Page 126 - Wire Bonding in Microelectronics

104 Cha pte r F o u r

were ~30 gf (~70% of values in Fig. 4-17 and closer to other predic-
tions) [4-12].
Mixed units are used in the industry for shear strength (SS)
because SI units (µm or mm) result in values so high or low that they
are not intuitively comprehendible. The reader can use the conver-
sion factor above to convert data. Note: All machines, except one in
the 100 µm process, bonded with 60 kHz US energy and on Al chip
pads. Observation of numerous cross-sections of ball bonds at this
laboratory (NIST) indicates that the outer boundary of the tool imprint
on top of the ball corresponds closely to the perimeter of the actual
bonded area. It is, therefore, recommended that the welded area
should be estimated by measuring the outer diameter of the tool
imprint in the ball, rather than the outside diameter of the (non-fine
pitch ∗ ) ball. This value should then be used to obtain the maximum
expected shear force in Fig. 4-17, and it may be 15 to 20% lower than
that obtained by using the outer diameter of the ball, as was used in
Table 4-1. However, some fine pitch capillaries do not produce bonds
with the traditional squashed-ball shape. They may be an inverted
conical shape (see Fig. 4-18), or a minimally squashed classical shape.
Fine pitch balls tend to be welded over a very high percentage of the
interface, and the outside perimeter is typically used for calculations.
The curves of Fig. 4-18 can be used to establish the maximum shear
force obtainable in as-made ball bonds. However, no minimum accept-
able value can be deduced from that curve, but it has been addressed
by comparing wire pull and ball-shear data [4-19]and in one standard

(Sec. 4.4.11). If the mean pull force is plotted against the mean shear
force, a minimum shear force will be found in which no ball bond lifts
occur during pull testing. Such a procedure requires thousands of
bonds to obtain meaningful data. In this study with large diameter
balls by today’s standards, the mean shear force of good bonds was
approximately 80 gf, and the crossover shear force, where some ball-
lifts occurred during pull testing, was 40 gf shear force. Therefore, by
this criterion, a minimum acceptable shear force appears to be approx-
imately one-half the indicated shear force value in Fig. 4-17. [However
this does not consider the thermal stress reliability requirements of Au-
2
Al bonds, which require higher shears of about 5.5 g/mil (84 MPa). See
Fig. 4-23 and its top curve along with its discussion.] Fine pitch appears
to result in higher values, but is more subject to thermal degradation.
The microelectronics industry has been moving toward fine-pitch
ball bonding (<50 µm pitch), as well as experimenting with high-
frequency ultrasonic energy. To compare the shear strength (amount
of welding) from one ball size (or US frequency) to another, the industry
has begun to use normalized shear-test values [4-25, 4-26, 4-27]. For this

∗ Capillary imprints on fine pitch balls often extend to the outer diameter of the
ball, depending on the specific tip design. For these, the outer diameter should be
used in calculations.