Page 174 - Wire Bonding in Microelectronics

Gold-Aluminum Intermetallic Compounds 151

amounts of free Br and cause bond corrosion [5-35]. In addition to
eliminating Sb O , recent compounds have included proprietary “ion
2 3
scavengers” to eliminate background levels of free Br and Cl. Devices
molded with such compounds were reported to withstand 1400 h of
autoclave [207 kPa (30 psi) at 135°C] without bond failure.
There is still a lack of agreement in the literature in both observa-
tions and interpretation, for example, a corrosion mechanism producing
Al(OH) [5-33], metallurgical phase separation) [5-30], oxidation of the
3
Al in the intermetallic (Al O ) [5-36], and volatile metal halide removal
2 3
[5-32]. It is quite possible that all of these mechanisms occur under vari-
ous conditions that have not been clearly defined, or understood.
The role of H O in the bond-degradation process is not clear,
2
Klein [5-34] did the most conclusive work on this (≈ 10,000 ppm
required) and suggested that H O serves as a catalyst. However, it
2
could also be an oxidant, resulting in Au-Al voiding or a lamellar
structure to proceed at the lower temperatures of an autoclave. Even
if no autoclave is used, the high-temperature (approximately 180 to
200°C) begins the breakdown of epoxy encapsulants which release
water [5-37]. Thus, Thomas’ [5-30] sealed-device experiment could
have contained enough released H O (from the epoxy) to affect the
2
results. Other cases that introduced pure gasses were presumed to be
dry, so this is not a complete explanation. A summary of bond failures
resulting from halogens is given in Table 5-5. Fluorine, Cl, Br, and C
were introduced onto bond pads before bonding (without subsequent
plastic encapsulation), and only the normally expected interface deg-
radation after thermal stress tests was observed [5-46].
Very large contamination was required to cause failures, and gen-
erally these occurred when the contamination layer was thick enough
to limit bond formation, rather than to chemically degrade a well-
made one. Most of the stress tests were run at high temperatures
where no liquid H O is possible. Most were baked near 300°C and in
2
N , with one run at 175°C. High-humidity environments encountered
2
in plastic devices, HAST, or in hermetic enclosures containing H O
2
were not included. Nevertheless, by not finding halogen degradation
under clearly stated conditions, this work is supportive of the require-
ment for significant H O (or vapor) before halogen failures will occur.
2
For a discussion of how Au-Al bond failures may occur when non-
halogen films are in the interface, see App. 5A.
Many experiments have been run in an effort to understand the
complex Au-Al plastic-induced contamination interactions. No com-
plete understanding has emerged. Thus, more work is still needed.

5.2.2 Recommendations for Removing or Avoiding
Halogen Contamination
It has been well established that halogens in an Au-Al bond interface
or even in the environment after bonding (as long as some moisture