Page 84 - Wire Bonding in Microelectronics

Bonding W ir e Metallur gy and Characteristics 63

that bondability decreases as the metallization hardness increases, all
other conditions being equal. (See Chap. 9 for an example of bond-
ability as a function of hardness.) In addition, others found that the
best bondability occurs when both the wire and the metallization are
about equal in hardness [3-11]. Typically pure bondable Au and Al
wire and films range in hardness from ~50 to 90 HKn, (a metal hard-
ness scale, see glossary) but can increase in hardness rapidly with
impurity and gas (oxygen) content.

3.7 The Effect of EFO Polarity on Gold Wire
and Its Metallurgy
Around 1984, the industry began to change from using a positive
electronic flame-off spark (EFO) to a negative EFO for gold ball for-
mation [3-12]. One reason for the change was that the negative EFO
resulted in more uniform ball formation (important for today’s high
yield and fine pitch bonding requirements). Another was that foreign
(carbonaceous) particles were not attracted to the wire and the capil-
lary. Also, gold is not sputtered from the wire and deposited on the
capillary. Thus, using the negative EFO stopped the deposits, and sig-
nificantly increased capillary life, minimizing capillary-relatedmachine
down-time. In addition, extensive theoretical studies of ball forma-
tion at the University of Pennsylvania [3-13] showed that a negative
EFO produced more effective and uniform heat transport from the
spark to the wire.
There were also some “claimed” benefits for the EFO polarity
change. As an example, it was claimed that the ball was softer and,
therefore, resulted in less cratering. However, limited studies of ball
hardness resulting from positive and negative EFOs showed that
the latter actually resulted in a slightly harder ball, probably because
it produced smaller grain structure [3-14]. Values reported were
(average) 39.3 HKn for −EFO and 37.3 HKn for +EFO. Thus, any
EFO polarity effect on cratering, which was never documented,
remains unexplained.

3.8 Metallurgical Fatigue of Bonding Wires
Wire bond reliability problems resulting from temperature and power
cycling are extensively discussed and explained in Figs. 8-17 to 8-19
(Sec. 8.4). However, no metallurgical stress versus number of cycles
failure data are given there. Metal fatigue is defined and some typical
S-N failure curves are shown. Several authors have studied the wire
fatigue problem in Au and Al wires, and some have given S-N curves
in their publications [3-15 to 3-20]. The information presented below
covers Au and Al wire alloys. Copper-alloy wire S-N experiments
have also been described [3-19].