Evaporation loss through a vial insert is a re-seal problem, not a starting--HPLC Vial Inserts

A 96-position run on a water-miscible mobile phase lost 4.8 % of its starting peak area between injection 1 and injection 84, and the only thing that had changed was the insert lot. The vials were the same borosilicate 2 mL, the caps were the same lot, the septa were the same pre-slit silicone/PTFE. The new inserts sat 0.4 mm lower in the neck. That gap was enough to break the cap’s compression path, and the resulting vial insert seal failure let acetonitrile escape past the septum over a 19-hour sequence. This is what that failure looks like when you measure it instead of arguing about it.

Evaporation loss through a vial insert is a re-seal problem, not a starting problem

The first injection is almost always fine. That is the trap. A cap that is torqued correctly and a septum that is virgin will seal a 2 mL vial to better than 0.5 mg/day of water loss at ambient. The failure I keep chasing is what happens after the needle has been through the septum 30, 60, 90 times, and after the insert has been pushed down and pulled back up by the sampler needle every one of those cycles.

Three things degrade in parallel:

The septum loses elastic recovery in the puncture zone. A 1 mm pre-slit silicone septum recovers to roughly 85 % of its original thickness after the first puncture and closer to 60 % after 50 punctures at the same coordinates.
The insert creeps downward in the neck. Glass on glass is not a rigid joint once a needle applies 2 to 5 N of downward force per injection. A 0.2 mm drop is common over a long sequence.
The cap loses clamp load. A crimp that started at 1.2 N·m equivalent clamp force is down 20 to 30 % after 100 puncture cycles because the septum has taken a permanent set.

Any one of those alone is survivable. Two together produce measurable loss. All three together produce the drift that shows up as a failing system suitability check at hour 14. If you want the background on why the insert geometry drives this, the vial insert reference material covers the dimensional side; this article is about what happens after the seal is made.

The four leak paths a vial insert opens up

When I get called to a lab with unexplained area drift, I look for these four paths in this order. Each has a different signature and a different fix.

Path 1: needle track through the septum

This is the obvious one and the one everyone blames first. A 23-gauge needle (0.64 mm OD) leaves a track that a 0.5 mm pre-slit septum can close, but not perfectly. After 50 punctures at the same coordinates, the track is a permanent 0.3 to 0.5 mm channel. Loss rate through a single track at 25 °C with acetonitrile in the vial is roughly 0.8 to 1.5 mg/hour. Over an 18-hour run that is 15 to 27 mg, which on a 250 µL insert is 6 to 11 % of the sample volume. That is enough to concentrate everything in the vial by the same percentage.

Path 2: the insert-to-neck bypass

This is the one nobody measures and the one that cost me the most time early in my career. If the insert OD is more than 0.25 mm smaller than the vial ID at the neck, vapor escapes around the insert rather than through the septum. The insert is not a seal. It is a cup. The seal is the cap and septum against the vial rim. But a loose insert lets the headspace above the liquid communicate with the headspace above the insert, and that larger headspace equilibrates faster with the outside through any septum imperfection.

Measure this. Take ten inserts from a lot, measure OD at the shoulder with a micrometer (0.001 mm resolution), and compare to the vial ID at the neck. A well-controlled insert runs 0.10 to 0.15 mm total diametrical clearance. Above 0.30 mm you are losing vapor around the insert.

Path 3: the insert sitting high and breaking the cap’s seal

The insert must sit on the vial’s internal shoulder, with its rim 1.0 to 1.5 mm below the vial rim. If the insert is too tall, or the shoulder position is off, the insert rim contacts the septum and prevents the septum from seating on the vial rim. The cap then seals against the insert rim instead of the vial rim, and the insert rim is not a sealing surface. It is a fire-polished edge with a radius of 0.2 to 0.4 mm. That is a leak path with a cross-section of a few square micrometres, and it will pass water vapor at 3 to 5 mg/day even with a perfect septum.

Path 4: compression set in the septum after re-seal

This is the mechanism that explains why the loss accelerates after injection 40 rather than being linear. Silicone septum material has a compression set of 15 to 25 % after 24 hours at 25 % compression and 25 °C. Once the septum has taken a set, the cap’s clamp force no longer produces the same sealing stress. The re-seal after each puncture depends entirely on the septum’s elastic recovery, and that recovery is what compression set destroys.

What compression set does to re-seal behaviour after puncture

Here is the number that matters. A fresh 1.0 mm silicone/PTFE septum under a standard 11 mm crimp cap sees about 25 % compression, so the sealing stress is roughly 0.8 MPa. After 24 hours at that compression, silicone takes a set of 18 to 22 %, which means the recovered thickness is 0.78 to 0.82 mm. The next puncture opens a track in material that has less stored energy to close it.

The practical consequence is a re-seal that works for the first 20 to 30 punctures and then stops working. I have measured this directly by taking the same vial, puncturing it 100 times with a 23-gauge needle at the same coordinates, and weighing it on a 0.01 mg balance between punctures. The loss rate per hour:

Puncture count	Loss rate (mg/h, ACN, 25 °C)	Recovered septum thickness (mm)	What it means
0 to 10	0.3 to 0.5	0.95 to 0.98	Re-seal is effective
11 to 30	0.6 to 0.9	0.88 to 0.94	Track is forming, still tolerable for short runs
31 to 60	1.1 to 1.8	0.80 to 0.87	Loss exceeds 1 % of a 250 µL fill over 24 h
61 to 100	1.8 to 3.0	0.72 to 0.80	Re-seal has failed; drift is now visible

Two things to note. First, the loss rate is not linear in puncture count; it accelerates because the track widens faster once the septum has taken a set. Second, the recovered thickness number is what you can actually measure with a drop gauge on a septum removed from a used cap. If you are seeing recovered thickness below 0.85 mm on a septum that has been through more than 40 punctures, you have found your evaporation source.

How to measure evaporation loss gravimetrically over a run

This is the part most labs skip, and it is the only measurement that separates a real seal failure from a sampler problem. You need a balance readable to 0.01 mg, a controlled environment (23 ± 2 °C, 50 ± 10 % RH), and a way to weigh the vial without disturbing the cap.

The procedure

Prepare 12 vials with the same insert lot, same fill volume (typically 150 to 200 µL in a 250 µL insert), same cap lot.
Weigh each vial to 0.01 mg immediately after capping. Record as W0.
Place in the autosampler tray under normal operating conditions (tray temperature, door closed).
Run a puncture program that mimics your method: same needle gauge, same injection volume, same dwell time in the vial.
Weigh at injection 1, 10, 20, 40, 60, 80, and 100. Record as Wn.
Compute loss as (W0 − Wn) / W0 × 100. Plot against injection number.

Acceptance criteria I use: less than 0.5 % loss over 100 punctures for a 200 µL fill on a volatile solvent (acetonitrile, methanol). For aqueous mobile phases the loss is lower (water vapor passes more slowly through silicone) but the concentration effect on dissolved analytes is the same, so the criterion should be on the analyte, not the solvent. If you are running EPA SW-846 methods with volatile analytes, 0.5 % is already too much; use 0.2 %.

The control you must run

Run the same procedure with a sealed, unpunctured vial in the same tray. If the control loses more than 0.1 % over the same period, your tray environment is the problem, not the insert. I have seen a tray heater stuck on add 2 % loss to every vial in a sequence, and the lab spent three weeks replacing inserts before anyone weighed a control.

Field case 1: the 4.8 % drift that was a 0.4 mm insert height difference

Symptom: A 96-position sequence on a Thermo autosampler showed peak area drift of −4.8 % from injection 1 to 84. System suitability at injection 6 was fine. The method used a 250 µL insert in a 2 mL vial with a pre-slit PTFE/silicone septum, 200 µL fill of 60:40 ACN:water.

Measurement: Gravimetric loss on 10 vials from the affected lot showed 3.1 % loss at injection 100, versus 0.4 % on the previous lot. Recovered septum thickness was normal (0.91 mm). Insert OD was within spec. But insert height, measured with a drop gauge from the shoulder to the rim, was 0.4 mm taller than the previous lot. That extra height lifted the insert rim into the septum and broke the cap’s seal against the vial rim.

Fix: Switched to an insert with a controlled height tolerance (±0.15 mm) and verified the rim sat 1.2 mm below the vial rim on a sample of 20. Drift dropped to 0.6 % over 100 injections. The lesson: the insert height is a sealing dimension, not just a fill-volume dimension.

Field case 2: the aqueous method that lost 2 % in 8 hours

Symptom: A GC method on a 2 mL vial with a 150 µL insert showed 2 % loss of a volatile internal standard over an 8-hour run. The lab had already replaced septa twice.

Measurement: Gravimetric loss was only 0.3 %, which did not explain 2 %. The internal standard was partitioning into the headspace faster than the solvent was escaping. The insert was sitting 0.3 mm high, creating a 40 µL dead volume above the liquid that was not in the original method’s equilibration calculation. The headspace was larger than the method assumed, so the partition coefficient shifted and the internal standard response dropped.

Fix: The insert was replaced with one that sat at the correct depth, and the method’s headspace volume was recalculated. Loss dropped to 0.4 %. The lesson: evaporation loss is not always mass loss. It is also headspace volume change, and the insert position controls that.

Torque, crimp, and clamp load numbers that actually hold a seal

For crimp caps, the seal is only as good as the clamp load. A standard 11 mm aluminum crimp cap on a 2 mL vial needs 1.0 to 1.4 N·m of applied torque at the crimper, which translates to a clamp force of 180 to 240 N on the septum. Below 0.8 N·m the seal is marginal; above 1.6 N·m you deform the vial rim and create a leak path on the opposite side.

For screw caps, the numbers are different. A 9 mm screw cap with a silicone/PTFE liner needs 0.6 to 0.9 N·m to compress the liner to 25 %. Under-torque (below 0.4 N·m) is the most common cause of screw-cap evaporation loss I see, and it is almost always because the lab is using a hand driver with no torque control. Get a torque screwdriver with a 0.1 N·m scale and check it monthly.

For snap caps, there is no torque number. The seal depends on the cap’s molded snap feature engaging the vial’s bead. If the insert sits high enough to lift the cap 0.2 mm, the snap does not fully engage and the seal is compromised. This is why snap-cap systems are more sensitive to insert height than crimp systems.

A troubleshooting checklist for vial insert seal failure

Work through this in order. Each step rules out one path.

Weigh a control. Unpunctured vial in the same tray. If loss is above 0.1 % per day, the tray environment is the problem.
Weigh 10 vials at injection 1 and 100. If loss is above 0.5 %, the seal is failing. If loss is below 0.5 % but drift is present, the problem is not evaporation.
Measure recovered septum thickness on a used cap. Below 0.85 mm after 40 punctures means compression set has done its work.
Measure insert height (shoulder to rim) on 10 inserts. It should be within ±0.15 mm of the drawing, and the rim should sit 1.0 to 1.5 mm below the vial rim when seated.
Measure insert OD versus vial neck ID. Clearance should be 0.10 to 0.15 mm. Above 0.30 mm, vapor bypasses the insert.
Check cap torque with a torque screwdriver. Crimp: 1.0 to 1.4 N·m. Screw: 0.6 to 0.9 N·m.
Check needle gauge and puncture pattern. A 23-gauge needle at the same coordinates for 100 injections is the worst case. If your method allows, vary the puncture coordinates or use a 25-gauge needle.
Check tray temperature stability. A tray that swings 3 °C adds 0.5 % loss per day through thermal expansion of the headspace.

Selection criteria: what to specify so the seal survives the run

When you write the insert specification, these are the numbers that matter for seal integrity, separate from the dimensional numbers that matter for sampler pickup:

Height tolerance: ±0.15 mm, measured shoulder to rim. This is the single most important sealing dimension.
Rim position: 1.0 to 1.5 mm below the vial rim when seated. Specify it, and verify it on incoming inspection.
OD tolerance: ±0.10 mm, with a target clearance of 0.10 to 0.15 mm to the vial neck ID.
Glass type: Type I borosilicate, USP <660> compliant. Type II or III glass can have a higher coefficient of thermal expansion, which means more movement at the seal under temperature cycling.
Fire-polished rim: a rim with a radius above 0.5 mm is not a sealing surface. Specify a radius of 0.2 to 0.4 mm, or a flat-ground rim if the insert will contact the septum.

For the septum side, specify a pre-slit silicone/PTFE septum with a compression set below 20 % after 24 hours at 25 % compression. That number is in the supplier’s datasheet if you ask for it. If they cannot supply it, that tells you something. The HPLC consumables supply chain has enough qualified septum manufacturers that you do not need to accept an unmeasured compression set.

Re-seal after puncture: what to expect and what to reject

A good insert and cap combination should re-seal to within 0.3 mg/hour of loss after 30 punctures at 25 °C with acetonitrile. That is the number to put in your incoming inspection. If a lot cannot meet it, reject it.

The test is simple. Take 10 vials, puncture each 30 times with your method’s needle, weigh, wait 4 hours, weigh again. Loss should be below 1.2 mg per vial. This takes half a day and it is the only way to catch a lot that will fail at hour 14 of a 19-hour run. I have used this test to reject three lots in five years, and each one would have cost a full sequence of rework.

For GC methods, the same test applies but the acceptance limit is tighter: 0.15 mg/hour after 30 punctures, because GC headspace methods are more sensitive to volume change than HPLC. The Restek technical resources have good reference data on septum performance for GC that is worth reading alongside your own measurements.

The decision rule

If your drift is above 1 % over a run and your system suitability at injection 6 was fine, weigh the vials before you touch the method. A gravimetric loss above 0.5 % over 100 punctures means the seal is failing, and the fix is a height-controlled insert plus a septum with measured compression set, not a new column. If the gravimetric loss is below 0.5 % and the drift is still there, the problem is not evaporation and you should look at the sampler and the mobile phase instead. The insert is a sealing component as much as a volume-reduction component, and specifying it as one is cheaper than re-running a sequence.

When you source inserts, ask for the height tolerance, the rim position, and the compression set of the septum that ships with the cap. A supplier who can give you those three numbers is a supplier who understands that the chromatography supplier relationship is built on measured performance, not on a catalog photo. The Agilent consumables and Thermo Fisher sample handling documentation both publish septum compression data you can use as a benchmark when you evaluate a new lot.

References & further reading

Related resources from our lab consumables network:

External references and standards: