Page 134 - Wire Bonding in Microelectronics

112 Cha pte r F o u r

been corrected (see Secs. 4.3.2 and 4.3.11), and currently (2008) all
major semiconductor assembly and MCM/hybrid/SIP facilities are
using the shear test to control their ball bond production.
Historically, the first published use of the ball-bond shear test
(in 1967) was to monitor and control aspects of microelectronic (chip)
production rather than to evaluate bonding. Gill [4-14] used it to
monitor adhesion of the “then” new molybdenum-gold metallization
system. Weak adhesion of Au to molybdenum was revealed by the
Au peeling from the molybdenum during the bond-shear test. The
shear test is still an excellent method to determine the quality of met-
allization adhesion (see 4-15) and has also been used to evaluate the
tendency of different metal and bond systems to crater under plastic
package generated shear and thermal stresses.
It is apparent that the high stresses applied to ball bonds in plastic-
encapsulated devices during molding and later during thermal cycling
(and also the thermal shock during surface-mount soldering) make it
essential to use the ball-shear test for production-bond quality control.
When used as a continuing production test, the shear test will reveal the
degrading effects of recently introduced contamination, as well as any
variation in the metallization or glassivation removal process. This
information can be obtained quickly enough to take corrective action
before large amounts of a failed product are made. Several studies
have shown a correlation between ball-shear test results and the reli-
ability of the devices [4-40 to 4-43].
The nondestructive ball-shear test was investigated only twice
[4-19, 4-39], and it was found that balls can be stressed to 75% of their
destructive shear force without significant final shear force degrada-
tion. In principle, this test could be used to assure production-bond
quality as can be done with the nondestructive bond-pull test. How-
ever, great care is required in positioning the shear tool. As a result,
the test is slow and, therefore, costly, which has prevented its further
consideration. A worse problem, however, is probable damage to the
top surface chip passivation, and in the case of fine pitch, it becomes
impossible to use. Thus, although this was an interesting investiga-
tion historically, it cannot be recommended for production and abso-
lutely cannot be used on fine pitch bonds.

4.3.10 Shear Test for Wedge Bonds
The shear test is clearly useful for evaluating ball bonds, but is it also
useful on small-diameter ultrasonic Al wedge bonds? A cooperative
experiment between NIST and Sandia National Laboratories (reported
in [4-44]) was designed to determine this. The test consisted of three
groups of 25 µm diameter Al, 1% Si wire, [BL = 12 to 14 gf (× 9.8 for
mN)] ultrasonic wedge bonds on a single wafer substrate made at
NIST. Half of the bonds (randomly selected) were pulled to destruc-
tion, and then the wafer was sent to Sandia where the remaining
bonds were shear tested. The data are shown in Fig. 4-22.