You Can't Always Get What You Want....

Going back a few years (1969) I can recall a tune by the
Stones whose chorus went :

“You can’t always get what you want

 You can’t always get what you want
 You can’t always get what you want
 But if you try sometimes you just might find
You just might find
You get what you need………………. “

While
reading through the annals of rock-and-roll history I
have uncovered that this song evolved after the Stones
US tour in ’64. They had done a show in Excelsior MN
which was not very well attended or received. After the show Jagger
went to a local drug store soda fountain and ordered a cherry coke
and was told that they were out of cherry syrup. “Mr. Jimmy”, a
local Forrest Gump like character, who was sitting
beside Jagger leaned over and shared the now infamous line
 ”Well, you can’t always get what you want “ and thus
legend was born !

 Now you
might be wondering “…exactly what does that have to do
with semiconductor microelectronics in general or more
specifically 3D Integration”….Well, that’s the theme
that came to mind as I flew back to the East Coast after attending
the RTI 3D Integration Conference last week, and I’ll
explain why shortly.

Sr. Semi
International editor Alex Braun was in attendance last week in
Burlingame and was filing very accurate reports on what was going
on. I know you all were paying attention because they all ended up
in the top 5 stories of the week. Having said that, I still
feel compelled to give you my two cents on what I saw and heard
because its always good to have multiple perspectives on emerging
technologies that are this large and all encompassing….and make
no mistake 3D IC will soon be the new elephant on the block –
the only question we all have is  when ?

When I look back
on the conference a few major themes stand out all
of which are linked to the line of Mr Jimmy “You
can’t always get what you want”. Most of us came to the
conference looking for signs that 3D IC was taking off –
perhaps a major commercial announcement from Intel
or Micron or IBM ? Perhaps
an indication that the design community was finally issuing
software so that 3D could be laid out in a standardized fashion ?
Perhaps an indication that the industry was narrowing the process
technology choices to a manageable number ? Certainly we all wanted
to hear that despite what now looks like it will be a deep
recession, 3D IC would not be affected, but rather would move
forward, bucking current economic trends.

While there were
no blockbuster announcements last week (what we
want) we did get assurances that we are steadily, if not
rapidly, moving forward and that we are not wasting our time or
money chasing this technology (what we need).
Although we are living in the instant gratification generation, as
I have outlined before [see PFTLE “Time Isn’t
on Your Side” Aug 12, 2007] new technology usually takes a
decade to be commercially accepted. For those of us that are
impatient, that is not a pleasant concept, but it’s
true.

Alex did a
wonderful job covering the course I gave on Monday at the Symposium
in his piece “3-D Integration Lacking in Design and Test
Support”,so I won’t retread that ground. The only
point I want to leave you with is that we are not trying to
commercialize 3D integration because it is the next evolution of
wire bonded 3D stacks. We are doing this because it leads us
to repartitioning of chips so we can stack functions being built by
optimized processes (the opposite of SoC -  system on chip
solutions) . We are doing this so we can stack disparate
technologies to form completely new devices. We are doing this so
that we can build smaller higher performing devices without
continuing to shrink circuit size. We are doing this because
ultimately it will lead to lower cost. 

Standardization

Several of the
speakers including plenary lecturers Gurtej Sandhu
of Micron and Mike Shapiro of
IBM pleaded for more standardization. The IBM
slide shown below clearly outlines the issues. As we have mentioned
before, new technologies always bring out creativity of
technologists and therefore, early on in the cycle, we have many
solutions to choose from, usually way too many. That is the case
for 3D IC where the process choices still seem
overwhelming. History tells us that full commercialization
will not occur until these myriad of choices are narrowed down by
the industry. In fact, this narrowing down process will tell us
that we are getting close. 

It was / is hoped that consortia like EMC-3D,
Sematech and IMEC will help to narrow
down the choices, but it is clear that this has not happened
yet. 

Copper Vias

Another example of “you can’t always get what you
want” presented itself in the reliability data shown by Mike
Shapiro. Mike indicated that reliability of TSV was related to the
composition, size and shape of the TSV. For IBM’s annular via
structures W is significantly more reliable than Cu as shown in the
figure below. The have favored annular vias because “.. the
Si core results in improved mechanical reliability”.
 IBM has been showing this data since 2006 (see: “A
CMOS-compatible Process for Fabricating Electrical Through Vias in
Silicon” P. Andry et. al., ECTC 2006, p. 831) . Mike was not
able to divulge any further details as to the mechanism of what was
going on with the copper via failures. 

Later, private discussions with Tezzaron CTO Bob Patti confirmed
that Tezzaron had also seen Cu via failures thus prompting their
move to vias first ( after FEOL) W vias, which they announced
earlier in the year [ see PFTLE “More 3D IC
Integration from Ft McDowell “ March 30, 2008 ]. Bob was
willing to get more specific and pointed to < 10 um
diameter vias with AR > 6 as being in a dangerous zone when it
comes to
reliability.   

Although most of us are probably “pulling” for Cu vias
to be successful (what we want) this information indicating that we
should be very careful with via composition, size and shape is
really what we need !

Note - we now have another reason to keep the AR low in such
small vias !!

 ….more updates from the RTI conference are
coming later this week.

For all the latest in
3D IC integration stay linked to Perspectives From the Leading
Edge
(PFTLE)……………………………….