Compounding is the first control point, and it is where most OEM failures--Aijiren HPLC Vials

Last month I walked a line that had just rejected 14,000 molded septa in a single shift. The failure was not the silicone. It was a 4-second dwell time drift on the transfer press, and the operator caught it because the durometer reading on the hourly pull had moved from 48 Shore A to 41. That is the reality of how septa is manufactured and what the factory controls: the line either holds the setpoints, or it produces scrap that looks identical to good product until someone puts a needle through it. This article maps the stations, names the parameters that are closed-loop controlled versus the ones that only get checked after the batch, and tells you what to ask a septa factory before you qualify a lot.

Compounding is the first control point, and it is where most OEM failures begin

Before a single septum is molded, the factory has to compound the elastomer. For a standard GC inlet septum, that is usually a silicone base, often dimethyl or fluorosilicone, loaded with a filler and a peroxide or platinum catalyst. The compounder is not just mixing. It is setting the crosslink density that will determine resealability, hardness, and bleed. I have watched factories run this step on an open two-roll mill with no torque monitoring. You cannot hold a 2 Shore A window that way.

What a serious line controls at compounding:

Mix torque and time. A closed-loop internal mixer records torque in real time. When the torque curve flattens, the filler has dispersed. If the curve spikes late, you have scorch. If it never flattens, you have wetting failure. The line trips at a set torque-time integral, not at a timer alone.
Temperature of the batch. Platinum-cured silicones scorch above roughly 120°C during mixing. Peroxide systems are more forgiving but still have a defined ceiling. The mixer jacket is on a closed loop with a tolerance of ±3°C.
Filler content by weight. Diatomaceous earth or fumed silica changes puncture resistance and hardness. A 1% drift in filler loading moves Shore A by 3 to 5 points. Factories that weigh filler manually and do not log each addition are not controlling this.
Volatile content after mixing. This is an end-of-batch check, not a closed loop. A sample goes into a moisture balance or a headspace GC. If volatiles are above the spec, the batch is reworked or scrapped. You cannot fix this downstream.

I have specified septa where the compounding record was a single sheet of paper with a mixer number and a time. That is not a control plan. That is a diary. Ask for the torque curve. If the factory cannot produce it, assume the durometer spec is aspirational.

The molding press controls cavity pressure, and cavity pressure controls the septum

Compression molding, transfer molding, and injection molding are all used for septa. The difference matters less than the control philosophy. A compression press can make excellent septa if the factory controls shot weight, preform temperature, and clamp force. An injection press can make terrible septa if the cavity pressure sensor is ignored.

The parameter that predicts a good septum better than any other is peak cavity pressure. For a typical 11 mm crimp-top septum molded from 50 Shore A silicone, I want to see peak cavity pressure between 350 and 550 bar, depending on the gate and the part geometry. Below 300 bar, you get short shots, voids at the parting line, and low crosslink density at the center. Above 600 bar, you get flash, overpacking, and residual stress that shows up later as a split when the autosampler needle hits the same spot twice.

Closed-loop controls on a properly instrumented molding station:

Cavity pressure transducer. This is inside the tool, not inferred from hydraulic pressure. It triggers the switch from injection to hold. The switchover point is set by pressure, not by time or screw position.
Mold temperature. For platinum-cured silicone, the tool runs at 160 to 190°C. A 5°C drift changes cure time enough to move the compression set result. The mold has thermocouples in each plate and a hot oil or electric heater on a PID loop.
Clamp force. For a 64-cavity compression mold, clamp force is often 150 to 250 tons. The press measures actual tonnage at the tie bars. If one tie bar is 10% off, the mold deflects and the outer cavities produce thin septa.
Cure time. This is a timer, but it is interlocked with mold temperature. If the mold drops below setpoint, the timer pauses. A factory that lets the timer run while the mold cools is molding scrap.

What is not closed-loop controlled at this station: the actual crosslink density of each septum. You cannot measure that in the mold. You infer it from a rheometer curve on a sample pulled from the same batch, and you confirm it later with a compression set test. That is an end-of-line check, and it is one reason a lot can pass durometer and still fail reseal after 50 punctures.

Post-cure is where the volatiles leave, and where bleed performance is decided

A molded silicone septum is not finished when it comes out of the tool. It still contains low molecular weight siloxanes, residual catalyst byproducts, and unreacted monomer. If you put that septum into a GC inlet at 250°C, those volatiles bleed into the column and you see ghost peaks. The post-cure oven is the station that removes them.

For a septum rated for 300°C use, a standard post-cure is 4 hours at 200°C in a forced-air oven, followed by a ramp to 250°C for 2 hours. The exact profile depends on the elastomer. Fluorosilicone septa need a longer, hotter post-cure than dimethyl silicone. A factory that skips post-cure to save cycle time is selling you a septum that will fail a bleed test on day one.

