Plastic Welding Hermetic Seal Failures: Causes & Fixes

The part looked perfect: a clean flash ring, a smooth seam, no voids visible under a loupe. Then it hit the helium leak detector and the needle pinned to full scale. A quality engineer staring at a flawless-looking plastic weld that leaks is one of the most confusing failures in manufacturing — and it is far more common than most people assume, because plastic welding is a fusion process where the seal is created by heat, pressure and time acting on a moving, moisture-sensitive, dimensionally imperfect material.

Why a Perfect-Looking Plastic Weld Still Leaks

A molded reservoir came off the line, flash trimmed, seam glossy and continuous. Under a loupe it looked textbook. Then it went onto the helium detector and the needle pinned to full scale. The engineer could not point to a defect, and that was the problem: there was no visible defect to point at. The leak existed only to gas.

Hermetic seal failure in plastic welding is rarely a mystery and rarely random. It is almost always a root-cause trio: joint and part design that cannot physically produce a continuous molten path, material condition (hygroscopic moisture, mold release, regrind, thermal degradation) that stops the melt from wetting out, and slightly wrong process parameters, which leave you a partial fusion rather than a full one.

There is no filler metal to save you

Most people arrive at plastic welding carrying mental models from metal welding, and those mislead. In a steel weld, filler metal physically bridges the gap and fills the joint. A hermetic plastic weld is a fusion joint. Nothing is added. The seal exists only where molten polymer actually flowed, wetted and interdiffused across the interface between the two parts.

That reframes the question. For a thermoplastic weld the only question that matters is whether the resin went molten, moved under pressure, wet the mating surface and interdiffused across the entire seal land. Miss any stretch of the circumference and you have a joint that is merely two parts sitting next to each other. The fundamentals of continuous molten paths and wetting behaviour are laid out in this primer on hermetic seam sealing, and they apply to plastic just as much as to metal.

Why looking at the weld tells you almost nothing

A weld can look continuous and still contain a continuous micro-void running along the interface. It can also contain a cold joint with essentially zero bond strength. Both live inside the joint line, where the flash hides them. Neither produces a visible gap, burr or discolouration on the outside.

This is where visual inspection fails so badly. The only things that reliably expose these defects are pressure or vacuum decay testing and, for fine leaks, helium mass spectrometry, which remains the reference method for leaks down in the 10⁻⁶ to 10⁻⁸ standard cubic centimetres per second range. A part that looks fine tells you nothing about whether it is sealed, and nothing about whether gas is creeping through the bulk polymer itself, a separate mechanism even a good weld will exhibit.

Failures are process-caused, not random

Leaks do not appear because a weld went bad by chance. They appear because something upstream made a continuous molten path impossible, stopped the resin wetting out, or left the joint only partly fused. Fix the cause and the leak rate changes. Change parameters blindly and you waste weeks.

The evidence is strong. A technical report on hermetic connector welding, available through the U.S. Department of Energy’s technical information database, found that up to 70 percent of ten-pin connectors leaked above 0.003 mm³/s STP after welding with established processes and no epoxy pre-seal. The fix was not a new resin, part or seal geometry. It was heat sinking plus segmented welding with compressed argon cooling between segments, which brought acceptable rates below 0.00001 mm³/s STP. The process was the cause, so the process was the cure.

The same report shows a second truth: welding heat near a finished seal disturbs that seal. Research on laser welding of hermetic glass-to-metal connectors showed local weld heat reduced radial compression in the sealing glass by as much as 10 ksi. A weld can pass every test the day it is made and still be compromised by how much heat travelled. Anyone working with laser welding in an aerospace welding context will recognise that pattern.

What this article will give you

Over the next sections: the four cause families behind nearly every hermetic weld leak, the leak testing methods that actually find them, a repeatable root-cause workflow, and the controls that stop them coming back.

• Cause 1 — Joint design that cannot make a continuous seal. Step height, leak-off grooves, corner radii, projection flatness and tolerance stack-up.
• Cause 2 — Materials that are wet, dirty or degrading. Moisture in nylons and polycarbonates, mold release on the joint line, regrind and contamination, thermal breakdown.
• Cause 3 — Process parameters that produce partial fusion. Heat, force, time and, most often overlooked, the cooling hold before the part leaves the tool.
• Cause 4 — Fit-up variation, post-weld damage and ageing. Warpage, out-of-flat lids, impact and fastener loads, pressure cycling, chemical attack and creep.

The order mirrors how you should work a failed part: geometry first, then material, then process settings, then everything that happened after the weld. Work it in that order and the answer usually arrives faster than you expect. Skip it and you will spend a month tuning a machine that was never the problem.

What “Hermetic” Actually Means for a Welded Plastic Part

The most common mix-up in welded plastic specs: “hermetic” is not a yes/no property. It is a number, usually a maximum allowable leak rate. Engineers use these words loosely, and mixing them up is how a part passes one department and fails another.

• Gas-tight: holds pressure against gas. What a fuel system or pneumatic reservoir needs, and the loosest term here.
• Watertight: blocks liquid water under low pressure. Relevant to appliance housings and outdoor equipment, but silent about gas.
• Pressure-tight: holds a defined differential pressure, usually gas, with the spec naming pressure and duration.
• Leak-proof: the vendor’s word, not an engineering one. Treat it as a claim to be tested, not a requirement to trust.
• Hermetic: the strictest. Leak rate below a stated threshold, under a stated condition, by a stated method.

The unit everyone actually uses

Leak rates for welded plastic parts are quoted in atm·cc/s (atmosphere cubic centimetres per second), also written scc, most often with helium as the test gas. MIL-STD-883 Method 1014 defines a standard leak rate as the quantity of dry air at 25 °C, in atmosphere cubic centimetres per second, flowing through a leak when the high-pressure side is at one atmosphere and the low-pressure side is at not more than 1 mm Hg.

A leak rate means nothing without a reference pressure differential, temperature and gas. “10^-6 scc” without those conditions is not a specification, it is a guess.

Fine leakage spans many orders of magnitude, so engineers sort it into two bands:

• Gross leaks are large, obvious holes. A gross-leak bubble test typically uses a high-boiling-point perfluorocarbon such as Fluorinert FC-40 or FC-43 held at 125 °C in a back-lit bath, where a stream of bubbles both detects and pinpoints the leak. Hermetic seam sealing guidance calls this the standard first screen for gross leaks.
• Fine leaks are too small to bubble, sitting roughly in the 10^-6 to 10^-9 atm·cc/s range and belonging to helium mass spectrometry, which is why the reference method at Los Alamos National Laboratory accepts parts at limits around 1.0 × 10^-6 standard cc/sec.

Why “just barely sealed” is not good enough

The consequence of a fine leak is not the hole, it is what passes through it over time. Classic hermetic packaging analysis shows a leak rate of 10^-8 atm·cc/s would admit the equivalent of nearly 100,000 layers of water onto a chip over ten years, which is why 10^-13 atm·cc/s is often cited as the rate needed to truly limit exposure. Hermetic encapsulation literature works through that reasoning for electronic packages.

Apply it to a welded plastic fuel tank or a medical fluid reservoir and the arithmetic holds. A joint passing at 10^-6 in the lab may still admit enough moisture or volatile vapour to matter across a ten-year service life. The spec, not the pass/fail sticker, is the deliverable.

No welded thermoplastic is hermetic forever

This unsettles people who assume a weld is a barrier. A polymer is not a gas barrier by default. Molecules migrate through amorphous regions of the bulk material at a rate set by polymer family, temperature and pressure differential across the wall. Hermetic connector qualification guidance is blunt: seal performance degrades with temperature and time, and permeation through the polymer is separate from leakage through a joint.

A welded part can therefore be perfectly fused, with full bond strength and zero voids, and still slowly lose gas or take on moisture because the material itself lets molecules through. Design rules such as matched corner radii and uniform wall thickness (design guide material) cannot fix permeation, and extra welding heat will not help either.

Where hermetic plastic welds show up

• Automotive fuel systems: tanks, fuel rails, bladders and airbag bolsters, where slow pressure loss triggers a fault code.
• Medical fluid handling: IV containers, infusion sets, fluid reservoirs and diagnostic cartridges, where sterility dominates.
• Water and filtration: filter and RO membrane housings, usually needing watertight rather than gas-tight performance.
• Refrigerant and pneumatic components: small hermetic housings where refrigerant loss makes the assembly scrap.
• Inhalers and consumer appliances: canister valves and applicators where dose consistency depends on an intact seal.

Processes differ in what they can deliver. Emerson Branson’s plastics joining literature states that vibration welding creates high-strength, leakproof hermetic seals with most thermoplastics, while hot plate welding is chosen where both high-strength and hermetic seals are required.

Form one habit: whenever “hermetic” appears in a print, a quote or a supplier email, ask for the number, the gas, the pressure conditions and the test method behind it. “Hermetic” alone is a wish. “Less than 10^-6 atm·cc/s helium, fine leak, per MIL-STD-883 Method 1014″ is a requirement you can weld to, test against and hold someone to.

Cause #1 — Joint Design That Cannot Make a Continuous Seal

Some leaks are not weld problems at all. They are geometry problems wearing a weld problem’s clothes. If the joint was drawn wrong, no amount of dwell time, amplitude, platen temperature or clamping force will create a continuous molten path, and the machine will dutifully produce a beautiful, fully welded, entirely non-hermetic part. Rule this family out first, because it is the one parameter tweaking can never fix. The good news is that design failures are usually visible on the drawing long before anyone loads a leak detector, and each one can be audited in minutes in CAD rather than by trial and error on the production floor.

Step depth: give the melt somewhere to go

Most hermetic plastic joints work as a step, a shear or a small land that the molten material must completely fill. Step depth sets the volume of melt needed to displace air and bridge the gap. Too shallow and the molten reservoir has nowhere to go: material flashes at the surface, the seam looks continuous, and the interface underneath is only partially fused. You get a micro-void running around the joint that a bubble test will not find but a helium test finds instantly. Too deep and you get the opposite problem, excess melt, heavy flash, and more heat dumped into the part than the wall thickness can absorb. The rule of thumb is that step depth should comfortably exceed the wall thickness, and it should be checked against material shrinkage, since a step sized for a low-shrink resin can starve on a high-shrink one.

Leak-off grooves: decide where the gas goes

A pressure-relief or leak-off groove is a deliberately designed vent path that lets displaced air and moisture vapour escape as the melt closes the joint. Without one, gas is trapped in the weld and frozen in place, and an undersized groove is worse than none because the flash is forced to travel somewhere unintended. If the joint is pressure-loaded in service, that trapped gas is not just a void but a stored path for leak-driven stress relaxation. Design the groove before you design the flash pattern, and size it for the volume of material that will actually squeeze out.

Corners, radii and the overhang trap

Corners are where good hermetic designs go bad. A lid whose corner radius is smaller than the package radius sits correctly on the flats and bridges across the corner, with the lid’s step fillet radius landing directly on the seal ring and leaving a gap between them. That gap is not a manufacturing defect. It is a designed-in leak path, straight from inside the part to outside, and it will survive every parameter change you make. An overhanging lid fails the same way: the lip extends past the seal ring, so the flash forms outboard of the seal instead of over it, and the actual sealing surface is never compressed. The fix is matched radii and real step clearance, not a hotter weld.

Unwelded corners are a well-known potential leak source, and the practical fix is to weld beyond the lid size so a flash overlap forms at the corners. That overlap does more than seal. It gives you a visual indicator of a hermetic seam, which is worth having when you are sampling rather than testing every part. Hermetic seam sealing fundamentals cover the same principle in detail, and published guidance on avoiding seal failures in hermetic assembly processing makes the same design-first case.

Land width, projections and flatness

Three more geometry errors account for a large share of what shows up as “leaky welds.”

• Weld land or flange too narrow. There has to be enough flat surface for the process to land on and for the flash to sit without overhanging. Too narrow and the tool cannot hold the part square, the melt has no stable path, and the joint is a leak waiting for a pressure cycle.
• Non-continuous projections. Interrupted, inconsistent or broken projection rings leave unwelded gaps in the seal, which is a direct definition of a non-hermetic joint. If a projection ring is designed, every projection around the circumference has to be the same height, the same shape and continuous.
• Flatness and straightness. Projection flatness must be held to better than 0.002 in (50 µm). When the ring is not flat, the high point over-welds, blows out and sinks into the part while the low point never reaches melt temperature at all. The result is the classic failure: a beautiful, thick flash at one location and a leak at another. Lid edges must be straight within 0.002 in (50 µm) for the same reason, and the ultrasonic welding design guide spells out these tolerances and the mating-part geometry that goes with them.

Design flaw	What it produces	How it shows up
Shallow or absent step	Partial fusion, micro-void at the interface	Fine leak on a seam that looks perfect
No leak-off groove	Trapped gas and volatiles in the melt	Localised porosity, leak near the low point
Mismatched corner radius	Gap between fillet and seal ring	Leak that follows the corner exactly
Flatness out of tolerance	Over-weld at the high point, no weld at the low point	Heavy flash on one side, leak on the other
Broken projection ring	Unwelded gaps in the seal	Leak at an unrelated location along the ring

The design rule checklist

Before you touch a parameter, confirm the drawing does the following:

• Generous step depth, sized for the wall thickness and the material’s shrinkage.
• A designed leak path or pressure-relief groove with a real exit for displaced gas.
• Matched corner radii and no lid overhang beyond the seal ring.
• Uniform wall thickness at the joint, so heat input is the same all the way around.
• A defined shear-tab or knock-out feature, so the lid comes off without peeling the seal and the tear-off path is designed rather than improvised.
• A geometry that can be leak-tested at 100 percent, or through a sampling plan you have actually validated, with a port or a dunk-testable configuration. Process capability, including joint design guidance in the plastic joining design material, is worth reading before the tool is bought, not after the first leaks arrive.

Cause #2 — Materials That Are Wet, Dirty, or Degrading in the Weld

A flawless process can still produce a leaky part if the resin arrived wet, dirty or heat-damaged. A contaminated joint will not fuse, no matter how good the arc was, and that is the lesson metal fabricators learn the hard way. Plastic welding follows the same logic, only the contaminants are invisible. Absorbed water, mold release, regrind and thermal history are all inside or on the part before the first vibration, horn or laser pulse touches it.

Moisture: the most common material cause

Polyamide (nylon), polycarbonate, polycarbonate/polyester alloys and polysulfone are hygroscopic. They pull water out of the air and hold it in the bulk of the moulded part. At the weld that trapped moisture flashes to steam, bubbles out through the melt and leaves voids and porosity along the bond line. It is the porosity mechanism you would recognise in a metal weld, except here it hides along an interface nobody looks at while the outer surface stays presentable.

Moisture also changes the process, which is why it is misdiagnosed so often. Wet resin takes longer to reach welding temperature, so cycle times stretch, and the bond that does form is weaker because melt energy went into boiling water rather than wetting out the substrate. Ultrasonic welding is the fusion process most exposed to part condition, and the DuKane ultrasonic welding design guide puts it plainly: dry, clean, dimensionally consistent parts are a precondition for a repeatable joint, not a refinement of one.

Contamination on the joint line

Mold release agents and lubricants on a moulded joint land reduce friction and wetting. Weld energy that should have squeezed out the melt and interlaced it with the substrate gets spent sliding the parts against each other, leaving a partial bond instead of a continuous one and a leak path that survives. Ask your molder for silicone-free, weld-friendly releases, and specify whether the joint must be wiped with IPA, plasma treated or otherwise prepared. Our guide to prepping aluminum for welding makes the same point for metal: when you are fusing rather than brazing, joint cleanliness is not optional.

Metallised parts bring their own chemistry. Electro-plated nickel generally welds well, but organic brighteners left in the plating can generate gas in the weld and produce excessive porosity, so plating rinse and cleanliness requirements belong on the drawing. The NASA hermetic sealing guidance covers this for space-qualified assemblies, where trace contamination on a plated lead frame separates a qualified seal from a rejected one.

Regrind, recycled content, fillers and pigments

Regrind arrives with unknown thermal history, inconsistent melt flow index and contamination risk from other colours, flame retardants and foreign particles. Each of those changes melt viscosity, wetting and shrinkage, and therefore changes the joint. If regrind is permitted in a leak-critical part, cap the percentage, set a particle and contamination limit, and keep it in a controlled stream with dedicated storage and labelling rather than an open hopper.

Fillers, pigments and glass fibre change shrinkage, which changes fit-up, which changes the weld. Glass fibre also raises stress concentrations at the bond line and wears tooling. A batch that is “the same resin” to purchasing can behave like a different material on the floor.

Degradation, volatiles and outgassing

Excess temperature and dwell break the polymer chain. Chain scission lowers molecular weight, embrittles the joint and generates gas and volatile fragments with nowhere to go. Watch for yellowing to brown or tan flash, a burnt or stale smell, brittleness at the flash rather than in the bulk, and popping during the weld. The hermetic seam sealing fundamentals describe how such voids and contamination paths undermine an otherwise sound seal.

Material expelled inside or outside the weld perimeter usually points to flatness issues, insufficient welding force or excessive weld energy. Trapped inside the package it becomes loose debris that fails Particle Impact Noise Detection testing, so the fix is a process and part-flatness fix, not a filter.

Material handling rules that stop this class of failure

1. Traceability. Record resin grade, lot, molder, tool number, regrind percentage and conditioning history for every sealed part.
2. Moisture limits. Set a maximum moisture content or conditioning requirement per resin and verify it on incoming lots.
3. Drying parameters. Fix time and temperature per resin and cap how long parts may sit open before welding.
4. Incoming inspection. Check mold release, flash, flatness, colour and contamination, not just dimensions.
5. Cleaning. Specify cleaning method, frequency and validation, then confirm the bond improves.
6. Change control. A new resin, release agent, regrind source or supplier should trigger requalification.

Welding is generally more repeatable than soldering and brazing, which is why contamination and part flatness dominate material-side risk in plastic sealing. When a leak turns up, check the resin lot, the release agent and the conditioning record first.

Cause #3 — Process Parameters That Produce a Partial Fusion

A plastic weld doesn’t rely on filler metal to create its hermetic seal. It relies on molten polymer wetting out across the entire joint interface and solidifying as one continuous mass. If any of the process settings fall out of balance, that bond can be incomplete while still looking perfectly fused to the naked eye. The four parameters to watch are heat, pressure, time, and cooling hold.

Variable	Too little	Too much
Heat	Cold joint or partial fusion with micro-voids along the bond line.	Polymer degradation, excessive flash, shrinkage, and locked-in internal stress.
Pressure	Insufficient squeeze-out, trapping air and volatiles in the joint.	Material ejected from the seal land, thinning, and uneven bonding across the interface.
Time	Incomplete melt with little to no cohesive bond.	Excessive dwell, flash buildup, and heat damage to nearby features and tolerances.
Cooling hold	Part released while soft and moving, allowing the seal to open as it cools.	Wasted cycle time with little benefit to a joint already solidified.

Ultrasonic welding

Ultrasonic welding uses high-frequency vibrations, typically at 15 kHz or 30 kHz, under clamping pressure for a set weld time. As outlined in Dukane’s ultrasonic welding design guide, the energy is extremely concentrated. With such a tight energy budget, a small loss goes a long way. Worn or gouged horn faces, contamination, or incorrect amplitude, force, or weld time can all leave a clean-looking flash ring with no bond behind it. Trigger force is especially critical: too high can fire before full seating and capture a step-height error into the seal, while too low can allow variation to dictate the trigger point. Moisture is also a major issue, as vaporizing absorbed water pulls energy away from the interface. For those looking to compare this process to others, our guide to Alpha Laser Welding: Systems, Uses, and Safety Guide can help put its strengths into context.

Vibration welding

Vibration welding runs at a lower frequency with a larger amplitude, oscillating parts under pressure until collapse before a dwell and cooling hold. The most deceptive failure here is an attractive, even flash ring with voids beneath it. Collapse pressure set too low is often the cause, while excessive amplitude can degrade the surfaces and generate gas. Above all, releasing the part before the cooling hold is complete is a frequent error. The joint may appear stable at the tool, but movement as it cools is enough to break the seal.

Hot plate welding

Hot plate welding heats an entire joint face against a platen, making it well suited for applications that require both strength and a hermetic seal. As Emerson Branson notes in its plastic joining technical brochure, platen temperature, contact time, and the cooling hold must all be controlled. Unlike localized processes, the heat is distributed over the whole flange, which can cause it to dish or bow as it cools. A joint that was perfectly concentric on the tool can fall out of alignment once released, placing strain directly onto the seal.

Laser welding

With laser welding, focus position becomes a fourth variable alongside energy density and travel speed. If focus drifts or a mold-release film is present on the joint, fusion can be strong in the middle and cold at the edges, creating a narrow leak path around the perimeter. Heat can also affect neighboring features. In one hermetic connector study, a nominally 350 W laser weld could transition between keyhole and conduction mode, and the resulting shell contraction decreased the compression applied to the sealing glass. The fusion itself was sound, but the geometry the seal relied on had been moved by the weld.

Spin and infrared welding

Spin welding relies on frictional heat, so a controlled torque ramp is needed to generate an even melt around the full circumference. Releasing too early often produces a seal that leaks on just one side of the closure. Infrared welding eliminates contact altogether, but that benefit comes with a reliance on heat uniformity. Hot and cold spots across the joint can leave an incompletely fused ring that is nearly impossible to detect by visual inspection alone.

The Cooling Hold and Heat Transfer Matter Most

Across every process, the cooling hold is where many hermetic seals are lost. A joint pulled from the tool before it is fully solidified is still mobile, and any handling or flexing can open a leak path that won’t appear until leak testing. Time this interval for your process and, if in doubt, extend the hold before adjusting more aggressive parameters.

The same principle applies to heat near an already sealed feature. A report on hermetic connector welding found that up to 70 percent of connectors welded by established processes showed gross leaks above 0.003 mm3/s STP, traced to welding heat disturbing the seal (study abstract). The solution was not a weaker weld, but a massive copper heat sink with silicone thermal joint compound paired with a 12-segment GTA weld and compressed argon cooling between segments, bringing acceptable leak rates under 0.00001 mm3/s STP (the full report). The lesson is simple: you can over-weld for a neighboring feature and create a partial fusion failure by geometry alone.

Cause #4 — Fit-Up Variation, Post-Weld Damage, and Aging in Service

Some leaks never happened at the weld station. The joint was fully fused, the flash was clean, and the part passed its incoming helium test. Then it spent a weekend in a vehicle, a month in a chemical cabinet, or ten minutes under a clamp, and it started to leak. This cause family frustrates root-cause work most, because the failure looks like a weld defect and is not one.

When the two halves simply do not match

Fit-up variation is the quietest member of this family. Warpage in a moulded lid, sink marks near a thick boss, flash left on the parting line, a gate vestige on the seal land, or a lid that is out of flat all change where the load lands. A lid sitting even a fraction of a millimetre high at one corner drives the melt to fuse hard there while barely touching at the other: the classic over-weld at the high point, leak at the low point geometry. That is why a weld that looks beautiful along one side of the part can leak at the other. The hermetic seam sealing fundamentals from Amada Weld Tech put it plainly: seal reliability depends on how evenly the joint closes, not only on how hot the tool is. Two parts from the same tool, moulded minutes apart, can behave completely differently.

Damage that arrives after the joint is made

Once a welded part leaves the weld cell it is still fragile, and most damage happens afterwards: dropping it on a hard floor, impact from a fixture, over-tightening a fastener, clamping in a vice or an assembly jig, or stacking parts under load. Polymer flash is thin and brittle and peels easily, so a cracked flash is a direct path to atmosphere while the joint underneath still looks perfectly fused. Sequencing matters too. Fasteners, clamps and press fits applied while the joint is still hot and settling load a weld that has not finished crystallising, which is exactly when an apparently sound weld most often fails.

Hermetic feedthrough work shows how easily a good part is destroyed after the fact. The Glenair guidance on hermetic connector qualification describes side-loading a pin to fit a cable, which crushes the glass insulator, and soldering the connection for more than about five seconds, which lets heat travel back into the seal and cause thermal shock. Neither fault shows up in the weld, and both end in a leak.

What service does to a sound weld

• Pressure cycling and pulsation work the seam open slightly on every cycle.
• Thermal cycling and CTE mismatch load the joint each time temperature changes, because the welded part and its neighbouring assembly expand at different rates.
• Environmental stress cracking from solvents, fuels, cleaners and oils often starts at a flash edge, gate vestige or scratch rather than in the weld itself.
• Hydrolysis and UV attack polyamides and polyesters in hot water and degrade polyolefin surfaces over time.
• Creep and swelling under sustained internal pressure, or after the part absorbs solvent or water, can open a seam that was tight when it was made.

Reliability work on hermetic packages documents the same mechanisms, including how permeation and environmental stress cracking interact with a stressed joint.

A weld that passed and still leaked

This is why validation cannot stop at the weld station. A part that passes every incoming leak test can still fail in the field, so a qualification plan needs environmental conditioning after welding, and the leak test has to come last. Requirements and conditioning are laid out in NASA-STD-7012.

Take one caution from the SHP seals guidance on avoiding seal failures in hermetic assembly processing seriously. The vast majority of hermetic seal failures are thermally induced during subsequent manufacturing and test processes rather than in service, and some gross-leak test conditions are destructive for certain geometries because of thermal gradients.

Manufacturing failure or service failure?

Three questions separate the two, and the Los Alamos helium leak procedure is a good reference for locating the leak before you cut the part open.

• Where is the leak? A leak that follows the joint line, sits at a corner, or radiates from a fastener points to the joint. A leak through a wall far from the joint points to permeation or a moulding defect.
• When did it appear? Found at end-of-line test means manufacturing. Appearing after a shock, a chemical exposure or thermal cycling means post-weld damage or ageing.
• What does a cross-section show? A section from a part that never saw service reveals the process truth: lack of fusion, voids along the interface, a continuous cold-joint line, a cracked flash. A section from a field failure shows the service truth: crazing, stress whitening, creep distortion or swelling.

One more thing worth checking before you blame the weld: a neighbouring operation. A laser welding operation or a hot stamping step on the same assembly can soften a joint that was sealed perfectly minutes earlier.

How Leaks Are Found: Leak Testing Methods Compared

Every root cause in this article ultimately shows up as one number: a leak rate. Finding it means picking the right measurement, knowing what that measurement can and cannot tell you, and reading it without fooling yourself.

The menu for plastic hermetic joints is short. Four methods cover nearly everything: bubble or dunk testing for gross leaks, pressure decay, vacuum decay, and helium mass spectrometry. Each trades sensitivity against cost, speed and simplicity, and each fails in its own characteristic way. The most common mistake is not choosing the wrong test. It is choosing a fine test and believing a pass, or choosing a gross test and believing a fail.

Gross Leak Testing: The Bubble and Dunk Method

The oldest method is still the most useful first pass. The part is immersed in a perfluorocarbon bath, typically Fluorinert FC-40 or FC-43 held at 125 °C, pressurised with helium, and watched for bubbles. Or you simply pressurise the part with air and dunk it.

The strength of this method is localisation. When a bubble starts, you know exactly where the hole is, which is often the single most useful piece of information available. It is also fast, inexpensive, and needs no expensive instrumentation.

The limitation is equally clear. The bubble test only finds holes large enough to pass a visible bubble, so a leak small enough to matter for a fuel system or a medical reservoir slips through without a murmur. Treat a clean dunk as proof that there is no gross leak, never as proof of hermeticity.

Pressure Decay: Simple, Fast, and Prone to False Alarms

Pressure decay testing seals the part, pressurises it with air or nitrogen, isolates it, and watches the pressure fall. It works well on closed volumes and detects gross and moderate leaks reasonably well. It is the workhorse of high-volume production lines because it is quick and needs modest equipment.

Its weakness is thermal. Gas temperature inside the part changes as the part cools or warms, and pressure follows temperature. A part that has been out of the fixture for a minute will show a pressure drop that has nothing to do with a leak. That is why the method demands stabilisation, a good baseline on a known-good part, and a temperature compensation step. Get those wrong and you will spend a week chasing a leak that does not exist.

Vacuum Decay: Sensitive, and Confused by Outgassing

Pull a vacuum on the sealed part and measure the rate of pressure rise. It is more sensitive than pressure decay and is often the right answer for parts with small internal volumes.

Here the plastic itself becomes the problem. When you pull a vacuum on a moulded polymer, absorbed moisture and residual volatiles outgas into the chamber and pressure climbs even on a perfect part. Plastics therefore need a soak or equalisation period before the measurement means anything, and the test laboratory needs to know how long that is for your specific resin and conditioning state. A vacuum decay result without a documented soak time is not a measurement. It is a guess.

Helium Mass Spectrometry: The Reference Method

Helium is the tracer gas of choice because it is the smallest molecule, it passes through almost any leak path, and it is easy to detect at very low concentrations with a mass spectrometer. A fine leak of 10^-6 to 10^-9 atm·cc/s sits well within the reach of the method, which is why hermetic connector and aerospace programmes specify it.

There are four practical modes, each standardised separately:

• Inside-out — the part is pressurised with helium and the spectrometer sits in a vacuum chamber surrounding it. ASTM E493/E493M covers this technique. Most sensitive, but the whole part must go in a chamber.
• Tracer probe — helium is sprayed on the part and a probe sniffs for escaping tracer. ASTM E498/E498M covers the method.
• Detector probe — the part is pressurised and a probe moves along the joint searching for helium. ASTM E499/E499M covers this mode, and it is the one most often used on welded plastic parts because it localises the leak precisely.
• Hood mode — a sealed hood encloses the part. ASTM E1603/E1603M covers it, and it is fast for high-volume screening of one specific area.

Accepted practice is worth copying. Los Alamos National Laboratory’s helium leak testing procedure sets a maximum allowable leak rate of 1.0 × 10^-6 standard cc/sec for acceptance, with helium sprayed over the area and the nozzle held no more than 1 mm from the surface, moving at roughly 24 inches per minute. Calibration uses helium standard leaks typically in the 10^-6 to 10^-8 scc range. Follow the nozzle geometry and the sweep rate exactly, or your reading is not comparable to the limit you are testing against.

