What an HPLC septum is actually doing

I have pulled more failed septa out of autosampler trays than I care to admit. Some were split clean across the center after three injections; some had a neat 0.5 mm hole punched through the PTFE face and were bleeding plasticizer into every blank; a few were the wrong durometer entirely and simply refused to seal against the vial rim. Almost none of those failures were random. They traced back to a specific mismatch between the septum, the closure, the injection conditions, and the way the tray was handled. This article is the field guide I wish I had handed to every new analyst on day one.

If you buy septa by the thousand, the difference between a good lot and a bad one is not usually the price per unit. It is the number of injections you get before the seal fails, the size of the coring particle that ends up in the needle, and the ghost peak that shows up at 2.1 minutes in your blank. Here is how those failures actually happen and how to stop them.

A septum is a small elastomer disc, usually 11 mm diameter for standard 2 mL vials and 8 mm or 9 mm for short-bed and crimp formats, that sits inside a screw cap or crimp seal and provides a resealable barrier between the sample and the atmosphere. It has to do four jobs at once:

  • Seal against the vial rim at the correct compression so solvent vapor does not escape over a 24 to 48 hour run.
  • Allow a needle, typically 22 to 23 gauge (0.64 to 0.34 mm OD), to pass through without tearing.
  • Reseal behind the needle so the vial stays closed between injections.
  • Stay chemically inert to the sample and mobile phase, so nothing leaches into the vial and nothing absorbs out of it.

Those four jobs pull in opposite directions. Softer silicone reseals beautifully but cores easily and leaches siloxanes. Harder PTFE-faced silicone resists coring and chemical attack but can fail to seal on a damaged rim. Every failure mode below is a consequence of one of those trade-offs being pushed too far for the application.

The septum is only one part of the closure. The cap, the rim geometry, and the thread pitch all matter. If you are re-evaluating your whole closure stack, the vial closures reference material on thread and crimp formats is worth bookmarking alongside your septum spec sheet.

The seven failures I see most often

1. Coring (the punched-hole failure)

This is the number one failure in high-throughput labs. The needle punches a small disc of septum material free, and that disc either falls into the sample or rides up the needle and deposits in the injection port, the rotor seal, or the column inlet frit. Symptoms: a sudden pressure spike, a partial blockage that shows as a split or shouldered peak, or a black particle visible in the vial under a loupe.

Root causes, in order of frequency:

  • Needle gauge too large for the septum durometer. A 21 gauge needle (0.51 mm OD) through a soft 40 Shore A silicone septum will core within 5 to 10 injections. Move to a 23 gauge needle or a PTFE-faced septum with a higher durometer core.
  • Needle not centered. If the autosampler needle is off-center by more than about 0.5 mm, it hits the septum at an angle and shears rather than pierces. Check the needle guide and the vial seat alignment.
  • Blunt or burred needle. A needle that has seen 5,000 injections has a rolled tip. Replace it on a schedule, not on failure.
  • Septum pre-pierced in the wrong place. If you pre-slit septa for a manual injection, the slit must be dead center. Off-center slits cause the needle to tear the edge of the slit.

Prevention is mostly about matching needle to septum. For a 23 gauge needle, a PTFE-faced silicone septum with a 45 to 55 Shore A silicone core and a 0.010 inch (0.25 mm) PTFE film will typically survive 100+ injections before coring becomes visible. For a 21 gauge needle, drop to a 22 gauge or accept a shorter septum life and change it more often.

2. Leaching and ghost peaks

Silicone septa contain low-molecular-weight siloxanes and plasticizers that migrate into the headspace and, in a sealed vial, into the sample itself. In reversed-phase HPLC at UV wavelengths below 220 nm, this shows up as a rising baseline, a ghost peak at a consistent retention time, or a blank that is never truly blank.

The classic offender is the dimethylpolysiloxane family. If you see a peak at m/z 73, 207, or 281 in your LC-MS blank, that is siloxane. The fix is not to abandon silicone; it is to use a PTFE-faced septum where the sample only contacts the PTFE film, and to use a septum that has been extracted and baked by the manufacturer. A good extraction protocol (usually a solvent wash followed by a 200 °C bake) removes most of the volatile fraction. Ask your supplier for the extraction certificate, not just a datasheet.

For trace analysis at ppb levels, I have seen a 10-fold improvement in blank cleanliness simply by switching from an unextracted natural silicone septum to a pre-extracted PTFE-faced one. That is a real number from a real method validation, not a marketing claim.

3. Failure to reseal (the “leaky vial” problem)

A septum that does not reseal after needle withdrawal lets solvent evaporate, concentrates the sample, and changes your quantitation. The failure is silent unless you weigh vials before and after a run. A 2 mL vial filled with acetonitrile that loses 0.5 mg over 24 hours is a septum that is not sealing.

Root causes:

  • Wrong durometer. A 60 Shore A septum on a vial with a slightly out-of-round rim will not conform. A 45 Shore A will.
  • Over-torqued cap. Yes, you can over-tighten a screw cap. More on this below.
  • Damaged vial rim. A chipped or out-of-round rim is a guaranteed leak regardless of septum quality. Inspect rims under a loupe on a sample of every lot.
  • Septum too thin. A 0.060 inch (1.5 mm) septum in a cap designed for a 0.125 inch (3.2 mm) septum will not compress enough to seal.

4. Chemical attack and swelling

Silicone swells in nonpolar solvents (toluene, hexane, dichloromethane) and can swell 20 to 40 percent by volume. A swollen septum pushes out of the cap, loses its compression, and leaks. PTFE is essentially inert to everything except molten alkali metals and fluorine at high temperature, which is why PTFE-faced septa are the default for aggressive solvents.

If you run a method in 100 percent dichloromethane or THF, use a PTFE-faced septum with a silicone core and accept that the silicone core will swell behind the PTFE face. The PTFE face keeps the sample away from the swelling core. Do not use a bare silicone septum in DCM; it will fail within a day.

5. Temperature-related failure

Autosampler trays are often kept at 4 °C for stability, and some methods run at 40 °C for solubility. Both extremes stress the septum. At 4 °C, a soft silicone septum becomes stiff and cores more easily. At 40 °C, it softens and can flow under compression, losing seal.

Standard silicone septa are rated for roughly -40 °C to 200 °C in continuous service, but the practical working range for a resealable septum in an autosampler is narrower: about 4 °C to 40 °C for reliable reseal over 100+ injections. If your tray runs colder, use a septum with a higher durometer core or a PTFE-faced design that does not rely on elastomer compliance for the seal.

6. Particulate shedding

Every needle pass through a septum sheds a few particles. Over 100 injections, that is hundreds of particles per vial, and some of them end up in the column. This is the failure mode that shows up as a gradually rising backpressure over a long sequence, not a sudden failure. The fix is a PTFE-faced septum (PTFE sheds far less than silicone) and an inline filter before the column.

7. Wrong septum for the closure format

An 11 mm septum does not fit an 8 mm short-bed cap. A crimp septum does not fit a screw cap. A septum with a molded center depression (for a specific autosampler needle guide) will not seat in a flat-bottom cap. This sounds obvious, but I have seen a whole tray of 11 mm septa forced into 9 mm caps, and the resulting leaks were blamed on the septa for a week before anyone checked the part number.

If you are standardizing across multiple instruments, the HPLC vial format guide is a useful cross-reference for matching septum diameter to cap and vial geometry.

Selection criteria with real numbers

Here is the decision table I use when specifying a septum for a new method. The numbers are typical values; your supplier’s certificate will have the exact figures for the lot.

Parameter Standard silicone PTFE-faced silicone PTFE/silicone laminated
Typical durometer (core) 40 to 50 Shore A 45 to 55 Shore A 45 to 55 Shore A
PTFE film thickness none 0.005 to 0.010 in (0.13 to 0.25 mm) 0.002 to 0.005 in (0.05 to 0.13 mm)
Recommended needle gauge 23 to 26 22 to 23 22 to 23
Typical puncture count before coring 30 to 50 100 to 150 80 to 120
Solvent resistance poor in nonpolar good good
Blank cleanliness (LC-MS) fair good good
Temperature range (practical) 4 to 40 °C 4 to 60 °C 4 to 60 °C
Cost index (relative) 1.0 1.8 to 2.5 2.0 to 3.0

My default recommendation for a modern LC-MS or UV method at 210 nm is the PTFE-faced silicone septum, 11 mm, 45 to 55 Shore A, 0.010 inch PTFE film, pre-extracted. It costs roughly twice a bare silicone septum and buys you 2 to 3 times the puncture life plus a cleaner blank. That is a straightforward trade.

For GC headspace, the calculus is different: you want the lowest-bleed septum you can get, and a PTFE/silicone laminated septum with a molded center is often the better choice because the laminate resists the higher headspace pressure. The Restek technical resources have good bleed data on septum types for GC.

The torque question, with actual numbers

Screw caps are the most over-torqued consumable in the lab. The correct torque for an 11 mm screw cap on a 2 mL vial is roughly 0.4 to 0.6 N·m for a glass vial with a standard thread. That is about a quarter-turn past finger-tight. If you are using a torque wrench (and for a validated method, you should be), set it to 0.5 N·m and check it monthly.

Over-torquing does three things: it compresses the septum beyond its elastic limit so it cannot reseal, it deforms the cap so it backs off after a few hours, and it can crack the vial neck. Under-torquing leaves the septum uncompressed and the vial leaks. The window is narrow, which is why hand-tightening is a source of variability you can actually measure. I have seen RSD on peak area improve from 2.8 percent to 1.1 percent just by standardizing cap torque across a tray.

Crimp caps are a different problem. The crimp height and the crimp diameter both matter, and a worn crimping tool will produce a visible “smile” or a loose seal. Check the crimp with a go/no-go gauge every 500 caps, and replace the crimping jaws when the gauge fails. The GPI 400 thread standard and the ISO 4796 vial dimensions are the references to cite in your SOP.

Operational precautions that actually move the needle

  • Buy pre-extracted septa. The extraction certificate is worth the premium. Un-extracted silicone will cost you a week of blank-chasing.
  • Store septa in the original sealed bag, away from solvent vapors. A septum that has sat in a lab drawer next to a bottle of toluene for a month is already contaminated.
  • Do not pre-slit unless your method requires it. Pre-slit septa reseal worse than intact ones.
  • Change septa on a schedule, not on failure. For a 23 gauge needle and a PTFE-faced septum, change every 100 injections. For a 21 gauge needle, every 50. Log it in the instrument maintenance record.
  • Inspect a sample of every lot. Check diameter (11.0 ± 0.2 mm), thickness (0.125 ± 0.005 in for standard caps), and durometer. A lot that is 0.010 in thin will leak.
  • Match the septum to the needle guide. If your autosampler has a needle guide that contacts the septum, a molded-center septum seats better than a flat one.
  • Keep a torque wrench at the bench. A $40 torque wrench saves $400 in re-runs.

Troubleshooting checklist

When a septum-related problem shows up, work through this list in order. It is ordered by frequency, not by severity.

  1. Is the blank dirty? Run a blank with a fresh septum and a fresh vial. If the ghost peak disappears, the septum was leaching. Switch to pre-extracted PTFE-faced.
  2. Is there a particle in the vial or on the needle? Inspect under a loupe. If yes, the septum is coring. Check needle gauge, centering, and needle condition.
  3. Is the peak area drifting up over a long sequence? Weigh a vial before and after. If it lost mass, the septum is not resealing. Check torque and durometer.
  4. Is the backpressure rising gradually? Particulate shedding. Switch to PTFE-faced and add an inline filter.
  5. Is the septum visibly swollen or pushed out of the cap? Solvent incompatibility. Switch to PTFE-faced or laminated.
  6. Is the failure happening only in one tray position? That is a mechanical problem (needle guide, vial seat, cap), not a septum problem. Swap the tray position and see if the failure follows.
  7. Is the failure happening across a whole lot? Check the lot certificate. If the durometer or thickness is out of spec, reject the lot.

For method-level confirmation, the analytical testing resources cover the blank and system suitability checks that will catch a septum problem before it reaches a reportable result. And if you are validating a new method against a regulatory framework, the U.S. EPA SW-846 test methods are a good reference for the blank and holding-time requirements that septa directly affect.

What I recommend

Standardize on a pre-extracted PTFE-faced silicone septum, 11 mm diameter, 45 to 55 Shore A core, 0.010 inch PTFE film, 0.125 inch total thickness, for all LC-MS and UV work. Pair it with a 0.5 N·m torque setting on a calibrated cap wrench, a 23 gauge needle, and a 100-injection replacement interval. For GC headspace, switch to a PTFE/silicone laminated septum and follow the instrument maker’s bleed specification. For aggressive solvents (DCM, THF, toluene), the PTFE-faced design is not optional.

If you are sourcing septa across multiple instruments and formats, start from the HPLC septa catalog to match diameter and thickness to your cap format, and cross-check the needle and column compatibility against the Agilent chromatography consumables and Thermo Fisher sample handling references before you commit to a lot. A septum is a $0.30 part that can cost you a $3,000 column and a week of re-runs if you get it wrong.

Takeaway

Septum failures are almost always specification failures, not manufacturing defects. Match the durometer to the needle gauge, the face material to the solvent, the thickness to the cap, and the torque to the vial thread, and you will get 100+ clean injections per septum instead of 30. The three numbers worth writing on the bench: 0.5 N·m torque, 23 gauge needle, 100-injection change interval. Everything else follows from those.

References & further reading

Related resources from our lab consumables network:

External references and standards: