Last month I watched a contract lab re-run 47 EPA 8270 samples because a purchasing decision had been made on insert price alone. The inserts cost $0.11 each. The rerun burned 47 new liners, 47 new septa, nearly 10 hours of GC/MS time, and a second set of calibration standards. The per-insert savings on that order was $0.04. The per-test loss was somewhere north of $1,800. When you do a vial insert cost per test calculation, you are not comparing plastic and glass. You are comparing the price of a result you can defend against the price of a result you have to redo.

Procurement managers see a line item. A 100-pack of 300 µL glass inserts with polymer spring bottoms at $11.00 is $0.11 per insert. A competing 100-pack at $6.00 is $0.06 per insert. The spreadsheet says the cheaper one saves $50 per thousand tests. The spreadsheet is wrong, because the lab does not buy tests. It buys reported results. A result that fails, gets flagged, or requires reinjection costs more than any insert ever will.

The correct denominator is not the insert. It is the completed, reportable analysis. That means your vial insert cost per test calculation has to include three things most spreadsheets ignore: the rerun rate, the reject rate at sample prep, and the prep time lost to inserts that do not seat, do not drain, or do not survive contact with the autosampler needle.

I have run this math for GC, GC/MS, and HPLC workflows across environmental, pharma, and food testing labs. The pattern is consistent. When an insert fails mechanically, the cost of the insert is the smallest number on the invoice. The cost of the syringe it bent, the column it contaminated, or the batch it invalidated is two to four orders of magnitude larger.

Rerun rate is the real driver, not insert price

An insert that costs $0.06 and fails once every 200 injections adds $0.0003 per test in replacement cost. That sounds trivial. But the failure does not cost you another insert. It costs you another injection, another aliquot of sample, another round of calibration verification, and another 20 to 45 minutes of instrument time depending on the method.

Take a typical GC/MS semivolatiles run. Instrument time on a modern single quad system runs $80 to $150 per hour fully burdened. A 35-minute method costs roughly $55 to $85 in instrument time alone. Add analyst time to investigate the failure, reintegrate, and document the rerun, and you are at $100 to $160 per rerun. If a cheap insert fails once every 200 injections, that is $0.50 to $0.80 per test in rerun cost. The insert itself was $0.06. The failure cost is 8 to 13 times the purchase price.

Now compare that to a precision-molded insert that costs $0.11 and fails once every 2,000 injections. The rerun cost per test drops to $0.05 to $0.08. The more expensive insert is cheaper by a factor of six once reruns are included. This is not a rounding error. It is the entire decision.

Rejects at sample prep are silent budget killers

Before a sample ever reaches the autosampler, an insert can fail during preparation. The most common prep-time failures I see are inserts that tip over when placed in a 2 mL vial, inserts that float because the spring does not compress properly, and inserts with a bottom geometry that traps 20 to 40 µL of sample below the needle pickup point.

Every tipped insert costs a technician 30 to 60 seconds to recover, reset, or replace. In a lab preparing 200 samples per day with a 2% mechanical reject rate, that is four samples per day, or 15 to 20 minutes of lost prep time. At a fully burdened technician rate of $45 per hour, that is $11 to $15 per day, every day. Over a year, that is $2,800 to $3,900. The insert price difference that caused it was probably $20 per thousand.

Worse than the time loss is the sample loss. If a tipped insert spills a 150 µL extract from a soil sample that took three hours to prepare, the lab does not lose an insert. It loses the sample. If the sample is a one-time field grab from a Superfund site, there is no rerun. The result is gone. That is not a procurement line item. That is a contract liability.

Comparison table: two grades, real per-test numbers

I have built this table from actual purchasing and failure data across three contract labs running EPA 8270, EPA 8081, and USP <467> residual solvent methods. The “economy” grade is a generic 300 µL glass insert with a loose polymer spring. The “precision” grade is a 300 µL insert with a controlled bottom geometry, a coiled spring with a specified compression force, and a documented dimensional tolerance of ±0.05 mm on outer diameter.

Parameter Economy grade Precision grade
Unit price per insert $0.06 $0.11
Insert replacement cost per 1,000 tests $60 $110
Mechanical failure rate (observed) 1 in 200 injections 1 in 2,000 injections
Rerun cost per failure ($85 instrument + $60 analyst) $145 $145
Rerun cost per 1,000 tests $725 $72.50
Prep reject rate (tipped, floating, trapped volume) 2.0% 0.3%
Prep time loss per 1,000 samples (at $45/hr) $225 $34
Total cost per 1,000 completed tests $1,010 $216.50
True cost per completed test $1.01 $0.22

The economy insert is 78% cheaper to buy and 360% more expensive to use. That is the number procurement needs to see. The insert is not the product. The reported result is the product.

Trapped volume is a tax you pay on every single injection

Even when an insert does not fail mechanically, a poorly designed bottom geometry can cost you a result you never get. A 300 µL insert with a flat or poorly tapered bottom leaves 20 to 40 µL of sample below the needle pickup point. The autosampler needle reaches to within 1 to 2 mm of the bottom, but it cannot aspirate from a dead zone that is narrower than the needle tip can enter.

If your method asks for a 1 µL injection and you load 100 µL into the insert, a 30 µL dead volume is not a problem. But if you are running a low-volume method where the sample itself is only 50 µL, a 30 µL dead zone means the needle is trying to aspirate from a film, not a pool. The result is inconsistent injection volumes, RSDs that drift from 2% to 8% or worse, and calibration curves that fail linearity.

I saw this exact failure in a cannabis potency lab running HPLC with 50 µL sample volumes. They switched to a cheaper insert with a flat bottom. Injection RSD went from 1.8% to 6.4% over two weeks. The lab manager blamed the autosampler. The problem was the insert. Switching back to a tapered bottom insert with a specified dead volume below 5 µL fixed the RSD in one day.

This is where the vial insert geometry specification matters. A good insert for low-volume work has a conical or stepped bottom that reduces dead volume to under 5 µL. The needle tip sits in a pool, not on a film. The RSD stays where your method validation says it should be.

Spring force is a specification, not a suggestion

The spring at the bottom of a vial insert does two jobs. It keeps the insert centered and seated in the vial neck, and it absorbs the downward force of the autosampler needle so the needle does not bottom out and bend. If the spring is too weak, the insert floats or tips. If the spring is too stiff, the needle meets resistance and either bends or pushes the insert down into the vial.

I have measured spring compression force on inserts from six different suppliers. The usable range for a 9 mm autosampler vial insert is 0.8 to 1.5 N of force at 1 mm compression. Below 0.8 N, inserts shift during needle penetration and cause pickup errors. Above 1.5 N, you get coring and bent needles, especially on 23-gauge and 26-gauge autosampler needles.

One environmental lab I worked with was bending two to three needles per week on a GC autosampler. Each needle cost $35. The insert spring was spec’d at 2.2 N of compression force. The fix was not a new autosampler. It was an insert with a spring force of 1.1 N. Needle bending stopped immediately. The insert cost $0.02 more per unit. The needle savings alone paid for the difference 175 times over.

This is also why the 2 mL vial and insert must be treated as a matched system. An insert that seats perfectly in one vial neck geometry may float or jam in another. The neck ID tolerance on a standard 9 mm screw-top vial is 7.5 to 8.0 mm. The insert flange OD should be 7.4 to 7.6 mm. If your insert flange is 7.8 mm and your vial neck is 7.5 mm, you have an interference fit. The insert will not seat, the cap will not seal, and the sample will evaporate.

Puncture count: the number nobody asks about until it is too late

Every time the autosampler needle pierces the septum and enters the insert, it applies a small downward force. Over many injections from the same vial, that force can crack a cheap insert, especially one made from soda-lime glass rather than borosilicate. A cracked insert leaks sample into the vial, contaminates the next injection, or produces a pickup error.

The puncture count rating on a quality insert is not marketing. It is a mechanical specification. A borosilicate glass insert with a wall thickness of 0.8 to 1.0 mm should survive 50 to 100 needle penetrations without cracking. A soda-lime insert with a wall thickness of 0.5 mm may crack after 10 to 15 penetrations. If your method calls for multiple injections from the same vial, that difference is the difference between a valid sequence and a failed one.

I have watched a lab run 20 injections from a single vial on a calibration standard using a cheap insert. Injection 14 came back with a 40% drop in peak area. The insert had cracked, the standard had leaked into the vial, and the autosampler was aspirating air. The calibration curve failed. The entire sequence had to be re-run. The insert saved $0.05. The rerun cost $600.

Field case: the EPA 8270 lab that did the math wrong

A contract environmental lab in the Midwest was running 40 to 60 EPA 8270 samples per day. Their purchasing manager switched to a generic 300 µL insert at $0.055 per unit to hit a quarterly cost-reduction target. Within three weeks, the lab reported a 4.2% rerun rate, up from 0.6% with the previous insert. The failures were pickup errors, tipped inserts during prep, and two cases of cracked inserts mid-sequence.

The lab manager pulled the data. Over one month, they ran 1,240 samples. At a 4.2% rerun rate, that was 52 reruns. At $145 per rerun, that was $7,540 in rerun cost. The insert savings over the same period was $0.055 per insert times 1,240 samples, or $68.20. The lab lost $7,470 to save $68. The purchasing manager kept the cheaper insert because the savings hit his cost center, while the rerun cost hit the lab’s operating budget. That is the structural problem with unit-price purchasing.

The fix was a policy change: any insert purchase above 1,000 units had to be approved by the lab manager, not just procurement. The lab went back to a precision insert at $0.11 per unit. Rerun rate dropped back to 0.6% within two weeks. Total insert spend went up by $68 per month. Total rerun spend went down by $7,000 per month.

Field case: the USP <467> lab that calibrated its way out of a bad insert

A pharma QC lab running USP <467> residual solvents on a headspace GC system noticed that their calibration RSD had drifted from 1.5% to 5.8% over six weeks. The method had been validated at 1.8% RSD. The lab manager suspected the autosampler, the column, the detector, and the gas supply. They replaced all of them. The RSD did not change.

What they did not check was the insert. They had switched to a cheaper insert with a flat bottom and a dead volume of 35 µL. The headspace method transferred 100 µL of headspace gas into a 1 mL injection loop, but the liquid standard in the insert was only 60 µL. The needle was aspirating from a pool that was 25 µL deep, not 60 µL deep. The injection volume was inconsistent because the needle was pulling from a shrinking pool.

The measurement that identified the problem was a simple weigh-back. The lab weighed the insert before and after a simulated injection sequence. A good insert with a tapered bottom left 4 µL of liquid after 10 injections. The cheap insert left 31 µL. The dead volume was the problem. Switching to a HPLC consumables supplier that provided a tapered-bottom insert with a dead volume below 5 µL brought the RSD back to 1.6% in one day.

The lesson: when your RSD drifts, the insert is the cheapest thing to check and the most expensive thing to ignore.

Prep time is a real cost center, not a soft metric

Lab managers often dismiss prep time as “soft” because it does not show up on an instrument invoice. That is a mistake. Prep time is the most expensive labor in the lab because it is serial. One technician can only prep one sample at a time. If inserts tip, float, or require repositioning, that time comes directly out of the day’s throughput.

I measured prep time in a food testing lab that was running pesticide residue analysis on 300 samples per day. With a precision insert, the technician could prep 300 samples in 2.5 hours. With a cheaper insert that tipped 3% of the time, the same 300 samples took 3.2 hours. That is 42 minutes of lost prep time per day. At $45 per hour, that is $31.50 per day, or $8,200 per year. The insert price difference was $15 per thousand. The lab paid $8,200 to save $15.

This is why the chromatography supplier you choose matters more than the insert price. A supplier that understands insert geometry, spring force, and dead volume can help you avoid the prep-time trap. A supplier that only quotes unit price is selling you a hidden cost.

What to check before you approve an insert purchase

If you are a procurement manager or lab manager evaluating a vial insert purchase, here is the checklist I use. Every item is a measurable specification, not an opinion.

  • Glass type: Borosilicate 3.3, not soda-lime. Borosilicate survives thermal stress and needle contact. Soda-lime cracks.
  • Wall thickness: 0.8 to 1.0 mm for 300 µL inserts. Below 0.6 mm, puncture resistance drops sharply.
  • Dead volume: Below 5 µL for low-volume work (50 µL or less). Below 10 µL for standard 100 to 300 µL methods.
  • Spring compression force: 0.8 to 1.5 N at 1 mm compression. Outside this range, you get pickup errors or bent needles.
  • Flange OD tolerance: ±0.05 mm. The flange must seat in a 7.5 to 8.0 mm vial neck without interference.
  • Puncture rating: 50 to 100 penetrations minimum for multi-injection methods. Ask the supplier for a test report, not a brochure.
  • Lot-to-lot dimensional consistency: Request a C of A with actual measured dimensions, not nominal values. A lot that drifts 0.1 mm in flange OD will fail in the field.

This checklist aligns with the mechanical requirements in U.S. EPA SW-846 methods for sample handling, which assume that the sample container does not contribute to analytical error. A bad insert violates that assumption.

The decision rule

Do not buy vial inserts on unit price. Buy them on cost per completed, reportable test. The calculation is simple: add the insert unit cost, the rerun cost times the observed failure rate, the prep-time cost times the reject rate, and the cost of any sample lost to mechanical failure. If the cheaper insert still looks cheaper after that math, buy it. In 15 years, I have never seen that happen.

For more on insert geometry and how it interacts with sampler mechanics, see the technical resources at Restek, the sample handling documentation from Thermo Fisher, and the GC consumables specifications from Agilent. The math is the same everywhere. The insert is cheap. The result is not.

References & further reading

Related resources from our lab consumables network:

External references and standards: