A 40 mL VOA vial with a 20 mm PTFE-faced silicone septum, run hot at 60 °C on an autosampler, gave me a 4.7-minute ghost peak at 0.9 ppb on a 8260 list, in every blank, for three consecutive lots. The instrument was clean. The column was new. The purge-and-trap trap had been baked that morning. The peak was coming out of the septum, and it had been there since the first injection. That is the failure mode this article is about: ghost peaks and extractables that originate from the voa vial itself, not from the sample, the plumbing, or the column.
If you already know what a voa vial is, you do not need the definition. What you need is a way to tell which material is bleeding, at what temperature it starts to matter, and how to prove it with two injections instead of two weeks of guessing. The primary search intent here is voa vial ghost peaks, extractables and contamination, and the useful answer is material-specific and temperature-specific. I will give you both, plus the blank and placebo protocol that isolates the source in under an hour.
When I trace a contamination event back to the vial, it lands on one of four surfaces. Each has a signature: a temperature threshold, a compound class, and a characteristic retention time window. Learn the signatures and you stop replacing whole lots.
Septum bleed: the silicone and the antioxidant package
A standard 20 mm silicone septum is a polydimethylsiloxane sheet with fillers and a cure system. Above about 40 °C, low-molecular-weight siloxanes start to migrate. The classic markers are hexamethyldisiloxane (HMDSO, boiling point 100 °C), octamethyltrisiloxane, and the cyclic D3 through D6 series. On a 60 m × 0.25 mm × 1.4 µm DB-624 column at 1.0 mL/min, HMDSO elutes early, often under 3 minutes, and D4 through D6 fall in the 5 to 9 minute window depending on ramp. On a mass spec they show as m/z 73, 147, 207, 281. If you see that ladder, the septum is the source, not the sample.
The second septum contributor is the antioxidant. Many silicone formulations carry a phenolic antioxidant that oxidizes to 2,6-di-tert-butyl-p-benzoquinone (BHT-quinone), which shows up as a distinct peak with m/z 220 and a fragment at 205. I have seen it at 0.5 to 2 ppb in blanks from septa stored above 30 °C for more than a year. The cure catalyst residues (tin carboxylates) are less volatile but can catalyze cross-reactions in reactive samples.
PTFE facing: usually clean, occasionally not
The PTFE face on a PTFE/silicone septum is the barrier that protects the sample from the silicone. Bonded PTFE is sintered, and a good bond is inert up to 200 °C continuous. What fails is the edge: the punched circumference where the PTFE is slit and the silicone is exposed. If the punch is dull, the PTFE delaminates in a ring and the silicone edge is now in contact with headspace. I have measured that as a 3 to 5× increase in siloxane response on the same lot versus a sharp-punch lot. The fix is not the material; it is the tooling.
Glass: sodium, boron, and the silanol layer
Type I borosilicate glass (USP voa vial grade, ASTM E438 Type I) is the correct material, but glass is not inert to everything. At pH above 8 or below 4, the silica network hydrolyzes and releases sodium and boron. In a VOA application the more relevant issue is the silanol surface: it adsorbs and later releases polar analytes. For 8260 compounds this is usually a loss, not a ghost, but for amines and phenols the carryover can appear as a ghost peak on the next injection. A deactivated glass surface (silane-treated) reduces this. Untreated glass does not.
Labels, adhesives, and the closure coating
This is the one people forget. A paper label with a hot-melt adhesive, applied to the outside of a 40 mL vial and then heated to 60 °C in a headspace oven, off-gasses. The markers are typically toluene, xylenes, and a styrene or terpene trace from the adhesive. I have seen a 1.2 ppb toluene ghost that disappeared when the labels were moved to the shipping carton. The same applies to the silicone-based release coating on some caps and to the ink on printed septa.
Temperature is the variable that decides which material bleeds first
Extractables are a kinetic problem. Every material has a temperature below which its contribution is under the detection limit for your method, and above which it is not. The table below is what I use as a first-pass screen. The numbers are approximate and method-dependent, but the ordering is reliable.
| Material / component | Onset temperature where bleed becomes method-relevant | Marker compounds | Typical blank level at 60 °C |
|---|---|---|---|
| Silicone septum (unfaced) | ~35–40 °C | HMDSO, D3–D6 siloxanes | 2–10 ppb total siloxane |
| PTFE/silicone septum (bonded) | ~60 °C at the punched edge | Siloxanes from exposed edge only | 0.2–1 ppb if punch is sharp |
| Phenolic antioxidant in silicone | ~50 °C | BHT-quinone (m/z 220) | 0.5–2 ppb after 12 months storage |
| Borosilicate glass, untreated | pH-dependent, not temperature | Na, B, silanol-adsorbed polar species | Carryover, not steady ghost |
| Label adhesive (hot-melt) | ~45 °C | Toluene, xylenes, terpenes | 0.5–3 ppb toluene |
| Cap liner / release coating | ~70 °C | Silicone oils, fatty acid esters | Method-dependent |
Two operational consequences follow. First, if your method heats to 60 °C or above, an unfaced silicone septum is not a candidate, regardless of price. Use a PTFE-faced septum and verify the punch quality. Second, if you run at 40 °C or below, the septum is rarely the problem and you should look at the label and the cap before you replace the lot. The septa you buy should come with a lot-specific bleed statement; if the supplier cannot tell you the antioxidant package, you cannot predict the BHT-quinone risk.
Isolating the source with blank and placebo injections
The protocol is simple and it is the only reliable way to separate vial-origin ghosts from instrument-origin ghosts. You need two injections and a clean reference. Do them in this order.
Step 1: the system blank with no vial
Run the method with a sealed, empty sample loop or a direct injection bypassing the vial entirely. If the ghost appears here, it is the instrument, the trap, the transfer line, or the column, and this article does not apply. If it does not appear, the source is somewhere in the vial-closure-sample system. Record the baseline for the retention window of interest.
Step 2: the vial blank with reagent water only
Fill a vial from the suspect lot with HPLC-grade water (or the matrix blank your method specifies), cap it with the suspect septum and cap, and run it. Compare against a vial from a known-good lot, same water, same closure. The difference between the two is the vial contribution. If the suspect lot shows the siloxane ladder and the good lot does not, you have localized the problem to the septum or the cap, not the glass.
Step 3: the placebo injection that separates septum from sample
This is the step most labs skip. Prepare two vials from the suspect lot: one with water, one with water plus a known concentration of an internal standard that is not expected to react with the septum. Run both. If the ghost peak is identical in both, it is a material bleed. If it scales with the internal standard, the septum is reacting with your sample, and the fix is a different closure material, not a different lot. This distinction saves a lot of unnecessary supplier complaints.
Step 4: the temperature ramp test
Take one vial from the suspect lot, run it at 40 °C, 50 °C, and 60 °C. If the ghost area grows nonlinearly (typically it doubles between 50 and 60 °C), you have a thermal desorption source, which is almost always the septum or the adhesive. If it is flat across temperature, suspect a non-volatile contaminant introduced during manufacturing, such as a mold release agent or a rinse residue.
Two field cases with the measurement that found the source
Case 1: the 0.9 ppb peak at 4.7 minutes
The symptom was a single ghost peak at 4.7 minutes, 0.9 ppb as toluene-equivalent, in every blank from one lot of 40 mL vials. The instrument blank was clean. The water blank from a good lot was clean. The suspect-lot water blank reproduced the peak. A temperature ramp showed the peak area going from 0.2 ppb at 40 °C to 0.9 ppb at 60 °C, a 4.5× rise, which pointed to a thermal desorption source. The MS scan gave m/z 91 and 106, consistent with xylene, not siloxane. The vials in that lot had been labeled with a hot-melt adhesive that was applied before the final rinse, and the adhesive was on the outside but the vials had been stored in a warm room at 32 °C for eight weeks. The fix was to move labeling to after the QC release and store at 20 to 25 °C. The peak disappeared on the next lot.
Case 2: the siloxane ladder that only appeared on the third injection
The symptom was a clean first injection, a small siloxane peak on the second, and a full HMDSO/D4/D5 ladder by the third, on the same vial. That pattern is a puncture problem, not a material problem. Each needle puncture opens the PTFE face slightly, and by the third puncture the silicone is exposed to headspace. The measurement was a puncture count: the septum was rated for two punctures at 20 mm, and the method was doing four. The fix was to switch to a pre-slit septum rated for the puncture count the method actually uses, and to verify the needle gauge (a 23-gauge needle at 0.64 mm OD opens a smaller hole than a 21-gauge at 0.80 mm). The vial septa selection should be driven by puncture count and needle gauge, not by the catalog’s default.
Storage, handling, and the torque value that prevents re-contamination
Most extractables problems I have investigated were created after the vial left the supplier. Three controls matter.
- Storage temperature: keep sealed vials at 15 to 25 °C, out of direct light, and away from solvent cabinets. A septum stored at 35 °C for a year will show BHT-quinone even if it was clean at release.
- Torque: cap to 1.5 to 2.0 N·m for a 20 mm closure. Under-torque lets headspace escape and pulls lab air in; over-torque deforms the PTFE face and creates a leak path at the crimp. Use a torque tester, not feel.
- Rinse protocol: if you rinse vials in-house, use HPLC-grade water and dry at ≤ 40 °C. Drying at 80 °C bakes the adhesive and the antioxidant into the septum and you will see it on the first injection.
For the closure dimensions and the crimp geometry, the GPI 20 mm finish (GPI 400 thread equivalent) is the reference. If your capper is set to a different finish, the seal is compromised before the sample is even added, and that is a mechanical fit issue covered elsewhere on this site, so I will not expand on it here.
A troubleshooting checklist you can run in one shift
Work top to bottom. Stop at the first step that reproduces the ghost.
- Instrument blank, no vial. Ghost present? Instrument problem. Stop.
- Vial blank, known-good lot, water only. Ghost present? Method or water problem. Stop.
- Vial blank, suspect lot, water only. Ghost present and absent in good lot? Vial material problem. Continue.
- Temperature ramp at 40, 50, 60 °C. Nonlinear growth? Thermal desorption (septum or adhesive).
- MS scan of the ghost. m/z 73/147/207/281? Silicone. m/z 91/106? Aromatic adhesive. m/z 220/205? Antioxidant.
- Puncture count test. Ghost grows with puncture number? Septum rating too low or needle too large.
- Label check. Peel the label and rerun. Ghost disappears? Adhesive source.
- Torque check. Measure with a torque tester. Outside 1.5 to 2.0 N·m? Re-cap and rerun.
If you get through all eight and the ghost is still there, the problem is upstream of the vial and you should look at the transfer line and the trap. The Restek technical library has good reference chromatograms for siloxane and antioxidant bleed that you can compare against your own scans.
What to specify so the ghost does not come back
The purchasing decision that prevents this class of problem is a specification, not a brand. When you write the spec for a voa vial, put these on the certificate:
- Septum material and facing: PTFE-faced silicone, bonded (not laminated), with the PTFE thickness stated (typically 0.010 in / 0.25 mm).
- Bleed statement: lot-specific, at the method temperature, with the siloxane and antioxidant markers quantified. If the supplier cannot provide it, that is the answer.
- Puncture rating: number of punctures at the specified needle gauge before the PTFE face is breached. Ask for the test method.
- Glass: ASTM E438 Type I, USP <660> compliant, with the hydrolytic resistance class stated.
- Label and adhesive: low-outgassing adhesive, or no label on the vial at all. This is a one-line spec change that eliminates a whole failure mode.
The Agilent consumables reference and the Thermo Fisher chromatography sample handling pages both publish bleed data you can use as a benchmark when a supplier’s certificate looks too clean. For the regulatory side, EPA SW-846 sets the container requirements for VOA methods, and the HPLC vial guidance on this network covers the liquid-injection side of the same material question.
The decision rule
If a ghost peak reproduces in a water blank from the suspect lot and not from a known-good lot, and its area grows nonlinearly between 50 and 60 °C, treat it as a septum or adhesive extractable and change the closure specification before you change the supplier. If it is flat across temperature, treat it as a manufacturing residue and reject the lot. If it grows with puncture number, the septum rating is wrong for your needle, and the fix is a pre-slit septum with a stated puncture count. Run the eight-step checklist once per lot change, keep the scans, and the next ghost peak will take an hour to identify instead of a month.
References & further reading
Related resources from our lab consumables network:
External references and standards: