Page 148 - Wire Bonding in Microelectronics

126 Cha pte r F o u r

unspecified solvents (see Chap. 7) is given as an option rather than a
requirement. Normal distributions are assumed, but may not exist in
many devices. In addition, many space parts are still made in small
numbers on manual bonders, but need wirebond yield loss and failure
rates to be in the low ppm (near 4.5 σ) range. Any statistical monitoring
system will, of necessity, have to assume normal distributions, but most
bond failures in a well-controlled high yield process are better described
as “freaks” or “outliers” (see discussion in Sec. 9.4 on small-sample sta-
tistics). Assessment of process capability depends on the normality of
the underlying failure mode distribution. If different simultaneous
failure modes are present, then it is unlikely that the normality
assumption will be fulfilled, and estimates of product quality based
on the expected failure mode will not necessarily reflect the true
defect levels in devices. Thus, it is not clear that the chosen SPC
approach and variables will yield the bond quality essential for high
reliability in the small quantities of individual devices or SIPs needed
for typical satellite or other space applications. The most encouraging
aspect of current bonding technology is that modern autobonders
make very reproducible bonds in high volume production. Unless
there is a metallization or cleaning problem, the bonds made will be
more uniform and reliable than obtainable with manual bonders. If
plasma or UV-ozone cleaning were added to the SPC preparation,
then that condition would be adequate for high-reliance use.

References
4-1 ASTM, 100 Barr Harbor Drive, West Conshohocken, Pennsylvania, 19428–2959.
[ASTM Standard Test Methods with Round Robin test verifications: F 459-06
(Pull Test), and F 1269–06 (Ball shear test).]
4-2 JEDEC, 2500 Wilson Blvd., Suite 220, Arlington, VA 22201–3834, USA.
4-3 Schafft, H. A., Testing and Fabrication of Wire-Bond Electrical Connections—A
Comprehensive Survey, National Bureau of Standards Tech. Note 726, Sept.
1972.
4-4 Albers, J. H., Ed., “Semiconductor Measurement Technology. The Destructive
Bond Pull Test,” NBS Spec., Pub. 400–18, Feb. 1976.
4-5 Harman, G. G. and Cannon, C. A., “The Microelectronic Wire Bond Pull Test,
How to Use It, How to Abuse It,” IEEE Trans. on Components, Hybrids, and
Manufacturing Technology CHMT-1, Sept. 1978, pp. 203–210.
4-6 John Beleran, Alejandro Turiano, Dodgie R. M. Calpito, Dominik Stephan,
Saraswati, Frank Wulff, Breach, C., “Tail Pull Strength of Cu Wire on Gold and
Silver-plated Bonding Leads,” Proc. Semicon, May 4–6, 2005, Suntech Center,
Singapore.
4-7 MIL-STD-883G, 28 February 2006, Test Methods and Procedures for
Microelectronics, and MIL-PRF-38534F, 2006. (Both have been revised in version
“H” November 2008 in Initial Draft of MIL-STD-883, Revision H.
4-8 Owens, N. L., “Wire Pull and Normality Assumptions,” 9th Ann. Proc. IEPS,
San Diego, California, Sept. 11–13, 1989, pp. 595–601.
4-9 Harman, G. G., Ed., “Semiconductor Measurement Technology. Microelectronic
Ultrasonic Bonding,” NBS Spec., Pub. 400–2, Jan. 1974.
4-10 Harman, G. G., “The Use of Acoustic Emission as a Test Method for
Microelectronic Interconnections,” Proc. International Conference on Soldering
and Welding in Electronics, Munich, Germany, Nov. 11–12, 1981, pp. 104–110.