Accelerated Moisture Resistance Test Chamber Per JESD22 A102

What A102 puts a part through

The autoclave test, also called the pressure cooker test, is the bluntest moisture stress a semiconductor package meets. JESD22-A102 places parts in saturated steam at a hundred and twenty-one degrees and a hundred percent humidity, held under roughly two atmospheres of pressure, with no electrical bias, for a span that commonly runs from ninety-six hours to a couple of hundred. There is no cycling and no voltage; the part simply sits in pure hot steam while the test waits to see whether moisture alone can break it. What it finds is package integrity failure, under conditions far past anything the field will ever apply.

Why it has to be pressurised

The pressure is not there to stress the part directly; it is there to make the steam possible. Water at ordinary atmospheric pressure boils at a hundred degrees, and air at that pressure can hold no more than a hundred percent humidity at a hundred degrees and no higher. To reach a hundred and twenty-one degrees while keeping the air fully saturated, the pressure has to climb, since raising the pressure raises the temperature at which water boils and lets the vessel hold liquid and saturated vapour together at that hotter point.

Two atmospheres lifts the boiling point to about a hundred and twenty-one degrees, the exact figure the test names, so the vessel sits at the boundary where water and saturated steam coexist at that temperature. In gauge terms that is roughly fifteen pounds per square inch above the air outside, about two hundred and five kilopascals absolute. The pressure is a consequence of wanting saturated steam well above a hundred degrees rather than a separate stress, and it is what forces the whole test into a sealed vessel instead of an open chamber.

The pressure is the price of saturated steam past boiling.

Saturated, and meant to be

This test embraces the very thing the biased humidity tests work to avoid: full saturation, a hundred percent humidity, with liquid water condensing freely on the part. Where HAST holds its air just below the dew point so moisture diffuses in as vapour, the autoclave floods the chamber with saturated steam and lets the parts come out dripping, and that is the test working as designed rather than a fault. The reasoning is brute simple. By driving the package with as much water as physics allows, at a temperature well above boiling held there by pressure, A102 forces moisture through the plastic and along every interface in days, exposing any path that a real climate would find only over years. Saturation also removes a variable, since there is no humidity setpoint to drift when the air is simply as wet as it can be at that temperature and pressure and the steam tables fix the rest. The cost of that ferocity is that the test can be unrealistically harsh, flagging parts that would never fail in service, which is why the trade has moved much qualification toward the gentler, biased, unsaturated runs and keeps the autoclave as a fast, severe screen. The chamber that runs it has one job above all, to hold the vessel saturated and steady at the named temperature so every part meets the same drowning, neither milder nor crueller than the standard intends.

The mechanism, with no current to help

No voltage is needed here.

Without bias the corrosion has to be driven by moisture and contamination alone, and that shapes what A102 detects. Water saturating the package mobilises whatever ionic contamination was left inside it from manufacture, the chloride from a flux residue or an etch, and that activated electrolyte attacks the aluminium bond pads and the fine metallization until they corrode open. The same moisture collects at the interfaces between the plastic and the metal of the lead frame, weakening the adhesion until the encapsulant delaminates, and at the test temperature the trapped water can flash to vapour and lift or crack the package outright, the popcorn fracture that names the related preconditioning test.

None of this needs a voltage, which is the point: A102 exposes packages that were built dirty or sealed poorly, the contamination and the weak interfaces that a cleaner unbiased path would never reveal so quickly. The harshness is what makes it a fast integrity screen.

It is an autoclave, not a chamber

The machine that runs A102 has more in common with a steriliser than with a humidity cabinet. It is a sealed pressure vessel, a heavy stainless chamber rated to hold two atmospheres of hot steam safely, with a heater and a reservoir of water in the bottom that boils to fill the space rather than a humidifier feeding a flowing airstream. The door seals and locks, and a safety interlock keeps it shut while the vessel is under pressure, since opening a hot pressurised autoclave would vent scalding steam in an instant.

