Why one lot can pass on instrument A and fail on instrument B-- HPLC Autosampler Vials

I have watched the same box of 2 mL vials run clean on a Thermo autosampler for three weeks and then generate a run of short injections, bent needles and a pressure fault on an Agilent 1260 sitting two benches away. Same lot number, same certificate of analysis, same septa. The instrument that failed was not broken. The vial was not out of specification. What changed was the set of mechanical tolerances the second instrument imposes on that vial, and the fact that the first instrument’s tolerances were loose enough to hide the problem. This article is about that gap: why lot-to-lot variation inside a valid specification band can pass one autosampler and fail another, and how to work out which instrument is telling you the truth.

An autosampler vial is a stack of tolerances, not a single dimension. The ones that matter in practice are the outside diameter of the glass body, the concentricity of the neck to the body, the height from the base to the crimp or screw finish, the internal depth from the top of the finish to the bottom of the well, the septum thickness, and the durometer and thickness of the septum’s centre versus its skirt. Any one of those can sit at the edge of its band and still be “in spec”. Whether that edge matters depends entirely on the instrument.

Consider a real example. A 12 x 32 mm vial made to the GPI 9 mm screw-thread finish is allowed a body diameter of roughly 11.6 to 12.0 mm. That is a 0.4 mm window. A tray that grips vials in a fixed nest with 0.15 mm clearance per side will accept a 11.95 mm vial and jam on a 12.02 mm one. A different tray with 0.25 mm clearance per side will take both. If your lot is running at 11.97 mm and your failing instrument has the tighter nest, you have found your answer without ever looking at the chromatogram.

The same logic applies to height. A vial that is 31.85 mm tall instead of 32.00 mm may still be picked up by a needle that descends to a fixed depth, but on an instrument that uses a bottom-sensing or a fixed-height needle guide, a 0.15 mm shortfall changes the puncture depth. That changes the number of times the needle passes through the same hole, which changes coring risk. The vial is the same on both instruments. The failure is not.

The four tolerance axes that actually move between instruments

When I audit a complaint that says “this lot worked before”, I work through four axes in this order.

Radial location: how the vial is held in the tray or nest, and how much lateral float the needle guide allows. Fixed nests with tight clearance are unforgiving of body diameter and concentricity.
Vertical registration: whether the instrument references the vial from the base, the shoulder, or the top of the cap. Base-referenced instruments are sensitive to overall height. Cap-referenced instruments are sensitive to crimp or screw finish height and to cap thickness.
Septum engagement: needle penetration depth, needle gauge, and whether the guide pre-pierces or the needle does all the work. A 22-gauge needle (0.72 mm OD) through a 1.0 mm septum has more margin than a 26-gauge (0.46 mm) through the same septum.
Closure force: crimp compression or screw torque, which sets how much the septum is pre-loaded. A septum compressed too hard is harder to pierce and cores more readily.

Two instruments that differ on any of these axes will disagree about a vial lot that sits near the edge of its band. That is not a quality failure. It is a specification mismatch.

The instrument-side variables nobody puts on the datasheet

Vendors publish needle gauge, injection volume range and tray capacity. They rarely publish nest clearance, needle descent tolerance, or the force the needle guide applies to the cap. Those are the numbers that decide whether your lot passes. I have collected them by measurement over the years, and they vary more than most people expect.

What I measure on a suspect instrument

Take ten vials from the lot, load them into the failing tray position, and measure the following with a digital caliper and a depth gauge. This takes twenty minutes and settles most arguments.

Nest clearance: the gap between the vial body and the nest wall at the widest point. Under 0.10 mm per side is tight. Over 0.30 mm is loose.
Needle descent repeatability: run the same vial ten times and check whether the puncture marks land within 0.05 mm of each other. If they wander, the guide is worn or the tray is not seating.
Guide pre-load: a spring-loaded guide should apply 2 to 6 N to the cap. Much more and it distorts the septum before the needle arrives.
Crimp or screw torque: for a 9 mm screw cap, 0.6 to 1.0 N·m is the working band. Above 1.2 N·m the septum starts to extrude and core. Below 0.4 N·m it leaks.

If your failing instrument has 0.08 mm nest clearance and your passing instrument has 0.22 mm, you already know why the same lot behaves differently. The vial did not change. The tolerance stack did.

