Reflow Soak Profile And Void Formation

A reflow profile has four parts: a ramp, a soak, a spike to the peak and a cool down. The soak is the part that engineers most often remove in the name of speed, and it is also the part that decides how much gas is trapped inside the finished solder joint. Understanding what the soak does makes the trade between cycle time and voiding explicit.

This article explains where the gas comes from, what the soak stage does about it, how ramp rate changes the result, and which voids no profile can remove.

Where The Gas Comes From

Every void in a solder joint began as something that was not solder. Volatiles from the solder paste solvent and from the flux are the most common source; water absorbed by the laminate and released through the pad is a second; and air trapped by the powder structure or by a paste deposit that sits over a via hole is a third. During reflow the flux becomes fluid and the gas it produces has to leave through the still-liquid alloy before the joint freezes.

The route out is through the top of the joint, and the time available is the interval between the melting of the alloy and the solidification at cool down. Anything that shortens that interval, or that raises the rate of outgassing, increases the amount of gas left behind. This is why voiding is a process signature rather than a material constant, and why two boards built with the same paste on the same machine can differ if the profiles differ. Where the joint sits over a via, the trapped volume is larger and the problem is worse, as described under via in pad or plated through.

What The Soak Profile Does

The soak profile holds the assembly at a temperature below the liquidus, usually between one hundred and fifty and two hundred degrees, for a period long enough that the whole board reaches a uniform temperature and that the solvent and the lighter flux fractions evaporate before the alloy melts. Gas that leaves during the soak escapes through the still-open paste structure and never enters the joint, which is why a soak reduces voiding rather than merely slowing the process.

A soak also reduces the spread of temperatures across the assembly. Without it, a board with a heavy component and a light one reaches the liquidus at different times, and the profile has to be set for the slowest point, which overheats everything else. With a proper soak the whole board converges on a similar temperature before the spike, so the peak can be lower and the time above liquidus shorter. That is the real argument for the stage: it is not only about volatiles, it is about making one profile work for an uneven assembly. Choosing the alloy that the profile is written for is covered under lead free versus leaded solder.

Ramp Rate And Its Trade-offs

The ramp into the soak controls how fast the paste warms. A fast ramp saves cycle time and is appropriate for a thin, uniform board where the whole assembly heats together. A slow ramp is kinder to components that are sensitive to thermal shock, particularly multilayer ceramic capacitors, and it gives the flux more time to begin working before the soak. The cost of a slow ramp is time and, in some cases, more oxide growth on the surfaces before the activator reaches them.

The ramp out of the soak to the peak is the one that has the greatest effect on voiding. A rapid rise to the peak shortens the time the alloy spends liquid, which leaves less time for gas to escape, but it also produces a more vigorous flux reaction that can generate gas faster. The optimum is usually found experimentally, and it is a combination of a moderate ramp and a peak that is above the minimum for the paste by a comfortable margin. Where the ramp is too fast, the joint often shows a burst pattern of small voids rather than one large cavity, which is a useful diagnostic when a profile is being tuned.

Where Voiding Persists

Some voids cannot be removed by any profile. A through hole that is not plugged under a ball grid array pad forms a reservoir that the solder cannot fill, and the gas from the flux has nowhere to go but into the joint. Ground planes and heavy thermal reliefs pull heat away from the pad so quickly that the joint is liquid for a much shorter time than its neighbours, and the void content in those joints is consistently higher. Both are layout decisions, and both are cheaper to change than the profile.

Paste selection sets a floor as well. A paste with a high metal loading and a low solvent content produces less gas, and a paste with a fine powder produces more surface oxide, which the activator has to remove and which can generate more residue. The deposit also matters: a thick deposit over a large area has more solvent to expel than a thin one, and the way the deposit is formed is a printing question rather than a reflow one, as described under placement order and pad positioning.

Measuring And Accepting Voids

Voiding is measured by X-ray in production and by microsection when a root cause is needed. The production measurement is an area percentage on a slice through the joint, taken at the widest plane, and it is compared against an acceptance limit that comes from the product specification. The limit is normally a total area percentage plus a maximum for a single void, and for a thermal pad it may also include a limit on the voiding at the interface with the pad.

The measurement has to be repeatable before it can be used as a control. Voxel size, threshold and the plane chosen all change the number, so the same equipment settings and the same measurement rule are used for every batch. Comparing a number produced at one setting with a number produced at another is one of the most common sources of confusion in voiding discussions, and it is worth fixing the rule in the process specification rather than in a person’s memory.

Changing The Profile Safely

When a profile is changed to reduce voiding, only one variable should move at a time. Extending the soak and reducing the ramp to peak are separate experiments; changing both together makes the result uninterpretable. Each change should be evaluated on joints that are representative of the worst case on the assembly, which means the thermal pad, the ground connected joint and the joint nearest the board edge, not a convenient signal joint in the middle.

The change should also be checked for its other effects before it is released. A longer soak increases the time the paste spends at temperature before melting, which can dry the deposit in an extreme case, and it exposes the components to a longer thermal excursion. A higher peak increases the intermetallic growth and the thermal load on the parts. The profile that gives the lowest void content is not automatically the correct one, and the process window approach that balances all the requirements is set out under design guidelines for manufacturability.

FAQ

Does a longer soak always reduce voiding? Up to a point. Beyond the point at which all the volatiles have been driven off, further soak time adds cycle time without changing the joint, and it begins to work against the paste by drying it before the alloy melts.

Why is one pad on the board consistently worse than the others? Look at what is connected to it. A pad tied to a ground plane or a thick thermal relief reaches the liquidus later than its neighbours, so the joint is liquid for less time and holds more gas. The fix is usually in the copper rather than in the profile.

Can voiding be eliminated entirely? In a real process it cannot. The practical aim is to keep the total area and the largest single void inside the acceptance limit, and to make the result repeatable so that a change in voiding indicates a change in the process.