A batch of 20 mm serum stoppers came off the line at 3:14 PM last Thursday with an average shore A hardness of 48.6. The specification window is 45 to 55. The batch was technically in spec, but the Cpk for that run was 0.82, and the shift supervisor knew the next lot of rubber compound would not fix itself. That is the reality of how vial closures is manufactured and what the factory controls: the line does not produce a single value, it produces a distribution, and the controls you do not see are the ones that determine whether your next fill line runs at 400 vials per minute or stops dead at 11:00 AM.
Most people assume a closure factory is a molding operation. It is, but the control logic is not what you think. The press operator sets a temperature profile across three zones: top platen, bottom platen, and the mold itself. The mold temperature is the one that matters for crosslink density. I have seen top and bottom platens hold 165 °C while the mold cavity sits at 158 °C because the mold was not preheated long enough after a shift change. The result is an undercured stopper that looks identical to a good one for the first 48 hours, then starts leaching extractables.
The parameter that separates a real OEM line from a job shop is cavity pressure control. Modern presses use a transducer inside the cavity to hold pressure within a ±2 bar band during the cure cycle. If pressure drops early, the flash thickens and the sealing ribs lose definition. You cannot see this on a finished closure without sectioning it. The factory controls it by closing the loop on the transducer signal every 50 milliseconds. Older presses control only hydraulic pressure, which is a proxy that drifts as the mold heats up. Ask your supplier whether they control cavity pressure or hydraulic pressure. If they pause, you have your answer.
Cure time is the third variable. A 20 mm chlorobutyl stopper typically cures for 4 to 6 minutes at 165 °C. Cutting cure time by 30 seconds to increase throughput changes the crosslink density enough to shift compression set from 18% to 27%. That is not a cosmetic change. Compression set above 25% means the stopper does not recover after needle puncture, and the vial loses seal integrity on day three of storage. The factory controls cure time with a timer tied to the mold close sensor, not to the operator’s watch. Some lines add a redundant thermocouple in the mold to confirm the cure only starts when the cavity reaches the setpoint, not when the press closes.
Trimming and washing are where surface contamination is either removed or baked in
After molding, closures go through a deflashing or trimming step. Cryogenic deflashing uses liquid nitrogen to freeze the flash and tumbling media to knock it off. The factory controls the temperature, the tumble speed, and the cycle duration. If the temperature is too warm, the flash bends instead of fracturing, leaving a ragged edge on the sealing surface. If the tumble speed is too high, you get surface scuffing that shows up later as particulate shedding in the fill line. A typical cryogenic deflash cycle for a 13 mm stopper runs at -60 °C for 90 seconds. I have audited lines where the nitrogen flow was reduced to save cost, and the cycle temperature crept up to -40 °C. The closures passed visual inspection but failed a particulate rinse test with counts three times the limit.
Washing is the next station. The factory controls wash water temperature, detergent concentration, rinse water conductivity, and the number of rinse cycles. For silicone-free closures, the wash water is typically USP purified water at 60 to 70 °C with a non-ionic detergent, followed by three rinses with water at less than 1.3 µS/cm conductivity. The conductivity sensor is the key control. If the final rinse water conductivity rises above 2 µS/cm, the line should divert the batch. Some factories use a cascade rinse system where the final rinse water is fresh and the earlier rinses reuse water from later stages. This saves water but requires tighter conductivity monitoring because detergent carryover into the final rinse is the failure mode.
What the factory does not control at the washing station is the incoming water quality upstream of their treatment system. That is a raw material input. A factory that draws municipal water with seasonal chlorine spikes will see variable rinse performance unless they have a carbon filter and monitor it. Ask for the rinse water conductivity log, not just the certificate of analysis. The log tells you whether the control is real or aspirational.
Silicone coating is a closed-loop process only if the factory invested in it
Silicone application on vial closures is not a dip-and-dry operation, or it should not be. The standard method is to spray a diluted silicone emulsion onto closures in a rotating drum or a fluidized bed. The factory controls the silicone concentration, the spray rate, the drum speed, and the drying temperature. The output variable is the silicone level on the closure surface, typically expressed in micrograms per square centimeter. For a ready-to-sterilize stopper, the target is usually 10 to 30 µg/cm². Below 10, needle penetration force rises and coring risk increases. Above 30, you get silicone transfer to the drug product and visible droplets in the vial.
