Steady State Temperature Humidity Bias Test Chamber Per JESD22 A101 THB

What the test asks

The THB test, set out in JESD22-A101, is a steady-state stress rather than a cycle. Devices sit at a constant eighty-five degrees and eighty-five percent humidity, the condition the industry calls eighty-five eighty-five, while a DC bias is held across them for a thousand hours. What it hunts is moisture-driven failure: water vapour works through the plastic encapsulant to the die, and with the bias present it corrodes the metallization and grows conductive filaments between the conductors. A part that survives the soak has shown a package that keeps moisture off its silicon for the years a humid service life would take.

Why the bias changes everything

Voltage is the difference.

Heat and humidity alone corrode slowly; add voltage and the chemistry turns electrochemical. A potential difference across two neighbouring conductors, with a film of moisture and a trace of ionic contamination bridging them, sets up a tiny electrolytic cell, and metal ions dissolve off the positive conductor, drift across the gap, and plate out on the negative one as a branching dendrite that creeps until it bridges the spacing and leaks or shorts.

The same bias drives the anodic corrosion of the fine aluminium tracks, eating them open where the moisture has reached them. None of this happens in a humidity soak without power, where a part can sit damp for weeks and merely come into balance with the air; the voltage is what turns dissolved water into an agent of failure, and it is the heart of the THB method rather than an accessory to it.

The test reproduces the exact condition a powered device meets in service, moisture and potential together, and the chamber has to deliver the humidity to the part while the part holds its voltage, both stresses present at once for the whole thousand hours.

Moisture corrodes; voltage decides where and how fast.

Condensation is the enemy here

This is where the THB chamber parts company with an ordinary humidity oven, because the one thing it must never do is let water condense on a powered part. With a voltage held across closely spaced leads, a film of liquid water is not a mild dampness but a conductive bridge, and a single drop sitting between two pins can short them, grow a dendrite, or corrode a track in minutes, turning a slow and fair corrosion test into a sudden, meaningless failure. Steady eighty-five degrees and eighty-five percent humidity sit only a few degrees from the dew point, so any surface that runs even slightly cooler than the air, a lead carrying its own current away, a cold spot near a wall, a device chilled by a draft, will pull dew onto itself while the air still reads a perfect eighty-five. The whole control strategy bends toward preventing that. The chamber keeps its air moving gently and evenly so no pocket stagnates and cools, it holds the walls and the load above the dew point as it ramps in and out, and it brings temperature and humidity up in the right order so the part stays warmer than the moisture around it. Hold the climate and keep every surface clear of dew and the test corrodes honestly; let one cold corner weep and the result is no longer about the part at all.

The eighty-five point and the dew line

The numbers leave little room. Air at eighty-five degrees and eighty-five percent relative humidity has a dew point close to eighty-one and a half degrees, so the parts sit barely three and a half degrees above the temperature at which water would condense on them. Any surface that drifts below that dew point, a cold corner, a chilled feedthrough, a part that lags the air on the way up, grows the film the test cannot tolerate.

The margin is so thin that the order of the ramp matters: the chamber brings the temperature to eighty-five and lets the load reach it thoroughly before the humidity climbs to its full eighty-five percent, so the parts are already warm and dry when the moisture finally arrives rather than cold and waiting to catch it. Pulling the humidity up too early, while the devices still lag the air, is the classic way to wet a biased part and ruin a run before it has begun. The same care governs the descent at the end: the humidity comes off first and the temperature follows, so the parts never pass through a moment where cooling metal meets air still saturated enough to wet it.

Bias feedthroughs that do not leak

Getting the voltage to the parts is its own piece of engineering. A THB load may be dozens or hundreds of devices, each needing wires for its bias, and all those conductors have to cross the chamber wall without letting humidity out or a cold path in. The chamber carries sealed electrical feedthroughs, ports built to pass many wires through a gland that stays vapour-tight at eighty-five eighty-five for weeks on end.

The feedthrough also has to avoid becoming a thermal short: a bundle of metal wires running to a cooler outside world can conduct heat away and chill the connector block inside, dropping it below the dew point and condensing water exactly where the bias is densest. Good designs warm the feedthrough or break the thermal path, keeping the entry point as warm as the rest of the space. The wiring inside is run in materials that resist the humid heat and do not corrode or outgas, since a connector that greens over with corrosion injects its own leakage into the measurement.

Keeping the device at eighty-five, not above it

The bias has to stress the part without heating it. JESD22-A101 wants the device held at eighty-five degrees, so the bias is chosen to draw almost no current, the parts reverse-biased or held in a state that self-heats only a degree or two. A device running warm under its own power would sit hotter than eighty-five and see a lower local humidity at its surface, quietly softening the very stress the test applies. The chamber holds eighty-five in the air; the bias scheme has to respect that number.

Uniformity and the cold corner

Tight uniformity is a corrosion safeguard here rather than a nicety. With only three and a half degrees between the set point and the dew point, a chamber whose corners run two or three degrees cool has corners that condense, and any biased part sitting in one is lost. A THB chamber holds its temperature uniform to within a degree or two across the whole working volume and its humidity to a few percent, often a target of plus or minus two degrees and five percent, moving the air gently and evenly so that no pocket of the load lags the rest toward the dew point.

The walls are kept at or above the air temperature so they never become the cold surface, the same logic as the feedthrough. Mapping the empty chamber with a grid of sensors before trusting it with a long run is routine, since a single cold spot found after six weeks of testing is six weeks wasted.

Watching the parts during the soak

How the devices are measured shapes the wiring. Some programmes pull the load at intervals and measure leakage on a bench, accepting the disturbance of cooling and rewarming. Others monitor in situ, reading leakage current continuously through the same feedthroughs and catching the moment a part begins to degrade. The in-situ route gives a far richer picture of when and how failure arrives, at the cost of a more crowded and more carefully warmed set of feedthroughs.

Running a thousand hours unattended

A thousand hours is roughly six weeks, and the chamber has to hold eighty-five eighty-five through every one of them without a lapse. That puts reliability ahead of almost everything: a humidifier with feed water enough for the duration, a control that does not drift across six weeks, and alarms that call a technician if the temperature or humidity strays toward the dew line. A datalogger trending both readings through the run gives the engineer the evidence that the conditions held, which a qualification report has to show alongside the pass or fail.

A brief power loss is its own hazard, since a chamber cooling with the bias still live can condense across the whole load at once, so many THB setups drop the bias safely or ride through short outages and recover the conditions before the parts cool to the dew point. The long unattended soak rewards a chamber built for endurance and punishes one that needs nursing.

THB set beside HAST

The THB test has a faster, fiercer cousin, and the chambers are not interchangeable. The highly accelerated stress test of JESD22-A110, HAST, drives the same moisture-and-bias mechanism at a hundred and thirty degrees and eighty-five percent humidity under about two atmospheres of pressure, compressing a thousand hours of THB into a hundred or so. That pressure means HAST runs in a sealed pressure vessel, a different and heavier machine, while THB lives in an unpressurised eighty-five eighty-five chamber at ordinary room pressure.

A lab choosing between them is choosing time against capital and complexity, and a chamber bought for one does not serve the other. THB remains the long, gentle reference the accelerated test is calibrated against, the slow truth that the faster method has to agree with.

Pulling it together

A THB chamber is a steady-state eighty-five eighty-five box built around one fear: water on a biased part. It holds the temperature and humidity rock steady for a thousand hours, keeps every surface and every feedthrough above a dew point that sits only three and a half degrees away, passes the bias wiring through warmed vapour-tight ports, and lets the devices reach eighty-five before the moisture arrives and never fall below it after.

The bias energises the corrosion the test is built to find, the uniformity and the warm walls keep that corrosion honest, and the endurance to run six weeks unattended turns the whole thing into a result rather than a gamble. Get the dew margin right and the chamber truly measures the package; get it wrong and it measures only its own cold corner.