Page 83 - Wire Bonding in Microelectronics

62 Cha pte r T h ree

would stretch rather than break and stop the bonding operation,
often loosing the threading of the wire. However, large-diameter wire
having high elongation is cut off by a specially designed bonding tool
and bonded with an inverted V-shaped, parallel-grooved tool that
limits deformation. Examples of bonded large-wire are given in
Chap. 2. Thus, large-Al wire can be fully annealed, have very high
elongations, and still be bonded, as shown in Chap. 2, Fig. 2-16. Most
large-diameter Al wire is either specified as 99.99% Al, or a few may
contain 0.5% Mg. However, 99.99% wires have been offered with
~50 ppm of Ni additive, which is intended to make them less subject
to corrosion in plastic packages as discussed previously. These wires
are reputed to withstand up to 700 h in a pressure cooker test (121°C,
100% RH). This wire alloy is not offered in small diameters. Any pos-
sible effect on Au-Al intermetallic compound formation has not been
a problem since large wire is usually bonded to Ni on the package
and Al pads on the device.

3.6 Wire and Metallization Hardness
The hardness values of wires and balls are often needed to assess the
possibility of cratering and to match the hardness of metallization for
best bonding.∗ Such data are scattered throughout the book and are
collected together in the cratering section (Sec. 5.1) in Table 5-2 for con-
venience. These data are often in different units and may not be directly
converted, since, necessary information is often not published. Values
of typical metallization hardness are not included in that table because
they depend on heat treatment and any alloying agents. For instance,
Al metallizations containing Cu can increase in nanohardness (UMH,
or ultramicrohardness, see glossary) by up to a factor of 4 (1 to 4 GPa)
as the concentration of Cu increases from 0 to 10% [3-7]. Another UMH
test [3-8] found that pure Al films ranged from ~0.45 to 0.6 GPa. The
2 to 4% Cu range (sometimes used in IC metallization) is about 2 GPa.
This alloy is noted for age hardening, so its actual hardness will depend
on heat treatment and age. Hardness measurements made on thin
(∼1 µm) metallization require special UMH testers using very low
forces (~0.2 g load) and cannot be performed without special knowl-
edge and training. However, hardness measurements on wires and
balls have been made with standard microhardness testers, typically
using loads of ~1 to 4 g.
Two studies have measured the correlation between metalliza-
tion hardness and bondability. Nabatian [3-9] (thick films and wedge
bonding) and Klein [3-10] (IC metallization and ball bonding) found

∗There is also a hardness difference between as-made undeformed balls and
bonded balls, with the latter being about 40% harder (Chap. 8, Table 8-3), but
differences depend on the metal (e.g., Au, Cu).