The control challenge is that silicone distribution is never uniform. The factory measures it by extracting the silicone from a sample of closures and analyzing it by atomic absorption or ICP. This is an end-of-line check, not a closed-loop control. The closed-loop part is the spray system: a mass flow controller on the silicone feed line and a tachometer on the drum. If the drum speed drops 5%, the closures tumble less and the silicone pools on one side. The factory can control the inputs tightly but cannot measure the output in real time. That is why a good OEM runs a silicone level check every 30 minutes during a run, not once per shift. The sampling frequency is a control parameter, and it is one of the first things cut when a factory is under margin pressure.
For HPLC vial closures, the silicone target is lower, often 5 to 15 µg/cm², because silicone extractables interfere with mass spec baselines. The same spray line can run both product families, but the changeover procedure matters. Residual silicone from a high-level run contaminates the first 200 closures of a low-level run if the line is not purged. A factory that controls this documents a changeover purge protocol with a minimum purge time and a verification sample before the run starts.
Needle puncture testing on the line catches what shore A hardness misses
Shore A hardness is the parameter everyone asks about, but it is a bulk property. Needle puncture force is a functional property. A stopper with the right shore A can still have a high puncture force if the surface skin is over-crosslinked or the silicone coating is uneven. The factory controls puncture force by testing closures on a tensile tester with a standardized needle. The needle is typically a 21-gauge or 23-gauge hypodermic needle mounted in a fixture that drives it through the closure at a fixed speed, usually 200 mm/min. The peak force is recorded. For a 20 mm chlorobutyl stopper, a typical puncture force specification is 5 to 12 N. Above 15 N, auto-sampler needles bend and fill lines jam.
What the factory controls at this station is the needle lot, the test speed, and the fixture alignment. Needle lots vary. A needle with a burr or a slightly different bevel angle changes the puncture force by 1 to 2 N, which is enough to shift a marginal batch from pass to fail. The factory should use a new needle for each test or at least verify the needle after every 50 punctures. The puncture count on a needle is a control parameter. I have seen a factory use the same needle for an entire shift, and the puncture force readings drifted upward as the needle dulled. The batch was rejected by the customer, and the factory’s own data showed the drift if you plotted puncture force against test sequence number. Nobody had plotted it.
The puncture test is also where coring shows up. Coring is not just a needle problem; it is a closure problem. A closure with a hard surface skin or a high crosslink density at the surface will core more readily. The factory controls this indirectly through the molding parameters and directly through a coring test where the needle is withdrawn and the number of rubber fragments in a rinse solution is counted. The limit is typically zero visible cores for a 10-puncture test. If a factory cannot show you a coring test result, they are not controlling the parameter that matters most for chromatography consumables users.
Residual moisture and endotoxin are controlled by the drying and packaging sequence
After washing and siliconization, closures are dried. The factory controls the drying temperature, the drying time, and the residual moisture level. For closures destined for steam sterilization, the residual moisture target is typically less than 0.5% by weight. For closures that will be gamma sterilized, the target is often lower, less than 0.2%, because residual water accelerates polymer degradation during irradiation. The drying oven is usually a convection oven or a vacuum dryer. The control point is the exhaust air humidity sensor. If the exhaust air humidity does not drop below a setpoint within a specified time, the cycle is extended or the batch is flagged.
Endotoxin control is a separate issue. The washing step removes most endotoxin, but the drying and packaging steps can reintroduce it if the air handling is poor. A factory that controls endotoxin runs a depyrogenation step for closures used in injectable applications. This is typically a dry heat treatment at 250 °C for 30 minutes, or a validated washing process with endotoxin reduction of at least 3 log. The factory controls the oven temperature with redundant thermocouples and the residence time with a belt speed sensor. If the belt speed drifts 2%, the residence time shifts enough to compromise the depyrogenation. This is a closed-loop control on the belt drive, but the endotoxin level itself is only checked on a sample at the end of the run. The factory controls the process, not the result.