A pressure relief valve guards against overpressure, and the build follows pressure-vessel codes rather than the lighter construction of an ordinary climate chamber. The vessel is built and stamped to a recognised pressure-vessel code with a rated working pressure and a tested margin above it, and the door gasket that seals against that pressure becomes the part inspected and renewed more often than any other. There is little air movement to manage and no tight humidity loop to tune; the physics of boiling water in a sealed space sets the condition, and the controller mainly holds the temperature that fixes the pressure. The engineering effort goes into containing the steam rather than shaping it.

The water it boils matters

Even here the feed water is not an afterthought. The steam in the vessel touches every part directly, so any impurity the water carries is delivered straight onto the devices, and A102 calls for high-purity water to keep the steam from depositing minerals or adding contamination of its own. Dirty water would seed corrosion the package never caused and confuse the result, laying its own scale and ions across surfaces the test is meant to read cleanly.

Deionised water, low in dissolved solids, keeps the steam a pure stress rather than a carrier of mess, and the reservoir is refreshed rather than left to concentrate over many runs. The purity matters less for the machine, which tolerates ordinary water, than for the honesty of the test.

Loading so the steam reaches everything

Getting an even result depends on the steam touching every part alike. Devices sit on open racks that let the saturated vapour circulate around each one, rather than stacked where a part in the middle would see less of the steam than a part at the edge. The racks hold the devices apart and clear of the vessel floor, away from any pooled water, so each meets the same saturated vapour. With no bias and no wiring to manage, the only craft in the loading is that spacing.

The cycle, from cold to cold

A run is bracketed by the pressurisation and the vent. The vessel is sealed and heated, the water boils and drives the air out and the pressure up until it reaches the soak condition, and the parts then sit in saturated steam for the prescribed hours. At the end the heat comes off and the vessel is vented and allowed to cool before the interlock releases the door, since the parts cannot be reached while the steam is still hot and pressurised.

The devices are dried and then measured, electrically and often by acoustic imaging that looks for the delamination the steam may have started, and the comparison of before and after settles the pass or fail. The cycle is short on control subtlety and long on the discipline of handling a pressure vessel safely.

A102 set against HAST

The autoclave test has a more modern relative that is often preferred, and the distinction is precise. The highly accelerated stress test of JESD22-A110, HAST, also pressurises the vessel, but it runs unsaturated, at a hundred and thirty degrees and eighty-five percent humidity, and it applies bias to the parts. A102 by contrast runs fully saturated at a hundred percent and carries no bias. The saturation is the sharp difference: A102 wets the parts completely while HAST holds them in very humid but unsaturated air, and HAST adds the electrical stress that A102 lacks.

HAST reproduces a powered, humid service condition more faithfully and has displaced the autoclave test in much qualification work, while A102 survives as a fast, fierce, unbiased screen of bare package integrity. The two share a pressure vessel but ask different questions of the part.

When the autoclave over-stresses

The harshness that makes A102 fast also makes it suspect. Saturated steam at a hundred and twenty-one degrees is more severe than anything a device meets in service, and the test can fail packages by mechanisms that would never trouble them in the field, condensing liquid water where a real environment would only ever reach high humidity. A failure in the autoclave is read with that in mind, as a sign of a weak or contaminated package rather than a prediction of field life in years.

The industry's drift toward HAST reflects exactly this, a wish for a stress hard enough to be quick yet close enough to reality to mean something. A102 keeps its place as a screen precisely because it overstresses, flushing out the worst packages without claiming to model their service.

Pulling it together

The A102 chamber is a pressure vessel that boils pure water into saturated steam at a hundred and twenty-one degrees and holds an unbiased package in it for days. The pressure exists only to keep the steam saturated above its ordinary boiling point, the saturation soaks moisture into the package faster than any gentler method, and the absence of bias lets the water stand on every surface without the electrolytic artifact that haunts the biased tests.

Built like a steriliser, fed with clean water, loaded so the steam reaches each part, and handled with the care a hot pressurised vessel demands, it screens package integrity hard and fast. It overstresses on purpose, and a part that walks out of it dry inside, its bonds bright and its plastic still bonded to the metal, has a package that will keep the weather out for a very long time.