Thermal Shock Test Chamber For Semiconductors Per JESD22 A106

A temperature cycle ramps slowly between extremes. Thermal shock throws the part across in seconds. The speed is the stress, since the surface reaches the new temperature while the core still lags behind.

A thermal shock test moves a part between a hot extreme and a cold one as fast as the equipment can manage. The standard behind it is JESD22-A106. It uses the same two extremes a temperature cycle would, so the difference is not the temperatures. The difference is the speed of the trip between them. That speed leaves the surface of the part at the new temperature while the core still sits at the old one, opening a steep gradient through the part for as long as it takes the heat to even out. That transient gradient is the stress A106 applies, the one a slow cycle never builds. The test finds the parts that crack under it.

Shock is the speed

Thermal shock and temperature cycling use the same two extremes. What differs is the speed of the trip between them. A temperature cycle ramps the air slowly from cold to hot, giving the part time to follow together. Thermal shock moves the part across in seconds. The surface of the part reaches the new temperature almost at once. The core lags behind, still sitting near the old temperature. For a moment the part holds a steep temperature gradient through its own body, a hot skin over a cold middle, or the reverse. That gradient is the stress the test applies. Every part of the material expands or shrinks to match its local temperature. The skin moves first. The core resists. The interface between them strains hard while the gradient lasts. A slow ramp leaves that gradient out, since the part stays near one temperature throughout. The shock builds it on every transfer, thousands of times over. So the damage from thermal shock runs ahead of the damage from a gentle cycle, even at the same two extremes. The test exists to find the parts that crack under that transient, the ones a slow cycle would pass. The equipment that runs it is built around one job, moving the part from a hot zone to a cold zone fast enough that the gradient bites before the part can even out.

The transient gradient

Heat moves through a solid at a finite speed. Drop a part into a cold zone. Its outside cools at once, touching the cold air. The inside cools later, as the heat works its way out through the material. For the time in between, the outside sits colder than the inside. That temperature difference across the part forms the transient gradient, the thing thermal shock creates that a slow ramp does not.

The faster the transfer, the steeper the gradient. A part moved in one second meets the full temperature step with no warning, so its surface swings the moment it arrives. A part eased over slowly sees almost no gradient, since its inside keeps pace with its outside. Thermal shock pushes the transfer as fast as the equipment allows, to drive the gradient as steep as it goes.

The gradient turns into stress through expansion. Every part of the material expands by an amount set by its local temperature. A hot region expands. A neighbouring cold region holds still. The two pull against each other across the boundary between them. The steeper the gradient, the harder the pull. A crack starts where the pull beats the strength of a bond or a joint.

Shock against temperature cycling

Temperature cycling and thermal shock look alike on paper. Both swing a part between a hot extreme and a cold one, many times over. The temperature cycle, the A104 method, ramps from one extreme to the other on a controlled slope, slow enough that the part stays near one temperature throughout. A105 adds power to that cycle. Thermal shock, A106, throws out the slow ramp. It transfers the part across as fast as it can.

The fast transfer is what sets shock apart. A slow cycle stresses the part through the difference between the two end temperatures. A shock stresses it through that difference, with the transient gradient added on every transfer. So a part can pass a temperature cycle, then fail a thermal shock at the same two extremes, since the shock adds a stress the cycle leaves out. Thermal shock makes the harder screen for a brittle interface.

What the shock cracks

The cracks land at the interfaces. A package joins materials that expand at different rates, silicon against die-attach, die-attach against a lead frame. Each boundary holds two materials that disagree on how much to move. The transient gradient drives them apart hardest, since the gradient itself sits across those boundaries. The interface is where thermal shock does its damage.

Die-attach cracking shows up early. The layer bonding the die to the case takes the gradient straight across it. Each shock strains it, growing a crack from the edge inward. The crack raises the resistance of the heat path, so the die runs hotter afterward. A cross section after the test shows the crack creeping across the attach.

Solder joints crack the same way. A solder joint under a chip, or under a module baseplate, joins two materials of different expansion. The shock works it harder than a slow cycle would, since the gradient adds to the end-to-end strain. Solder joints on a board fail this way too, the reason a board goes through thermal shock the way a part does.

Brittle materials crack outright. Glass or ceramic can fracture when the gradient beats its strength, the way a cold glass cracks under boiling water. A silicon die does the same. A package with a moulding flaw splits along it. The shock finds the weak interface, the marginal joint a gentle test would miss, faster than any slow test. What survives a thermal shock holds margin against a gradient a slow cycle leaves unprobed.

The conditions

The extremes look like a cycle’s. A common pair runs from minus 55 degrees Celsius to plus 125, the part held at each end long enough to reach that temperature throughout. The dwell at each extreme runs in the five to thirty minute range. The standard lists several combinations of extremes, with their dwells, for different severities.

The transfer is where the conditions get strict. The part has to cross from one zone to the other inside a short defined window, since a slow transfer would soften the shock into a cycle. The standard caps that transfer time. The cycle count runs from hundreds into the thousands, since the fatigue from the gradient still builds over many transfers, just faster than a gentle cycle builds it.

Two ways to shock

Two methods deliver the shock. Air-to-air uses two chambers, a hot zone over a cold zone, with a basket that carries the parts from one to the other. The basket drops from hot to cold in seconds, so the parts meet the new air at once. This is the common method for semiconductors, since it handles dry parts cleanly.

Liquid-to-liquid goes faster still. The parts plunge from a hot bath into a cold one, both filled with an inert fluid that carries heat far better than air. The liquid pulls the surface to its temperature almost instantly, so the gradient runs steeper than air can drive. The method stresses harder, at the cost of fluid handling, which is why air-to-air covers the bulk of semiconductor work.

What the chamber must do

The chamber holds two extremes ready at once. A hot zone stays at the high temperature, a cold zone at the low, both maintained whether or not a part sits in them. The two zones wait preconditioned, so a part transferred in meets the full extreme with no warm-up delay. Holding both zones at temperature all the time is the price of a fast transfer.

A transfer mechanism moves the load between the zones. A basket or an elevator carries the parts from one zone to the other inside the defined window. The mechanism runs fast, since the transfer time is part of the test. It seals each zone during the dwell, so neither zone loses its temperature while the parts soak.

Recovery time matters as much as transfer time. When a cold load lands in the hot zone, it drags the zone temperature down. The zone has to pull back to its set point fast, so the part sees the full extreme for its dwell. A zone with weak heating recovers slowly, softening the shock. The chamber sizes its heating, with its cooling, to recover quickly under a full load.

Load mass sets the real performance. A light load transfers fast, recovers fast. A heavy load of parts carries more heat into each zone, slowing the transfer effect, slowing the recovery too. The chamber gets rated for a load mass, so the test stays valid up to that load. Overload it. The shock then softens below what the standard demands.

Uniformity holds across each zone. Every part in the basket has to meet the same extreme, so the zone keeps an even temperature throughout its volume. A part in a cold corner of the hot zone gets a milder shock than one in the centre. Good air movement in each zone keeps the temperature even, so every part on the load takes the same transfer.

The transfer time sets the test

One number defines a thermal shock more than any other, the transfer time. It fixes how fast the part crosses from zone to zone, so it fixes how steep the gradient gets. A short transfer makes a hard shock. A long one softens toward a cycle. The standard sets a ceiling on it for that reason.

Recovery time backs it up. The transfer puts the part into the new zone. The part feels the full extreme only once the zone recovers from the load it just took. A fast transfer into a zone that then sags gives a weaker shock than the transfer time alone suggests. The real shock depends on the transfer time, with the zone’s recovery behind it.

Load mass ties the two together. A bigger load slows the transfer effect, since more heat has to move. It also slows the recovery, since the zone takes longer to pull back. So the same chamber gives a harder shock on a light load than on a heavy one. Rating the chamber by load mass keeps the shock honest across different batch sizes.

Reading the result

A thermal shock result reads out in cycles to failure. Parts come off the test at intervals for an electrical check. An open points to a cracked bond or a fractured joint. A rising resistance points to a degrading die-attach or solder. A part that survives the planned cycles passes.

The failure analysis confirms the mechanism. A cross section shows the crack at its interface, the die-attach delamination, or the fractured die. The pattern matches thermal shock, a crack driven from where the gradient sat hardest. The cycle count at failure feeds the part’s reliability rating, since it sets how many shocks the part takes before it cracks.

The rule in one line

Thermal shock lives in the transfer speed, far more than in the temperature gap. A106 crosses the extremes fast enough to crack what a slow cycle would pass.

Matching the chamber to the test

The test names a two-zone chamber. It has to hold a hot zone, with a cold zone, both at temperature at once. It has to transfer the load from one to the other inside the standard’s window. It has to recover each zone fast under the rated load. A single chamber that ramps one volume slowly cannot deliver a shock at all.

The transfer is what separates a shock chamber from a cycling one. Anyone specifying the test checks the transfer time against the standard, then the recovery time under load, then the rated load mass. A chamber that reaches the extremes without a fast transfer runs a cycle in disguise. The speed of the trip carries the entire point of A106.

Common questions

What is thermal shock testing per JESD22-A106?

A106 moves a part from a hot extreme to a cold one as fast as the equipment allows. The speed is the point. The rapid transfer leaves the part’s surface at the new temperature while the core lags, opening a steep gradient inside. That gradient stresses the interfaces harder than a slow temperature cycle does. JESD22-A106 fixes the extremes and the transfer limit.

How is thermal shock different from temperature cycling?

Both swing a part between the same two extremes. Temperature cycling ramps from one to the other slowly, so the part stays near one temperature throughout. Thermal shock transfers across in seconds, so the surface reaches the new temperature while the core lags. That lag forms a steep gradient, the stress shock adds over a cycle. A part can pass the cycle, then fail the shock at the same extremes.

Why does the transfer have to be fast?

The gradient depends on the speed. A fast transfer hits the surface with the full step at once, so the inside lags far behind, leaving the gradient steep. A slow transfer lets the inside keep pace, so almost no gradient forms. Speed drives the entire mechanism of the test, which is why the standard caps the transfer time.

What temperatures does A106 use?

A common pair runs from minus 55 degrees Celsius to plus 125. The part dwells at each extreme for five to thirty minutes, long enough to reach that temperature throughout. The standard lists several combinations of extremes for different severities. The transfer between them happens inside a short, capped window.

Air-to-air or liquid-to-liquid?

Both deliver a shock. Air-to-air uses a hot zone over a cold zone with a basket moving the parts from one to the other, the common method for semiconductors. Liquid-to-liquid plunges the parts from a hot fluid into a cold one, both inert, pulling heat faster than air, driving a steeper gradient. Liquid stresses harder. Air handles dry parts more cleanly, so it covers the bulk of the work.

Why run boards through thermal shock too?

Solder joints on a board face the same mismatch a package does. The joint bonds a component to a board of different expansion, so the gradient strains it on every transfer. Thermal shock cracks a marginal board joint the way it cracks one inside a package. A board headed for a wide temperature swing in service goes through the shock to prove its joints.

Part of the Envsin guide to semiconductor reliability testing. A thermal shock chamber lives or dies on its transfer time, so specify that against the standard before any other figure.