Reliability Test Chamber For CoWoS Advanced Packaging

A package of packages

CoWoS means chip-on-wafer-on-substrate. It is how the largest processors are built now.

What CoWoS holds

In a CoWoS package the working chips do not sit on the circuit board directly. They sit first on a thin slab of silicon called an interposer, which carries the fine wiring that links a logic die to the stacks of high-bandwidth memory beside it, and the interposer in turn rests on a larger substrate that fans the connections out to the board.

The result is a single component that holds many chips, often across an area several times the size of an ordinary processor, and the reliability question is whether that whole assembly holds together through heat, cold, and damp.

Why a body this large pulls itself apart

The trouble with CoWoS is that its layers are made of different materials that swell and shrink by different amounts as the temperature changes, and the package is wide enough that those small differences add up to real movement at its edges. Silicon expands little, the organic substrate beneath it expands several times more, and between them sit the tiny copper microbumps and the underfill that bond the chips to the interposer and the interposer to the substrate. When the package heats, the substrate strains to grow faster than the silicon it is glued to, the assembly bows like a bimetallic strip, and every microbump along the way is stretched and sheared by the mismatch; when it cools, the bow reverses and the bumps are worked the other way. Over enough of these swings a bump cracks, an underfill layer peels from a corner, or the interposer lifts at an edge, and a single broken link among the many thousands can take the whole package down. Heat alone does this through expansion, and moisture makes it worse by creeping into the underfill and along the interfaces, swelling them, weakening their grip, and flashing to steam if the part is heated fast while wet, so the chamber has to bring temperature and humidity to a package this large slowly and evenly, holding every corner at one point at one moment, since a chamber that warms one side ahead of the other bends the package on its own and writes a stress into the result that the part would never have met in service.

Why moisture finds it

A wide package gives moisture a wide door. The underfill and the substrate take up water across their whole area, and the larger that area the more water the part holds and the longer it takes to dry. That trapped moisture is what turns a quick heating into a crack, and it is why a CoWoS part is dried and sealed to a moisture level before it is ever soldered.

The humidity-and-bias trial

Much of CoWoS testing is humidity held against the part while a voltage runs through it, the biased damp-heat trial that drives moisture and current together. The chamber holds a warm, damp climate steady for a long soak, or a pressurised one for a shorter accelerated run, while the package sits powered on a fixture. Water in the dielectric and along the interposer wiring leaks current, corrodes a fine line, or migrates metal across a gap, and the test watches the leakage climb.

The chamber has to keep the climate flat across a large fixture and carry clean bias to a part with thousands of connections. Because the package is wide, the run is held long enough for moisture to reach its centre and not only its rim, and the leakage is logged through the whole soak so a slow climb shows itself rather than hiding between a start and an end reading.

Cycling the big body

The other half is temperature cycling, swinging the package between a deep cold and a high heat to work the bumps and the underfill by expansion alone. The chamber drives the swing on a controlled ramp and holds at each end long enough for the whole body to reach the set point, since a large package lags the air and a corner left behind sees a smaller swing than the standard intends. Run enough cycles and a weak bump or a poor underfill edge gives way, the same way it would over years of the chip warming under load and cooling at rest.

The number of cycles stands in for those years, and the spread between the cold and the hot end sets how hard each one pulls, so a chamber that cannot reach its low point or holds an uneven heat quietly softens the very stress the count was meant to apply.

Holding the climate flat around a wide part

A large package magnifies any unevenness in the chamber. If the air moves faster over one edge, that edge heats, cools, or dries ahead of the rest, and the reading turns into a verdict on the chamber. The chamber has to move its air gently and evenly, hold its temperature and humidity uniform across the whole working space, and give a large thermal mass the time it needs to settle, so that what the part feels at its centre matches what it feels at its corner.

Older chambers sized for small parts can leave a warm zone near a heater or a fast stream by a fan inlet, and a body wide enough to span those zones feels the difference as a bend, so a chamber for advanced packages is built with the airflow and the working space planned around a large, still load.

One weak link is the whole part

A CoWoS package fails as a unit. One cracked bump can darken every chip on the slab.

Reading the package after

After the run the package is judged. Electrical tests look for an open bump or a leaking line, and acoustic imaging looks through the package for a delamination at the underfill or a lifted interposer edge that the electricals have not yet caught. A cross-section can confirm a cracked bump. A part that holds its connections and shows no delamination passes; one that has opened, leaked, or delaminated points back to the bump, the underfill, or the warpage that let it move.

Where it sits among the tests

CoWoS reliability leans on the same family of trials as any package, the damp-heat soak, the biased humidity run, the accelerated pressurised test, and the temperature cycling, but it asks more of each because the body is so large and so layered. The chamber earns its place by holding those climates steady and even around a part that punishes any drift, and a lab that builds advanced packages leans on it to tell a sound design from one that will move itself apart in the field.

Drying before the run

Before any humidity work the package is baked to drive out whatever water it has already taken up, so the run starts from a known dry state. A part that goes in damp confounds the test, since the chamber cannot tell moisture it added from moisture the part carried in. After the bake the part is held sealed with a desiccant until it enters the chamber, and the time it spends in open room air is counted, because a wide CoWoS body soaks up moisture fast enough that an hour on the bench can shift where the run begins.

That clock on open-air time is tracked for every part in the lot, since one piece left out too long can drink in enough water to behave unlike the rest.

The fine wiring that cannot corrode

The interposer carries wiring far finer than the substrate below it, lines and spaces measured in microns that link the logic die to its memory at enormous width. That fineness is what makes the package fast, and it is also what makes it tender, since a trace this thin has little metal to lose before it opens and little spacing to spare before moisture bridges it.

The humidity run leans hardest on exactly these lines, and a chamber that holds a steady, clean, even climate is what lets the test show whether the fine wiring survives the damp or slowly fails in it.

Why an ordinary part skips this

A single small chip in its own package shrugs off much of what punishes CoWoS. It is light, it heats and cools almost as one piece, and its few connections leave little room for a mismatch to build. The advanced package inherits none of that ease: it is large, it is layered, it is studded with thousands of joints, and it carries far more value, so the same family of climate tests is run on it with more care and more weight on holding the chamber even, because the cost of a wrong call is a part that should never have shipped.

Watching the package bow

Some labs measure how far a CoWoS package bows as it heats and cools, mapping the warpage across its face at each step of a thermal run. That map tells whether the bow stays within what the joints can take and where the strain gathers, at a corner, along an edge, or over the widest span of the interposer. A chamber that drives a clean, even thermal swing is what makes the measurement mean something, since a reading taken while one side of the part lags the other measures the chamber as much as the package.

The slab that has to hold

A small chip forgives a chamber that runs a touch uneven; a CoWoS slab does not, since its size turns a small gradient into a real bend and a single bad joint among thousands is enough to lose it. A chamber for this work keeps heat, cold, and damp even across a wide body and moves between them slowly, so the package is tried by its own materials and not by the box.

Get that right and the slab that comes through clean is one that will carry its chips through years of heating and cooling; get it wrong and a crack waits in a package too valuable to lose.