At Lucent, we have also struggled with the optimal choice of a surface
finish. I will try to give you my thinking as to each coating.

Electroless Nickel Immersion Gold (ENIG) can be an excellent choice, but I
feel its success is very dependent on the board vendor. Controlling the ENIG
process is the most difficult of the surface choices about which you
inquired. If the nickel deposition is not well controlled, it can lead to
problems. If the bath is run in low volumes, I feel that control can be very
difficult. Also, the phenomena of "Black Pad" has caused us great concern.
It is a very difficult failure to detect because you can easily pass ICT
with these fractured joints, and then get into a an intermittent problem
once in the field. Also, our experience on wetting to nickel to form
nickel/tin intermetallics is only a fraction of that which we have attained
over the years for wetting to copper and forming copper/tin intermetallics.
Soldermask attack can also be an issue. Finally, during our evaluation of
ENIG we found that it could not hold up to temperature/humidity for even 24
hours. Although, this is not real production conditions, it indicated that
the coating wasn't as robust as the Immersion Silver (ImAg) tested in
parallel.

Immersion Tin (ImSn) is also felt to be suspect. With the thin coatings
expected on the board, we are concerned with shelf life. We do use ImSn for
press-fit backplanes, but do not recommend it for SMT or TH applications. We
completed assembly testing with ImSn a few years ago that indicated there
was a higher defect rate at assembly than with other finishes. However, I
want to stress that this was one formulation and that improvements may have
been made. However, there is still the issue of intermetallic formation
consuming the tin, reducing shelf life and making multiple thermal cycles
problematic. Also, the chemistry involved is more of a health hazard than
those used with OSP's or immersion silver.

We have used OSP's for almost two decades now and in general have had good
success. With proper care in handling, and usage of nitrogen blanketing to
reduce oxygen levels during thermal excursions, we have been able to
accomplish good soldering on multiple thermal cycles. The reason we no
longer prefer OSP is not because of the coating itself, but due to isolated
busts that have occurred. We have seen defects caused by very thin layers of
soldermask residue on the pads and also due to incomplete removal of tin
strip residues after etching. Also, with via in pad product we have
experienced where the inside of the small holes are not completely dried and
thereby break down the surface of the pad and reduce solderability. These
busts can't be detected until after expensive components have been attached
to the board. These are low level type situations, but when they occur they
cost a lot of money because we end up junking that product. I know you could
scrape the pad and repair the individual defect, but you can't determine if
there are other marginal areas on the board that meet visual requirements
yet may be marginal and be a reliability concern.

Immersion Silver (ImAg) is now the surface of choice for Lucent. We
underwent extensive testing during 1997 and introduced the finish into
production in 1998. A paper detailing our testing was presented at IPC
summit on Surface Finishes and PWB Solderability in September 1999. We have
since produced a large quantity of circuit packs with this finish and are
very satisfied with our results. Although there have been a couple of isolated
incidents, the problems encountered with this coating are less than any
other with which we have experience.

If you would like further information, please contact me.

[text/html]