Why the same lot can be “good” and “bad” at the same time

Lot acceptance is a statistical decision, not a binary one. If your supplier’s sampling plan draws thirty vials from a 50,000-piece lot and measures body diameter, the sample mean might be 11.94 mm with a standard deviation of 0.03 mm. That passes any reasonable band. But the tail of that distribution still contains vials at 12.02 mm. On the loose instrument, that tail never contacts anything. On the tight instrument, it jams. Same lot, two truths. The supplier is not lying. Your instrument is simply sampling the tail more aggressively than the supplier’s inspection did.

This is why “send me a new lot” often fixes the symptom for a quarter and then it comes back. The root cause is the mismatch between the tolerance band your instrument can absorb and the band your supplier is allowed to ship. Fixing that means either tightening the purchase specification or loosening the instrument’s sensitivity. There is no third option.

A decision checklist for diagnosing which side is at fault

Before you blame the vial, run this checklist. It is ordered so that the cheapest tests come first and the most expensive come last. I use it on every cross-instrument complaint.

Swap test: take ten vials from the failing lot and run them on the instrument that passes. If they pass, the vial is fine and the instrument is the variable. If they fail, the lot is the variable.
Control lot test: run ten vials from a known-good lot on the failing instrument. If they fail too, the instrument has drifted. If they pass, the failing lot is the variable.
Dimensional audit: measure body diameter, height, concentricity and septum thickness on twenty vials from each lot. Compare the means and the ranges, not just the pass/fail.
Puncture count: run a single vial through 100 injections and inspect the septum under a low-power microscope. More than three visible cores in 100 punctures is a septum problem. Zero cores but a wandering puncture pattern is a guide problem.
Torque check: for screw caps, verify closure torque with a calibrated torque screwdriver. For crimp caps, verify crimp height with a crimp gauge. Out-of-band closure force mimics a septum failure.
Pressure trace: log the system pressure during the failing injections. A pressure spike at the moment of needle descent points at the guide or the septum, not the pump.

If the swap test and the control lot test disagree, the answer is usually that both lots are fine and the failing instrument has a worn guide or a misaligned tray. That is the most common outcome I see, and it is the one least likely to be reported as an instrument fault.

Worked example: the 11.97 mm lot

Here is a case I worked through in detail. A contract lab ran a method on a Thermo Vanquish with a split-sample tray and an Agilent 1260 with a standard 100-position tray. The same lot of 9 mm screw-top vials, 12 x 32 mm, clear glass, with a silicone/PTFE septum, ran clean on the Thermo for six weeks. On the Agilent it produced short injections on roughly 4% of positions, always in the outer ring of the tray.

We measured twenty vials from the lot. Body diameter mean was 11.97 mm, range 11.92 to 12.01 mm. Height mean was 31.92 mm, range 31.88 to 31.97 mm. Septum thickness was 1.02 mm nominal, range 0.98 to 1.06 mm.

The Thermo tray had 0.24 mm clearance per side. The Agilent tray had 0.11 mm. The outer ring of the Agilent tray was also slightly out of round from years of use, adding another 0.05 mm of variance at the worst positions. The combination of a vial at 12.01 mm and a tray position at the edge of its tolerance produced a jam that the needle interpreted as a short injection. The vial was in specification. The tray was in specification. The combination was not.

The fix was not a new lot. It was a tray replacement on the Agilent and a purchase specification that tightened body diameter to 11.90 to 11.98 mm for that method. The lot continued to run on both instruments. The 4% failure rate dropped to zero.

What this example teaches

Three things. First, the failing instrument was the one with the tighter tolerance stack, not the one with the better reputation. Second, the vial lot was not defective, and a supplier who replaced it would have been solving the wrong problem. Third, the durable fix was a change to the purchase specification, not a change to the lot. That is the pattern in almost every cross-instrument complaint I have investigated.

How to write a purchase specification that survives two instruments

If you run the same method on more than one instrument, your vial specification has to be tight enough to survive the tighter of the two. That is a different document from the one most labs write. Here is the structure I use.

Body diameter: specify a nominal and a tolerance band, for example 11.94 ± 0.04 mm. Do not accept “meets GPI 9 mm” as a substitute for a number.
Height: specify from base to top of finish, for example 32.00 ± 0.10 mm. State the reference plane explicitly.
Concentricity: specify neck-to-body runout, for example ≤ 0.10 mm TIR. This is the number that decides whether a tight nest will accept the vial.
Septum thickness: specify nominal and tolerance, for example 1.00 ± 0.05 mm, and specify durometer, for example 45 ± 5 Shore A for silicone.
Closure: specify crimp height or screw torque band, for example 0.6 to 1.0 N·m for a 9 mm screw cap, verified with a calibrated tool.
Puncture performance: specify a minimum puncture count before coring, for example ≥ 100 punctures with ≤ 3 cores at 22-gauge.

That document is longer than most labs want to write. It is also the only thing that stops the same complaint coming back every six months. If you want a starting point, the dimensional conventions in the GPI finish standards and the USP <661> container guidance give you the reference planes. The rest is your own instrument data.

Troubleshooting checklist: cross-instrument failure

When a lot fails on one instrument and passes on another, work through this list in order. Stop when you find the divergence.

Confirm the failure mode. Short injection, bent needle, pressure spike, or carryover. Each points at a different tolerance axis.
Run the swap test. Ten vials, both instruments, same method. Record which positions fail.
Map the failures to tray positions. If failures cluster in one ring or one quadrant, the tray is the variable, not the vial.
Measure nest clearance on both trays. Under 0.10 mm per side is a red flag.
Inspect the needle guide. Look for a worn bore, a bent tip, or a spring that no longer returns. A guide that applies more than 6 N pre-load distorts the septum.
Check closure torque or crimp height. Out-of-band closure is the second most common cause after tray wear.
Measure twenty vials from each lot. Compare means and ranges. If the ranges overlap, the lots are statistically indistinguishable and the instrument is the variable.
Run a 100-puncture test on a single vial. Count cores. More than three is a septum problem. Zero cores with wandering punctures is a guide problem.
Log system pressure during injection. A spike at needle descent points at the guide or septum. A spike at injection points at the sample or the needle bore.
Decide the fix. Tighten the purchase specification, replace the tray or guide, or both. Do not simply request a new lot.

I keep this list on a single page and hand it to anyone who calls me about a cross-instrument failure. It has never failed to find the divergence within a day.

Where the vial supplier fits, and where they do not

A good supplier will tell you the truth: their lot is in specification, and the failure is on your instrument. A great supplier will help you measure the divergence and adjust the purchase specification so it does not recur. That is the difference between a vendor who ships boxes and a partner who understands that an HPLC autosampler vial is a mechanical component with a tolerance stack, not a commodity.

When I specify vials for a multi-instrument lab, I work with suppliers who publish their dimensional bands and can hold a tightened band on request. Aijiren, for example, will quote a tightened body-diameter band on a defined MOQ and supply the inspection data with the lot. That is the kind of supplier behaviour that makes a cross-instrument specification possible. You can see their range at Aijiren.

For the broader consumables picture, including septa and caps that have to match the vial’s closure band, the chromatography products catalogue is a useful reference. For pricing on the vial itself, sample vial price comparisons are worth running before you commit to a tightened band, because the cost of a tighter tolerance is real and you should know what you are paying for it.

On the instrument side, the vendor technical resources are the place to find the tray clearances and needle guide specifications that decide whether your lot will pass. Agilent publishes autosampler and tray specifications in its chromatography consumables and columns documentation. Thermo Fisher does the same for its sample handling range at Thermo Fisher chromatography and sample handling. Restek’s technical resources are useful for septum and closure guidance. If your method is regulated, the sample handling requirements in U.S. EPA SW-846 give you the documentary framework for the purchase specification.

Operational precautions that stop the problem recurring

Once you have found the divergence, the fix has to be operational, not just documentary. These are the precautions I insist on in any lab running the same method on more than one instrument.

Qualify the tray, not just the instrument. Measure nest clearance on every tray at installation and every twelve months. A worn tray drifts silently.
Log tray position with every failure. A failure map over three months will show you whether the problem is the tray, the guide, or the lot.
Hold a reference lot. Keep 200 vials from a known-good lot in a sealed box. Use them as the control in every cross-instrument investigation.
Verify closure torque weekly. A torque screwdriver costs less than a single lost run. Check the band, not just the average.
Inspect the needle guide monthly. Look for a worn bore and a spring that no longer returns. Replace at the first sign of wander.
Write the purchase specification around the tighter instrument. The looser instrument will always accept it. The tighter one will not accept anything else.

None of this is expensive. All of it is cheaper than a run of short injections on a Friday afternoon.

Takeaway

The same vial lot fails on one instrument and passes on another because the two instruments impose different tolerance stacks on the same vial, and the lot is sampling the edge of its band. The vial is usually in specification. The instrument is usually in specification. The combination is not. The durable fix is a purchase specification written around the tighter of your instruments, with explicit numbers for body diameter, height, concentricity, septum thickness, closure torque and puncture count, backed by a tray and guide qualification programme that catches drift before it shows up as a failed run.

References & further reading

Related resources from our lab consumables network:

External references and standards: