edfas.org
ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 18 NO. 4
62
GUEST COLUMNIST
TRAINING IN AN ERA OF INDUSTRY CONSOLIDATION
AND AUTOMATION
Christopher Henderson, Semitracks, Inc.
henderson@semitracks.comW
riters have spilled a lot of ink during the past two
years regarding the consolidation of the semi-
conductor industry. Of coursewe read the usual
comments: jobswill be lost due to redundancy, growth for
our industry is over, etc. How is this era of consolidation
going to affect those of us working in failure analysis labs
throughout the industry? Here ismy take on this concern.
While the industry has indeed consolidated, it is
not completely clear how many engineering jobs will
be lost. Certainly, there are economies of scale within
the finance, human resource, purchasing, and logistics
operations of these companies. There may also be some
economies of scale (in the long run) on themanufacturing
side as well. However, product engineering is one area in
which it is difficult to achieve an economy of scale. The
activities associated with new product introduction, test
engineering, reliability, qualification, and failure analysis
have been, and remain, difficult areas to consolidate. As
companies combine, they generally do not reduce the
number of products they sell. This should bode well for
our discipline.
The more interesting trend is automation. In theory,
automation could lead to a reduction in jobs. The automa-
tion of tasks should allow fewer analysts to accomplish
more analysis work. Is this really the case though? While
we have benefited from the development of new failure
analysis techniques that have helped productivity, there
has not been wholesale automation of failure analysis
work. The fault isolation of complex systems-on-chip can
run intoweeks for more challenging problems. Part of the
reason for this is that the chips we are analyzing are also
increasing in complexity. The integrated circuitry is more
complex, the packages are more complex, and the test
routines aremore complex. Automation in this area simply
does not keepupwith the increase in complexity. This phe-
nomenon is also true in the electronic design automation
industry. The design of chips has not decreased because
the complexity has increased as much or faster than we
can improve the design tools, algorithms, anddesignflow.
Here is another example: We have not increased the
throughput on decapsulation. The reason for this is pri-
marily due to the increase in complexity of our products
from a packaging perspective. We simply do not under-
stand all of the parameters around package geometries
and materials to do this effectively. Furthermore, we
continue to introduce new packages and materials at an
accelerating rate. Therefore, I don’t see failure analysis
succumbing to automation anytime soon, as might be
the case with, for example, Fed-Ex or UPS couriers or taxi
drivers. If the rate of change in our industry slows mark-
edly, then wemay seemore automation, but that is likely
to be some time from now.
So, if we don’t see bigworkforce changes in our indus-
try due to consolidation or automation, how do we think
about our jobs? Theone thingwe cannot be is complacent.
If we do not work to improve our skills, thenwe do run the
risk of job loss through the failure of our company due to
unsolved problems that leave our company uncompeti-
tive. Even though automation may not directly take our
jobs, we do need to embrace automation of various tasks.
Right now, this may mean learning about automatic test
patterngenerationdiagnosis, or learninghowtoautomate
a portion of the transmission electronmicroscope lift-out
process. This involves learningmore software. An increas-
ing amount of our analysis work will be done through a
computer screen, rather than throughmanipulation with
our hands. Computer skills and awillingness to learn new
computer-based skills will be essential.
For many of you, consolidationmeans that youwill be
“HOW IS THIS ERA OF CONSOLIDATION
GOING TO AFFECT THOSE OF US
WORKING IN FAILURE ANALYSIS LABS
THROUGHOUT THE INDUSTRY?”
(continued on page 64)