A 40-vial sequence for a volatile organic compound screen started at 06:10. By 14:40, the internal standard response in vial 36 had dropped 18% relative to vial 4, and the quantitation for benzene in that same vial read 12% low against the calibration curve. The run did not fail an audit. It failed a mass balance. The vials were not cracked, the caps were not loose by hand feel, and the autosampler never flagged a pressure error. The problem was analytical testing seal failure, evaporation and sample loss, and it had been happening for weeks before anyone weighed a tray.

Most people picture a leak as a visible gap. In analytical testing, the relevant leak is a diffusion path measured in micrograms per hour. A septum can look perfectly seated and still lose enough solvent to shift concentrations. The failure modes that matter are permeation through the elastomer, interfacial leakage between the septum and the vial rim or cap, and re-seal failure after the needle withdraws.

Permeation is not a defect. Every elastomer has a finite permeability to organic vapors. What changes is the rate. A butyl/PTFE septum at 40 °C will lose methylene chloride through the butyl layer at a rate I have measured gravimetrically at 0.8 to 1.2 mg/hour per vial when the cap is under-torqued by 20%. The same septum at proper torque loses 0.2 to 0.4 mg/hour. That is a threefold difference driven entirely by compression, not by the material itself.

Interfacial leakage is simpler. If the septum is not compressed evenly against the vial rim, vapor escapes along the contact line. This happens when the cap is cross-threaded, when the vial rim has a molding flash line, or when the septum is off-center. You will not see it. You will see a drift in peak areas that tracks with vial position in the tray, because earlier vials have less time to lose mass.

The two numbers that predict re-seal behaviour: puncture count and compression set

A septum is a spring. You compress it, it pushes back. You puncture it, it must close around the needle during injection and then close again when the needle leaves. Two material properties decide whether that works over a long sequence: compression set and puncture resilience.

Compression set is the permanent deformation after a compressive load is removed. For a silicone/PTFE septum, a compression set above 35% after 22 hours at 70 °C means the septum has lost most of its ability to maintain interfacial pressure. I have pulled septa from multi-injection vials after 8 hours and measured a visible annular groove where the cap compressed the elastomer. The center never recovers. The result is a slow, continuous loss path that gets worse with every injection because the needle hole is no longer under radial compression.

Puncture count matters more than most lab managers admit. A standard 8 mm PTFE/silicone septum in a 9 mm screw cap is rated for roughly 3 to 5 clean punctures with a 23-gauge needle before re-seal integrity degrades measurably. A pre-slit septum can handle 10 to 15 punctures because the slit is designed to close under compression rather than tear. If your method calls for 8 injections from the same vial and you are using a standard septum, you are running the last four injections on a compromised seal. The loss per puncture is nonlinear: the first three punctures lose almost nothing, the fourth loses maybe 0.1 mg over the remaining run, the sixth can lose 0.5 to 1.0 mg depending on solvent volatility.

Compression set is temperature-dependent, and most methods ignore that

Compression set accelerates with temperature. A silicone/PTFE septum that holds a 20% compression set at 23 °C will hit 45% after the same time at 60 °C. Headspace methods run hotter. GC inlet temperatures do not directly heat the septum in the vial, but tray compartments in some autosamplers sit at 35 to 45 °C, and that is enough to matter over a 12-hour sequence.

The practical rule I use is this: if your run exceeds 6 hours and your solvent has a vapor pressure above 30 kPa at 20 °C (methylene chloride, acetone, ethyl acetate), you should assume seal loss is a variable in the method. That means you either weigh vials before and after, or you change the closure system. Guessing is not a control strategy.

For methods using HPLC vials with snap caps, the problem is worse because snap caps apply lower and less uniform compression than screw caps. A snap cap with a PTFE/silicone septum typically applies 40 to 60 N of axial force. A properly torqued 9 mm screw cap applies 90 to 120 N. That difference shows up as faster interfacial leakage and higher compression set in snap-cap systems. If your method tolerates it, switch to screw caps for anything volatile or long-running.

How to measure loss gravimetrically without a dedicated setup

You do not need a microbalance in a humidity-controlled room. A standard analytical balance with 0.01 mg resolution is enough, and the procedure is straightforward.

Weigh each vial before capping. Record the mass to 0.01 mg. Fill, cap, and torque. Weigh again. After the run, weigh a third time. The difference between the second and third weighings is your total loss, minus any sample withdrawn by injection. You must correct for injected volume: a 1 µL injection from a vial containing 1.5 mL of solvent removes 0.065% of the liquid mass, which is negligible for most purposes but not for a 24-hour study.

The more useful measurement is the loss rate. Divide total loss by elapsed time. A loss rate below 0.05 mg/hour per vial is acceptable for most quantitative methods with non-halogenated solvents. Between 0.05 and 0.2 mg/hour, you have a seal problem that will show up as drift in long sequences. Above 0.2 mg/hour, you have a closure failure that needs immediate correction, not a method adjustment.

For a tray-level check, weigh the entire tray before and after. A 40-vial tray losing 20 mg over 8 hours averages 0.5 mg per vial, but that average hides the real story: vials 1 through 10 may lose almost nothing, while vials 35 through 40 lose 1.5 mg each. The tray average tells you there is a problem. The per-vial weights tell you which closure type or position is failing.

Field case: the 18% internal standard drift that was not the detector

A contract lab running EPA Method 8260 for volatiles saw internal standard response drop by 18% between the first and last calibration check in a 12-hour sequence. The instrument passed a leak check. The inlet was clean. The column was new. The first instinct was detector drift, but the detector response for the continuing calibration standard was stable when injected from a fresh vial.

The measurement that found it: we weighed 10 vials from the sequence after completion. Vials injected early in the run lost 0.8 to 1.1 mg each. Vials injected late lost 3.2 to 4.5 mg each. The closure was a 9 mm screw cap with a PTFE/butyl septum, torqued by hand. Hand torque varies from 40 to 80 N depending on the operator. The fix was a calibrated torque driver set to 0.9 N·m (8 in-lb) for 9 mm caps and a switch to a PTFE/silicone septum with lower compression set at tray temperature. Internal standard drift dropped to 3% over the same sequence length.

The lesson is not that butyl is bad. Butyl has lower permeability than silicone to many solvents. The problem was uneven compression. A butyl septum under-torqued by one operator loses more than a silicone septum properly torqued by another. The closure system is only as good as the torque applied to it.

Field case: the headspace run that lost 22% of ethanol in three hours

A food testing lab ran residual ethanol in packaged goods by headspace GC. Samples were prepared in 20 mL headspace vials with 20 mm crimp caps and PTFE/silicone septa. The ethanol peak area for the same standard concentration dropped 22% over a 3-hour run. The oven temperature was 80 °C.

The measurement: we weighed six sealed vials before and after a 3-hour hold at 80 °C. Average loss was 6.8 mg per vial. That is enormous. The crimp caps had been applied with a hand crimper that produced inconsistent crimp depth. Some caps were over-crimped, which crushes the septum and creates a radial tear. Others were under-crimped, which leaves an interfacial gap. Both fail.

The fix was a calibrated crimper set to a crimp depth of 8.0 mm for a 20 mm cap, verified with a crimp gauge, plus a switch to a septum rated for 120 °C continuous use. Loss dropped to 0.9 mg per vial over the same hold. The method recovered its precision. The root cause was not the septum material. It was the crimp geometry.

Torque values, dimensions, and closure specs that actually matter

If you are sourcing closures for volatile or long-run methods, here are the numbers I use as acceptance criteria. They come from working with multiple suppliers and measuring failure rates in the field, not from a single vendor datasheet.

The torque values assume a GPI 9-425 or 10-425 thread finish. If your vials use a different thread standard, the numbers shift. Lab vials from different manufacturers have different rim profiles, and a torque value that works for one may not work for another. The only way to know is to measure loss gravimetrically with your specific vial, cap, septum, solvent, and temperature combination.

Why re-seal failure shows up as position-dependent drift, not random error

Random error scatters. Seal loss accumulates. That is the diagnostic difference. If your peak areas drift monotonically with vial position in the sequence, and the drift is larger for later vials, you are looking at a time-dependent mass loss mechanism. Seal failure is the most common one.

Re-seal failure after puncture is position-dependent because each puncture adds a small leak path. Vial 1 gets one puncture and sits for 11 hours. Vial 40 gets one puncture and sits for 20 minutes. If the septum re-seals perfectly, both vials lose the same amount per hour. If it does not, vial 1 loses more total mass because it has more time to leak. The drift pattern will show early vials with lower response and later vials with higher response, which is the opposite of what most people expect from evaporation.

That inversion is the tell. Evaporation from an open container is fastest at the start. Seal-mediated loss is slow and continuous, so the cumulative loss is greatest for the vials that have been sitting longest. When I see early vials reading low and late vials reading normal, I check seal integrity before I touch the instrument.

A troubleshooting checklist for seal-related loss in a running method

When you suspect analytical testing seal failure, evaporation and sample loss, work through this sequence before changing the method or the instrument.

• Weigh three vials from the beginning, middle, and end of the sequence. If the loss rates differ by more than 0.1 mg/hour, the closure is the variable.
• Check torque on 10 caps with a torque driver. If the standard deviation exceeds 15% of the mean, hand torquing is your problem.
• Inspect septa under magnification after the run. Look for an annular compression groove deeper than 0.2 mm, radial tears, or a puncture hole that does not close.
• Count punctures per vial. If any vial exceeds the septum rating, replace the septum or reduce injections per vial.
• Check tray temperature. A tray at 40 °C doubles the loss rate of a tray at 25 °C for most volatile solvents.
• Verify crimp depth with a gauge if using crimp caps. Reject any cap outside the 7.8 to 8.2 mm range.
• Run a blank vial with solvent only through the same puncture schedule. If it loses mass, the problem is the closure, not the sample matrix.

This checklist takes about 30 minutes and answers the question of whether the closure is the failure point. Most labs skip it and spend days chasing detector drift or column degradation. The closure is cheaper to test and more often the cause.

What to specify when you order closures for volatile methods

Do not order “PTFE/silicone septa” and assume the problem is solved. The specification needs numbers. I specify the following for any method with volatile solvents or run times over 4 hours:

• Septum material: PTFE face, 0.010 in (0.25 mm) thick, bonded to a silicone core with Shore A hardness 45 to 55. Softer silicone conforms better to the vial rim but sets faster. Harder silicone resists set but seals less well at low torque.
• Compression set: maximum 30% after 22 hours at 70 °C, tested per ASTM D395 Method B.
• Puncture rating: minimum 5 clean punctures with a 23-gauge bevel needle without visible coring or tearing.
• Temperature rating: continuous use at 100 °C for headspace, 60 °C for autosampler trays.
• Cap material: polypropylene with a GPI 9-425 or 10-425 thread, torque range 0.8 to 1.2 N·m.
• Vial rim: flat, fire-polished, free of molding flash. A rim with a 0.05 mm flash line will leak at any torque.

For suppliers, I have used closures from chromatography consumables vendors who publish compression set and puncture data. If a supplier cannot tell you the compression set of their septum, assume they have not tested it. That is not a supplier you want for a volatile method.

For method development work, the Restek technical resources include application notes on septum selection and closure torque that match what I have measured in the field. The Agilent chromatography consumables pages list septum materials with temperature and puncture ratings for their inlet configurations. Thermo Fisher chromatography and sample handling resources cover vial and closure compatibility for their autosampler lines. For regulated environmental work, U.S. EPA SW-846 methods specify holding times and preservation requirements that assume a sealed container; if your seal fails, your holding time claim fails with it.

The decision rule

If your method runs longer than 4 hours, uses a solvent with vapor pressure above 30 kPa at 20 °C, or pulls more than 3 punctures from a single vial, weigh your vials before and after one full sequence. If the loss rate exceeds 0.05 mg/hour per vial, fix the closure before you adjust anything else. Torque to 0.9 N·m for 9 mm caps, verify crimp depth to 8.0 mm for 20 mm caps, and switch to a septum with compression set below 30%. That one measurement will save more time than any detector diagnostic. The seal is the cheapest component in the system and the most common reason a validated method drifts.

References & further reading

Related resources from our lab consumables network:

• Analytical Testing at testinganalytical.com
• Hplc Vial at vialforhplc.com
• Chromatography Consumables at chromatographyconsumables.com
• Lab Vials at labvials.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