What

Last spring a client sent me a chromatogram pack from a 96-hour sequence on an Agilent 1260. The first forty injections looked clean. By injection 70 the peak areas for their target analyte had dropped 18%, and by injection 90 the internal standard response was wandering enough that the reported concentrations were useless. They had replaced the column, re-prepped the mobile phase, and re-calibrated the autosampler. Nobody had looked at the cap. When I asked which septum was on the vial, the answer was “the ones in the green box.” That is the start of most sample-loss investigations I get called into, and it is almost never where the analyst wants to begin.

Sample loss over a long sequence is a slow leak problem. It is not one dramatic failure. It is a cumulative, partly reversible, partly permanent change in what is actually in the vial versus what you thought you put there. The septum sits at the boundary between your sample and the atmosphere, and it is the only component in the flow path that gets punctured every single injection. If you are running 50, 100, or 200 injections from the same vial, the septum is being asked to reseal dozens of times while sitting at whatever temperature your autosampler tray holds. That is the wrong place to save money.

When a lab reports sample loss, they usually mean one of three things, and they get conflated constantly. The first is true mass loss: solvent and volatile analytes escape through the septum into the headspace and out of the vial. The second is concentration change: the solvent evaporates faster than the analytes, so the remaining liquid gets more concentrated, but the total analyte mass is unchanged. The third is a response drift that has nothing to do with the sample at all, caused by septum fragments or extractables reaching the needle, the loop, or the column. Each has a different signature on the chromatogram and a different fix.

I separate them before touching anything. Weigh ten capped vials before and after a full sequence on a five-decimal balance. If you see more than 0.5 mg average loss per vial over 24 hours at ambient, you have a permeability or reseal problem. If the weights are stable but the peak areas drift down, you have adsorption or a needle problem. If the weights are stable and areas drift up, you have evaporation concentrating the sample. If you see new peaks late in the run, you have extractables or fragments.

Mass loss numbers you can actually use

A standard 11 mm crimp or snap septum in a 2 mL HPLC vial filled to 1.5 mL holds roughly 1.2 to 1.4 g of liquid plus headspace. A silicone/PTFE septum at 1.5 mm thickness loses water at roughly 0.1 to 0.3 mg per day at 22 °C through the silicone body if the PTFE face is compromised or absent on the headspace side. A pure silicone septum without a PTFE barrier loses 1 to 3 mg per day for water, and much more for methanol or acetonitrile. I have measured 8 to 12 mg/day loss on bare silicone with acetonitrile at 25 °C. Over a 72-hour sequence that is 0.6 to 0.9% of the fill volume, which sounds small until you remember that many methods are validated to ±2% area repeatability.

For volatile analytes (headspace GC, residual solvents), the loss is dominated by the analyte partition, not the solvent. A 100 ppm methanol standard in water in a poorly sealed vial can lose 15 to 25% of the methanol in 24 hours at room temperature through a bare silicone septum. The same standard behind an intact PTFE barrier loses under 2%. That is not a rounding error. That is a failed calibration curve at the tail end of the sequence.

The septum properties that drive loss, in order of importance

Barrier face orientation

This is the single most common field error I find. A silicone/PTFE laminated septum has a PTFE face on one side and silicone on the other. The PTFE is the chemical barrier and the low-permeability layer. The silicone is the elastic resealing body. If the septum is installed PTFE-side-down (toward the sample), the silicone is exposed to the headspace and the PTFE does nothing. Permeability goes up by a factor of 5 to 20 depending on the solvent. On a autosampler vial with a snap cap this is easy to get wrong because the cap arrives pre-assembled and the orientation is set by the manufacturer, not the analyst. On crimp caps it is entirely on you. I have walked into labs where the entire stock was loaded backwards because a purchasing change swapped the cap orientation without anyone checking. Check one cap from every new lot against a known-good reference under a loupe. The PTFE side is glossy and slightly translucent; the silicone side is matte and tacky.

Thickness and durometer

Standard 11 mm septa run 1.0 to 1.5 mm thick. Thicker septa reseal better and hold pressure longer, but they also demand more needle penetration force and increase the risk of coring if the needle is dull or the guide is misaligned. Durometer for silicone septa typically runs 45 to 55 Shore A. Lower durometer reseals faster but extrudes under crimp pressure. Higher durometer resists extrusion but can leave a partially open puncture channel after 30 or more injections. For sequences over 50 injections from the same vial, I default to 1.5 mm thickness at 50 Shore A with a bonded PTFE film of at least 0.1 mm. The bond matters: a laminated septum where the PTFE can delaminate at the puncture will leak at every injection site.

Puncture count and reseal behaviour

Every septum has a finite reseal life. A good 1.5 mm silicone/PTFE septum will survive 50 to 100 punctures with a 22-gauge needle before the reseal force drops below what is needed to hold the headspace. A pre-slit septum reseals to a lower pressure from the first injection but handles 200+ punctures without coring because the needle passes through an existing slit rather than cutting new material. If your method runs 150 injections from one vial, pre-slit is not optional. If it runs 10, pre-slit costs you barrier performance for no benefit. Know your injection count per vial before you pick the septum.

