. A 11 mm crimp needs the aluminum seal deformed so the septum is compressed about 30% to 40% of its thickness. Too little and the interface is a leak. Too much and you crush the septum and create a permanent track that never re-seals.

Path 3: the cap thread on screw-top vials

Screw caps leak at the thread, not the septum, when the cap is not fully seated. A 9 mm screw thread with a PTFE-faced silicone liner needs the cap turned until it seats against the finish, typically 1.2 to 1.5 N·m for a 9 mm cap on a glass vial. Hand-tight is not a torque value. In a tray that cycles between 4 °C and 25 °C, a cap that was only finger-tight backs off as the liner relaxes, and the leak rate climbs through the run.

Path 4: permeation through the septum material itself

Even a perfect seal permeates. Silicone is the worst offender for small polar molecules like methanol and acetonitrile; PTFE-faced septa cut permeation dramatically on the face side but the silicone body still transmits. For a 24-hour run at ambient, permeation through a bare silicone septum can account for a measurable fraction of total loss. This is why the PTFE face matters and why a bare silicone septum on a long run is a poor choice regardless of how well it re-seals.

Re-seal behaviour after puncture is a material property, not a manufacturing defect

I get asked constantly why a septum “doesn’t close.” It does close. The question is how much. Re-seal is governed by the elastomer’s elastic recovery, and that varies by formulation and by how the septum was made.

Bare silicone recovers fast and well, but it corks easily and permeates. Silicone with a PTFE face recovers well and permeates less, but the PTFE face does not self-heal, so the track in the PTFE layer stays open and the silicone underneath has to do the sealing. Natural rubber and butyl recover more slowly and take a compression set faster. For a run with more than about 50 punctures in the same position, I specify a septum with a laminated PTFE face and a silicone body, and I accept that the PTFE track is permanent. The loss then depends on the silicone body closing under it.

The practical number: a good silicone/PTFE septum re-seals to better than 85% of its sealing capacity after the first puncture, and that figure drops with each subsequent puncture in the same spot. By puncture 10 in the same location you may be at 70%. If your method punches the same vial more than a dozen times, plan for it in the loss budget or move to a fresh vial. The mechanical sequence behind that degradation is covered elsewhere on the site; here I care about the loss it produces.

Compression set is what turns a good seal into a slow leak

Compression set is the permanent deformation an elastomer takes when held under load. A septum under a crimp is held under load for the entire run. Over hours at elevated tray temperature, the material relaxes and the sealing force drops. That is compression set, and it is the reason a vial that sealed perfectly at injection 1 is leaking at injection 200.

Two numbers govern it. First, the compression ratio: the septum should be compressed to roughly 25% to 35% of its free thickness by the cap. Below 20% you have insufficient sealing force. Above 40% you accelerate set and make the track worse. Second, temperature. Standard silicone is rated to about 200 °C continuous, but compression set accelerates well before that. In a 4 °C tray set, the temperature is low so set is slow; in a heated tray at 40 °C, set is fast. I have measured a crimp septum that held 0.4 mg/h loss at hour 1 and 1.3 mg/h at hour 12 on the same vial, on the same balance, with nothing changed but time.

The fix is not a thicker septum. A thicker septum under the same crimp gets more compression, not less set. The fix is a septum formulated for low compression set (peroxide-cured silicone rather than platinum-cured for long static seals) and a crimp that hits the 25% to 35% window, verified with a crimper that has a depth stop rather than a hand tool.

Measuring the loss gravimetrically: the only number that settles the argument

If you want to know whether your seal is losing sample, weigh it. Not the peak area, not the internal standard ratio, the vial. A balance readable to 0.1 mg is enough. Here is the procedure I use, and it takes about an hour to set up for a whole tray.

1. Label and weigh 12 vials empty, cap and septum in place, to 0.1 mg. Record the tare.
2. Fill each with a known mass of your actual sample solvent, typically 1.0 g of the mobile phase or diluent, and reweigh. The difference is your starting liquid mass.
3. Puncture each vial the number of times your method will puncture it, using the actual needle gauge and the actual autosampler.
4. Place the tray at the run temperature and reweigh at 0, 2, 6, 12, and 24 hours, or at whatever interval brackets your run.
5. Compute loss rate in mg/h per vial and convert to percent of starting mass.

Run the same tray with a control set: same vials, same septa, no punctures. The difference between the punctured and unpunctured loss is your puncture-driven leak. The unpunctured loss is permeation plus interface leak. That split tells you whether to change the septum, the crimp, or the cap torque.

A worked example from a recent tray: unpunctured control lost 0.2 mg/h per vial at 25 °C. Punctured once, 0.5 mg/h. Punctured ten times, 1.1 mg/h. On a 1.0 g fill over 24 hours, the ten-puncture vial lost 2.6% of its mass. That is enough to move an internal standard area by several percent and enough to fail a system suitability check on precision. The same tray on a laminated PTFE/silicone septum lost 0.6 mg/h at ten punctures, and the run passed.

Field cases: two trays, two different leak paths

Case 1: the under-crimped lot

Symptom: a 48-vial run on a GC headspace method showed internal standard area falling steadily from injection 20 onward, ending 9% low. The column was replaced, the inlet liner was replaced, the drift persisted. Gravimetric check on the same lot: punctured vials lost 1.4 mg/h against a control of 0.15 mg/h, a 9x ratio that no normal septum produces. Inspection of the crimp with a magnifier showed the aluminum skirt had not rolled under the vial finish on about a third of the lot. The crimper depth stop was set 0.3 mm too shallow. Fix: reset the crimper to spec, re-crimp the lot, and the drift dropped to 1.5% over the same run length. The septa were fine. The crimp was not.

Case 2: the wrong septum for the temperature

Symptom: a 12-hour run at 40 °C tray temperature on a bioanalytical method lost 4.1% of sample mass, and the late-eluting peaks lost area faster than the early ones, which is the signature of differential evaporation by volatility. Gravimetric check split the loss: control vials lost 0.3 mg/h, punctured vials lost 1.8 mg/h. The septum was a bare silicone, not laminated. At 40 °C the permeation term dominated. Fix: switch to a PTFE-faced silicone septum and drop the tray to 10 °C where the method allowed it. Loss fell to 0.5 mg/h and the volatility-dependent drift disappeared. The lesson: at elevated tray temperature, permeation through the septum body, not the puncture track, is the dominant path.

Decision checklist: match the seal to the run before you buy

Use this as a specification sheet, not a shopping list. The right column is where most seal failures start.

Two more selection notes. First, the septum thickness must match the cap and the vial finish; a 1.3 mm septum in a cap designed for 1.0 mm will over-compress and take set fast. Second, if you are buying vials and caps as a set, buy them as a set. Mixing a cap from one lot with a vial from another is where the interface leak in Path 2 comes from. The dimensional stack that decides fit is covered on Vials With Caps, and the lot-level consistency you should demand is covered on Vial Supplier. For the seal itself, the material and finish data you need to specify against is on Aijiren.

Operational precautions that cut loss without changing parts

• Punch a fresh spot each injection. Most autosamplers can be programmed to index the needle across the septum. Ten punctures in ten locations beat ten in one.
• Keep the tray cold. Every 10 °C drop in tray temperature roughly halves the vapor pressure driving the loss. If the method allows 4 °C, use it.
• Seat the cap to torque, not to feel. A 9 mm screw cap at 1.2 N·m is repeatable. Hand-tight is not, and it varies by analyst.
• Verify the crimper weekly. A depth stop drifts. Check it against a go/no-go gauge and log the reading.
• Pre-weigh a control vial on every lot. One unpunctured vial and one punctured vial on the balance at receiving tells you if the lot is in spec before it reaches the autosampler.
• Do not re-use septa across runs. A septum that took set on Monday leaks on Tuesday. The cost of a septum is trivial against a failed run.

The reference material for the mechanical and material side of this is worth having on the bench: Restek technical resources covers septum selection and re-seal data, Agilent’s GC consumables pages list septum specs by instrument, and Thermo Fisher’s chromatography and sample handling documentation covers cap torque and vial finish compatibility. For regulated methods, the container and holding-time rules that make seal integrity a compliance issue, not just a precision issue, are in U.S. EPA SW-846. If your lab runs chromatography day in and day out, the seal is the part of the workflow that Lab Chromatography practitioners argue about most, because it is the cheapest part and the one that quietly decides whether the run holds.

When to change the part and when to change the procedure

Here is the decision rule I use. If the gravimetric loss on a punctured vial is more than 0.5 mg/h above the unpunctured control, the seal is the problem and the fix is a part change: laminated PTFE/silicone septum, correct crimp depth, correct cap torque. If the punctured loss is within 0.2 mg/h of the control but the run still drifts, the seal is fine and the problem is procedural: too many punctures in one spot, tray too warm, or a cap that was not seated to torque. Measure first, then change one variable. A seal failure and an evaporation loss look identical on the chromatogram. They are told apart on the balance, in mg/h, before you touch the instrument.

References & further reading

Related resources from our lab consumables network:

• Aijiren at aijirentech.com
• Vials With Caps at vialswithcaps.com
• Vial Supplier at vialsupplier.com
• Lab Chromatography at labchromatography.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