Page 114 - Wire Bonding in Microelectronics

W ir e Bond Testing 93

machine setup parameters, are also sufficient to set up the ball-bonding
machine parameters. However, considering that most ball bonds can
have interfacial welded areas in the order of 3 to 6 times the cross-section
area of the wire (fine pitch ball bonds excepted), it is apparent that the
wire will break in pull testing before even a poorly welded ball will
lift ∗ [4-11, 4-12]. In addition, the wire just above the neck of the ball
bond is totally annealed, recrystallized (the heat affected zone), and
generally becomes the weakest part of the ball-bond wire wedge-
bond system. In a pull test, the wire often breaks at this point, depend-
ing on the placement of the hook and the bond geometry (see Sec. 4.2).
Thus, little information is gained on the breaking strength of the ball-to-
bonding-pad interface if the ball is welded over more than 10 to 20%
of its interfacial area.
Considering the above, it is apparent that some type of push-off or
shear test offers the best possibility of assessing the quality of a ball-
bond interface and, thus, properly setting up a ball-bonding machine.
The ball-shear test was independently introduced to the micro-
electronics industry in 1967 [4-13, 4-14]. However, it appears to have
been ignored or forgotten for almost 10 years until Jellison [4-15, 4-16]
and, later, Shimada [4-17] designed precision shear testers and used
them in a series of laboratory experiments that clearly demonstrated
the usefulness of the test. Since that time, there have been numerous
studies of the ball-shear test. Today, it has been almost universally
adopted for production control, and there is excellent equipment
available, as well as several published standards (see Sec. 4.3.11).

4.3.2 Apparatus
The equipment used to perform the ball bond-shear test has ranged

from tweezers and other hand-held probes [4-11, 4-18] to dedicated
shear-test machines with strain-gage force sensors, automatic height
positioning, and various electronic methods of data recording/
presentation. In principle, the ball-shear test is simple and consists of
bringing some form of a shear tool up to the side of a bonded ball,
applying a force sufficient to push it off, and recording that force. The
test is illustrated in Fig. 4-10.

∗ Consider a ball bond on a 25 µm (1 mil) diameter gold wire. The diameter of
a well-bonded normal pitch ball is usually in the 65 to 90 µm (2.5 to 3.5 mil)
diameter range, and the wire-to-bonded-area ratio in this case is from about
4 to 10 in favor of the ball. (It would be lower for fine-pitch ball bonds, see
Table 4-3.) Thus, in a wire-bond pull test, the wire will break even if the ball is
weakly welded. Arguments similar to this were first pointed out by Gill [4-14],
and variations of them have appeared in most of the later studies involving the
ball-shear test. For instance, Stafford [4-12] calculated the force that the wire can
apply on the ball-metallization interface in the pull test and verified the above
conclusion quantitatively.