Reflow Void Reduction Guide
A reflow void is a bubble of gas trapped inside a solder joint. It is invisible from the outside, it reduces the area that carries the current and the heat, and it is measured by X ray rather than by the eye. The gas comes from the flux and from the surface, and the way to keep it out of the joint is to let it escape before the alloy solidifies.
What a Void Is
A void is a volume of gas that was unable to leave the joint before the solder froze. It often has a smooth spherical shape, because the gas was compressed by the liquid metal while it cooled.
The effect on the electrical performance is small in most joints, because a void rarely spans the whole section. The effect on the thermal performance can be large, because the thermal path is short and a void in the middle of a thermal pad blocks it.
The acceptance criteria are quoted as a percentage of the joint area, and the value depends on the product. A general purpose board may accept twenty five per cent, while a power device may require less than ten.
The measurement is by X ray, and the comparison of the available methods is in the X ray and AOI guide.
Where the Gas Comes From
The flux contains solvents that evaporate during the preheat and activators that release gas when they react with the oxide. Both are sources of vapour inside the joint.
The powder in the paste carries an oxide layer, and the flux that surrounds each particle has to remove it. The reaction produces gas at the worst possible moment, while the alloy is melting.
The board and the components also release moisture. A laminate that has absorbed water from the air gives off steam at soldering temperature, and a plastic package does the same.
A plated through hole that is not filled lets air out of the barrel, and a via in a thermal pad is a direct path for the gas into the joint.
Paste Volume and Aperture Design
The volume of flux in a joint is proportional to the volume of paste, so a joint that receives too much paste has too much gas to expel.
A large single aperture on a thermal pad is the classic case. The paste is thick in the middle, the flux cannot escape sideways, and the void forms under the centre of the device.
Dividing the aperture into a grid of smaller openings with gaps between them gives the gas a route out. The total volume is kept the same and the voids fall by a large factor.
The stencil thickness also matters, because a thicker stencil deposits more paste for the same area. A thick stencil with a large aperture is the worst combination.
Profile and Outgassing
The profile has to give the gas time to leave before the alloy solidifies. A soak that is too short reaches the melting point with the flux still wet, and the gas is trapped.
The peak also matters, because a low peak gives a short time above liquidus and the bubbles have less time to rise through the molten metal.
The ramp down is the critical part in many cases, since a fast cooling rate freezes the joint while the gas is still moving. A slower cooling rate allows more of it to escape.
The board and the component also act as a heat sink, so the joint under a large device cools more slowly than a small joint. That difference is why one area of the board can show voids and another does not.
Thermal Pads and Large Joints
A thermal pad is the largest joint on most boards, and it is the one where voids matter most. The pad is under the device, the gas cannot escape sideways and the thermal resistance rises with every void.
The paste pattern on a thermal pad is usually a grid of small deposits rather than one large area. The gaps between the deposits act as channels for the gas and as the place where the flux collects.
The pad also has to be kept clean and flat, because a contaminated surface produces more gas and a warped pad produces an uneven joint with trapped air at the edges.
The thermal management work and the void reduction work meet at this point, since a void and a thin joint both raise the thermal resistance of the same path.
Measurement and Acceptance
The measurement is done by X ray, and the result is reported as a percentage of the joint area. The instrument has to be calibrated, because the apparent void fraction depends on the threshold that is used.
The acceptance criterion should be part of the assembly specification, with a different value for a thermal joint and for a signal joint. A single figure for the whole board is usually either too loose for the power device or too tight for the rest.
Samples should be sectioned to confirm what the X ray image shows, since a void and a thin joint look similar in a two dimensional image and different in a section.
The result should be recorded by area of the board rather than as a single number, because a void under a power device matters far more than one under a resistor.
Practical Rules
Design the paste pattern for a large pad as a grid rather than as one opening, and keep the total volume the same.
Give the profile enough soak to dry the flux and enough time above liquidus for the gas to escape, and slow the cooling if the voids persist.
Keep the board dry before assembly, since moisture from storage is a hidden source of gas that appears only at soldering temperature.
Set a void criterion per joint type and check it by X ray, and record the result with the assembly documentation.
Process Control and Verification
On a design of this kind, paste volume is the item that decides how the rest of the board is arranged. Keeping a sample from the panel turns a dispute into a measurement, because the same coupon can be re-examined by both parties without rebuilding the batch. Reviewing the design before the data is released is cheaper than correcting it after the panel is in the tank, because every step downstream inherits the decision made at the front end.
Related reading: our fabrication notes, board quality and design release notes cover the same ground.
Process Control and Verification
On a design of this kind, paste volume is the item that decides how the rest of the board is arranged. Keeping a sample from the panel turns a dispute into a measurement, because the same coupon can be re-examined by both parties without rebuilding the batch. Reviewing the design before the data is released is cheaper than correcting it after the panel is in the tank, because every step downstream inherits the decision made at the front end.
FAQ
Are voids always a defect? Not always. A small void in a signal joint is usually acceptable, while the same void under a power device raises the thermal resistance and is a defect.
What causes most voids? The flux and the surface condition. The gas comes from the solvents and from the reaction with the oxide, and it is trapped when the joint freezes before it can escape.
Does a longer profile always help? Up to a point. More time above liquidus lets the gas rise, but too long a time grows the intermetallic layer and reduces the joint strength.