The factory controls at post-cure:

Oven temperature profile. Multi-zone ovens log temperature at 3 to 5 points. The tolerance is ±5°C. A single-zone oven with one thermocouple cannot guarantee that the septa in the center of the tray see the same temperature as the ones at the edge.
Airflow. Forced air, measured in m/s or air changes per hour. Volatiles have to be carried away. A static oven post-cures the top layer and leaves the bottom layer loaded with siloxanes.
Load density. Trays are loaded to a maximum grams per tray. Overloading turns the oven into a heat sink and the profile no longer matches the validated cycle.
Time at temperature. This is a timer, but it starts only when the coldest probe in the load reaches setpoint. Not when the oven door closes.

End-of-line check: a headspace GC or thermogravimetric analysis (TGA) on a sample from the lot. The spec I use for a low-bleed septum is less than 0.1% weight loss by TGA at 300°C over 30 minutes, and no individual siloxane peak above 10 ppm by headspace GC. If the factory does not run this, they do not know what they are shipping.

This is the station where the difference between a genuine septa manufacturer and a reseller becomes obvious. A reseller can buy molded parts and call them post-cured. They have no oven log, no TGA trace, and no idea what the volatiles profile looked like before the parts went into the bag.

Slitting and punching: the station that determines coring and needle drag

Many septa are molded with a slit or a cross-cut, but the slit is often added after molding. The slitting station is deceptively simple: a blade, a fixture, and a press. The control problem is that a blade dulls, and a dull blade tears the elastomer instead of cutting it. A torn slit closes differently than a cut slit. It reseals worse, and it cores more.

The factory controls at slitting:

Blade sharpness. Measured by cut force on a test strip, not by feel. A fresh blade cuts with less than 2 N of force. When cut force exceeds 4 N, the blade is changed. This is a scheduled replacement, not a reactive one.
Blade depth. For a 3 mm thick septum, the slit depth is typically 80 to 90% of the thickness. Cut through completely and the septum leaks. Cut too shallow and the needle has to tear the remaining web, which increases coring.
Slit alignment to the mold mark. If the septum has a molded orientation feature, the slit has to align to it. Misalignment causes the needle to enter off-axis, which is the failure mode covered elsewhere on this site, but the root cause here is a fixture that does not locate the part.
Punch diameter and die clearance. For punched septa, the die clearance between punch and die is 0.02 to 0.05 mm. Too tight and the edge shows micro-tears. Too loose and the edge is ragged. The factory checks punch wear with a go/no-go gauge every shift.

This station is mostly operator-dependent, which is why it gets the most attention during an audit. A good factory has a slitting station with a force gauge, a blade log, and a first-piece approval every shift. A bad factory has a box of blades and a guy who changes them when the parts start looking fuzzy.

For autosampler applications, the slit geometry is not generic. A vial septa supplier that understands autosampler needle forces will specify a slit length, a slit pattern (single, cross, or Y), and a slit depth for a specific needle gauge. A 23-gauge needle needs a different slit than a 26-gauge needle. If the factory cannot tell you the slit depth in millimeters, they are not controlling it.

Washing and packaging: the contamination control step most factories underrate

After slitting, septa are washed to remove mold release, cutting debris, and any particulate from the slitting station. The wash is usually a multi-stage process: a detergent wash, a rinse, and a final rinse in deionized water or a solvent. Then the septa are dried in a clean-air dryer.

The controls at washing:

Water quality. The final rinse uses water at 18 megohm-cm resistivity or better. If the factory uses tap water for the final rinse, the septa will carry chloride and sulfate residues into your inlet.
Detergent type and concentration. The detergent has to be compatible with silicone and completely rinseable. A nonionic surfactant at 0.1 to 0.5% is typical. Residue from the detergent is as bad as the contamination it removes.
Drying temperature and time. Silicone absorbs some solvents. The dryer runs at 60 to 80°C, below the point where the septum would begin to re-cure or harden. Drying time is validated by weight loss, not by a fixed timer.
Particulate count in the packaging area. The packaging area should be at least ISO Class 7. I have seen septa packaged in a room with an open loading dock door. The particulate count on those septa was 10 times the spec.

Packaging is also a control point. Septa are packed in clean, low-outgassing bags, often nitrogen-flushed for low-bleed grades. The bag material matters. A polyethylene bag can transfer plasticizers to the septa during storage. The factory should specify the bag material and the seal integrity check.

This is where the HPLC vial and GC consumable supply chain diverges from other molded rubber products. A septum is not a gasket. It sits in a heated inlet and directly contacts the sample path. The cleanliness spec is not optional.

Which parameters are closed-loop controlled, and which are only checked at the end

This is the distinction that separates a controlled factory from a hopeful one. Closed-loop means the process variable is measured continuously and the machine adjusts itself to hold a setpoint. End-of-line means the factory measures the finished part and decides whether the lot ships. Both are legitimate. Confusing them is how you buy bad septa.

