A 12 x 32 mm clear borosilicate run drifted out of spec at 2:14 PM on a Tuesday. The glass tubing OD had crept from 11.98 mm to 12.04 mm over 41 minutes, and the forming station kept accepting it because the laser micrometer’s alarm threshold was set to ±0.08 mm. The shift lead caught it only when the annealing lehr’s end-of-line camera flagged a 0.3 mm shoulder height variance. That is what a voa vial factory actually controls on the line: not just the final dimensions, but the process drift between stations that lets bad vials get made for 41 minutes before anyone notices.
Most lab buyers think of vial manufacturing as a molding step plus a QC check. It is not. A properly run voa vial line is a sequence of closed-loop and open-loop controls, and the difference between those two categories determines whether you get 100 leak-free vials or 94 good ones and 6 that fail under pressure. This article maps the line station by station, names the parameters that are actively controlled versus only inspected at the end, and gives you the numbers to ask for when you audit a supplier.
Vial manufacturing starts with borosilicate glass tubing, usually 51 expansion or 33 expansion depending on the vial grade. The tubing arrives in 1.5 meter lengths with a spec for OD, wall thickness, and camber. The factory’s first real control point is the conversion lathe or vertical forming machine, where the tubing is cut, heated, and formed into a vial body with a closed bottom.
Three parameters are closed-loop controlled here. The first is tube OD, measured by a laser micrometer at 200 to 400 readings per second. The feedback loop adjusts the forming mandrel position and the burner gas flow to hold the OD within ±0.05 mm of nominal. The second is wall thickness, measured by a capacitive sensor after forming. Wall thickness variance above 0.02 mm across the vial body creates stress points that crack under autosampler clamping force. The third is bottom thickness, controlled by the dwell time of the glass in the forming zone. Bottom thickness below 0.8 mm is a puncture risk; above 1.5 mm wastes glass and shifts the center of gravity.
What the factory does not control in real time at this station is the glass composition itself. That is locked in by the tubing supplier. A factory that does not verify the expansion coefficient of each incoming lot is gambling. I have seen a lot of 51 expansion tubing mixed with 33 expansion cullet during recycling at the tubing plant, and the resulting vials cracked at the shoulder when heated to 300°C in a GC headspace run. The fix was not on the forming line. It was a lot-level incoming inspection using a dilatometer, which is a destructive test done on a sample, not a line control.
Annealing is a time-temperature integral, not a temperature setting
After forming, vials pass through an annealing lehr. The glass enters at around 565°C and exits at room temperature over a controlled ramp. The parameter that matters is the annealing point soak time, not just the peak temperature. For 51 expansion borosilicate, the annealing point is 565°C, and the vial needs to hold within ±5°C of that for at least 20 minutes to relieve forming stress. The lehr belt speed and zone temperatures are closed-loop controlled, but the actual time-temperature integral for each vial is only estimated, not measured individually.
This is where a lot of factories cut corners. They run the lehr at 580°C to speed up throughput, which reduces soak time but leaves residual stress. The end-of-line check for stress uses a polariscope, which shows birefringence patterns under polarized light. A well-annealed vial shows a uniform gray field. A stressed vial shows rainbow fringes at the shoulder and bottom radius. The problem is that the polariscope check is a go/no-go inspection at the end, not a closed-loop control. If the lehr drifts, you find out after 3,000 vials are already annealed wrong.
The number to ask for: residual stress below 4 nm/cm optical retardation, measured per ASTM C978 or equivalent. Anything above that and you will see spontaneous cracking in storage, especially in vials shipped through cold climates where thermal shock adds to residual stress.
Shoulder and neck forming is where the autosampler fit is decided
The shoulder profile and neck finish are formed either in the same conversion step or in a secondary forming station. This is the station that determines whether the vial seats correctly in the autosampler tray and whether the cap seals evenly. The critical dimensions are the neck ID, the neck OD, and the shoulder angle.
Neck ID is controlled by a plug mandrel during forming. For a standard 9 mm screw thread finish, the neck ID spec is 7.6 mm ±0.1 mm. The neck OD is controlled by the forming rollers and checked with a go/no-go ring gauge. The shoulder angle, measured from the vial body axis, should be 30° ±2° for most autosampler-compatible vials. A shoulder that is too steep causes the needle to deflect when it descends; too shallow and the needle hits the vial wall instead of the bottom center.
