Chip Scale Review May • June • 2019[ChipScaleReview.com]
After the EMC pellets are loaded, they
are melted and injected into the cavity with a
pressure limit of 4MPa. During this operation,
the sample is clamped at 950kN, which
increases to 1100kN if the mold opening
exceeds a threshold of 10µm, or when the
compaction phase is activated. This high pre-
clamping force, or relatively low increase in
clamp force of 16%, is possible due to the
simple layout of the demonstrator sample.
As explained earlier, the die population and
interconnect density determines the maximum
amount of clamping that can be applied.
Although this simple demonstrator product
has a die population of only 35%, the pressure
that can be applied on the die is high as the
die is directly supported by the bond layer and
carrier (P3 and K3 in
). Therefore, a
high pre-clamping force can be used, which is
beneficial for the process.
The final result of the molded demonstrator
is shown in
. The injection point is
located at the lower right side, and the EMC
flows from the lower right to the upper left.
The notch for rotational alignment is on the
left side. Further, one can see the clamp edge
of themold to the glass carrier, (P1 in
and an overview of the exposed dies. Clearly,
the die surfaces are kept exposed. All dies are
visible, but more important, however, is that no
discoloration is visible, and therefore, no signs
of flash. This indicates a good uniformity of
how the mold tool clamps onto the die, and
thereby the achievement of a stable clamping
process, for which the dynamic clamping
was setup. Secondly, it is visible in the lower
left picture that the mold cap and die are on
the same height level, which is beneficial for
further processing, but also shows that the
loading on the die was not too high, as the die
would then sink into the release foil leaving
an imprint in the mold cap. So this operation
again shows the purpose of applying dynamic
clamping. Finally, it is important to notice that,
although clamping of the mold tool occurred
directly onto the glass carrier, this surface is
clean and shows no damage.
• This paper explains that both EMC
properties and clamp force control
are dominant topics in the field of 12”
wafer-level transfer molding.
• A basic method to estimate the effect
of EMC choice on the resulting
wafer-level warpage after molding is
• The impact of clamping dynamics
during exposed die transfer molding
were discussed and accompanying
control strategies were introduced.
• Finally, the process of exposed die
wafer-level packaging is demonstrated
on a 12” exposed D2W sample.
• With both warpage control and the
exposed dies capabilities, transfer
mol d i ng a t t he wa f e r l e ve l i s
shown to be a serious alternative to
We wou ld l i ke t o t ha n k Hit a ch i
Chemical Co. Ltd. for delivering epoxy
molding compounds and release film for
1. K. Kwon, J. Y. Chung, D. Lee, S.
K. Kim, “Wafer warpage control by
epoxy molding EMCs for wafer level
package,” IEEE 67th Elec. Comp.
and Tech. Conf. (ECTC) (2017).
2. S. P. Timoshenko, Jour. Opt. Soc.
Am. 11, 233 (1925).
3. S. H. M. Kersjes, J. L. J. Zijl, W.
G. J. Gal, H. A. M. Fierkens, H.
Wensink, “Exposed die wafer-level
encapsulation by transfer molding,”
Elec. System-Integration Tech. Conf.
4. W. Gal, et al., “Wafer-level packaging
(WLP) by transfer molding,” ISIT
Conf., Belgium, 2009.
Molding parameters used for the
A molded D2W demonstrator sample. Us-
ing wafer-level transfer molding, an exposed die 12”
FOWLP on a glass carrier was achieved.
Sebastiaan H.M. Kersjes is the Manager of the process technology group at Besi Netherlands B.V., The
Netherlands. He has over 12 years of experience with back-end processes for the semiconductor industry, focusing
on packaging processes such as molding, trim and form, and singulation. Next to managing the process technology
group, he maintains the international R&D network for the packaging division of Besi, and has cooperated in
starting up several relevant European projects. Mr. Kersjes holds a MSc degree in Mechanical Engineering from the
Technical U. of Eindhoven; emailSebastiaan.email@example.com
Jurrian L.J. Zijl is a Process Scientist at BESI Netherlands B.V. He has 18 years of research experience in the back-end semiconductor
industry. He contributed to many new developments on molding, trim and form, and laser singulation equipment. He holds a MSc
degree (’97) in Mechanical Engineering and a PhD (’01) in fluid dynamics from Delft U. of Technology in the Netherlands.
Niels de Jong is Product Manager Molding at Besi Singapore, Taiwan branch. After a brief start as a physics teacher, he made the
jump to production machine building and has been doing that ever since. He has been working for Besi for 8 years now, mostly in the
Asian region, where he is currently also stationed.
Henk Wensink has been a Senior Manager at Besi Netherlands B.V., with 14 years of experience in the semiconductor packaging
industry. He currently holds the position of team leader at the Wageningen U. and Research the Netherlands. He has MSc and PhD
degrees in Applied Physics.