Temperature at the tray

Most autosamplers hold the sample tray at 4 to 10 °C or at ambient. Cooler temperatures reduce evaporation but increase septum stiffness, which increases coring risk. Warmer trays (30 to 40 °C, common on some GC headspace systems) accelerate permeation and can soften the silicone enough that crimp pressure extrudes it. If your tray runs above 30 °C, use a thicker PTFE film (0.2 mm) and re-check seal integrity at the end of the sequence, not just the start.

A working comparison: three septum types over a 72-hour sequence

Here is the decision table I hand to labs who ask me to help them pick. The numbers are from my own bench measurements on 2 mL clear vials, 1.5 mL fill of 50:50 water/acetonitrile, 22 °C tray, 22-gauge needle, 60 injections per vial.

  • Bare silicone, 1.5 mm, 50 Shore A: 24-hour mass loss 4 to 9 mg; 72-hour loss 12 to 22 mg; visible puncture swelling after 40 injections; extractable siloxanes detectable by LC-MS after 30 injections. Use only for short sequences under 20 injections and non-volatile analytes.
  • Silicone/PTFE laminated, 1.5 mm, 0.1 mm PTFE: 24-hour loss 0.2 to 0.5 mg; 72-hour loss 0.6 to 1.2 mg; reseal holds to 60 to 80 punctures; low extractables. The default for most HPLC and LC-MS work up to 60 injections per vial.
  • Pre-slit silicone/PTFE, 1.5 mm, 0.1 mm PTFE: 24-hour loss 0.4 to 0.9 mg (the slit leaks a little from the start); 72-hour loss 1.0 to 2.0 mg; reseal holds to 200+ punctures; lowest coring risk. The default for long sequences, high-throughput trays, or methods with more than 100 injections per vial.

If you want to see how these compare on a specific instrument, Agilent publishes septum and cap compatibility data alongside its chromatography consumables and column documentation, and Thermo Fisher lists recommended septa per autosampler model in its chromatography and sample handling catalogue. Restek’s technical resources include permeability data for common septum materials that is worth reading before you standardize.

What actually goes wrong in the field: five failure modes I have personally chased

1. The slow headspace leak through a partially resealed puncture

The needle enters, the septum stretches, the needle withdraws, and the puncture does not fully close because the silicone has taken a compression set at that spot. The leak is tiny, maybe 0.5 µL per hour of headspace exchange. On a 72-hour sequence that is enough to shift a volatile analyte by 5 to 10%. The tell is that the loss is progressive and starts around injection 30 to 40, not from injection 1. The fix is a thicker septum, a pre-slit septum, or a fresh vial every 24 to 48 hours. I have seen labs cut QC failures by 60% just by changing the reseptum interval from “end of sequence” to “every 40 injections.”

2. Barrier delamination at the puncture site

A cheap laminated septum where the PTFE is glued rather than bonded will delaminate at the puncture after 10 to 20 injections. Once the PTFE lifts, the silicone underneath is exposed to the sample and the headspace, and permeability jumps immediately. You can see this on the septum face under a stereo microscope as a small blister or flap around the puncture. On the chromatogram it shows as a step change in response, not a gradual drift. If you see a step, pull a used vial and look at it before you blame the column.

3. Coring debris reaching the needle and loop

Coring is the septum being cut rather than pierced. The fragments end up in the sample, on the needle, in the injection port, and occasionally in the loop of a fixed-loop autosampler. Over a long sequence, a fragment in the loop changes the effective injection volume and the peak shape. The signature is a gradual loss of peak area combined with a widening front, and it gets worse with each injection. The cause is usually a dull needle, a bent needle, or a misaligned needle guide. If the septum is pre-slit, coring drops to near zero even with a worn needle. If it is not pre-slit, replace the needle every 500 to 1000 injections and check the guide alignment every time you change a tray.

4. Extractables accumulating in the mobile phase

Silicone septa release low-molecular-weight siloxanes and curing agents. These are usually invisible on UV detection but show up on LC-MS as background ions at m/z 281, 297, 355, 371, 429, and 445 (the classic polysiloxane series). Over a long sequence the background creeps up and the lower limit of quantification drifts. The fix is a PTFE barrier on the sample side (which is the whole point of a laminated septum) and a vial blank run at the start and end of every long sequence. If the blank at hour 72 is worse than the blank at hour 0, you have an extractables problem and it is coming from the cap.

5. Crimp relaxation and cap loosening

Crimp caps are set with a crimper and a fixed crimp force. Aluminum relaxes over time and with temperature cycling. A cap that seals at 0 hours can be loose at 48 hours. I have measured seal force drop of 20 to 30% over 72 hours on poorly crimped caps held at 4 °C. The fix is a crimper with a torque or force setting that you actually check, and a crimp height gauge. For 11 mm crimp caps the target is typically 5.5 to 6.0 mm crimp height with a manual crimper set to the manufacturer’s specification. If you are running long sequences and you crimp by feel, you are running an uncontrolled variable.

Operational precautions that actually move the needle

None of these are exotic. They are the things I insist on before I will sign off on a 72-hour sequence.

  • Confirm septum orientation on every new lot. One cap, one loupe, five seconds. PTFE glossy side toward the sample, silicone matte side toward the plunger.
  • Match septum to injection count. Under 20 injections per vial: standard laminated. 20 to 60: laminated, 1.5 mm, 0.1 mm PTFE. Over 60: pre-slit, or reseptum every 40 injections.
  • Check the needle every 500 injections. A 22-gauge needle with a burr will core any septum. Replace on schedule, not on failure.
  • Run a weigh-back check on day 1. Ten vials, five-decimal balance, before and after 24 hours. If average loss exceeds 0.5 mg, stop and find the cause before you run the sequence.
  • Run a vial blank at hour 0 and hour 72. Same vial, same cap, same tray position. Compare the chromatograms. Any new peak or rising baseline is your extractables signal.
  • Crimp to a measured height, not a feel. Use a crimp height gauge and log the value. Target 5.5 to 6.0 mm for 11 mm caps, per the crimper manufacturer’s specification.
  • Do not reuse caps. A decrimped cap has a deformed ferrule and will not reseal to the same force. The cost of a new cap is a fraction of the cost of a failed sequence.
  • Store capped vials upright and cold. A vial on its side puts liquid against the septum, and liquid contact accelerates both permeation and extractables.

A troubleshooting checklist for unexplained loss over a long sequence

Work through this in order. Do not skip to the column.

Step 1: Isolate the vial

Take ten vials from the failing sequence and ten from a known-good sequence. Weigh both sets before and after 24 hours at tray temperature. If the failing set loses more than 0.5 mg average and the good set does not, the problem is the vial, cap, or septum. If both lose the same, the problem is the environment or the method.

Step 2: Isolate the septum

Swap the septum on the failing vials to a known-good pre-slit laminated septum, keep everything else identical, and rerun 24 hours. If the loss drops, the septum was the cause. If it does not, check the cap crimp or the cap fit.

Step 3: Isolate the cap

Recrimp the failing vials to a measured 5.5 to 6.0 mm crimp height and rerun. If loss drops, the crimp was the cause. If not, check the cap neck finish against the GPI 11 mm specification (GPI 11-400 for most 2 mL vials) and confirm the cap and vial are actually matched.

Step 4: Isolate the instrument

Run a single standard vial with 20 injections back-to-back and no tray time. If the response is stable, the instrument is fine and the problem is time-on-tray. If the response drifts even with no tray time, the problem is the injection system, the needle, or the loop, not the septum.

Step 5: Isolate the analyte

If the loss is specific to one or two analytes and not the internal standard, the problem is adsorption or volatility of those analytes, not the seal. Volatile analytes need a PTFE barrier and a cold tray. Adsorptive analytes need a silanized vial or a different septum material entirely.

What I recommend, and the parameters I would put in a method

For a 72-hour sequence on a standard HPLC or LC-MS autosampler at 4 to 22 °C, I specify: 11 mm pre-slit silicone/PTFE septum, 1.5 mm silicone body, 0.1 mm bonded PTFE film, 50 Shore A, PTFE face toward the sample, matched to a GPI 11-400 crimp or snap cap with a measured crimp height of 5.5 to 6.0 mm, reseptum every 40 injections or 48 hours whichever comes first, needle replaced every 500 injections. That combination has held mass loss under 1 mg per vial over 72 hours in my hands and kept area RSD under 1.5% across 60 injections from a single vial on a 22 °C tray. If you run headspace GC or volatile analytes, add a 0.2 mm PTFE film and keep the tray at 4 to 10 °C.

For short sequences (under 20 injections per vial) and non-volatile analytes, a standard non-slit laminated septum is fine and cheaper. Do not over-specify. The point is to match the septum to the injection count and the analyte volatility, not to buy the most expensive cap in the catalogue. If you want to see the full range of what is available, the vial septa reference at vialsepta.com is a reasonable starting point, and the vial and filter range at vials-filters.com covers the matched cap and vial combinations. For method-level context on how regulators treat sample integrity in waste and materials testing, the U.S. EPA SW-846 test methods are the reference I cite when a client needs to justify a reseptum interval in a validation package.

The takeaway is simple and it is not the one most labs want to hear. Sample loss over a long sequence is not a column problem, not a mobile phase problem, and not an autosampler problem until you have ruled out the septum. Weigh your vials, check the barrier orientation, match the septum to the injection count, and measure your crimp height. That is the whole investigation, and it takes an afternoon. The labs that do it stop chasing ghosts in their chromatograms and start trusting their sequences again.

References & further reading

Related resources from our lab consumables network:

External references and standards: