Very good point. Even if there was criteria, how do we inspect for that?

-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of David Hillman
Sent: Thursday, September 06, 2018 10:50 AM
To: [log in to unmask]
Subject: Re: [TN] R: [TN] Ni intermetallic thickness target

Hi Guy - I guess I am confused as how the IMC could be unacceptable if
there are not industry defined standards and there is no linkage to a
failure mode. Seems very very subjective.

Dave

On Thu, Sep 6, 2018 at 9:15 AM, Guy Ramsey <[log in to unmask]> wrote:

> Not linked to a root cause, but observed as unacceptable,  too thin and
> inconsistent.
> soak time (150 to 200) = 150 sec
> reflow time (above 217) = 75 sec
> peak temp = 250
> Seven thermal couples placed at various places on and in the board all
> within 3 degrees delta.  The board is heavy, Megtron 6 with 34 copper
> layers. Most of the thermal load is the board. There would be very little
> difference from pad to pad on any given component.  Solder mask defined
> pads conform to MFG recommendations.  We learned that the DDR packages are
> several years old. Maybe the solder spheres are hosed?
>
> On Thu, Sep 6, 2018 at 9:38 AM Wayne Showers <[log in to unmask]
> >
> wrote:
>
> > IMC has been the bane of my existence more than once.  So here is my take
> > from experience backed up with some lab evidence.
> > I have had failures attributable to insufficient IMC.  Almost all of
> these
> > were corrected by either ramp-to-peak or increased time above liquidus /
> > intimate contact time / temperature.
> > For SAC, the intimate contact temperature I have been able to dial in on
> > is time above 208C after liquidus is achieved.
> > For SnPb, the intimate contact temperature I have been able to dial in on
> > is time above 177C after liquidus is achieved.
> > I have found that if I can back end a little more time above these
> > temperatures on the profile, I have more consistent joints.  I have not
> > devoted a whole lot of 'science' on these numbers just 20+ years of trial
> > and error coupled with white papers, IPC BOK, and other research.
> >
> > I have yet to have a problem or a return tied to excessive IMC.  I am
> sure
> > that at some point excessive IMC may cause or have caused a failure
> mode, I
> > just have not seen it.  The few times where it may have been a
> contributing
> > cause, I also cooked the parts making cross-sectioning of the IMC moot.
> >
>