Parameter	Control type	Typical setpoint or spec	What happens when it drifts
Mix torque during compounding	Closed-loop	Torque-time integral setpoint	Poor filler dispersion, hardness drift
Mold cavity pressure	Closed-loop	350-550 bar peak	Voids below 300 bar, flash above 600 bar
Mold temperature	Closed-loop	160-190°C, ±3°C	Cure time drift, compression set failure
Post-cure oven temperature	Closed-loop	200°C for 4 h, ±5°C	High bleed, ghost peaks
Blade sharpness at slitting	End-of-line check	Cut force below 4 N	Coring, poor reseal
Durometer (Shore A)	End-of-line check	±3 points from nominal	Needle force changes, leak path opens
Compression set	End-of-line check	Below 20% at 200°C for 22 h	Permanent deformation, no reseal
Volatiles by TGA or headspace GC	End-of-line check	Below 0.1% at 300°C	Column bleed, baseline rise
Particulate count	End-of-line check	ISO Class 7 packaging area	Inlet contamination, detector spikes

The table tells you something important. The parameters that are closed-loop controlled are the ones that determine whether the septum is molded correctly. The parameters that are only checked at the end are the ones that determine whether the septum will perform in your instrument. A factory can hold every closed-loop setpoint and still ship septa that fail compression set, because compression set is a material property that depends on the cure system and the post-cure, and it is only verified after the fact.

Two field cases that show where the line actually fails

Case one: the septum that passed durometer and still leaked. A contract lab running EPA 8260 volatiles on a autosampler vial system started seeing pressure drops after 20 injections. The septa were 11 mm, 50 Shore A, PTFE-faced. Incoming QC showed durometer at 49, right in spec. The failure was compression set. The septa had been post-cured at 180°C for 2 hours instead of 200°C for 4 hours. The factory had shortened the cycle to increase throughput. The septa looked fine, felt fine, and passed the durometer check. They took a permanent set after 20 needle hits, and the inlet lost seal. The fix was to reject the lot, audit the post-cure oven logs, and require a compression set certificate with every shipment. The oven logs showed the shorter cycle immediately.

Case two: the septum that cored on a 23-gauge needle but not on a 26-gauge. A pharmaceutical QC lab switched autosampler needle gauges and suddenly saw coring in 1 out of every 40 injections. The septum was a slit design, molded with a single slit. The slit depth had been set for a 26-gauge needle at 85% of thickness. The 23-gauge needle is stiffer and larger in diameter. It hit the remaining web at the bottom of the slit and tore it instead of parting it. The tear left a small plug of silicone that ended up in the inlet liner. The fix was to specify a deeper slit, 90% of thickness, and to run a needle-force study across the gauge range before qualifying the new septum. The factory had not changed anything. The application had changed, and the slit geometry no longer matched.

Both cases are specific to manufacturing control, not to general septum failure. The first was a post-cure deviation that no incoming durometer check could catch. The second was a slit geometry mismatch that only appeared when the needle gauge changed. A factory that controls the line would have caught the first with an oven log audit and the second with a slit depth specification tied to needle gauge.

What to ask a septa factory before you qualify them

When you are qualifying a septa manufacturer for OEM supply, do not ask for a certificate of analysis and stop there. The CoA tells you what the lot tested at. It does not tell you whether the factory controls the process that made the lot.

Ask these questions, in this order:

Show me the compounding torque curve for the last three lots. If they cannot produce it, they are not controlling dispersion. Walk away.
What is your cavity pressure setpoint and what is the alarm band? A factory that molds septa without cavity pressure transducers is guessing. The alarm band should be ±10% of setpoint.
What is your post-cure oven profile, and can I see the temperature trace for the lot I am buying? The trace should show the load temperature reaching setpoint before the timer starts. If the trace is a flat line at the setpoint, it is fake. Real oven traces have a ramp.
How do you decide when to change the slitting blade? The answer should be a force measurement or a part count, not “when it looks dull.”
What is your compression set spec and how often do you test it? The spec should be below 20% at 200°C for 22 hours, tested on every lot. If they test once a quarter, they are not controlling the material.
What is the particulate class of your packaging area? ISO Class 7 minimum for clean septa. If they do not know, the septa are not clean.

These questions map directly to the stations in this article. A factory that controls compounding, molding, post-cure, slitting, and packaging will answer all of them without hesitation. A factory that does not will start talking about how long they have been in business.

For reference points on what a septum has to survive in a real GC inlet, the Agilent GC consumables documentation lists inlet temperature ratings and recommended septum change intervals. Thermo Fisher publishes septum compatibility guides for their inlet designs. For methods that push septa hard, EPA SW-846 methods like 8260 and 8270 specify inlet conditions that a septum must survive without bleeding or leaking. Restek has technical notes on septum bleed and reseal that are worth reading before you write a spec.

The decision rule is simple. A septa factory either controls the line at the station level, with closed-loop feedback on compounding, molding, and post-cure, plus defined end-of-line checks on slit geometry, compression set, and volatiles, or it does not. If the factory cannot show you the torque curve, the cavity pressure trace, and the oven temperature log for the exact lot you are buying, you are not buying a controlled product. You are buying a hope. Specify the control evidence in your purchase order, not just the durometer and the dimensions. That single change eliminates most of the septa failures that show up later as ghost peaks, inlet leaks, and cored liners.

References & further reading

Related resources from our lab consumables network:

External references and standards: