SMT007 Magazine


Issue link:

Contents of this Issue


Page 19 of 101

20 SMT Magazine • August 2015 THE WAR ON SOLDERING DEFECTS uNDER AREA ARRAY PACkAGES continues and bump, even in the face of warpage, is criti- cal to preventing the formation of that defect. HiP-resistant pastes are also designed to have the ability to remain continuous as the paste is stretched during warpage under reflow. This is analogous to the elasticity of a rubber band when stretched. Ensuring the paste de- posit remains continuous is critical to ensuring that the two solder volumes are able to coalesce once reflow has been achieved. Unlike very familiar tests performed on solder paste materials (e.g., voiding resistance, slump resistance, and tack testing), there is no standard test for HiP and NWO resistance. Direct testing of physical properties (viscosity, tack, etc.) cannot be directly correlated to HiP and NWO performance. This leads to solder manufacturers creat- ing their own proprietary test methods to characterize and benchmark paste performance with respect to these defects. There are some pitfalls to this type of testing and it is impor- tant to discuss with your sol- der paste manufacturer how the test data is generated and analyzed to ensure the data provides a good reflection of actual performance. The major difficulty in de- veloping HiP and NWO charac- terization tests is that these defects can occur with very low frequency, and it can be difficult to ensure formation of these defects in a repeatable fashion. Simply reflowing a test vehicle with a variety of area array packages is not sufficient to demonstrate robustness to HiP and NWO. Test of this nature, which are likely to result in zero HiP or NWO defects under typi- cal conditions, fall prey to a data fallacy: If a test sample with no defects is a common outcome of a test, how can two test samples be compared against each other when they both result in zero defects? Is sample A better, worse, or the same as sample B, as both have zero defects? In addition, can either sample A or sample B be advertised as being completely resistant to the defect? In this situation, samples A and B can- not be differentiated from each other and there is no guarantee that either formula has solved the problem. One hallmark of a good characterization test is create a test that is "designed to fail" by cre- ating conditions that are at the very limits of performance expectations. In other words, de- signing a test where defects are assured to occur then changing inputs and measuring the rate at which defects occur allows for relative com- parison of each sample. These types of tests are excellent at identifying when perfor- mance has been improved or dep- recated when comparing mul- tiple test samples. Another consideration when analyzing test data is to be skeptical of claims to "eliminate HiP defects" by cit- ing test data. Characterization tests are excellent tests when used as a tool to compare per- formance across similar condi- tions. Using a characterization test to make a claim of com- plete elimination of a defect is an example where a high risk of an error of the first kind (a false positive) exists. The old adage that you cannot prove a negative holds here. A claim that a solder paste eliminates HiP and NWO defects is a strong claim, but basing that claim on testing performed under controlled conditions only proves the claim until the first example from the field where HiP or NWO is discovered. A claim that a particular solder paste is more re- sistant to or less likely to experience HiP/NWO defects than other solder pastes is a more appro- priate claim to make based on characterization testing. A third consideration when analyzing a solder paste manufacturer's HiP and NWO test data is sample size. All statistics are estimates to some degree, and those estimates gain precision with increased samples. Characterization test- ing for HiP and NWO should be performed in relatively large sample sizes in order to increase FeAture the major difficulty in developing HiP and nWo characterization tests is that these defects can occur with very low frequency, and it can be difficult to ensure formation of these defects in a repeatable fashion. " "

Articles in this issue

Archives of this issue

view archives of SMT007 Magazine - SMT-Aug2015