Sunday, June 24, 2018
Publication Date: 05/1/2008
Archive >  May 2008 Issue >  Special Feature: Test and Measurement > 

X-Ray Inspection of BGA Solder Joints
X-ray image of BGA with bridges after popcorning.

The moisture sensitivity levels (MSL) of semiconductor devices can increase one to three levels from their current values as the temperature goes up for lead-free solders. This means there is an increased danger for most moisture-sensitive devices, including ball grid arrays (BGAs) being used outside of their safe operating envelope during reflow processing.

As a result, popcorning, and the resulting failure of BGA devices becomes more likely with the related effects on product reliability, throughput and end-user satisfaction.

The term "popcorning" is used within the electronics assembly industry to define a failure mode associated with moisture entrapment in surface mount components. Popcorning occurs when a relatively small amount of moisture trapped within the component is converted into a large volume of gas or steam during the reflow process, causing the package to expand like a kernel of cooked popcorn.

This sudden expansion can be fatal to the component; inside, the die and the wire bonds together with the package itself, will be distorted by the popcorning. The resulting failure modes for popcorned devices can include broken die and broken wires either at the die interface or within the wire length, causing a total failure of the device.

However, popcorning can also cause intermittent failures when the product is tested. Such intermittent faults are the most difficult to diagnose; by their very nature they are inconsistent in their effect. The likely failure mode for the intermittent faults within a popcorned device occurs when the wire bonds within the package separate from the die interface because of package expansion. Whether the failure is total or intermittent, the component will ultimately have to be replaced.

The most typical cause of popcorning is the hygroscopic sensitivity of the molding compound used to protect the die. Device manufacturers have been aware of package moisture sensitivity levels (MSL) for many years and have procedures in place to treat components prior to their use in the reflow process. However, with the move to lead-free solders that require higher reflow temperatures, it makes packages even more susceptible to the threat of popcorning. This is due to the peak temperature that a package will be exposed to for lead-free alloys is approximately 230 to 250°C compared with 215 to 230°C commonly used for lead-tin eutectic solders. As a result, a typical MSL will increase between one and three levels for the same devices when used in a lead-free process.

Detecting BGA Defects
Although the effect of popcorning is destruction of the device, identifying the failure is not necessarily easy to do. The failure locations are internal to the package and as such are typically hidden. X-ray inspection allows non-destructive internal investigation of the device and its solder joints. The presence of broken or destroyed wire bonds within the package is usually very clear with the high resolution and high magnification commonly available with today's x-ray systems.
Typical x-ray image of completely reflowed solder balls under BGA.

However, cracks in the die are very difficult to see with any x-ray system since silicon is transparent to x-rays and the density difference in a crack will have to be seen against the various densities of the package and board. Popcorning does not occur just during initial printed circuit board assembly. Any device will be at risk of popcorning if it has not been handled correctly whenever it goes through a reflow cycle. Thus there is as much risk of popcorning occurring during the rework process as during the initial manufacture, even if the cause of the rework was not caused by popcorning in the beginning. This is why it is recommended best practice to have x-ray inspection following rework. This is especially true for BGAs so that reflow quality can be confirmed and popcorning or other potential problems can be detected before the product is delivered to the customer.

Bridging Under BGAs
The popcorning of BGAs is often indicated by the presence of bridging under the device. This is caused when the expansion of the package occurs during reflow causes it to dish, where the bottom of the package deforms and presses down onto the solder balls. Because the solder is liquid at the time, it allows the solder from adjacent balls to combine and produce bridges. While bridging between solder balls is a very common indicator of a popcorned BGA, bridging does not always occur. In such cases, there will still be evidence of the package deformation because the ball diameters across the BGA will not be consistent. Instead, solder balls at the center of the package where they have been pressed down would be expected to be larger in diameter, with the other solder balls reduced in size as the edges of the package have been lifted. This analysis can be confirmed through an investigation of the solder balls at oblique angle views with an x-ray system. Where reflow is complete there is a consistency of the solder joints with a well defined shape.

By contrast, the center solder balls when seen at an oblique angle have a totally different appearance. Although there is no bridging present, the difference in solder ball diameter between the inner and outer solder balls becomes very clear. This variation in solder ball diameter is sufficient to flag the BGA for investigation, or more likely, potential replacement. Thus measuring and comparing the BGA solder ball diameters offers the opportunity to identify popcorning even if the most obvious signs of bridging are not present in initial x-ray images.

Developments in digital x-ray technology allow a relatively inexperienced operator to quickly assess and quantify the analysis within a production environment. With lesser x-ray inspection equipment that typically lacks good magnification, resolution and contrast sensitivity, the clarity of the analysis may be more difficult to achieve. In addition, newer x-ray developments enable other aspects of the quality of the lead-free process to be further investigated.

Measuring BGA Joints
During a recent workshop, lead-free test boards were produced which contained a variety of components including BGAs. Boards with four different surface finishes were produced using the same lead-free solder paste. The finishes used included immersion gold, immersion silver, immersion tin and organic solderability preservative (OSP) over bare copper. Vapor phase reflow as well as convection reflow was used to reflow the boards. Once the boards had been produced, they were analyzed with a variety of inspection techniques including x-ray inspection. The x-ray inspection system used had an open x-ray tube with sub-micron resolution that provided 16-bit grayscale sensitivity with an x-ray image size of 1.3megapixels and acquired at 25 frames per second. The system was able to provide oblique angles of up to 70° at any point 360° around any position on the board without compromising the available magnification achieved by tilting the x-ray detector instead of tilting the board. Software was available on this x-ray system to provide automatic BGA measurements of solder ball diameter and solder ball void percentage.

To evaluate the tendency of solder ball diameter variation in post-reflow popcorned BGAs, some BGAs were deliberately exposed to moisture for a considerable period of time to exacerbate the likelihood of popcorning. This approach worked very well and, irrespective of the board finish and reflow method used, popcorning did occur. During x-ray inspection the diameter of each solder ball within the BGA was measured, together with total void percentage within each solder ball and the solder ball area. These measurements were taken using the standard supplied software functions of the x-ray system.

Dimensional Relationship
The technique of exposing some of the BGAs to moisture prior to reflow resulted in all of the BGAs exhibiting some bridging under the central part of the BGA, irrespective of board finish and reflow method. These bridges could be clearly seen in the x-ray images and this observation would be sufficient on its own to demand rework and replacement of the affected BGA. Apart from the obvious bridging, variation of the solder ball diameters between the inner and outer balls was not that obvious. To check the ball diameters, the x-ray system software looked at the BGA under higher magnification so that the BGA was split into a number of smaller areas. This was a trade-off between magnification and the speed of throughput. Using higher magnification for the analysis area within the BGA provided fewer solder ball images, but each solder ball had a relatively high number of pixels, thereby enhancing measurement precision.
Typical X-ray image of popcorned solder balls seen under center of BGA.

The results of the diameter measurements for each solder ball in each board were analyzed with the identification of each solder ball noted schematically. The solder ball designation could be defined in the software to coincide with on-board labeling. Diameter measurement limits can be set easily within the software, such that any solder ball that would measure outside of a user defined range can be indicated within the schematic diagram as a failure and therefore fail of the entire device. This feature was not used in this study. Instead, the measured ball diameters were exported into spreadsheets for data manipulation and evaluation.

As all the boards provided similar results, the data from a selected few were analyzed to illustrate the results. The data from each of the four board finishes was averaged to a nominal reference diameter of 840 microns so that any diameter variation can be seen clearly and the potential for underlying natural solder ball variation between devices is removed. The result clearly showed a marked difference in the measured solder ball diameters between the center solder balls and those on the outside of the BGA. This can be shown more clearly when the center solder ball measurements have been placed alongside, and on the same scale, as solder balls along two outside rows. Looking at the data more closely, the average ball diameter for all of the measurements under the central area was approximately 925µ. By contrast, the average ball diameter size in rows C and T of the BGA device was approximately 835µ. In other words, there is an approximate increase of around 11 percent in the central solder ball diameter compared to the outer balls. Such a diameter difference in percentage terms is large but this difference would not necessarily be obvious to an operator when viewed at low magnification. The operator would immediately see the bridge, however if the bridge was not present, then a clearly faulty board could escape if BGA measurements had not been taken.

It is recommended that a variation in solder ball diameter greater than 7 percent compared to the average would strongly indicate that a potential problem may exist. This method will be more reliable than looking at delamination or die cracks within an x-ray image because of the lack of available density variation in the board material and device packaging. It is recommended that automatic BGA solder ball measurements be taken periodically during board assembly so that a database of measurements can be built up for particular products. In this way, any subsequent change or variation in the trend of solder ball diameters can be easily identified and quickly highlight potential process failures.

Contact: Dage Precision Industries, Inc., 48065 Fremont Blvd., Fremont, CA 94538-6541 510-683-3930 Web:

search login