Temperature & Humidity / Semiconductor
Preconditioning Chamber for SMT Components MSL 1 to 6
Moisture soak, reflow, and the MSL grade a surface-mount package earns
A preconditioning chamber drives a measured amount of moisture into a plastic surface-mount package, holding it at a set temperature and humidity until the body has taken up as much water as it would sitting on a factory floor. A reflow profile follows. The whole exercise hunts, in a few days of soak plus a pass through soldering heat, for the packages that crack in real assembly.
Why a wet package fails
The plastic over a chip is not waterproof. Epoxy mould compound is hygroscopic. It pulls water vapour from humid air, the water working into the bulk to collect along the internal interfaces: the die-attach, the leadframe, the face of the die.
None of that shows until the board reaches reflow. That moment is what the soak is built around. The package carries its worst moisture into the heat, the heat does the damage, the test sees it before a customer would.
The pop
Reflow takes the whole package from room temperature to the solder peak in under two minutes. For lead-free SAC alloys that peak sits near 245 to 260 degrees, the part held above the roughly 217-degree liquidus for tens of seconds. The exact ceiling tracks the body: a thin, small package can be classified up near 260; a thick, voluminous one is capped lower, nearer 245, its mass unable to shed heat fast enough to keep a higher peak safe. The water trapped inside reaches that temperature too. Saturated steam at 260 degrees carries a vapour pressure of several megapascals, tens of atmospheres, pressing outward on interfaces whose adhesion was never built for it. The pressure does not build gently. The part crosses the boiling point of the trapped water in seconds, so the steam forms in a rush. A package that might bear the same pressure applied slowly is split by the suddenness of it. The faster the body reaches the peak, the less time the vapour has to bleed out through the mould and the higher the pressure climbs before anything gives. The weakest bond goes first: a void left in the die-attach gives the steam a ready start, a thin skin of mould over a die corner concentrates the stress until it tears; the leadframe, smooth metal against epoxy, holds least of all where water has already crept along the join. The die-attach lifts. The mould compound peels off the leadframe. In the loudest version the body swells, then splits with an audible report, the pop that named the effect.
The quiet version does the real harm. A delamination that never breaks the surface still tears a wire bond at its heel, fractures the die where the swelling package drags on it, or opens a path that lets moisture corrode the metal months later in the field. One reflow can manage all of it. A board sees reflow two or three times over, between double-sided assembly and rework, so the package has to come through wet on every pass, not just the first. A marginal interface that held on the first heating can let go on the third, the damage compounding pass by pass until the joint inside finally opens.
The reflow profile, step by step
The heat that finds a wet package is no single jump to the peak; it is a shaped profile whose shape decides how hard the moisture hits.
It opens with a preheat, the board climbing at a degree or two a second to drive off solvents and wake the flux. Then comes a soak zone, a hold in the range of 150 to 200 degrees for a minute or two, where the flux does its work and the assembly evens out in temperature. Only then does the profile ramp to the peak, crossing the solder’s liquidus near 217 degrees for the common lead-free alloys, then staying above it for a controlled stretch, the time above liquidus, held to something like 60 to 90 seconds so the joints form without cooking the parts.
The peak itself is brief, a few seconds near 245 to 260 degrees, before a cooling ramp brings the board back down. That short time at the top is when the trapped water is hottest and its vapour pressure highest, the instant the popcorn failure lives or dies.
Preconditioning runs the soaked part through this whole profile, to the peak its package class allows, more than once over. A board meets reflow two or three times across a full build, double-sided assembly and any rework among them, so the test repeats the profile to match, each pass another chance for a marginal interface to let go.
MSL 1 to 6
The grade that comes out the far end is the Moisture Sensitivity Level, defined by the IPC/JEDEC standard J-STD-020. A single question fixes it. Out of the dry-pack, in ordinary factory air taken as no warmer than about 30 degrees at no more than 60 percent humidity, how long can the part sit before it has drunk enough to risk the pop?
The scale runs from forgiving to fragile. MSL 1 never needs a bake. MSL 2 buys roughly a year of bench life; 2a, about four weeks; 3, about a week, where a great many plastic parts settle. Past there it tightens hard: 4 near 72 hours, 5 near 48, 5a near 24, then 6, which has to be baked immediately before every reflow.
Deeper number, thirstier body.
A maker earns the grade by running the preconditioning sequence at that level’s soak. Claim MSL 3, soak to the MSL 3 condition, reflow, inspect; come through clean, the claim holds. A failure drops the part to a thirstier rating, a shorter floor life, a heavier handling burden on whatever line buys it. That burden is real: an MSL 1 part sits in the open, ignored; an MSL 3 part lives on a floor-life clock, sent back to a dry cabinet near a few percent humidity once the clock runs low; an MSL 5a or 6 part is handled close to a perishable, its exposure logged to the hour, a bake before the oven, nitrogen-dry storage between. Each step deeper costs the line time and discipline, the reason a maker fights to certify the lowest level a package can honestly carry.
How much water, how deep
Two separate things decide the state of a soaked package: how much water it holds at equilibrium, plus how far that water has reached into the body.
The amount is set by the humidity through a sorption relationship. Raise the humidity and the epoxy holds more water at equilibrium, climbing along a curve that steepens at the damp end, which is why a few percent of drift near the top of the range matters far more than the same drift lower down. Each level’s soak humidity is chosen to bring the package to the moisture its floor life would.
The depth is set by diffusion and by time. Water creeps inward at a pace that rises sharply with temperature; the time to soak a body through grows with the square of its thickness, so a thick package takes far longer to wet to the core than a thin one. The standard pairs a warm push with a long dwell precisely so the water reaches the centre, never the skin alone.
Both have to be right for the test to mean anything. A part wet only on the outside, or wet to the wrong total, reads as a different level than it carries when the reflow heat arrives.
The soak conditions, level by level
J-STD-020 does not leave the soak to judgment; it tabulates a condition for every level, then offers two routes to reaching it.
The standard soaks sit cooler and run longer, on the order of 30 degrees at 60 percent humidity for the middle levels, stretched across days. The accelerated equivalents trade a hotter, damper hold for a shorter one, clustering around 60 degrees at 60 percent humidity for tens of hours, or reaching 85 degrees at 60 or even 85 percent for the harshest ratings. Either route is meant to leave the package at the same moisture.
The soak times track the severity in reverse. A thirsty deep level needs less soak to reach its modest floor-life moisture, while a robust level soaked to a longer floor life needs more, so the numbers run from a few tens of hours up to the better part of a week, read off the table for the level being qualified.
The dry-out that comes first
Before any moisture goes in, whatever is already there has to come out.
A part arriving for preconditioning carries the moisture it picked up since it was made, an unknown starting point that would muddy the soak. So the sequence opens with a bake, a dry-down at around 125 degrees long enough to drive the part to a known dry state, the same kind of bake J-STD-033 uses to recover an overrun part. Only from that clean baseline does the controlled soak begin, adding back exactly the moisture the level calls for.
Skip the dry-down and the soak starts from a guess, so a part that arrived a little damp would finish the soak wetter than its level intends. The bake is what makes one soak comparable to the next, one lab’s grade comparable to another’s.
The soak is the test
Everything turns on the soak. The chamber dries the part first, often a spell at 125 degrees, so it starts from a known empty state. Then the hold begins: one temperature, one humidity, one length of time, set so the package takes up exactly the water its level stands for.
J-STD-020 fixes those conditions level by level, broadly between 30 and 85 degrees at 60 to 85 percent humidity, the soak running from tens of hours to the better part of a week.
The reason the chamber has to be exact sits in the physics of uptake. Water enters the epoxy faster than in proportion to humidity, closer to a curve than a straight line, so a drift of a few percent across a soak that lasts days changes how much the package finally carries. Change the water, and the reflow that follows bites harder or softer than the grade intends. Cut the soak short and only the outer shell takes on water. The soak runs in days, not minutes, for a plain reason: moisture crosses the mould compound by diffusion, a slow crawl whose pace rises with temperature, so the standard pairs a warmer push with a long dwell to carry water all the way to the centre of the body. Cut the soak short and only the outer shell takes on water, which reads back as a part sturdier than the field will find it.
A chamber that wanders mis-grades the part.
Uniformity carries the same weight across a full load. Let the humidity run high in one corner of the working space, the parts there drink more than the ones in a drier pocket. The grade becomes a record of where a part sat when it ought to record the part itself.
Reading the damage with sound
The instrument that sees inside a sealed package without opening it is the acoustic microscope. It works on a simple fact: sound reflects hard off a gap.
A focused ultrasonic pulse, sent in through a water couplant, passes cleanly through solid epoxy and bonded interfaces, while a delamination the steam has opened throws the pulse straight back with its phase flipped, because the pulse has met air. Scanned across the package, those echoes build a plan-view map where a delamination shows as a bright, sharp patch over the die or along the leadframe. The map is read against the standard’s limits, which cap how much delamination is allowed at each interface, the critical faces of the die surface and the die pad held tightest of all.
The acoustic map does not work alone. An electrical test catches a bond lifted or a die cracked through; a cross-section, destructive but final, settles a borderline call by putting the interface under a microscope. A delamination past the limit fails the part even when it still tests good, since the opened interface is a latent failure the field will finish.
The packages that drink hardest
Not every package carries the same risk; the geometry tells the bulk of the story.
A large, thin package is the hard case. It has a lot of interface area for the steam to work on, a thin mould cap that flexes and tears easily, with a long path for any bake to drive moisture back out. Big plastic ball-grid arrays on laminate substrates, thin quad-flat packs and leadless packages with a large exposed pad all tend to sit at the thirstier levels for this reason.
The ratio of die to package matters too. A big die under a thin skin of mould concentrates stress at its corners and gives a delamination a wide interface to spread along, while a small die buried in plenty of epoxy is better held. The substrate adds its own term: a laminate base soaks up moisture of its own and adds an interface the simple metal leadframe does not.
None of this changes the test, only the result. The same preconditioning sequence runs; the geometry decides what level comes out the far side.
The lead-free penalty
One change swept across the industry and made moisture sensitivity worse everywhere, with nothing to do with the packages themselves.
The move away from tin-lead solder, driven by the restriction of hazardous substances, raised the reflow peak. The old eutectic alloy melted and flowed near 220 degrees; the lead-free alloys that replaced it need a peak some 25 to 40 degrees hotter to do the same job. That hotter peak makes the trapped steam fiercer and the mould compound weaker at the same instant.
The upshot was that many parts dropped an MSL level when their makers requalified them for lead-free assembly. A package comfortable at one level under the cooler old reflow could no longer pass the same soak under the hotter new one; its floor life shortened to match. The preconditioning that grades a part today runs to the lead-free peak, which is why the levels read tighter than they once did, why a part carried over from the leaded era cannot keep its old rating without passing the soak again under the hotter profile.
Out of the bag
A moisture-sensitive part ships sealed against the same water the soak later forces in. The packaging that does the sealing is engineered, not incidental.
The bag is a moisture-barrier laminate, a foil layer between plastic films so that almost no vapour crosses it, heat-sealed around the parts. Inside go units of desiccant, sized to the bag and the time it must guard, soaking up the trace moisture sealed in. Beside them sits a humidity indicator card, a row of spots printed with a salt that shifts colour at set humidities, often near 5, 10 and 60 percent. Inside the bag the part stays dry, its floor clock stopped.
The card is the line’s proof on opening. Spots still dry mean the seal held and the clock starts fresh from the label; spots tripped to a damp colour mean the bag leaked or aged. The parts owe a bake whatever the label allows.
Open a sound bag and the clock starts, the level on the label the number it counts down. A line drawing MSL 3 parts has on the order of a week of bench time before those parts go back to a dry cabinet or into the oven. Reseal with fresh desiccant in time and the clock pauses; overrun it and the part owes a bake.
The dry line the level governs
The grade a preconditioning chamber assigns turns into a set of rules on the assembly floor. The deeper the level, the heavier those rules weigh.
A moisture-sensitive part lives between dry storage and the oven. Dry cabinets, held at a few percent humidity with a nitrogen purge or a desiccant wheel, stop the clock between uses; a floor-life log at the bench tracks how much exposure each reel has spent; the indicator card in every bag is checked on opening to prove the seal held. A part that overruns mid-build is pulled and baked before it goes near the oven.
All of that discipline exists because the part is drinking moisture from the moment its bag is opened, exactly the moisture the preconditioning soak reproduced. The level is the line’s instruction sheet. A part mis-graded by a wandering chamber would have the line either wasting bakes it does not need or skipping ones it does. Either way the cost lands as scrap at assembly or as a field return months later, which is why a maker pays for a chamber it can trust to grade a part right the first time. It is also why the soak conditions are pinned to the degree and the percent in the standard the chamber answers to.
First in the sequence
Preconditioning rarely runs alone. In a device qualification it goes first, ahead of the temperature cycling, the biased humidity, the high-temperature storage that follow. Those later stresses are meant to act on a part that already carries the moisture and the micro-damage of real assembly, with the strain a part fresh from the mould has not yet taken on.
So the soak does double duty: it grades the moisture level on its own, then primes every part for the battery to come.
What the soak chamber must hold
The chamber behind all of this is a temperature-and-humidity box asked to be unusually steady for an unusually long time. A few features decide whether it can grade a part honestly.
