First, the choice of the word ‘embrittlement’ is unfortunate, because it creates a mental image that is misleading. The pictures below of the crystalline AuSn-IMC and AgSn-IMC platelets and their impact on the solder joint structure illustrate the phenomenon much better.
“Au/Ag-embrittlement” is the weakening of the SJ structure by too many weak interfaces with the crystalline AuSn-IMC platelets and the solder. Larger number of such IMC platelets result in higher density of such interfaces and an increased SJ weakness. There is an interfacial loading that comes from a CTE-difference between these IMCs and the surrounding solder volume. There is little practical difference in the effect of either AuSn-IMCs, AgSn-IMCs or even a combination of them.
The ‘famous’ 3 to 4 wt.% rule-of-thumb is just that. It applies to the total solder volume and requires that the IMCs are uniformly distributed throughout the solder volume—this however, is rarely the case and problems with ‘embrittlement’ have occurred with less than 3 wt.% of Au.
It is important to put the issues into the correct perspective. Au- and Ag-embrittlements occur with the thick and somewhat uncontrolled deposits that can result from electroplating of Au and Ag; they do not occur with immersion plating of either Au for ENIG or imAg as a surface treatment.
‘Au-embrittlement’ is of course much more well know because of its long history. ‘Ag-embrittlement’ never reached the same level of notoriety as ‘Au-embrittlement.’ The whole 5-year effort of the IEEE Compliant Lead Task Force (CLTF) was started by ‘Ag-embrittlement’ when TI started production in Singapore and their J-leads were Ag-electroplated. It all hit the fan when the first customer in 1982, IBM in Austin, had PLCCs dropping off the PWBs on bed-of-nail testing.
The dull/grainy/lumpy appearance of solder joint surfaces stems from these IMC-platelets pocking through the SJ surface.