Page 31 - Chip Scale Review_January February_2023-digital

the mold material and therefore, the full
Figure 2: Representative 3-D warpage image for: a) pre- and b) post-mold wafers; and
c) warpage value for pre- and post-mold wafers
Figure 3: a) Bump height measurements pre- and post-mold; b) Mold height wafer map calculated from bump
height measurements; c) Mold thickness distribution and range (TTV) for the different skews.
Chip Scale Review
coverage of the wafer. All the different
process skews and their mold coverage
were tabulated (Table 3). Results
were further discussed in the section
entitled, “Results and discussion.”
W a r p a g e me a s u r e me n t s .
Representative 3-D warpage images
are shown in Figure 2. As can be seen
in the figure, the warpage values are
similar for both the pre- and post-
mold wafers, with a difference on the
order of ~100µm. Despite the CTE
mismatch between the mold and the
silicon substrate, the low thickness
of the mold (on the order of ~30µm)
relative to the substrate thickness leads
to better control of the entire package
temperature during the processing and
therefore, lower warpage. In addition,
both granular and liquid resin mold
wafers show similar warpage values,
which suggests that there isn’t any
specific benefit of one over the other in
terms of warpage. The combination of
the two allow better temperature control
and lower wafer warpage than seen in
other compression molding studies.
None of the processing skews showed
any warpage value different from
each other, suggesting that the reasons
listed previously play a greater part in
warpage control than modification of
the material or processing conditions.
Mold height measurements. Bump
height measurements were carried out
on a TAKAOKA multi-substrate bump
inspection system. Figure 3a shows
the pre- and post-mold bump height
distribution of every bump on a wafer.
The difference of the pre- and post-
mold bump height was used to calculate
the mold thickness distribution. Using
the wafer coordinates of the bumps,
Figure 3b shows the wafer map of
the mold thickness distribution over
an entire post-mold wafer. Wafers
with incomplete fill at the edges had
a mold height that was significantly
different from the rest of the wafer—
and that was clearly visible in the wafer
map. Interestingly, the wafer map of
the mold height suggested regions of
higher mold height near the center of
the wafer and slightly lower height as
we moved towards the edge.
Given the above considerations,
the mold thickness range could be
calculated f rom the dist r ibution
(Figure 3c). Because of the significant
29
January • February • 2023 [ChipScaleReview.com] 29