BGA Voiding: Causes, Measurement and Process Control
A void is a space inside a joint where alloy should be, and in a ball grid array it is measured as a percentage of the joint area on an X-ray image. Whether that percentage matters depends on where the void sits, and the causes are spread across the paste, the profile and the board rather than sitting in one place.
How Voids Form in a BGA Joint
The gas that becomes a void comes from the flux volatiles, from air trapped under the ball when it is placed and from the particles of powder that carry an oxide shell. During reflow the alloy is liquid and the gas has to escape through the surrounding paste before the joint freezes, and any gas that does not get out remains as a sphere.
The three sources behave differently. A volatile that is generated early escapes easily, while one generated at the peak is trapped by the solidifying alloy, and air trapped under a ball at placement has no route out at all until the paste around it liquefies. The shape of the void depends on surface tension, since the alloy pulls the gas into a sphere, and a small sphere has a different effect on the joint than a flat layer of the same volume.
Where the Void Sits Matters
A void in the middle of a joint away from either interface has little effect on the electrical path, while a void at the pad interface or at the package interface reduces the area that carries current and the area that carries heat. The position is therefore reported with the size rather than being averaged into a single figure.
In practice the interface void is the one that causes failures, and it is also the one that is harder to see, because an X-ray image is a projection and a void at the top of a ball overlaps the one at the bottom. A tilted view or a computed tomography scan separates them, and the section is the final confirmation.
How Voiding Is Measured
X-ray inspection reports the void area as a percentage of the joint area, usually with a maximum for a single void and a limit for the total. The measurement depends on the grey level threshold that separates alloy from void, and the threshold is a setting that has to be fixed and recorded rather than adjusted per board.
A joint with 25 percent voiding distributed as many small spheres behaves differently from one with a single 25 percent void across the interface, and a single number cannot distinguish them. That is why the criteria are written with both a size limit and a position rule, and why the inspection programme reports the largest void as well as the total. A reported figure is therefore always quoted with the equipment and the threshold, because changing the inspection system changes the numbers without any change to the joints.
Paste and Its Contribution
The paste volume deposited on the pad sets how much flux is present to generate gas and how much alloy is available to fill the joint. A deposit that is too small leaves a joint with little alloy and a high void fraction, while one that is too large traps flux under the ball with no route out.
The powder size and the oxide content of the powder matter as well, because an oxidised powder carries its own gas into the joint. A paste that has been open too long has a higher oxide load, which is one reason voiding can appear late in a shift without any change to the printer.
Profile and the Escape Path
The profile controls how long the alloy is liquid and how gradually it solidifies, and therefore how much time the gas has to leave. A longer soak and a slower ramp through the flux activation range reduce voiding, while a fast ramp to the peak traps the volatiles in the joint.
The cooling rate has an effect too, because a joint that freezes quickly locks in whatever gas is still present. Where voiding is the problem, the profile is changed in small steps with the X-ray result measured after each one, since the relationship is not linear and is not the same for every paste. Soak time is the easiest parameter to change and the one that most often produces a measurable improvement, because it acts on the volatiles before the alloy melts.
Stencil, Pad and Placement Effects
The stencil aperture and the pad size control the deposit, and a pad that is too small for the ball leaves the flux with nowhere to escape around the joint. The placement force and the height at which the ball is released also change how much air is trapped underneath it.
Where a via sits inside the pad, the via has to be filled and capped or the void will form inside the via and be counted as part of the joint. That is a design decision rather than a process one, and it is made before the board is fabricated rather than during the void investigation.
Reducing Voiding in Production
The practical measures are a fresh paste with a known open time, an aperture sized to the pad, a soak long enough to drive the volatiles off and a placement that does not press the ball into the paste. Each is measurable, and each is changed one at a time.
Where the target is aggressive, a vacuum assisted reflow or a profile with a longer time above liquidus is used, and both are verified with the same measurement. Our paste volume notes describe how the deposit is measured, which is the first place to look before the oven is touched. Where a change is made, the board is X-rayed before and after so that both results come from the same threshold and the same viewing angle, which is the only way the comparison carries any information.
Acceptance Criteria and Their Limits
Acceptance is written as a maximum total voiding, a maximum single void and a rule about interface voids, and the figures come from the product’s thermal and mechanical duty rather than from a general standard. A power device and a signal ball do not have the same requirement.
The criteria also have to be measurable with the equipment in use, because a limit of 10 percent cannot be enforced on a machine whose repeatability is 15 percent. The measurement capability is therefore established before the limit is written, and our X-ray and AOI notes describe where each method is applied.
Records and Process Control
The record carries the paste lot and open time, the aperture set, the profile with the measured peak, the placement settings and the X-ray result with its threshold. Those five items describe the process that produced the joint, and a change to any of them is a change to the voiding result.
Where a build is at the limit, the sampling is increased rather than the limit relaxed, because a process at the edge of its capability is one whose results will move. Our board failure notes describe how a void that later becomes a crack is recognised, which is the outcome the limit exists to prevent.
FAQ
Are voids always a defect? No, and the position matters more than the percentage. A void away from both interfaces with a moderate total is usually accepted, while a void across the pad interface is a defect at a much smaller size.
Can voiding be fixed by changing the paste? Sometimes, and the paste is usually the second thing to change rather than the first. The aperture, the paste open time and the profile are cheaper to adjust and are more often the cause.
Why does the X-ray result vary between machines? Because the threshold that separates alloy from void is a setting, and two machines with different settings report different areas for the same joint. The threshold is fixed, recorded and repeated before results from two lines are compared.
