There are some additional peculiarities of the free Radical cure system. The
atmosphere at Die Attach cure needs to be controlled to not allow the redeposition
of the out gassing products of the adhesive, and in some cases it also
needs to be controlled to limit the oxidation of the leadframe. Epoxy Die Attach
Cure ovens can often utilize Clean Dry Air (CDA) instead of CO2 or Nitrogen as
cost a savings measure. Free Radical Cure systems can be sensitive to the
Oxygen content of the atmosphere during cure, and the compatibility of either Air
or CO2 should be investigated carefully. The Oxygen in the cure oven can
compete with the cure chemistry, and reduce the effectiveness of the cure. In this
case, Nitrogen is used to purge the Die Attach Cure ovens.
Another issue occurred with this program during start-up. When the Leadframe
Supplier changed to a different manufacturing location for production, sporadic
cases of delamination were seen. The majority of these were detected after
board mount during system level check-out of the prototype designs. This is the
worst possible time and place to discover a packaging failure. It should be noted
that the failure mechanism was delamination in the thermal path only (between
the die and the exposed die pad) and there was no delamination of the top side
die surface, or effect on the electrical functionality of the device. There is no
screening (100% of product) for thermal performance before board mount, so it is
not known if the failure existed before board mount. A number of tests (most
notably board assembly with and without dry baking) that were run to try and
identify the source of the problem. The "Bake / No-bake" split lots showed no
sensitivity (improvement) to package moisture. Separate sample monitoring
using Thru-SAM showed some instances if initial die to Leadframe delamination.
This was not seen during the previous Thru-SAM evaluations. A Designed
experiment was then conducted which isolated the initial delamination cases to
the production leadframe, and not on the leadframe from a different
manufacturing location. The "hunt" for the root cause uncovered that a process
had been implemented at the Leadframe Supplier without notification. This
process was implemented to control Resin Bleed Out (RBO) for one of the lead
frame supplier's other customers. RBO is the separation of the Die Attach Resin
before cure (see Photo #1). This phenomenon is widely recognized, and TI was
aware of RBO, but this was not a problem for this program. There are a number
of conditions that affect the bleed rate, and the problem is often more
pronounced with NiPdAu plated leadframes and rough surfaces. In this instance,
the chemical used for "Anti-RBO" treatment was the same as is commonly used
for Silver Spot leadframe tarnish control. The leadframe Supplier did not
recognized that this could cause problems with adhesion. Further investigation
verified that the BMI/Acrylate Die Attach chemistry is much more sensitive to this
type of contamination than Epoxy based systems 1. Anti-tarnish and "Anti-RBO"
treatment residue is not easily detectable. AUGER, FTIR, and some wetting tests
are effective for detecting these materials, but are not commonly used for
leadframe testing. Even with these sophisticated analysis tools, there is some
inconsistency in the results (or non-uniformity in the contamination distribution).
The use of any "post plating" treatment on NiPdAu plated leadframes needs to
be closely controlled as the interaction with different Die Attach and Mold
compounds can be problematic. Once the "Anti-RBO" treatment was eliminated,
the delamination & thermal path issues stopped. With "clean" lead frames and
the new Die Attach and Mold compounds, the MSL performance was improved
by 3 levels at a higher reflow temperature (from MSL5a @ 235 C to MSL3 @
260C).
The use of non-epoxy based die attach adhesives allows for significant
improvement in MSL performance, but additional controls need to be recognized
and implemented to realize their full potential. |