One field case I dealt with involved a customer whose HPLC vial caps were cross-threading on about 3% of vials. The cap supplier swore their vial septa and cap dimensions were in spec. We measured the neck OD on 200 vials from the customer’s stock and found a bimodal distribution: one population at 8.95 mm and another at 9.05 mm. The forming station had two mandrels, and one was worn by 0.1 mm. The line’s SPC chart showed the combined distribution in control because the two populations overlapped enough to hide the shift. Fix: replace the worn mandrel and add a per-mandrel measurement channel instead of a pooled one.
The bottom radius is not cosmetic
The bottom radius, where the vial wall transitions to the flat bottom, is formed during the initial glass conversion. It is easy to overlook because it is not printed on most spec sheets. But it matters for two reasons. First, a sharp inside corner creates a stress concentration that fails under thermal cycling. Second, a radius that is too large reduces the flat bottom area, which makes the vial unstable in the autosampler tray.
The spec I hold suppliers to: inside bottom radius of 1.5 mm to 2.5 mm for a 12 mm OD vial. This is measured by sectioning a sample vial and using an optical comparator, which is an end-of-line destructive test, not a line control. The forming station controls the radius indirectly through the mandrel shape and the glass temperature at forming. If the glass is too cold, the radius comes out sharp. If too hot, the bottom sags and the radius becomes too large.
A GC lab once reported that their 2 mL vials were tipping over in the autosampler tray during high-throughput runs. The tray was a 100-position design with tight pockets. We measured the bottom flat diameter on 50 vials and found it ranged from 8.2 mm to 9.8 mm against a spec of 9.0 mm ±0.3 mm. The forming station’s bottom mold was worn, producing a convex bottom that reduced the contact area. The lab had been compensating by running the autosampler at 80% speed. The fix was a new bottom mold, and the vials returned to flat-bottom spec.
Surface treatment is a chemical process, not a rinse
After annealing and forming, vials go through a surface treatment line. For most HPLC and GC applications, the treatment is a deionized water wash followed by a silanization or surface passivation step for specific applications. The wash line controls water resistivity (typically above 18 MΩ·cm), rinse temperature (60°C to 80°C for effective removal of forming lubricants), and air knife drying pressure (0.4 to 0.6 MPa). These are closed-loop controlled with conductivity sensors and pressure transducers.
What is not controlled in real time is the residual particulate count on the vial interior. That is measured at the end of the line using a liquid particle counter on a rinse sample from a statistically selected vial. A spec of fewer than 10 particles ≥ 10 µm per vial is achievable on a well-run line. The wash station’s filter integrity and rinse nozzle alignment are the upstream controls, but the final count is an inspection, not a closed-loop variable.
For silanized vials, the treatment is a vapor-phase deposition of a silane such as dimethyldichlorosilane or a liquid-phase coating. The process controls are silane concentration (typically 2% to 5% in a solvent carrier), exposure time (30 to 60 seconds), and cure temperature (100°C to 120°C for 30 minutes). The output is checked by measuring the water contact angle, with a spec of 95° to 110° for a properly silanized surface. This is an end-of-line test on a sample, not a continuous monitor.
Cap and septa assembly is a torque and puncture problem
Many factories ship vials and caps separately, but some lines assemble them. When assembly happens on the line, the two parameters that matter are cap application torque and septa compression set. For a 9 mm screw cap, the application torque should be 0.09 to 0.11 N·m. Below 0.09 N·m, the seal is unreliable and the cap can back off during shipping or autosampler vibration. Above 0.11 N·m, the septa is over-compressed, which reduces its resealing ability after needle puncture.
The septa material matters here. A PTFE/silicone septa has a compression set spec of less than 15% after 22 hours at 70°C per ASTM D395. A septa that exceeds this will not recover after the needle is withdrawn, and the vial will leak on the second or third injection. The assembly station controls torque with a servo-driven capper that has a torque transducer feedback loop. The compression set is a material property tested on incoming septa lots, not on the assembly line.
A field case: a customer running 96-well plate format autosampler with 2 mL vials reported intermittent injection volume errors. The autosampler’s needle was piercing the septa but the system detected a pressure drop on the second draw from the same vial. We measured the septa thickness before and after puncture. The septa was 1.3 mm thick, but after one puncture the thickness at the puncture site was 0.7 mm, a 46% compression set. The septa lot was out of spec on compression set. The fix was replacing the septa lot and adding a compression set test to the incoming inspection plan, which the factory had skipped because the septa supplier’s COA looked fine.
End-of-line inspection is a filter, not a control loop
Every station I have described so far has a mix of closed-loop controls and end-of-line checks. The end-of-line inspection system typically includes camera-based dimensional checks, a leak test, and a visual defect sort. The camera system measures height, OD, neck ID, and shoulder profile at rates up to 600 vials per minute. The leak test applies a pressure differential or vacuum to the sealed vial and measures the decay rate. A typical spec is no leak greater than 0.02 sccm at 0.3 bar differential.
These end-of-line systems are good at catching gross defects: a missing cap, a cracked vial, a grossly out-of-spec dimension. They are not good at catching slow drift in a forming station, because they are sampling or inspecting at line speed without feedback to the upstream process. A factory that relies on end-of-line inspection to guarantee quality is running an open-loop process with a filter at the end. A factory that uses closed-loop controls at each forming station and uses end-of-line inspection only for final verification is running a controlled process.
The distinction matters when you audit a supplier. Ask to see the control charts for the forming station, not just the final inspection pass rate. A final inspection pass rate of 99.8% sounds good until you learn that the forming station’s OD chart shows a 0.06 mm drift over the shift that the end-of-line camera was not programmed to catch. The camera was set to reject at ±0.1 mm, so the drift never triggered a rejection. But the vials shipped to you were at the edge of the spec, and your autosampler’s needle guide is less forgiving than the camera’s tolerance band.
Decision checklist for auditing a voa vial line
When you visit a factory or review their process documentation, use this checklist. Each item is a concrete control point with a number. If the factory cannot answer the question with a number, that is your answer.
- Tube OD control: Is the forming line using a laser micrometer with a closed-loop feedback to the mandrel? What is the alarm threshold? It should be ±0.05 mm or tighter.
- Annealing soak time: What is the lehr belt speed and zone temperature profile? Can they show a time-temperature integral calculation for the last production lot? Residual stress should be below 4 nm/cm.
- Neck ID and OD: Are neck dimensions measured per mandrel or pooled across all mandrels? Per-mandrel is required to catch wear on a single forming head.
- Bottom radius: What is the inside bottom radius spec and how often is it sectioned and measured? A destructive test frequency of one vial per shift per forming head is a minimum.
- Wash water resistivity: What is the resistivity set point and is it alarmed? 18 MΩ·cm is the target; anything below 10 MΩ·cm means the rinse is leaving ionic residue.
- Cap torque: For assembled vials, what is the torque set point and what is the feedback loop? 0.09 to 0.11 N·m for 9 mm caps.
- Septa compression set: Is compression set tested on incoming lots per ASTM D395? What is the reject limit? 15% maximum after 22 hours at 70°C.
- Leak test sensitivity: What is the leak rate reject threshold and how is it calibrated? 0.02 sccm at 0.3 bar is a reasonable spec.
This checklist is not a substitute for a full audit, but it separates factories that control their process from factories that sort their output. The difference shows up in your lab as a 0.3% failure rate versus a 3% failure rate, and that difference costs you more in troubleshooting time than the vial price difference ever will.
The decision rule: buy from a factory that can show you closed-loop control charts for the forming and annealing stations, not just a final inspection certificate. The certificate tells you what they caught. The control charts tell you what they prevented.
For further reading on vial specifications and autosampler compatibility, the technical resources at Agilent and Thermo Fisher cover instrument-side requirements. The U.S. EPA SW-846 methods define the performance criteria for vials used in environmental analysis, and Restek publishes practical guidance on vial selection and handling.
References & further reading
Related resources from our lab consumables network:
External references and standards: