Mark Lettang asked:


>We had a PWA fail in the field after approximately 6-7 months.  The unit
>was powered on approximately 1000-2000 hours.  Analysis of the PWA
>revealed a short had developed between two parallel traces underneath an
>IC.  The traces are .006" wide and .010" apart.  The short was not visible
>(even under a microscope).  The short was approximately 2 1/2 ohms in
>value.  Once a razor was run between the traces the short was eliminated
>and the PWA was functional once again.

>Someone I spoke to at a PWB Fabricator suggested it might have been
>dendritic growth.
>
>Questions:
>
>What is dendritic growth?

dendrites are threads of metal which can form and grow on various metals
like silver, copper, tin, zinc etc.; especially plated coatings with high
stress and which have chromate conversion coatings. It is essentially ionic
removal and redeposition of a the metal under conditions favorable to the
basic electrochemical process.

>Does causes it on a PWA?

I think you are seeing what I have heard called metal migration. It is
caused by DC potentials, humidity, temperature, and time. If you have ionic
contamination, the process can accelerate.

>Where can I get more information about it?

Read "The Contamination of Printed Wiring Boards and Assemblies" by Carl J.
Tautscher, Omega Scientific Services, Bothell, Washington 1976. (What a
great book!) Check out _Metal Finishing_ magazine, they publish an annual
handbook and probably have relevant articles in back issues.

>Can it be attributed to the PWB fabrication?

If there are ammonium ions present on the board, perhaps. If you have
slivers or something else, more likely. I would not suggest this is the
primary cause.

>Any and all assistance is GREATLY APPRECIATED!

Silver migrates at a rate 1,000 times greater than copper or tin according
to Tautscher. I would suggest you look at the finish of the device leads.
And this is one reason I have never liked the idea of using SN62 or other
silver bearing solders. Back in the late 70's at Mostek, we used Ag plated
RAM DIP leads (at the IC packaging level) and had some complaints from
customers who could show us migration across the package and their boards
causing shorts. There are those who say that the alloying with tin greatly
reduces the tendency to migrate. Tehy may be right. Lots of people use
SN62, and Ag is commonly used to metallize ceramic packages, usually with a
barrier coating of Ni.

Pure tin can migrate dependent on the residual stresses of the plated
coating and the environmental exposure.

If you still have that PWA, electron microscopy could probably tell you
what element you had. They would have to destroy the board to put the
sample in the chamber, though.

good luck,

Jerry Cupples
Interphase Corporation
Dallas, TX
214-919-9150