Packaging is the last station. Closures are packed in double polyethylene bags inside a fiber drum or a carton. The factory controls the bag sealing temperature, the seal integrity, and the cleanliness of the packaging area. A bag seal that fails after shipping lets moisture and particulates in. The factory should run a seal integrity test on a sample of bags from each lot, typically a dye penetration test or a vacuum decay test. The packaging area should be an ISO 8 cleanroom at minimum. If the closures are packed in a general warehouse area, the particulate load on the closures will be higher than the specification allows, and no amount of washing upstream will fix it.
What the factory checks only at the end, and why that matters for your audit
Some parameters are controlled in real time. Cavity pressure, mold temperature, cure time, wash water conductivity, spray rate, drum speed, drying temperature, and belt speed are all closed-loop or at least continuously monitored. The factory can show you a trend chart for these. Other parameters are only checked on a sample after the run. These include shore A hardness, silicone level, puncture force, coring, residual moisture, endotoxin, particulate count, and extractables. The factory cannot adjust these in real time because the measurement is destructive or slow. The control is statistical: the factory runs the process within validated limits and verifies the output on a sample. The sampling plan is the control, and that is a topic for a separate discussion on analytical testing.
When you audit a closure factory, ask which parameters are closed-loop and which are end-of-line. Ask to see the trend charts for the closed-loop parameters. A factory that cannot produce a cavity pressure trend for the lot you are buying is not controlling the process. A factory that produces a trend chart with a flat line is either faking it or has a sensor that is not connected. Real process data is noisy. The noise is the proof that the sensor is live.
The distinction matters because end-of-line checks are a snapshot. A lot that passes shore A hardness can still have a bimodal distribution where half the closures are at 42 and half at 52. The average is 47, the spec is 45 to 55, and the lot ships. But the 42 shore A closures will seal differently than the 52 shore A closures, and your fill line will see variable crimp forces. The factory controls this by controlling the inputs tightly enough that the output distribution is narrow. That is what a real OEM does. A job shop controls the inputs loosely and hopes the end-of-line check catches the bad lots. Most of the time it does not.
Field case: the 13 mm stopper lot that passed every test but failed on the fill line
A contract manufacturer running a lyophilized product called me about a 13 mm stopper lot that passed all incoming QC. Shore A was 49, puncture force was 8 N, silicone level was 18 µg/cm², residual moisture was 0.3%. The lot was within spec on every parameter. But on the fill line, the stoppers jammed in the vibratory bowl feeder. The feeder would run for 20 minutes, then stop. The operator cleared the jam, and it ran again. The fill line throughput dropped from 300 vials per minute to 180.
The measurement that identified the problem was a surface roughness profile. The stoppers had a higher surface roughness on the side wall than previous lots, enough to increase the coefficient of friction against the feeder bowl. The roughness came from the cryogenic deflash step. The factory had changed the tumbling media from a polycarbonate bead to a cheaper ceramic bead. The ceramic bead left micro-scratches on the closure surface. The scratches were invisible to the eye and did not affect any of the standard QC tests. But they changed the frictional behavior enough to jam the feeder. The fix was to revert to the original tumbling media and add a surface roughness check to the incoming QC protocol. The factory had not controlled the deflash media as a raw material, and it cost the customer three days of lost production.
This is the kind of failure that never shows up on a certificate of analysis. The closure was in spec. The process was not. The factory controls the process, and the process includes the tumbling media, the changeover purge, the needle lot, the rinse water conductivity, and a dozen other variables that never appear on a spec sheet. When you buy closures from a factory that controls these things, you are buying the absence of surprises. That is what the price difference pays for.
Field case: the 20 mm stopper lot with a silicone level that drifted within spec
A pharmaceutical company sent me a data set from a 20 mm stopper lot that showed silicone levels ranging from 12 to 28 µg/cm² across 12 samples taken over a 6-hour run. The spec was 10 to 30, so the lot was technically in spec. But the within-lot variability was 16 µg/cm², which is a relative standard deviation of about 35%. The customer’s own data showed that stoppers with silicone levels above 25 µg/cm² caused visible silicone droplets in the vial after 6 months of storage. The lot was released, but the customer had to sort incoming closures by silicone level, which is not practical at scale.
The root cause was the spray system. The factory was using a single-nozzle spray system in a rotating drum. As the drum loaded with closures, the spray pattern changed. The nozzle was positioned to spray the center of the drum, but the closures at the drum wall received less silicone than the closures in the center. The factory had not validated the spray pattern for a full drum load. The fix was to switch to a multi-nozzle system with individual flow control on each nozzle and to run a silicone level check at the start, middle, and end of each drum load. The factory also added a drum load sensor that adjusted the spray time based on the actual closure weight in the drum. This is a closed-loop control on the silicone application, and it reduced the within-lot variability to less than 5 µg/cm².
The lesson is that a spec is not a control. A spec is a boundary. Control is what keeps the process inside the boundary with a margin. A factory that controls the silicone application process can hold a Cpk of 1.33 or better. A factory that only checks the silicone level at the end of the run is gambling. The gamble shows up in your stability data six months later.
Decision checklist: what to ask a vial closure factory before you qualify them
Use this checklist when you audit a closure factory. Each item is a control point that separates a real OEM from a reseller with a molding machine. If the factory cannot answer the question with data, not an anecdote, flag it.
- Cavity pressure transducer: Is cavity pressure controlled to ±2 bar during cure? Ask for the trend chart from the lot you are buying.
- Mold temperature: Is the mold temperature measured directly, or only the platen temperature? The difference is 5 to 10 °C in the cavity.
- Rinse water conductivity: What is the final rinse conductivity limit? It should be less than 1.3 µS/cm for silicone-free closures and less than 2 µS/cm for silicone-coated closures.
- Silicone spray system: Is the silicone feed rate controlled by a mass flow controller? Is the drum speed monitored? What is the changeover purge protocol between high and low silicone runs?
- Puncture test needle: What needle gauge and lot is used? How many punctures per needle before replacement? The limit should be 50 or fewer.
- Deflash media: What is the tumbling media material and size? Is the media lot-controlled? Has the factory changed media in the last 12 months?
- Drying oven humidity: Is the exhaust air humidity monitored and used to extend the cycle if the setpoint is not reached? What is the residual moisture spec for your closure type?
- Depyrogenation: For injectable closures, what is the depyrogenation cycle? The minimum is 250 °C for 30 minutes or a validated washing process with 3 log endotoxin reduction.
- Packaging area classification: Is the packaging area ISO 8 or better? Is the bag seal integrity tested on every lot?
- Trend data access: Will the factory provide process trend data for cavity pressure, mold temperature, and rinse conductivity for your lot? If the answer is no, walk away.
These questions map directly to the stations on the line. A factory that controls the line can answer all of them in 10 minutes. A factory that does not will start talking about their quality policy and their ISO certificate. The certificate tells you they have a quality system. The data tells you whether the system controls the process or just documents it.
For closures used in GC and HPLC applications, the stakes are higher. A closure that sheds particulates or leaches extractables ruins a chromatography run and wastes a column. The sample handling chain starts with the closure, and the closure starts with the factory controls. When you specify a closure for an EPA SW-846 method, the closure is part of the sample integrity chain. A factory that cannot show you the cavity pressure data for your lot is a factory that does not know what its own line did that day.
The Restek technical resources library has useful reference data on closure compatibility with common solvents, but the compatibility data assumes the closure was manufactured within the process limits. If the factory drifted outside those limits on your lot, the compatibility data does not apply. That is the practical consequence of poor process control. The closure is not a catalog item. It is the output of a process, and the process is what you are really buying.
When you qualify a closure supplier, do not qualify the sample. Qualify the line. The sample is a snapshot. The line is the distribution. A factory that controls the line gives you a narrow distribution and a predictable fill line. A factory that controls only the final inspection gives you a wide distribution and a surprise. The decision rule is simple: buy from the factory that can show you the cavity pressure trend for your lot, the rinse conductivity log, and the silicone level chart. If they cannot, you are buying a lottery ticket, not a closure. And lottery tickets do not hold a crimp seal at 400 vials per minute.
References & further reading
Related resources from our lab consumables network:
- Vial Closures at vialsclosures.com
- Analytical Testing at testinganalytical.com
- Hplc Vial at vialforhplc.com
- Chromatography Consumables at chromatographyconsumables.com
External references and standards: