Automotive Grade IC Qualification Chamber Per AEC Q006

AEC-Q006 is the stress-qualification standard the Automotive Electronics Council wrote for copper wire interconnects. Inside almost every chip, fine wires carry the signals from the silicon die out to the legs of the package. For decades those wires were gold. Gold is soft, easy to bond, and chemically inert, so a gold bond sits in a package for the life of a car without changing. The trouble with gold is its price. As chips grew and wire counts climbed, the cost of gold became too much to bear. The industry switched to copper. Copper conducts better and costs a fraction as much. Copper also brings a problem gold never had. It corrodes, and the place it corrodes is the bond itself. The switch was not a small one. A modern chip can carry thousands of bond wires, so the gold in them added a real cost to every part, multiplied across the billions of chips an industry ships in a year. Copper cut that cost sharply and conducted current a little better besides. The saving was too large to pass up. The corrosion copper brought was the bill that came with it. AEC-Q006 is how that bill is paid in testing, on the bench, where the failure costs nothing.

Why copper needed its own rule

The bond between a wire and a chip is not a simple touch of metal on metal. When the bonding tool presses a copper ball onto the aluminium pad of the die, the two metals diffuse into one another and form a thin layer of copper-aluminium compound. That layer, the intermetallic, is what holds the wire to the chip. A gold bond forms a gold-aluminium intermetallic that is stable and inert. A copper bond forms a copper-aluminium intermetallic that is not. Given moisture and a trace of the right contaminant, the copper-aluminium layer corrodes. A corroded intermetallic is a bond on its way to opening.

This is the reason copper needed a standard of its own. AEC-Q006 is an addendum to the main chip qualification, written to cover the one thing the move to copper changed: the reliability of the interconnect. It does not replace the chip standard. It adds to it, with environmental stresses run longer and harder than the chip standard asks, because the copper bond takes more proving than a gold bond ever did. A device can pass the general qualification and still hide a copper bond that will corrode, so the addendum exists to find that bond before a car does.

The corrosion that hides in the bond

The corrosion that AEC-Q006 hunts is a slow, hidden thing. It begins at the edge of the bonded ball, the rim where the copper-aluminium intermetallic meets the air trapped in the package. Moisture works through the moulding compound over the days of a soak and reaches that rim. On its own, clean water would do little. The danger comes when the water carries a halide, a chlorine or bromine ion left in the moulding compound, because the halide attacks the aluminium and the intermetallic in a way pure water cannot. The corrosion eats inward from the rim, replacing sound metal with a brittle oxide that carries little current and holds little strength.

For a long while the bond still works. The corrosion takes the outer ring of the intermetallic first. The centre still carries the wire. As the soak continues, the corroded ring grows inward, the conducting area shrinks, and the resistance of the bond climbs. Left long enough, the corrosion reaches the centre. The bond lifts away from the pad. A chip with a bond corroding in this way passes every test on the day it ships. It fails a year or two into service, on a damp morning, as an intermittent open that comes and goes with the weather. The entire purpose of the qualification is to make that slow corrosion happen fast, in a chamber, where it can be measured before the part ever reaches a car.

The reading of it is a rising resistance. A healthy bond carries current with almost no resistance of its own. A corroding bond adds resistance as its conducting area shrinks. That climb, measured across the soak, is the signal the test watches. A bond that has lifted entirely shows as an open circuit. Between sound and open lies the slow rise that tells the engineer the corrosion is at work and how far it has gone.

Gold sat in a package for the life of a car. Copper has to be proven it will.

The reflow that comes first

Before any wet test runs, the part goes through a preconditioning that mimics the heat of being soldered to a board. The device is moisture-soaked, then run through a reflow profile two or three times, the way it would be during assembly and any rework. AEC-Q006 sets this preconditioning to a defined moisture sensitivity level, so the bonds are judged after the heat of assembly has worked on them, the way a real board would deliver them. The reflow matters for a copper bond in particular. The heat of soldering can grow the intermetallic a little. A marginal bond, or a cratered one, can take its first real harm there. Preconditioning puts that harm in before the corrosion test begins. The soak that follows then works on a bond carrying the stresses of a real assembly. The chamber that runs the preconditioning has to hold the soak and the reflow to the level the standard names, since a precondition run too gently would let the wet test judge a bond easier than the one a car will carry.

HAST, the pressure that speeds the corrosion

The corrosion of a copper bond would take years to show at the gentle warmth and damp of a car. The qualification cannot wait years, so it uses pressure to compress the time. The highly accelerated stress test, HAST, holds the part at around a hundred and thirty degrees and eighty-five percent humidity, under a pressure of roughly two atmospheres, with a bias across the device. The pressure is what allows that much humidity at that high a temperature, since water would otherwise boil away. The detailed physics of how a pressurised vessel raises the boiling point to keep the humidity in place belong to the HAST method itself. What matters for a copper bond is the result: a soak that would take a thousand hours at the usual eighty-five and eighty-five is driven into a long weekend. The corrosion that would take field years is forced out in days.

The bias matters as much as the heat and the damp. A voltage across the device drives the corrosion at the bond, pulling ions along the field and speeding the attack on the intermetallic. AEC-Q006 leans on HAST harder than the plain chip standard does, running it longer because the copper bond is the thing being judged. A gold bond would shrug off the same soak. A copper bond, fed by the halide in its own moulding compound, corrodes under it. The chamber that holds the soak steady is what makes the result mean something. There is a reason the bias points the way it does. A device under HAST is held with a voltage across its pins. That voltage sets up a field at the bonds. Where the field drives metal ions away along its direction, the corrosion runs faster, so the bonds at one polarity tend to fail before the others. A copper-bond qualification reads every bond, at both polarities, so the worst-placed bond in the device sets the result. The chamber has to carry that bias to every part without a break, in an atmosphere of hot, pressurised steam that would corrode a careless feedthrough as surely as it corrodes a weak bond.

Halide and the moulding compound

The corrosion needs a partner. It finds one inside the package. The moulding compound that seals a chip is a complex resin. Older compounds carried halogen, a chlorine or bromine added as a flame retardant. A trace of that halogen, freed by heat and moisture, is exactly the contaminant a copper bond cannot tolerate. The halide ion shuttles back and forth at the intermetallic, carrying the corrosion forward without being used up, so a tiny amount does lasting harm. This is why a copper-bonded device is so sensitive to the cleanliness of its own materials. A part built with a low-halogen, green moulding compound corrodes far slower than one built with an older resin. AEC-Q006 puts both through the same soak so the difference shows. The chamber does not change the chemistry inside the part. It holds the heat and the damp so the chemistry, whatever it is, runs to its conclusion within the test. The move to green compounds has eased the problem without ending it. A low-halogen resin starves the corrosion of its partner, so a bond built with one survives the soak far longer. No resin is perfectly clean. A copper bond stays more sensitive to what surrounds it than a gold bond ever was. AEC-Q006 holds the part to the soak regardless of the compound it is built with, so a maker cannot lean on a clean resin to skip the proving. The standard assumes the worst the materials might bring and asks the bond to survive it.

Copper is harder to bond to begin with

Corrosion is not the only thing copper brought. Copper is a far harder metal than gold. That hardness shows at the moment of bonding. To weld a copper ball to an aluminium pad, the tool presses it down with force and adds ultrasonic energy to make the metals flow together. A soft gold ball deforms easily and bonds gently. A hard copper ball needs more force and more ultrasonic power. That extra energy goes somewhere. It can crack the silicon under the aluminium pad, a defect called cratering, where the pad and the dielectric beneath it fracture from the violence of the bond. A cratered bond may pass on the day it is made. The crack it leaves gives moisture a faster road to the intermetallic later, so a bonding defect becomes a corrosion path under a humid soak. AEC-Q006 catches both the cratering and the corrosion it leads to, which is the reason the standard pairs its environmental soaks with the mechanical checks that pull and shear the bond.

High-temperature storage and the growing intermetallic

Humidity is one face of the copper-bond problem. Heat alone is the other. The copper-aluminium intermetallic that holds the bond does not stop forming once the bond is made. At a high temperature it keeps growing, thickening over time. As it thickens it changes. The copper and the aluminium diffuse into one another at unequal rates. The imbalance leaves tiny voids behind, a weakening of the layer from within. A thick, voided intermetallic is brittle and resistive, a bond aged toward failure by heat with no moisture involved at all. AEC-Q006 runs a high-temperature storage test, holding parts at a hundred and seventy-five degrees for two thousand hours, to drive that growth and find the bonds that cannot take it. For the chamber the demand is a steady, accurate, high heat held for the long weeks the test runs, so every bond ages by the amount the standard intends. The voids are the heart of why heat alone can kill a copper bond. As the copper and the aluminium diffuse, the faster-moving metal leaves vacancies behind. The vacancies gather into voids along the intermetallic, a row of tiny gaps where solid metal should be. The voids carry no current and bear no load, so a bond riddled with them grows resistive and weak even though no moisture ever touched it. A storage test long enough to grow the intermetallic to its limit is what separates a bond that will last from one that will void itself open over the years of a hot engine bay.

The bond is the thing being judged

A point worth keeping in view is what AEC-Q006 qualifies. The general chip standard asks whether the silicon works and keeps working. AEC-Q006 narrows the question to the interconnect: does the copper wire stay bonded, with low resistance, through years of heat and damp. A device can compute perfectly and still fail this, if a bond corrodes or a wire lifts. So the reads that decide the test are reads of the bond. The environmental soaks corrode and age it. The mechanical checks test its strength directly, a wire pull that measures the force to tear the wire away and a ball shear that measures the force to push the bonded ball off the pad. A bond that has corroded or grown brittle gives way at a lower force than a sound one. The drop in strength is as telling as a rise in resistance.

Reading whether the bond held

Reading a copper bond combines the electrical and the physical. Through the soaks, the resistance of the bond chain is watched. A climb says corrosion is shrinking the conducting area. An open says a bond has lifted. After the stress, the mechanical strength is measured by pull and shear. A weak result says the intermetallic has corroded or voided even where the bond still conducts. Where a number raises a question, the bond is cross-sectioned and examined, the package ground away until the joint can be seen, so the corroded ring or the voided intermetallic shows under a microscope. The chamber’s part in all of it is the trust behind the numbers. A resistance that climbed under a clean, steady HAST soak points at the bond. A number read after a soak that lost its pressure or its humidity points nowhere, since a drift in the bond cannot be told from a fault in the vessel. The cross-section is the final word. When the resistance and the strength leave a doubt, the part is potted, ground down and polished until the plane of the bond is exposed. The joint is examined under a microscope. A sound bond shows an even intermetallic across the foot of the ball. A corroded one shows a ring of dark oxide eating in from the rim. A voided one shows a line of gaps along the layer. The picture confirms what the numbers suggested. It tells the maker how a bond failed, which is what a process engineer needs to fix the cause.

Graded across the range it serves

Because AEC-Q006 is an addendum, its grades follow the chip standard it extends. A copper-bonded device claims the same temperature grade an integrated circuit would, from the mild interior ranges up to the harshest under-bonnet grade that reaches a hundred and fifty degrees. The grade sets how hard the environmental stresses run. A part bound for the engine bay is aged and soaked at the higher temperatures, where the intermetallic grows fastest and the corrosion runs hardest. A part for a milder location faces a gentler version of the same battery. The chamber serving the top grades has to hold the higher temperatures of the storage test and carry the bias through the pressurised soak, all while keeping the conditions even across a loaded rack of devices.

The longer durations copper demands

The clearest mark of AEC-Q006 is how much longer it runs than the chip standard. Copper needs more proving, so the durations stretch. The high-temperature storage runs to two thousand hours where a chip standard might stop sooner. The temperature cycling runs to a high count, into the thousands of cycles, to work the bond through repeated expansion. The unbiased accelerated soak runs to the better part of two hundred hours after a moisture preconditioning. These longer times are not arbitrary. They are set to expose the copper bond’s slower failure modes, the corrosion and the intermetallic growth that a shorter soak would leave hidden. For the chamber the cost is occupancy. A copper-bond qualification holds its chambers longer than a gold-bonded part ever would. A lab planning the work sizes its fleet around the longer runs. A storage test of two thousand hours is twelve weeks. A HAST campaign and a long cycling run stack onto that. A single copper-bonded part can occupy a bank of chambers for months between its first read and its last, far longer than a gold-bonded part would have. A lab that qualifies copper sizes its schedule around runs that do not end quickly.

Turning HAST hours into field years

A long weekend in a pressurised chamber stands in for years of a car’s damp life. The bridge is acceleration. Heat, humidity, pressure and bias together drive the corrosion of the intermetallic far faster than the mild warmth and damp of service would, along the same chemical path. A model ties the hours of the soak to the years in the field. The model rests on the chamber having held the full condition without a break. An hour where the pressure dropped, or the humidity fell, or the bias failed, is an hour the corrosion was not driven. A field-life claim built on a soak with such gaps claims a durability the copper bond never earned. The chamber is the one part of that chain a lab fully controls, so its steadiness is where the credibility of the entire result begins.

Why a wrong call lands hard

A copper bond that corrodes does not fail loudly. It fails as an intermittent open, years out, on a cold or a damp day, in a part that tested perfectly when new. That is the hardest kind of failure to trace in the field, because it comes and goes with the conditions and leaves little behind. A qualification that passed a marginal copper bond would let exactly that fault into a car. The cost of finding it later, scattered across a fleet, dwarfs the cost of the longer soak that would have caught it on the bench. This is the reason AEC-Q006 runs its stresses longer and reads the bond directly, by strength as well as resistance, and cross-sections what the numbers leave in doubt. The standard would sooner reject a sound bond than pass a corroding one into service.

What the pressurised chamber must do

What AEC-Q006 asks of its chambers is steadiness across long, harsh runs, on a failure that hides until it is too late. The HAST vessel has to hold around a hundred and thirty degrees and eighty-five percent humidity under roughly two atmospheres of pressure, with a bias carried cleanly to every device, for days at a stretch. The storage oven has to hold a hundred and seventy-five degrees flat for two thousand hours. The cycling chamber has to swing the parts through thousands of turns without losing its extremes. Each has to keep its conditions even across a loaded rack, so the bond in the worst corner sees the same stress as the bond in the best. A set of chambers that holds all of that lets a maker put a copper-bonded part in a car and stand behind it. The grade is a promise that the bond will hold through the heat and the damp of a car’s life. The chamber is what makes the promise honest, one steady soak at a time.

Envsin reliability and environmental test chambers for automotive copper wire-bond qualification.