The Standards Behind the Menu

• MIL-STD-883 Method 1014 — the fine leak test for electronic packages, using the vacuum or bombing technique with tracer gas.
• MIL-STD-202 Method 112 — leak test methods for sealed and hermetic units, covering pressure and vacuum decay approaches.
• ASTM E1316 — the standard Leakage Testing Handbook, a good primer on method selection and terminology.
• ANSI N14.5 — leak testing for nuclear confinement welds, and a useful reference even outside nuclear work.
• NASA-STD-7012 — leak test requirements for space hardware, with a method menu for qualification and acceptance testing that includes vacuum exposure. It cites ASTM E1316 and the ASNT Nondestructive Testing Handbook as conversion references.

Sequencing: When to Test and in What Order

Order matters. Fine leak testing comes first because the exposure to tracer gas and vacuum stresses the seal, and any damage it causes will then show up in the gross leak test that follows. A part that passes fine and fails gross has told you something important: the test stressed the joint.

The second rule is more about manufacturing than testing. Hermeticity is normally verified after mechanical and environmental testing, not before. Leak test at the weld station, then let the part spend three weeks in a vibration, humidity and thermal cycle campaign, and you will find failures with no obvious cause. The weld was fine; the journey broke it. That is why guidance on avoiding seal failures in hermetic assembly processing stresses that the vast majority of hermetic seal failures are thermally induced during subsequent manufacturing and test processes rather than evidence of poor field reliability.

One practical warning from the same guidance: some gross leak test conditions are themselves destructive for certain geometries because of thermal gradients. A part with a thin web or a long unsupported wall can distort under a hot bath or a pressure ramp and open a leak that was not there on the bench.

Choose the test that answers your actual question. Does the part have a hole you can see? Use the bubble test and localise it. Does the part need to hold pressure in service with moderate leakage tolerance? Use pressure or vacuum decay with a documented baseline and soak. Does the part need to be qualified to a fine leak rate specification? Use helium mass spectrometry with the mode that suits your geometry, and the nozzle discipline that makes the number meaningful.

A Root-Cause Workflow: From a Failed Part to the Actual Cause

When a plastic part starts leaking, most teams do one of two things: change a weld parameter and hope, or change the resin and hope. Both are guesses. A repeatable sequence takes an afternoon and usually finds the answer in the first two steps.

The logic here is not new. Amada’s hermetic seam sealing guidance sorts every seal failure into four buckets: process, materials, part design and equipment. Those map almost directly onto this article’s cause families, with equipment folding into process. Work the sequence in order and you avoid the expensive habit of changing three things at once and learning nothing.

1. Confirm the failure is a leak. Before blaming the weld, check the test system: put a known-good part on the fixture, or test a sealed blank. A leaking connector, a contaminated port or a vacuum line with its own small leak will manufacture failures out of perfect parts. Then let outgassing settle. Plastics release absorbed gas into a vacuum, so a reading taken minutes after pump-down can look like a leak that disappears later.

   Separate a leak from permeation. Permeation scales with area, wall thickness, temperature and pressure differential, and it does not follow the seam. If the same part passes at one pressure and fails at another, you are measuring gas transport through the wall, not a joint defect.

2. Locate the leak. Sweep a helium probe along the seam, or dunk the part and mark the bubbles. Location is diagnostic on its own. A leak that follows the joint line points at fusion or contamination. A leak at one corner points at geometry. A leak at a fastener points at assembly torque or clamping. A leak through the wall, away from the weld, is a molding defect such as a void or short shot.

   Trust the method before you trust the result. National laboratory practice on helium leak procedure at Los Alamos uses calibrated reference leaks to prove the setup can detect a leak at or below the requirement before the test is accepted as valid.

3. Classify it against the four cause families. Design, material, process, or post-weld and service. If the leak appears only on parts from one mold tool or one resin lot, the answer is already in front of you.

4. Open the joint. Section it and look under a microscope for full fusion versus lack of fusion, voids and porosity along the interface, a continuous cold-joint line, flash and squeeze-out evidence, fibre or particle entrapment, and brown or yellow discoloration from thermal degradation. This is destructive work, so decide what you need before cutting. If a full section is not required, dye penetrant or a targeted gas exposure can reveal the same channel without destroying the part. Air arc welding and carbon arc gouging are standard tools for removing a joint when you need a clean cut edge for metallographic-style sectioning of plastic assemblies.

5. Pull the process data. Weld energy, collapse distance, trigger force, weld time, platen temperature, cooling hold, ambient humidity, part moisture content and conditioning history. One reading rarely proves anything, but a trend against a control chart will. Remember that a weld can also be “too good”: excessive energy drives material expulsion and flatness-driven over-welding, which both open leak paths and raise particle contamination risk.

6. Test the hypothesis, one variable at a time. Change a single factor, re-run a designed experiment with enough parts to see a difference, and keep samples from every run. If the leak rate moves, you have your cause.

Symptom to Probable Cause

What you observe	Probable cause	Check first
Leak at one corner only	Corner radius mismatch or step clearance gap	Joint design geometry
Leak along the whole interface	Lack of fusion, or contamination on the land	Surface cleanliness, weld energy
Bubbles inside only, on a humid day	Moisture in the part	Drying and conditioning records
Weld looks perfect but leaks	Micro-void or cold joint inside the interface	Sectioned joint
Leak only after cycling or months in service	Creep, environmental stress cracking, fatigue	Chemical exposure and duty cycle
Leak rate rises with pressure	Permeation through the wall	Rate versus pressure differential

Three Misdiagnoses to Avoid

• Blaming the resin when the geometry was wrong. No resin change fixes a shallow step or a mismatched corner radius.
• Adding weld time when the real problem is contamination. More heat on a dirty land burns polymer instead of fusing it.
• Rejecting a leak result because the part looks fine. Visual indicators such as material expulsion, weld overlap at corners and fillet formation around the perimeter are clues only, and they never replace a leak test.

Once the cause is confirmed, move to the controls that keep it from returning.

Prevention: Controls That Actually Stop Hermetic Failures

Diagnosis tells you why the last part leaked. Controls are what stop the next one from doing it. The encouraging part of this whole subject is that almost every hermetic plastic weld failure we have walked through in this article traces back to something a written procedure could have caught: a step depth that was never specified, a resin that arrived wet, a weld time nobody had qualified, or a mold release the purchasing department never asked about.

What follows is a control framework in four parts: design, material, process and organization. None of it is exotic. All of it works.

Design Controls: Get the Geometry Right Before the Tooling Is Cut

The cheapest hermetic seal is the one that needed no rework. Hold a joint design review before the mold exists, and do it with three parties in the room: your polymer supplier, the welding equipment maker, and the person who will run the leak tester. Each sees a different failure mode.

• Step geometry and leak path. Decide where the molten resin goes, where displaced gas can escape, and how the seal ring will be inspected. A designed pressure-relief path is the difference between a weld that vents and a weld that traps.
• Tolerance stack-up that includes molding shrinkage and warp. If the lid is 0.4 mm out of flat on the day of welding, no weld schedule compensates for it.
• Design for testability. Provide a port or a geometry that allows 100 percent dunk or helium testing. A part that cannot be tested in full is a part you are sampling forever.
• Avoid high local heat beside a finished seal. Anything that reheats an already-welded region in a later operation is a leak waiting for a customer.

Material Controls: Dry, Clean, Traceable

Incoming moisture is measurable with Karl Fischer titration, but most shops do just as well with a written conditioning protocol: defined drying time and temperature per resin, humidity-controlled storage, sealed line-side containers, and desiccant in reopened bags. Then lock down the rest of the material variables. Restrict or eliminate regrind in leak-critical parts, specify weld-friendly mold release agents or require joint cleaning before welding, and choose the resin with the coefficient of thermal expansion and chemical resistance the service environment actually demands, not the cheapest grade on the datasheet.

Process Controls: A Qualified Window, Not a Setting

ISO 23512, the standard coveringplastics joining of thermoplastic moulded components and the specification of variables for thermal joining processes, requires the qualified range of each variable to be documented. That is what keeps component quality inside the manufacturing quality agreement with the customer rather than inside the memory of a process engineer who left three years ago.

Build that window with a designed experiment, then hold it with data. Monitor-based weld data collection, with alarm limits on collapse distance, energy and trigger force, catches a drifting joint long before the leak detector does. Keep the tools honest: horn faces wear, platens drift in temperature, laser optics foul. Run SPC on the actual leak rate distribution rather than a pass/fail column, because a creeping average is the early warning. For ultrasonic welding, ISO 17608 quality levels B, C and D give production a shared vocabulary for weld imperfections, with B the most stringent.

Organizational Controls: The Layer Most Shops Skip

Train everyone on the line to tell a leak from permeation, so the same part does not get reworked three times. Run a corrective-action loop when a leak is found: quarantine, locate, confirm with data, correct, verify, and close. And put change control in writing, so a new resin grade, a new molder, a reworked tool or a second-source machine triggers requalification instead of quietly changing your process.

Matching Equipment to the Job, Honestly

Process choice follows the joint. Hot plate welding suits large parts and hermetic seals with simple tooling. Vibration welding handles large or irregular shapes but produces heavy flash and is slower on small components. Laser welding, covered in more detail in the alpha laser welding systems guide, delivers the cleanest cosmetics for medical and filtration devices but demands good focus and clean, consistent materials. Ultrasonic is the fastest option for small, high-volume assemblies yet is extremely sensitive to moisture and horn condition. A bench-top helium tester costs far more than a pressure-decay unit and needs operator skill; a pressure-decay tester is cheap but tells you less. Buy the tester that answers the question your specification actually asks, and note that we may earn a commission on some product links, at no extra cost to you.

Which Plastic Welding Process Gives the Most Reliable Hermetic Seal?

There is no single “best” plastic welding process for hermetic work. The right answer for a fuel bladder is wrong for a medical reservoir. What matters is whether joint geometry, material condition and the process window line up.

Emerson Branson’s plastics joining guidance puts it plainly: hot plate welding suits thermoplastics where high-strength and hermetic seals are required, vibration welding creates high-strength, leakproof hermetic seals with most thermoplastics and handles large, irregularly shaped parts, and laser welding is used in medical applications and filtration devices with excellent cosmetic results (Emerson Branson plastics joining brochure). Spin welding, the same source notes, was applied to a screw cap where a watertight, clean and aesthetically pleasing seal was needed at 20 units per minute.

Process-by-process comparison

Process	Typical joint	How heat is generated	Hermetic seal capability	Main failure causes	Cycle time	Best fit
Hot plate	Large flat joints, flanged shells	Heated platen into the interface	Excellent reliability, simple tooling	Whole face heated causes distortion; needs cooling hold	Seconds plus cool-down	Fuel systems, reservoirs, bladders
Vibration	Large, irregular parts	Frictional vibration	Very high; solid continuous flash	Excess flash, poor collapse, part pulled off early	Seconds, fast on large parts	Filter housings, tanks, ducts
Ultrasonic	Small parts, short seams	High-frequency horn vibration	Good when tightly controlled; no flash ring	Amplitude, force, weld time and trigger drift; horn wear; moisture	Sub-second to seconds	Medical fluid paths, filters, enclosures
Laser	Seams needing precision and appearance	Absorbed laser energy	Cleanest, most controllable; minimal flash	Focus, energy density, speed; clean surfaces, often inert atmosphere	Fast, slower on thick walls	Medical devices, filtration, housings
Spin	Round closures	Frictional rotation	Good on simple round closures	Torque and ramp profile; uneven pressure	Short, seconds	Screw caps, bottles, canisters
Infrared	Long or wide seams	Infrared radiators	Good where heat uniformity is the goal	Hot and cold spots, uneven shrinkage, contamination	Moderate, scales with length	Long seams, panels, packaging

The practical trade-offs

Vibration welding is the workhorse for large, irregular hermetic parts because it builds a solid, continuous flash you can inspect. The catch is flash you may need to trim and a cooling hold so the weld solidifies before the part leaves the tool. Hot plate welding offers excellent reliability with simple tooling, the default for fuel tanks and reservoirs for decades, but it heats the whole joint face, so cooling and distortion control matter. Ultrasonic welding is fast and clean but limited to small parts, and its key variables, amplitude, force, weld time and trigger, make it the most parameter-sensitive process for hermetic duty, so moisture and horn face condition demand tight incoming material control.

