Avoiding the Solder Void

Solder voiding is present in the majority solder joints and is generally accepted when the voids are small and the total void content is minimal. X-ray methods are the predominate method for solder void analysis but this method can be quite subjective for non grid array components due to the two dimensional aspects of X-ray images and software limitations. A novel method of making a copper “sandwich” to simulate under lead and under component environs during reflow has been developed and is discussed in detail. This method has enabled quantitative solder paste void analysis for lead free and specialty paste development and process refinement. Profile and paste storage effects on voiding are discussed. Additionally an optimal design and material selection from a solder void standpoint for a heat spreader on a BCC (Bumpered Chip Carrier) has been developed and is discussed. Introduction Solder voids in solder joints are a common occurrence in SMT assemblies. Their origins are not well understood but are typically faulted as a failure of the solder fillet to thoroughly expel flux remnants during the reflow process. The amount of solder voiding can vary significantly within an assembly, between different flux formulations, solder alloys, board and component metalizations. Reflow profiles as well as stencil aperture designs can often affect the overall level of voiding. Adding to the mystery of solder voiding is a lack of quantitative measurement tools in the industry with few exceptions. BGA void analysis software is one of these exceptions. This software uses gray level pixel analysis to determine the perimeter of the solder sphere and the internal perimeters of the voids. Once the perimeters are established the areas within these structures can be measured and an overall percent voiding can be calculated. This type of measurement works well if the voids are large or found on the outer edges of the sphere but if the void is small and centrally located where the sphere density is the greatest then the void may be invisible due to its relatively similar gray level to the surrounding material. Increasing the X-ray power will reveal the small void but also shrink the measured area of the sphere and yield an inaccurate and inflated percent voiding. This problem is even more complicated in a chip or a leaded component solder joint. When Xraying a completed assembly, internal traces, vias and even components on the backside of the board that intersect the image of the solder joint confound the software algorithms ability to accurately determine the perimeter of the solder joint. In simple terms the X-ray image is two-dimensional and the ideal structure must be symmetrical about the Z-axis such as a box or a cylinder. Novel Approach Based on the assumption that the ideal quantitative void measurement method will utilize BGA analysis software and a symmetrical Z-axis reflow structure, the “sandwich” concept was developed. This a novel approach simulates the worst conditions of a solder joint for voiding, under the component where flux evacuation is the most difficult while maintaining the same reflow thermal environment and metallurgies if desired. This idea was born out of a quest for a quantitative method of determining the percent voiding on a Ceramic Column Grid Array (CCGA). In the CCGA the columns are 10/90 Sn/Pb and cover about 45% of the solder fillet. If enough power is used to see through these dense columns, the perimeter of the solder joint is washed out and 55% of the total fillet is invisible. If adequate power is used to see the perimeter of the circular fillet, the area under the columns is invisible. The effort is complicated by column parallax, internal traces and vias as can be seen in Figure 1. With the thought of a column the same diameter as the solder pad that is thin enough to be X-rayed without excessive power, a solder preform was selected. In this application the preform alloy was selected to be the same as the 10/90 columns to minimize the variables that could contribute to solder voiding. Several thicknesses were tested with a 30 mil diameter by 5 mil thick as the final solution. Figure 1 CCGA X-ray Presented at IPC SMEMA Council APEX 2003 www.GoAPEX.org