May_EDFA_Digital

43 / 58

edfas.org
43
ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 19 NO. 2 be localized in 3-D space, which is useful for stacked dice
analysis. Preliminary results were obtained by locally
heating the substrate or the exposed copper paddle. This
method overcomes the pitfalls of using thermal cameras
tomeasure the phase of heat at the surface. The assump-
tion is that enough power could be poured into a localized
defect to reach the surface, but it quickly devolves into
providing hundreds of milliwatts of required power with
lengthy time constants and risks overstressing the defect.
However, withTOFSIFT the device temperature is changed
by externalmeans, and the resulting time variant response
is used to trilaterate temperature-sensitive localized func-
tional or parametric failures. Bothmethods are affectedby
variations in thermal conductivity, especially air gaps, and
generally assume a localized failure. In the event where
it is not practical to do TOFSIFT, a thermal gradient can
be forced from one end of a device or wafer to the other.
The point where the gradient temperature matches the
defect is the location in theX-plane of thedefect. The same
method is applied to the Y-plane to localize thedefect. This
is especially useful with temperature-sensitive failures on
higher-power devices such as computer processing units.
Temperature-forcing the opposing ends of a large metal
block as a heatsink provides the required cooling with a
controllable gradient tomap sensitivities. In recent years,
stacked dice can include as many as 15 dice, with each
die on the order of 30 µm thick. This is quite conducive to
forcing thermal gradients because the silicon substrate
is minimally present. A commercial product has not yet
been produced for TOFSIFT, but I am interested in a col-
laborative development of such a product.
PATENT: US6245586
We know that not all packages are createdwith failure
analysis in mind. Ball grid array products or stacked dice
products with wire bonding embedded in epoxy cannot
be analyzed through the backside if the die is facing up.
Pocketmilling severs balls and traces in the packagewhen
trying to access the substrate. While it is possible to probe
the exposed bonds, it is not realistic for more than a few
connections. The solution is to polish away the package
and backside polish the die, disregarding the electrical
connections. Once the desired polish is achieved, the
severed and polished wires become the new bond pads.
Thediecannowbewirebondedtoanewpackageor carrier
for test.
PATENTS: US9465049, US9157935,
US9034667
Sample preparation involving ultrathin silicon is of
current interest. Being able to target to 1 µm of remain-
ing silicon pushes the need for thinning tools capable
of system temperature control and electrical endpoint
detection during preparation. Modeling the top surface
and hoping the part will not deform further is unrealistic.
Polynomial curve models are used to define the final
surface a priori. This work won the Best Paper award at
ISTFA 2012 for the development of electrical endpoint
techniques on silicon and at the package/board trace level
for circuit surgery/access.
I’ve found that I love the inventing side much more
than the administrative side of business. I enjoyed devel-
opingportable emissionmicroscopy andvibration coupler
applications as well as SIFT fault isolation methods,
wire-to-wire bonding, and, more recently, integrated
parametric and thermal capabilities into sample ultrathin-
ning, to name a few of my 15 issued patents. However,
starting a manufacturing business is quite different from
consulting for other companies. Valuable advice given to
me by a chief executive officer friend of mine was: Be in
charge of your operation or walk away. I have taken this
lesson to heart. I’ll probably write a nail-biting book on
this topic after I retire. Being an inventor means wanting
to invent and avoid the politics. However, business rela-
tionships will always be part of the package. With all the
consolidation of companies, the playing field is chang-
ing, but from the failure analysis perspective, there will
always be problems to solve and, as such, the need for
technological advancements. Be sure your management
sees you as an indispensable resource to your company
because of your problem-solving skills. Being successful
in failure analysis requires you to be inventive with the
tools on hand (and sometimes the political situation) to
solve complex problems. The impossible answers usually
come in the next fewdays. Always let the data lead you by
using the process of elimination to distill out a solution,
and enjoy the challenge.
The author may be contacted for questions at
. Publications and patents can be
referenced at fainstruments.com in the “About Us” and
“Library” links.
Advertise in Electronic Device Failure Analysis magazine!