Used to reball parts, the process eliminates two thermal cycles and enhances component reliability.

Solder ball attachment to BGA and CSPs is seldom performed without at least some negative impact on component reliability. Most solder ball attachment processes in use are convection processes. These require heating the entire component to achieve good solder sphere wetting to the BGA pads during original ball attachment.

Many BGA or CSP rework situations occur where the original solder spheres are lost in the process of removing a non-conforming component and must be replaced to salvage the BGA. There are various methods of doing this, such as solder balls poured into stencil apertures, soluble paper preforms with embedded solder spheres, and similar methods. Again, all these use convection processes to heat the entire BGA to remove old solder and reflow new spheres in place as part of a rework process.

Since the RoHS Directive, most BGAs are no longer produced with standard SnPb37 solder spheres. Instead, a Pb-free alloy such as SAC 305 or similar alloy is used. Component vendors cannot feasibly manufacture additional smaller lots of a given BGA sans solder balls or with standard SnPb37 rather than unleaded solder balls. Having to set up their factory with different processes and a segregation system to provide smaller quantities of BGAs for both RoHS-exempt and RoHS-compliant customers is not a cost-efficient business process.

Many high-reliability companies are exempt from RoHS requirements, and for good reasons. Reliability constraints such as tin whisker formation, alloy mixing issues or the requirement to build fully qualified legacy products prevent them from using Pb-free components. They are required to use leaded (SnPb37) solder paste, and it has been shown that attempting to solder BGAs having Pb-free solder balls with a SnPb37 solder paste or profile (or vice versa) will nearly always negatively affect BGA solder joint reliability.^1,2 This is because the solder joint formed by this mixed alloy is seldom homogenous; the alloy blend of tin and lead must be mixed thoroughly to avoid regions rich in lead (Figures 1 and 2). Complete homogeneity can be achieved with mixed alloys, but requires both a higher reflow temperature and a longer dwell at that higher temperature.


This is well documented, even within many of the BGA component manufacturers’ specification sheets. Some specifically state soldering with mixed alloys or at higher temperatures than recommended voids the warranty on the component. Most warn the package will not withstand more than one or two rework cycles, especially at higher temperatures.

Other components on the assembly, and the board itself, are usually not made to handle these higher reflow times and temperatures, which may result in different reliability issues.³ To avoid these issues, a reballing process should be used to remove the unleaded solder balls and replace them with SnPb37 solder spheres, or vice versa as the case may be. For a rework situation, four additional thermal cycles are required if convection is used: one to remove the BGA from the assembly, the second to strip off the old solder balls, a third to solder the new balls on, and a fourth to reattach the BGA to the assembly. The result is inevitable, a further decrease in solder joint and BGA substrate via reliability, as well as additional cost.^4,5

Rework causes. Regardless of the RoHS-instigated alloy mixing issues, there are many situations where rework is required on a BGA component no matter which process or alloy is used: Pb-free or standard 63/37. Rework is often required because of hiccups such as misalignment during placement or reflow, excessive voiding after reflow, solder bridging, improper reflow profiles, poor solder paste stencil registration, improper solder paste handling practices, wrong components loaded into the placement machine, and many other factors. BGAs often are replaced to upgrade the performance of the assembly by using one with a higher operating speed or some other improved feature. Retaining the old components for other applications is often desired.

Laser Reballing

As stated, if a BGA needs to be unsoldered, the original solder balls are lost. In many cases these components can be very costly; some BGAs on the market today cost several thousands of dollars each.To salvage these components, new solder balls are attached with convection heat. Thus begins the vicious downward spiral of the component’s reliability. Hence, it is advantageous to perform initial ball attachment and reballing using laser energy to heat only the BGA pad and solder ball, not the component or internal die or bonds.

During qualification of the oxygen-evacuated laser ball attachment process, thermocouples were embedded within the package to compare the temperature gradients seen with convection reflow versus laser (Figures 3 to 5).


Methods have been developed using laser energy to remove old solder and perform attachment of new solder balls without heating the component itself, in an oxygen-free chemical bath without nitrogen. This process results in smooth, perfectly round, void-free solder spheres completely wetted to the BGA pads. These spheres are optimal for solder connection to the PWB pads in subsequent processes. When comparing laser-attached solder balls with balls attached using a convection soldering process, even those soldered within a nitrogen blanket, the difference is immediately noticeable (Figures 6 and 7).


Microsections were made to determine if there were any appreciable differences in initial solder attachment of BGA solder balls received from different vendors where the attachment was done using convection, on BGAs with solder stripped off as part of rework and new balls attached with convection, convection with benefit of added nitrogen, and laser. While major differences were seen in the grain structure and IMF, the microsections showed no sign of innerlayer substrate damage or resin recession cause by the localized heating of the pads and balls with laser (Figures 8 and 9). No substrate anomalies were seen in the cross-sections of either convection reballed or laser reballed samples (Figures 10 to 14).


Shear force test results. Shear testing was used as an indicator for comparison between reballing methods. Shear force testing of all three methods indicated a range from 1.2 to 2.9 Kg. The as-received BGAs failed under the lowest force range; convection and laser reballed BGAs were in the middle of the force range, and originally attached balls soldered with laser in an oxygen-free chemical bath were at the high end of the range.

Shear force testing indicated there is virtually no difference in the shear strength for spheres reballed using convection with nitrogen blanket and laser processes (Figure 15). However, initial laser attachment shear forces were higher than those of reballed spheres using both the convection process and the laser attachment with the proprietary process, and much higher than as-received BGAs with the original attachment done using convection.


Summary

Reballing BGAs due to rework and conversion from RoHS-compliant solders to SnPb (or vice versa) typically requires the component be subjected to additional reflow cycles beyond the component manufacturer’s recommendations. Laser attachment or laser reballing in an oxygen-free bath eliminates two additional temperature excursions and provides improved component reliability by facilitating the use of standard assembly reflow temperatures without mixed alloys.

References

1. Fay Hua et al, Raiyo Aspandiar, Cameron Anderson, Greg Clemons, Chee-key Chung, Mustapha Faizul, “Solder Joint Reliability Assessment of Sn-Ag-Cu BGA Components Attached with Eutectic Pb-Sn Solder,” IEEE 6th International Conference on Electronic Packaging Technology, August/September 2005.
2. David Hillman, Matt Wells and Kim Cho, “The Impact of Reflowing a Pb-free Solder Alloy Using a Tin/Lead Solder Alloy Reflow Profile On Solder Joint Integrity,” Centre for Microelectronics Assembly and Packaging International Lead-free Conference, May 2005.
3. Werner Engelmaier, “Printed Circuit Board Reliability: Needed PCB Design Changes for Lead-Free Soldering,” Global SMT and Packaging, September 2005.
4. Bala Nandagopal, Zequn Mei, Sue Teng and Mason Hu, “A Novel Approach to Evaluate the Impact on Solder Joint Reliability due to Multiple BGA Rework Cycles."
5. Werner Engelmaier, “Printed Circuit Board Reliability: Loss of Life During Soldering,” Global SMT and Packaging, October 2006.

Joshua Muonio is a senior process engineer at Analog Technologies/Lumagine Inc. (analog-tech.com); jmuonio@analog-tech.com. Richard D. Stadem is a principal process engineer at General Dynamics (gd.com); richard.stadem@gd-ais.com.

• Prev
• Next