It has to generate and hold humidity to a fraction of the band across a soak that runs days without a gap, since the moisture the part reaches is the controlled variable the whole result rests on. It has to hold that humidity evenly through a full working space, so a packed load soaks as one, never as a fed front with a lagging back. It has to log the condition through the soak, so a qualification can show the part met the level it claims, with any excursion caught and judged.
The same box usually does the dry bake too, the 125-degree dry-down before the soak and the moisture-removal bake after a part overruns, so it has to reach a hot, dry corner of its range as readily as a warm, damp one.
Doing the dry-down and the soak in one machine is a quiet convenience the sequence leans on. A part is baked dry, then soaked wet, then handed to reflow. A chamber that does the first two without the parts ever moving keeps the sequence clean and the timing tight, part of why a dedicated preconditioning chamber pays its way in a qualification lab. A chamber that does all of this is what lets the grade it certifies mean the same thing on every line that reads it.
Baking it back out
A part past its floor life is not scrap. J-STD-033, the companion handling standard, lets it be baked to drive the water back out before soldering.
The bake is slow because diffusion is slow. A common recipe holds 125 degrees for a stretch set by package thickness, the heaviest bodies wanting hours that run toward days, the moisture crawling out the way it crawled in.
The numbers come from a table read by package thickness. A thin body might clear in a day or so at 125 degrees; a thick one can want several days. Where 125 is too hot, a bake near 90 degrees runs much longer. For parts left on tape-and-reel, which cannot take real heat at all, a long, low bake near 40 degrees in tightly dried air is the only route, measured in weeks.
Where the part or its reel cannot take 125, that cooler bake does the same job over far more time. The recipe reads the package as closely as it reads the moisture, with the part resealed in a fresh dry bag the moment it comes out, its floor clock back to zero.
The floor clock the chamber sets
A clock starts the moment a moisture-sensitive part leaves its bag. The MSL is what it counts down. Every dry cabinet, indicator card and floor-life log on an assembly line exists for that reason. The preconditioning chamber is where the clock gets set: soak a sample to the worst its floor life permits, prove it lives through the reflow. The grade it earns tells the line just how long it has. Holding humidity is the easy part. Holding it steady and even across a soak of days, study after study, is what makes the grade mean the same thing on every line that reads it.
Questions that come up on preconditioning
What is the popcorn effect?
Moisture absorbed by a plastic surface-mount package flashes to steam at the reflow peak, near 245 to 260 degrees for lead-free solder. The steam’s vapour pressure, several megapascals, delaminates the internal interfaces, and in the worst case bulges and splits the body with an audible pop. Even with no visible crack it lifts bonds and fractures the die.
What does MSL mean?
Moisture Sensitivity Level, set by J-STD-020. It rates how long a part can sit out of its dry-pack in factory air, taken as about 30 degrees at 60 percent humidity, before it must be reflowed or baked. MSL 1 is unlimited; MSL 2 about a year; MSL 3 about a week; MSL 6 must be baked before every reflow.
What does the preconditioning soak do?
It loads the package with the moisture its floor life would, at a temperature and humidity fixed by the level, broadly 30 to 85 degrees at 60 to 85 percent for tens of hours to the better part of a week. The part is then reflowed and inspected. Uptake climbs steeply with humidity, so the soak has to be exact.
Why must the chamber’s humidity be so steady?
Moisture uptake rises with humidity faster than in proportion, so a drift of a few percent over a long soak changes how much water the part holds and shifts the severity of the reflow. An uneven chamber over-doses parts in a damp pocket, under-doses those in a dry one, and the grade stops being a property of the package.
How is a part read after preconditioning?
By acoustic microscopy for delamination inside the package, electrical test for lifted bonds or a cracked die, and a cross-section to settle a disputed result. Delamination past the standard’s limits fails the part even if it still works, since the opened interface is a latent failure.
Why did lead-free soldering make MSL ratings worse?
Lead-free alloys reflow some 25 to 40 degrees hotter than the old tin-lead eutectic, near 245 to 260 degrees against about 220. The hotter peak raises the steam pressure and weakens the mould compound at once, so many parts dropped an MSL level when requalified for lead-free assembly, with a shorter floor life to match.
Can a part over its floor life still be used?
Yes, after a bake. J-STD-033 lets a moisture-soaked part be baked to drive the water out, commonly 125 degrees for a time set by package thickness, or a cooler bake near 90 or 40 degrees where the part or its reel cannot take 125. Its floor clock then resets, and it is resealed in a fresh dry bag.