Laser welding gives the cleanest, most controllable seal with no contact contamination and minimal flash. The trade-offs are good focus, clear materials and often a controlled atmosphere for reactive or medical applications. Spin welding suits simple round closures and can hit respectable throughput, as the Emerson Branson screw cap example shows at 20 units per minute.

That last point is worth pressing. The most effective prevention measure across every joining process is a qualified process window paired with documented joining process specifications. The BSI specification of variables for plastics joining covers this for thermal joining processes, and ISO 23512 gives you the framework for listing qualified ranges for every essential variable.

Conclusion

Hermetic plastic weld failures are diagnosable and largely preventable. A weld that leaks is rarely random. It is telling you that one of four things went wrong, and you can usually find which one in an afternoon.

Use this triage whenever a part fails a leak test:

1. Design — Could this joint produce a continuous molten path? Check step depth, corner radii, leak path and projection consistency.
2. Material — Was the part dry, clean and in spec? Consider moisture, mold release, regrind and degradation.
3. Process — Was the weld fully fused at the right energy and pressure, and solid before release?
4. Post-weld and service — Did something damage the joint later, or did pressure cycling open it up?

Then take exactly one action before changing anything: stop the line, quarantine the parts, locate the leak with a helium probe or bubble test, and confirm the cause with data (weld energy, collapse distance, platen temperature, part moisture) before you touch a parameter. Changing weld time when the real problem is contamination on the seal land just burns a cycle and teaches you nothing.

Also worth changing on paper: stop specifying “leak-proof” and start specifying a leak rate with a stated test method. A number plus a method, such as a maximum fine leak rate verified by helium mass spectrometry per the helium leak test procedure used at Los Alamos National Laboratory, gives engineers something to design to and test labs something to quote. “Hermetic” without a number is a conversation, not a specification, and the hermetic leak rate testing limits and qualification guidance shows how much specification detail is normal in regulated industries.

For the diagnostic detail behind all of this, revisit weld inspection methods compared and the root-cause workflow, then read the leak testing and process control sections before you re-qualify anything. Diagnose before you adjust, specify before you build, and verify with a number.

Frequently Asked Questions

What is the most common cause of a hermetic seal failure in plastic welding?

In practice, joint and part design is the leading cause, followed by material condition and process parameters. If the step geometry, corner radii or flange land cannot produce a continuous molten path, no setting on the machine will close the leak. Many shops chase parameters first because it is faster, which is why so many leaks are ‘fixed’ temporarily by adding heat and then reappear in service.

How do you tell the difference between a leak and permeation through the plastic?

A leak is a discrete path — a void, a cold joint, an unsealed corner, a crack — and it shows up as a localized signal on a helium probe or as bubbles in a dunk test. Permeation is gas passing through the bulk polymer wall, so the whole part shows elevated helium with no localized hot spot, and the rate rises with temperature and pressure differential. Permeation is a material property and cannot be welded away; it is managed by resin choice, wall thickness and time-dependent testing.

Can a welded plastic part be made truly hermetic, or is that just marketing?

Not absolutely — no welded thermoplastic is hermetic forever, because polymers permeate gases. What the industry means by hermetic is a specified maximum leak rate, such as 10^-6 standard cc/sec of helium, verified by a defined test method. Quoting a leak rate and the test method in the purchase specification is far more useful than the word ‘leak-proof’.

Which leak test should I use for a welded plastic reservoir?

Use a helium mass spectrometer test when you need sensitivity or need to locate the leak, following ASTM E493, E498, E499 or MIL-STD-883 Method 1014. Use a bubble test with perfluorocarbon for fast gross-leak screening and localization, and pressure decay or vacuum decay for inexpensive line-side monitoring, allowing for outgassing and temperature stabilisation. Fine-leak testing is normally done before gross-leak testing and after mechanical and environmental testing.

Does moisture in nylon or polycarbonate really cause leaks?

Yes. Polyamide, polycarbonate, polycarbonate/polyester alloys and polysulfone are hygroscopic, and moisture absorbed from the air vaporises at the weld interface, leaving voids and porosity along the bond line while also lengthening the weld cycle. Drying, conditioning and humidity-controlled storage are process controls, and moisture content should be specified and monitored like any other incoming material property.

Why does my ultrasonic weld pass the leak test in the lab but leak in production?

The lab sample is usually conditioned and clean, while production parts carry mold release, dust, packing oil and variable moisture. Ultrasonic welding is unusually sensitive to interface friction and moisture, so the same settings can produce good bonds on clean conditioned parts and cold joints on dirty production parts. Monitor collapse distance and trigger energy per part rather than relying on settings alone.

Can I fix a leaky plastic weld by welding over it or adding sealant?

No. Epoxy bonds are inadequate hermetically, and re-welding the same area usually adds heat to a joint that already has voids, worsening degradation and warpage. Use a seam-weld-plus-sealant construction only when the design is validated that way from the start; a failing seal should be traced back to design, material or process and corrected there.

How much extra heat can I put into a joint near an existing plastic seal?

Treat it as a hard limit. In hermetic connector work, welding heat near a finished seal caused gross leaks on up to 70 percent of units with established processes, while heavy heat sinking and segmented welding with compressed argon cooling between segments brought acceptable leak rates well under the gross-leak threshold. If a second weld is unavoidable, use heat sinking, split the weld into short segments and allow full cooling between them.

What should I document so a hermetic plastic weld stays in control on the shop floor?

Create a thermal joining process specification in the style of ISO 23512 that lists every essential variable and its qualified range, add a process window built from a designed experiment, capture weld monitor data such as collapse distance and energy with alarm limits, run SPC on leak-rate results rather than pass/fail only, and put change control in place so a resin, molder, tooling or machine change triggers requalification.

Are vibration or hot plate welding better than laser for a hermetic plastic seal?

Hot plate and vibration welding are the traditional choices for large leak-critical parts such as fuel systems and reservoirs: the heat is applied to the whole joint face and the resulting solid flash gives a reliable, inspectable seal. Laser welding gives the cleanest, most controllable seal with minimal flash and no contact contamination, and is widely used for medical and filtration parts, but it needs accurate focus, suitable materials and often a controlled environment. The right answer depends on part size, geometry, volume and cosmetic requirements, not on a universal ranking.