The Technical Services Group at Alpha Metals has spent a fair amount of time looking at wave soldered Pb-free connections. We too have observed fillet tearing, fillet lifting, and pad lifting. I have concluded that without major changes in the materials used to construct PCB's the high-Sn Pb-free alloys are not viable alternatives to eutectic Sn-Pb for plated through-hole soldering. (Please note that this is my personal opinion and that it may not reflect that of Alpha Metals, Inc.)
In the past three years, we (Alpha Metals' Technical Services Group) have examined PCB's wave soldered with various Sn-Cu, Sn-Ag-Cu, and Sn-Ag-Cu-Sb alloys. Fillet lifting was observed with all of these alloys. Fillet tearing and pad lifting were observed less frequently. I have concluded that the cause of all three of these phenomena is the same. I offer the following as an explanation.
Large stresses develop in the through-hole solder connections as the assemblies cool. These stresses develop because the Z-axis contraction of the PCB (the plated through-hole) is much greater than the contraction of the solder (filling the plated through-hole). If sufficient stress exists when a connection is partially liquid the solid portions of the connection are forced apart from each other and fissures or tears are formed (in interdendritic regions or at colony boundaries - the last portions of the connection to solidify). In the casting industry these are called "hot tears". For a connection that becomes completely solid, stresses continue to build and become quite high (particularly at the rims of the through-holes) upon further cooling. In some cases, these stresses are great enough to tear Cu pads from the laminate or the solder from the pad. While these are referred to as pad lifting and fillet lifting, respectively, no upward movement actually occurs. Instead, the PCB laminate "shrinks" away.
In the case of pad lifting, the adhesion of the Cu to the laminate is the weak link in the system. It is not sufficient to bear the load.
In the case of fillet lifting, we have observed that the failure occurs at the bulk solder/IMC interface. For reasons unknown, this interface is the weak link in the system. It may be that "junk" in the plating and solder accumulate at this interface as the IMC forms. This "junk" could be organics from OSP's and platings, trace elements in the plating/pads or solder, intentional alloy additions to the plating/pad or solder, fluxing by-products,... It is unknown (at least to me) whether the lifting occurs while the assembly is still relatively warm or when it is near room temperature. Cooling rate could be a controlling factor.
Please note that while I call the adhesion of the Cu to the laminate or the solder/IMC interface the weak link, they may actually be quite strong. The forces exerted on them, however, are quite great (much greater than those exerted on 63Sn/37Pb plated through-hole connections).
We have observed no fillet lifting, pad lifting, or fillet tearing of surface mount component connections (formed via wave soldering) or non-plated through-hole connections formed with the high-Sn Pb-free alloys. (I not aware of anyone reporting such defects.) Like rafts on the sea these connections simply rise and fall with the Z-axis expansion and contraction of PCB's, so one would not expect a great Z-axis force exerted on them.
After observing and characterizing the defects, our analysis focused on characterizing the thermal expansion/contraction of the PCB's and the high-Sn Pb-free alloys. A fair amount of thermal mechanical analysis (TMA) was performed. To summarize the following was found. The typical FR4 PCB has a glass transition temperature of about 130 C with a coefficient of thermal expansion/contraction of about 60 um/mC before the transition and about 250 um/mC after the transition. (all Z-axis measurements) The solder alloys, including 63Sn/37Pb, have about the same coefficient of thermal expansion/contraction, about 25 um/mC. All of the alloys solidify at temperatures above the glass transition temperature of the FR4. Given that the high-Sn Pb-free alloys solidify at temperatures 30 C or more greater than 63Sn/37Pb they must accommodate 40% or more Z-axis strain upon cooling to room temperature than the Sn/Pb eutectic. This additional strain produces the higher stresses in the Pb-free connections.
Will the transition to Pb-free soldering mark the beginning of the end of plated through-hole technology, or the end of FR4 technology? Something has to give!
One final item, I have not observed fillet lifting with 96.5Sn/3.5Ag. Have you observed it? Are you aware of anyone who has? (Of course, if it doesn't so much for my theory.)
Please feel free to pass this on to anyone you think may be interested.
Denis O'Connell Northern Kentucky University Highland Heights, KY Formerly Sr. Applications Engineer Technical Services Laboratory Alpha Metals, Inc Jersey City, NJ