Where the sample actually leaves the vial--HPLC Vials Supplier

A 96-hour method run on a 6890 GC lost 11.4% of its internal standard peak area between injection 40 and injection 380. The autosampler was fine. The inlet was fine. The 2 mL crimp vials, drawn from a lot we had used for six months, were the problem: the septa had a compression set of 0.31 mm, versus 0.14 mm on the previous lot. Same catalog number, same supplier name on the box, different molding tool. That is the shape of most seal failure and evaporation loss cases I have investigated in twenty years of running a consumables operation, and it is why the phrase “vial supplier seal failure, evaporation and sample loss” belongs in the same sentence. The vial is a pressure vessel with a lid you puncture on purpose. Every supplier choice you make about that lid is a decision about how much of your sample is still there at the end of the run.

Evaporation loss is not one mechanism. It is at least four, and each has a different fingerprint on a gravimetric curve. Before you chase the supplier, you need to know which one you have.

Leak path 1: the septum-to-glass interface

This is the seal between the underside of the septum and the land of the vial neck. It is a face seal, held by crimp or screw torque. It fails when the land is out of flatness, when the septum is too hard to conform, or when the crimp is uneven. The signature is a steady linear loss from the moment of capping, before any puncture. If your gravimetric test shows loss on unpunctured controls, stop here. That is a capping problem, not a puncture problem.

Leak path 2: the cap-to-septum interface

On a crimp vial, the aluminum skirt compresses the septum against the glass. On a screw vial, the cap’s internal shoulder does the compressing. Either way, the septum has to be squeezed enough to seal and not so much that it is destroyed. A 11 mm crimp cap with a 1.0 mm PTFE/silicone septum needs roughly 0.7 to 0.9 mm of compression to seal reliably. If the tool is worn, if the cap skirt is thin, or if the septum is undersized, you lose that compression and the seal is marginal from day one.

Leak path 3: the puncture itself

A 23-gauge needle is about 0.64 mm outside diameter. It punches a hole that does not close. Silicone has enough elasticity to close most of a 0.64 mm puncture, but not all of it, and not forever. The re-seal is the part most suppliers will not talk about, and it is the part that decides whether your 96-hour run survives.

Leak path 4: permeation through the septum body

This is the slow one. PTFE has excellent chemical resistance but poor re-seal memory. Silicone has excellent re-seal but moderate permeation. A laminated PTFE/silicone septum uses PTFE on the sample side for chemical inertness and silicone on the cap side for re-seal. The laminate bond is where cheap septa fail: the PTFE layer delaminates after a few punctures and the re-seal drops off a cliff. If you are running volatile analytes in a headspace or long GC run, permeation through a thin silicone layer is a real loss path, not a theoretical one.

Re-seal behaviour after puncture: what the number should be

I test re-seal the same way every time. Cap 20 vials with the septum under test, fill each with 1.5 mL of a volatile solvent (methanol works; it is aggressive enough to show loss fast and cheap enough to throw away). Weigh to 0.1 mg. Puncture each vial 1, 3, 5, 10, and 20 times with a 23-gauge needle, holding the needle in place for two seconds per puncture to simulate a real injection. Weigh again after 24 hours at 25 °C. The percentage loss is your re-seal number.

Here is what I expect from a septum that is fit for a 24-hour run, and what I expect from the ones that are not:

Puncture count	Acceptable loss (24 h, 25 °C, methanol)	Marginal (investigate)	Fail (reject lot)
1	< 0.5%	0.5 to 1.0%	> 1.0%
3	< 1.0%	1.0 to 2.0%	> 2.0%
5	< 1.5%	1.5 to 3.0%	> 3.0%
10	< 3.0%	3.0 to 5.0%	> 5.0%
20	< 6.0%	6.0 to 10.0%	> 10.0%

Those numbers are for a 1.0 mm PTFE/silicone septum with a crimp cap. Screw caps with a 1.5 mm silicone septum will do better on re-seal (thicker material, more elastic) but worse on permeation of very volatile analytes. There is no free lunch. You pick the septum for the analyte and the run length, not for the catalog number.

Compression set is the number your supplier will not give you

Compression set is the permanent deformation a septum takes after being squeezed. You measure it by compressing a sample of septum to 25% of its thickness at 70 °C for 22 hours, releasing it, and measuring how much it fails to spring back. ASTM D395 covers the method. A good silicone septum comes back to within 15 to 20% of its original thickness. A cheap one comes back to 30 to 40% and stays there.

Why it matters on a vial: the septum is compressed by the cap and stays compressed for the life of the vial. If it takes a set, the compression force drops, the seal at the glass land goes marginal, and the vial leaks. It also means the septum cannot re-seal a puncture properly, because the material has already lost its elastic memory. I have seen lots where the compression set was 0.28 mm on a 1.0 mm septum. That is 28% set before the vial has even been used. No amount of careful capping will fix that.

Ask your vial supplier for the compression set data on the septum lot. If they cannot produce it, that is your answer. A supplier who molds septa in-house will have the data. A distributor who buys from a broker will not. The vial supplier you choose should be able to trace the septum back to a molding lot, not just a catalog number.

How to measure loss gravimetrically over a real run

You do not need a research budget to do this. You need a balance that reads to 0.1 mg, a timer, and a way to keep the vials at run temperature. Here is the protocol I use.

Step 1: prepare the controls

Fill 10 vials with 1.5 mL of your actual sample matrix (or a representative volatile solvent if the matrix is too precious). Cap them with the lot you are testing. Weigh each to 0.1 mg. Record the tare weight of the empty vial plus cap before filling so you can back-calculate fill volume.

Step 2: puncture on a schedule

Puncture 5 of the vials once per hour for 8 hours, simulating an autosampler sequence. Leave the other 5 unpunctured as a capping control. This separates capping loss from puncture loss. If the unpunctured controls lose more than 0.5% over 24 hours, your capping is the problem and the puncture test is not telling you anything yet.

Step 3: weigh at intervals

Weigh at 0, 2, 8, 24, and 72 hours. Keep the vials at the temperature your autosampler tray actually runs at. If your tray is at 4 °C, test at 4 °C. If it is ambient, test at ambient. Do not test at 25 °C and assume the tray is the same, because most trays are 2 to 4 °C warmer than the room due to the instrument’s own heat.

Step 4: calculate loss as a percentage of fill mass, not peak area

This is the part people get wrong. They watch the internal standard peak area drop and call it evaporation. Peak area drop is the symptom. Mass loss is the measurement. If you lose 2% of the fill mass over 24 hours and the internal standard area drops 2%, you have a seal problem. If the mass holds steady and the area drops, you have an injection or detector problem, and the vial is not your suspect. The gravimetric number is the one that tells you whether to call the supplier.

For a 96-hour run, I want the cumulative loss under 3% on the punctured vials and under 1% on the controls. Anything above that and the run is at risk of failing its system suitability check on the last injections. If you are running EPA SW-846 methods under SW-846, the calibration criteria are tight enough that a 3% loss will show up as a failed continuing calibration verification. The vial is the cheapest thing to fix and the last thing people check.

Field case 1: the lot that drifted at hour 60

Symptom: A contract lab running a 72-hour pesticide residue method on a Thermo Trace 1310 with a TriPlus RSH autosampler saw the internal standard area drop 8% between injection 50 and injection 400. The method had passed system suitability on day one.

Measurement: Gravimetric test on the current lot showed 4.1% loss on punctured vials at 24 hours, versus 1.2% on the previous lot. Compression set on the current lot was 0.26 mm versus 0.11 mm on the previous lot. The re-seal at 10 punctures was 5.8% versus 2.1%.

Fix: The lab had switched to a lower-cost supplier six weeks earlier. The septa were a 1.0 mm PTFE/silicone with a thinner silicone layer (0.6 mm instead of 0.9 mm), which explained the compression set and the re-seal. Switching back to the original spec, and adding a 10-puncture re-seal check to incoming inspection, fixed it. The lesson: the catalog number was the same. The silicone layer thickness was not.

Field case 2: the crimper that was not the problem

Symptom: A pharma QC lab running USP <621> residual solvents on an Agilent 7890 with a 7693 autosampler saw the headspace vial septa leaking at the crimp. They blamed the crimper and bought a new one. The leak persisted.

Measurement: Gravimetric test on unpunctured controls showed 2.8% loss over 24 hours. That is a capping loss, not a puncture loss. We measured the crimp diameter on 30 vials with a digital caliper. The spec for an 11 mm crimp is 10.9 to 11.1 mm after crimping. The lab’s crimper was producing 11.4 mm, which is over-crimped. But the root cause was the vial: the neck land on the new lot was 0.15 mm lower than the previous lot, so the same crimper setting produced a different compression. The supplier had changed the glass mold.

Fix: The lab switched to a supplier who could supply the neck land dimension on the certificate of analysis, and set the crimper to 11.0 mm on that lot. Loss dropped to 0.4%. The crimper was never the problem. The glass was.

A troubleshooting checklist for seal integrity

Work through these in order. Do not skip to the supplier call until you have the numbers.

Unpunctured controls: loss > 1% over 24 h means capping or land flatness, not puncture. Check crimp diameter (10.9 to 11.1 mm for 11 mm crimp) and neck land height.
Punctured vials: loss > 3% over 24 h at 10 punctures means re-seal failure. Check compression set and silicone layer thickness.
Septum durometer: a 45 to 55 Shore A silicone re-seals well. Above 60 Shore A it is too hard to conform to the land. Below 40 it is too soft and extrudes under crimp.
Septum thickness: 1.0 mm is standard for 11 mm crimp. 1.5 mm gives better re-seal but needs a deeper cap and will not fit all autosampler trays.
Temperature: silicone septa are rated to 200 °C continuous, but re-seal performance drops above 80 °C. If your tray runs hot, test at tray temperature, not at 25 °C.
Needle gauge: 23-gauge is standard. A 21-gauge needle (0.51 mm ID, 0.82 mm OD) makes a bigger hole and will roughly double the re-seal loss. If your method uses a 21-gauge, budget for it.
Lot traceability: ask for the molding lot on the septum, not just the vial lot. They are often different.
Permeation check: for very volatile analytes (dichloromethane, MTBE), run a 72-hour blank and look for the analyte in the blank. If it is there, the septum is permeating, not leaking.

If you are running HPLC rather than GC, the same seal logic applies to your autosampler vials, and the mobile phase evaporation is the loss path. A chromatography method with a volatile mobile phase (acetonitrile, methanol) will show drift on a long run if the vial seal is marginal. The fix is the same: measure the mass loss, then talk to the supplier with the number in hand. For sample prep upstream of the vial, a filter that sheds particles into the vial will make the seal look worse than it is, because the particles sit on the land and prevent a flat seal. Use a PTFE membrane filter for solvent compatibility and check the filtrate for particulate before you blame the septum.

What to put in the purchase specification

This is the part that stops the problem from coming back. A vial purchase spec that only says “2 mL clear crimp vial, 11 mm” is not a spec. It is a wish. Here is what I put in mine, and what I expect a supplier to sign up to.

Septum material and construction: PTFE-faced silicone, 1.0 mm nominal, silicone layer 0.9 mm minimum. Specify the durometer (45 to 55 Shore A).
Compression set: ≤ 20% per ASTM D395, method B, 22 h at 70 °C.
Re-seal: ≤ 3% mass loss at 10 punctures with a 23-gauge needle, 24 h at 25 °C, methanol fill.
Neck land: flatness within 0.05 mm, land height within 0.1 mm of drawing.
Crimp diameter after capping: 10.9 to 11.1 mm for 11 mm crimp, verified on 30 vials per lot.
Lot traceability: septum molding lot and glass lot on the certificate of analysis, not just the fill lot.
Change notification: 90 days written notice before any change to septum compound, glass mold, or molding tool.

The change notification clause is the one that saves you. Most of the failures I have seen were not bad suppliers. They were good suppliers who changed a tool or a compound and did not tell anyone. The Agilent and Thermo Fisher consumables catalogs both publish septum specifications down to the durometer and thickness, which is a useful benchmark when a lower-cost supplier’s datasheet is vague. Restek publishes re-seal and compression set data on their septum lines, and it is worth comparing your supplier’s numbers against theirs before you sign a contract.

The decision rule

If the unpunctured control loses more than 1% in 24 hours, the problem is capping or glass, and you fix it with a crimp diameter check and a neck land measurement before you call anyone. If the punctured vials lose more than 3% at 10 punctures, the problem is the septum, and you fix it with a compression set number and a re-seal number from the supplier. If the supplier cannot produce both numbers on the lot you are buying, change supplier. The gravimetric test costs you half a day and a balance. A failed 96-hour run costs you the run, the re-injection, and the client. Measure the loss, put the number in the purchase spec, and hold the supplier to it.

References & further reading

Related resources from our lab consumables network:

External references and standards: