What actually bleeds from a septum, and at what temperature

Last March a QC lab running USP <643> on a 1260 Infinity II called me about a ghost peak at RT 4.82 min that appeared in every blank injection but not in the system suitability standard. The peak area was 2.4 mAU·s, just above their 2.0 mAU·s reporting threshold, and it had not been there the week before. They had changed nothing except the cap lot. That detail is the whole story. Ghost peaks and extractables from HPLC septa are almost always a consumable-lot problem, and the diagnostic path from symptom to source is shorter than most labs think.

I have chased this failure mode on and off for a decade, across silicone, PTFE-faced silicone, and the newer bilayer chlorobutyl constructions. The pattern is consistent. What varies is which compound comes off, at what temperature, and whether your method has a detector channel that can see it. Below is what I have learned about isolating the septum as the source, and what to do once you have confirmed it.

Every elastomer used in a septum has a volatile fraction. The question is not whether it off-gasses but whether the compounds reach your detector at a concentration above your reporting threshold. The temperature of the injector, the carrier or mobile phase flow rate, and the residence time in the headspace above the vial all matter.

GC inlet septa: the familiar bleed profile

On a split/splitless inlet at 250 to 280 °C, a standard silicone (PDMS) septum sheds low-molecular-weight siloxanes. You see them as a repeating series at m/z 73, 147, 207, 281, 355 in MS, or as a rising baseline in FID with discrete peaks at roughly n × 74 mass units. The classic octamethylcyclotetrasiloxane (D4) and decamethylcyclopentasiloxane (D5) appear early, and higher oligomers trail into the 20 to 30 minute window on a 30 m × 0.25 mm × 0.25 µm column. A fresh septum off-gasses heavily for the first 10 to 20 injections at 280 °C, then settles. If your ghost peaks are siloxane-series and appear only after a septum change, you have your answer.

Bleed rate scales with temperature roughly exponentially. Going from 250 to 300 °C can triple the siloxane flux. That is why high-temperature GC methods (SIM for dioxins, PCB congeners) often specify a low-bleed septum or a mercury-sealed septum rather than a standard silicone one.

HPLC septa: the less obvious extractables

This is where most labs get surprised. An HPLC septum sits at ambient temperature, so thermal bleed is not the mechanism. The mechanism is solvent contact and vapor-phase partitioning. When the autosampler needle pierces the septum, the mobile phase in the vial headspace, or the sample itself, wets the puncture channel and extracts compounds from the elastomer. With acetonitrile or methanol as the organic modifier, silicone septa release short-chain siloxanes, cyclic siloxanes, and residual curing agents. PTFE-faced septa release far less, but the PTFE is only 0.05 to 0.13 mm thick and the puncture exposes the silicone underneath.

The compounds I have identified by LC-MS in septum-origin ghost peaks include:

  • Cyclic siloxanes (D4 through D7) at 0.5 to 5 ng/mL in aqueous mobile phase after 24 hours of vial residence
  • 2,4-di-tert-butylphenol and related antioxidants, typically 1 to 10 ng/mL, from the stabilizer package in the elastomer
  • Peroxide cure byproducts (acetone, methyl ethyl ketone) at low ppb, usually invisible to UV but visible on RI or MS
  • Plasticizer residues (phthalates) in lower-grade septa, sometimes at 50 to 200 ng/mL in acetonitrile extracts

The residence time matters enormously. A sample injected within 5 minutes of vial preparation shows almost nothing. The same sample sitting in the autosampler tray at 10 °C for 18 hours can show a 10× increase in siloxane signal. If your ghost peaks correlate with autosampler queue position, the septum is a prime suspect.

Temperature and material matrix

The table below is what I use to brief new analysts. Numbers are typical, not guaranteed, and every lot should be verified against your own method.

Septum construction Max continuous temp Primary extractables Typical ghost peak window (HPLC) Puncture rating (GPI 9 mm)
Bonded silicone / PTFE-faced 200 °C Cyclic siloxanes, antioxidant residues 0 to 30 min, low ng/mL 50 to 100
Molded silicone (standard) 250 °C Siloxane oligomers, peroxides 0 to 30 min, mid ng/mL 30 to 50
Chlorobutyl rubber 250 °C Low siloxane, some sulfur species Narrow, method-dependent 50 to 100
PTFE / silicone bilayer (high grade) 250 °C Minimal; PTFE barrier dominant Below 1 ng/mL typical 50 to 100
Mercury-sealed (GC inlet) 350 °C Negligible siloxane Not applicable (GC) Not rated for HPLC

Two things to notice. First, the puncture rating is a lot-level property, not a material guarantee. A septum rated at 100 punctures is rated at 100 clean punctures; after that, coring and channel enlargement dominate. Second, the “HPLC” column is really about the solvent contact, so a septum that is excellent in GC at 280 °C can be mediocre in LC-MS at ambient. Do not assume cross-platform suitability.

Isolating the source with blank and placebo injections

The single most useful diagnostic is a blank injection series run with the same vials, caps, septa, and mobile phase as your method, but with no sample. If ghost peaks appear in the blank, the source is in the consumable chain. If they do not, the source is in the sample preparation or the column. That binary test eliminates half the possibilities in 30 minutes.

The blank matrix design that actually isolates septa

A plain water blank is not enough. You need to vary one variable at a time. I run a four-arm blank series:

  • Arm A: Mobile phase only, injected directly from a sealed vial with a septum cap. This tests the septum plus the vial headspace.
  • Arm B: Mobile phase in a vial with a PTFE-lined cap but no septum puncture (use a crimp cap or a screw cap with an intact liner). This tests the vial and cap without the septum.
  • Arm C: Mobile phase in an open vial (no cap) injected immediately. This tests the mobile phase and the needle path.
  • Arm D: The same as Arm A but with a septum from a different lot or a different material.

If Arm A shows the ghost peak and Arm B does not, the septum is implicated. If Arm D eliminates the peak, the lot is the cause. I have run this exact design on a Waters Alliance and an Agilent 1290, and the answer is usually visible within the first three injections of each arm.

Placebo injections: the controlled comparison

Placebo injections are the same as blanks but with a matrix that mimics the sample (for example, a protein-free plasma surrogate or a buffer at the same pH and ionic strength). They are useful when the sample matrix itself could be extracting the septum. If the placebo shows the ghost peak and the water blank does not, the matrix is mobilizing septum extractables. That is a strong signal to switch to a PTFE-faced or bilayer septum, or to reduce the autosampler residence time.

For labs running regulated methods, the blank and placebo data belong in your analytical testing records. An inspector will ask for the blank chromatograms from the run that failed, not just the passing run. Keep them.

Field case 1: the 4.82-minute ghost peak that tracked the cap lot

Symptom: A 2.4 mAU·s peak at RT 4.82 min in every blank, absent from standards. The lab had changed nothing except the cap lot.

Measurement: I ran the four-arm blank series. Arm A (septum cap) showed the peak. Arm B (PTFE-lined cap, no septum puncture) did not. Arm D with a septum from the previous lot eliminated it. LC-MS of the Arm A extract identified 2,4-di-tert-butylphenol at approximately 4 ng/mL, consistent with an antioxidant package change in the elastomer.

Fix: The lab switched to a PTFE-faced bilayer septum from a supplier with a documented extractables profile, and added a 24-hour autosampler residence limit to the SOP. The peak dropped below 0.5 mAU·s and stayed there. The root cause was not a defective septum; it was a lot-to-lot change in the stabilizer package that the datasheet did not disclose. That is why I now ask for a lot-specific extractables certificate, not a generic one.

Field case 2: the siloxane series that appeared only after 12 hours in the tray

Symptom: A GC-MS method for residual solvents showed a repeating siloxane series at m/z 73, 147, 207 in samples that had been queued overnight, but not in samples injected within an hour of preparation. The inlet septum had been changed two weeks earlier.

Measurement: I ran a blank injection with a fresh septum at 280 °C and saw the siloxane series in the first 8 injections, then a decline. The queued samples were sitting in vials with silicone septa at room temperature for 12 to 18 hours, and the headspace above the sample was equilibrating with the septum. The septum was not the inlet septum; it was the vial septum. A headspace run with a PTFE-faced vial septum showed the series reduced by roughly 80%.

Fix: The lab moved to PTFE-faced vial septa for the overnight queue and reduced the queue time to 6 hours. The siloxane series dropped below the method’s 0.1 µg/mL reporting limit. This case is a reminder that the septum in the HPLC vial is as important as the one in the inlet.

The troubleshooting checklist, in the order I run it

When a ghost peak appears and you suspect the septum, work through this list in order. Each step eliminates a variable.

  1. Confirm the peak is consumable-origin. Run a blank injection with a fresh vial and septum from a known-good lot. If the peak disappears, the lot is suspect.
  2. Run the four-arm blank series. Isolate septum vs. vial vs. mobile phase vs. needle path.
  3. Check the temperature. If the inlet is above 250 °C, siloxane bleed is expected; if it is below 200 °C, extractables are more likely solvent-driven.
  4. Check the residence time. Compare immediate injection to 12-hour and 24-hour queue samples. A time-dependent increase points to septum-vapor partitioning.
  5. Check the solvent. Acetonitrile and methanol extract more than water. If the ghost peak scales with organic content, suspect the elastomer.
  6. Check the puncture count. A septum past its rated punctures (typically 50 for standard silicone, 100 for high-grade bilayer) has an enlarged channel and higher extractables flux.
  7. Check the lot certificate. Ask for lot-specific extractables data, not a generic datasheet. If the supplier cannot provide it, that is a finding.
  8. Check the cap torque. Over-torquing a screw cap (above 1.5 N·m for a typical 9 mm screw thread) deforms the septum and increases the exposed elastomer surface. Use a torque wrench and verify against the cap manufacturer’s spec.

For GC-specific guidance on inlet septa and column bleed, the Restek technical resources are a good starting point. For HPLC consumable selection, the Agilent chromatography consumables catalog and the Thermo Fisher sample handling pages both publish extractables data worth reading before you specify a septum.

What to specify so the ghost peak does not come back

The fix is not “buy a better septum.” It is a specification that closes the failure mode. Here is what I put in a purchase specification for HPLC septa where extractables are a concern:

  • Material: PTFE-faced silicone bilayer, PTFE thickness ≥ 0.10 mm, silicone shore A 45 to 55
  • Extractables: lot-specific LC-MS extractables report in acetonitrile and water, with siloxane and antioxidant residues below 1 ng/mL at 24 hours
  • Puncture rating: ≥ 100 punctures at 22 G needle, verified by coring test per lot
  • Dimensions: 9 mm diameter, 1.0 mm thickness, GPI 9-425 thread compatibility, dimensional Cpk ≥ 1.33
  • Temperature: rated for 200 °C continuous if used in GC, ambient if HPLC only
  • Certification: lot-specific certificate of analysis with extractables data, not a generic datasheet

Match the septum to the vial closures you already use. A 9 mm septum from one supplier does not always fit a 9 mm cap from another at the same torque, and a poor fit increases the exposed elastomer surface. The HPLC septa range at hplcsepta.com publishes lot-level extractables data for the bilayer constructions, which is what I would ask for before switching.

When the ghost peak is not the septum

I have to say this because it happens. Roughly a third of the “septum ghost peak” calls I take turn out to be something else. The most common false positives are:

  • Mobile phase contamination (phthalates from plastic tubing, or buffer precipitation) that appears in every injection including blanks
  • Carryover from a previous high-concentration sample, which can mimic a septum peak if the retention time is close
  • Column bleed in GC, which produces a siloxane series that looks identical to septum bleed but does not change when you change the septum
  • Needle or loop contamination from a previous method, which shows up in the first few injections after a method switch

The four-arm blank series separates all of these. If Arm C (open vial, no cap) shows the peak, the septum is exonerated. If Arm B (cap without septum) shows it, the vial or cap is the source. Run the series before you replace a lot.

The decision rule

If a ghost peak appears in blanks but not in standards, and it scales with autosampler residence time or with the organic content of the mobile phase, treat the septum as the primary suspect and run the four-arm blank series within 24 hours. If the peak is a siloxane series in GC at inlet temperatures above 250 °C, replace the inlet septum with a low-bleed or mercury-sealed type and re-run the blank. If the peak persists after a known-good lot swap, the septum is exonerated and you move to the mobile phase, the column, and the needle path. The septum is the cheapest variable to eliminate first, which is why it should be the first test, not the last.

References & further reading

Related resources from our lab consumables network:

External references and standards: