Page 133 - Wire Bonding in Microelectronics

W ir e Bond Testing 111

Currently most organizations use design of experiment methods
(DOE) obtained for ball-shear force encompassing machine variables
are described in Chap. 8, App. 8B, DOE for modern autobonder setup
by Lee Levine. There are significant differences between the older
results on manual bonders and ones obtained using modern auto-
bonders. This is not resolved, but may result from autobonders using
short bonding times (8 to 15 ms vs. 50 or so ms for manual ones, and
the force may change) whereas current autobonders use high fre-
quency US energy. At any rate DOE setup (above) must be used for
current autobonders!

The order of priority of the bonding machine parameters
Power is the most important variable.
Bond force and impact are important on ball diameter, especially in
ultrafine pitch, and maybe for reliability.
Temperature is important, especially for second bond (crescent)
strength.
Time is a minor variable.

Although values are different, their relationship is approximately similar for 60
and 120 kHz. One study, Charles, IMAPS 2002, found that for short thermal aging
times, 100 kHz gave better shear strength than 60 kHz. Thus the effect of fre-
quency on reliability of Au-Al bonds must still be studied to find its limits.

Earlier investigators have described DOE methods for setting up
TS bonders [4-37, 4-38, 4-39] and obtained differing coefficients. It is
clear that, at the present time, each machine type may have enough
variability to require its own DOE setup procedure. Currently trans-
ducers within a single manufacturer are very similar. However, dif-
ferent manufactures often run at different frequencies (90 to 135 kHz
typical) with experimental transducers above 250 kHz. They may use
shorter bonding times for higher speed/throughput, etc.
For TC bonding, it was possible to assemble published data and
give typical bonding parameters. This is not possible for TS bonding,
partly because the published data or parameters do not overlap, and
partly because of the reasons cited above and further development into
fine pitch. Since there are no clearly defined universal parameters, it is
extremely important to use DOE along with the ball-shear test to opti-
mize bonding machine setup which is done with modern software.

Evaluation of Production Bond Quality
The major use of the ball-shear test, as with the pull test, lies in the
area of production quality control. This test was slow to gain accep-
tance after its introduction in the 1970s, due to the lack of commercial
testing equipment and standards for its use. Both problems have long