Metal-based wafer level packaging
CB8 shown in Figure 7, and measured
independently on a freestanding IR mi-
croscope. The average overlay accuracy
of all wafers measured in three locations
each, was +0.89 µm in x and -1.03 µm in
y overlay accuracy.
A similar analysis was done with a
copper-to-copper bond on 200 mm wa-
fers. These wafers did not have backside
polished surfaces, and the IR alignment
was compromised as a result. In addi-
tion, the additional temperature, approx-
imately 100˚C, would have increased
the radial expansion of the wafer by over
50 µm and would have exacerbated any
upper and lower heater deviation. Never-
!
Figure 8. Left is a blanket layer of aluminum on a 200mm wafer successfully bonding with Al diffusion techniques.
theless the data in Table 5 shows that the
On the right is a 150mm device wafer also bonded with aligned Al to Al diffusion methods.
post-bond alignment accuracy for copper
wafers bonded at 450˚C was an average
of 1.5 µm microns, well within needed
requirements and specifications.
Figure 8 shows an example of 150
mm successful aluminum diffusion
bonding. These wafers were also bonded
in the SUSS MicroTec CB8 chamber
using 80KN of force at ~400˚C. Both
the blanket layers of aluminum as well
as the patterned device wafer structures
were successfully bonded in a void free
fashion.
The eutectic bonds are used in
a variety of applications. The lower
temperature processes make these bonds
appealing to optical markets such as
LEDs. The LED device is grown by
epitaxial methods on lattice-matched
!
Figure 9. Au-Sn eutectic bond between silicon and sapphire. Gray areas are the seal rings in this Sonoscan
substrates, such as InP, GaAs or sap-
acoustic image.
phire. The growth substrates are gener-
ally removed to access the backside of
ers such as the SUSS MicroTec CB8 (or ing and non-planar force application.
the device and improve output efficiency
CB200 and CB300 in 200 mm and 300 When the temperature is too high
and brightness. Bonding is used to attach
mm cluster tools) and have led to tempera- above the eutectic, the alloy has a very low
the LED structure to a lower-cost substrate
ture uniformity levels with <1% difference viscosity and flows readily with even the
such as silicon, and the growth substrate
within wafer or between upper and lower lightest application of force. This can lead
is removed by mechanical methods (grind
heaters. The force uniformity in advanced to the formation of bubbles at the interface
an etch) or by laser lift off
10
.In all these
bonders is now 5% and results from the and areas with unwanted metal. Figure 6a is
process flows, the thermal mismatch be-
fact that in these bonders the upper and an optical micrograph looking through the
tween the growth substrate and the silicon
lower pressure plates establish parallelism upper glass substrate. The unconfined flow
is extremely severe. This stress often cracks
by a WEC (wedge error compensation) is clearly visible and there are numerous
the epitaxial layers and ruins the device.
operation before bonding begins. During bubbles. In Figure 6b, the flow direction
Figure 9 shows the results obtained us-
the WEC operation the lower bond head is very obvious, and this is a tip off that
ing an AuSn eutectic alloy to bond silicon
assembly is floated on a bed on N2 air and the bond heads were not parallel to one
to sapphire at 290˚C for 20 minutes. The
allowed to rotate on a spherical bearing. another as the contact was made.
bonds are void free and the wafers did not
This eliminates any non-parallelism be-
crack. The ability to reduce the processing
tween upper and lower plates and ensures results
temperatures minimized the time needed
the applied force is perpendicular to the Gold-to-gold bonding is the lower-tem-
to heat and cool the bonder and the
bond interface at all locations. perature diffusion bonding method of
sample. The decreased sensitivity to ther-
A study was performed in which a glass the three options discussed above. Table
mal gradients (less CTE mismatch) also
wafer and a silicon wafer were bonded 4 is the post bond-alignment accuracy of
enabled faster cooling rates and improved
using eutectic alloy balls
9
. The balls were several wafers aligned with face-to-face ISA
the throughput and COO for this metal-
placed between the two substrates and put alignment on the SUSS MicroTec BA200
lurgical choice.
into the bonder. Several parameters were Gen2 aligner. The wafers were subse-
This is in direct contrast to the results
explored including the effect of overheat- quently bonded on the SUSS MicroTec
shown in Figure 10 for a gold diffusion
14 – Global SMT & Packaging – June 2009 www.globalsmt.net
Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48  |  Page 49  |  Page 50  |  Page 51  |  Page 52