Hi Greg - actually the answer is Yes and Yes. Werner and I talked about
QFNs in depth many moons ago. Many folks assume that a QFN will act like a
CLCC, primarily due to the low standoff height. However, a QFN has two
huge advantages over the old CLCC; (1) a QFN has a plastic body rather
than a ceramic body so the CTE mismatch is orders of magnitude lower; (2)
the "center pad" or "ground pad" or "slug pad" - you pick the descriptor -
is a dominate solder surface for the component. Werner's equations (which
I still use for many calculations) doesn't take that solder attachment
into account. The NASA DoD project test results showed that the QFN was
one of the more reliable component technologies on their test board. The
SMTAI Conference proceedings has some very good QFN investigation papers.
I would rather process a BGA than a QFN but that is due to producibility
reasons, not solder joint integrity reasons.
Dave
From: Gregory Munie <[log in to unmask]>
To: <[log in to unmask]>
Date: 02/25/2013 08:15 AM
Subject: Re: [TN] 48-pin QFN via-in-pad solder slug problem.
Sent by: TechNet <[log in to unmask]>
OK. Now I know the answer about QFN reliability: It's Yes. And No.
Starting off Monday more confused than finish of last Friday . . .
:-)
Greg Munie PhD
IPC Technical Conference Director
630-209-1683
[log in to unmask]
http://www.ipcapexexpo.org/
http://www.ipc.org
-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of Wayne Thayer
Sent: Monday, February 25, 2013 7:46 AM
To: [log in to unmask]
Subject: Re: [TN] 48-pin QFN via-in-pad solder slug problem.
QFNs are here to stay, and like Chinese semiconductors, are now or soon
will be used in all of our military systems. Years from now, while
picking through the rubble of our civilization, anthropologists will
speculate on what it was which made us all so stupid. Not Pb in the
water, like the Romans. Maybe proof will arise that television really
does suck out your brain!
Anyway, I had a good back & forth with Werner on QFNs (before he passed,
of course!). They are very interesting parts, and as noted below and
elsewhere, they don't fall into standard analysis techniques because the
solder joint is too thin. There's no real solder in the joint--all
intermetallics. Intermetallics are very strong and can be brittle. I
don't think they stress relieve/creep like solder, which can be both good
and bad for reliability.
On the other hand, the joint height on chip capacitors is also negligible.
We tell ourselves that the strength comes from the fillet.
Werner claimed his formulae were only useful for solder joints greater
than about 30 microns (my recollection--I have to dig that up if it is
really of interest).
Regarding the actual subject at hand:
From the provided XRAY, I agree with Steve that the part has been
reworked, or the solder deposition has poor control. The little blobs of
solder not wet to larger blobs (for example, 2nd pad down on right) or
evenly distributed around the wetted areas indicates a poor reflow profile
or exhausted/inadequate flux. I presume that hazy blob on the lower right
of the slug is solder that went through the via onto the pad for the via
on the other side (IF you insist on using unfilled vias under a QFN slug,
then at least encroach the mask on the opposite side via pad!). The voids
on the slug are irrelevant (not enough to make any significant thermal or
electrical difference).
The uniformity of the solder deposition as seen both in the XRAY and in
the pried off version suggest that the solder was in full contact with the
device pad.
Hence, either the reflow profile/flux system failed, or the slug on the
component just won't wet. You have not responded to my earlier suggestion
that you attempt to wet the part you pried off by placing it "slug up" on
a hot plate and trying to get paste to flow onto it.
Wayne Thayer
-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of Stadem, Richard D.
Sent: Monday, February 25, 2013 8:08 AM
To: [log in to unmask]
Subject: Re: [TN] 48-pin QFN via-in-pad solder slug problem.
Greg is correct: QFNs are not recommended for high-reliability
applications.
With no compliant leads, and a fairly large component body, the delta CTE
between the component and the PWB is usually too large to withstand very
many service cycles before failure due to fatigue-induced cracking of the
solder joints. Smaller ones are often used in commercial designs, etc, but
they are too problematic for high-rel applications. There are also issues
with large solder voids, solder balls under the component, flux
entrapment, flexural stress cracking, etc.
We do not use them in our designs, except for in cases where we perform a
patented compliant lead attachment process (of which we do a LOT) so those
are technically not really QFNs.
Werner did write a few papers on the issues with these components, and
because the solder height "h" is a prominent part of the fatigue formulas,
QFNs simply do not do well. For some low-power devices they may perform
adequately, but I would never use them if some other package was available
no matter what the product is. Remember, even if high-reliability is not
required, do you want to deal with the potential fallout of these parts
from a high-volume failure standpoint?
I understand many have developed various processes for using them without
any issues, but they nearly always require some special attention,
including 100% X-ray process control.
Designers need to think about the cost of that when they calculate the
cost savings of the QFN versus other compliant-leaded or BGA package
styles.
They are seldom cheaper.
-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of Gregory Munie
Sent: Saturday, February 23, 2013 1:40 PM
To: [log in to unmask]
Subject: Re: [TN] 48-pin QFN via-in-pad solder slug problem.
I have been following this QFN discussion as about a year ago I worked a
reliability problem for a QFN user.
One thing I noted is that the standoff is (of course) low. For anybody who
is using QFNs: Have you done a Werner Engelmaier style mechanical
reliability analysis on the parts?
Just asking. I saw them being used for avionics and was a little concerned
about whether any planes I was to be flying on used QFNs.
Greg Munie PhD
IPC Technical Conference Director
630-209-1683
[log in to unmask]
http://www.ipcapexexpo.org/
http://www.ipc.org
-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of Dave Schaefer
Sent: Saturday, February 23, 2013 11:33 AM
To: [log in to unmask]
Subject: Re: [TN] 48-pin QFN via-in-pad solder slug problem.
I am currently working on a design with 2 similar QFNs requiring via in
slug for thermal and electrical reasons.
IPC-4761 gives a good summary of methods for handling these components;
IPC-4761 Type VII (filled and conductively capped) appears to be the only
sure bet solution.
______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please contact helpdesk at x2960 or [log in to unmask]
______________________________________________________________________
______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please contact helpdesk at x2960 or [log in to unmask]
______________________________________________________________________
______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please contact helpdesk at x2960 or [log in to unmask]
______________________________________________________________________
______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please contact helpdesk at x2960 or [log in to unmask]
______________________________________________________________________
______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please contact helpdesk at x2960 or [log in to unmask]
______________________________________________________________________
|