Page 143 - Wire Bonding in Microelectronics

W ir e Bond Testing 121

impossible, to nondestructively pull the overlapping wires without the
hook damaging some wires and/or causing shorts between wire layers.
Thus, NASA and the U.S. military undertook an alternative statistical
process control (SPC) method for wire bonding as a substitute for the
NDPT. SPC was allowed for devices with packages having 84 or more
external terminations and with nominal bonding wire pitch at the pack-
age post of less than or equal to 305 µm (12 mils). Early evaluation stud-
ies of the NDPT were carried out by a number of organizations, but often
the information was obtained for in-house purposes and remains unpub-
lished. The several early published reports [4-51 to 4-54] indicate that the
NDP test is valid under the specific conditions of each particular experi-
ment. Of these, only Polcari [4-51]recognized and discussed the impor-
tance of bond geometry. He also repeatedly stressed a number of bonds
to their chosen NDP force and found that some did not fail during 100
applications of this force, whereas others failed after only four or five
trials. The average bond withstood about 50 successive applications of
force before failure. However, the standard deviations of the destructive
bond pull force for the bonds available to them were quite large. Many
of the bonds would have been stressed beyond their elastic limits (see
Sec. 4B.2). All of the bonds were stressed at forces higher than those rec-
ommended in the present work.
The nondestructive pull force is usually specified for a given wire
diameter and metallurgy (see ASTM F 458-06, MIL–STD-883 G/H,

Method 2023.5 [4-55] and MIL-PRF-38535). Typical values for 25 µm
(1 mil) diameter wire are 2.0 gf for Al and 2.4 gf for Au. Various in-
house specifications have ranged from 0.8 to 3 gf for the same wire size.
However, such specific values make no allowance for bonds having
widely different geometries. The test will break a strongly welded wire
bond when (because of package or other limitations) it has a very low
loop. Likewise, fixed-pull values apply relatively less testing force to
bonds with high loops.
A more scientific approach to deriving the NDP force is to consider
the metallurgical characteristics of the specific wire (obtainable from the
manufacturer) in addition to the bond geometry. Figure 3-1 in Chap. 3
showed two differing elongations of wire used for bonding. Although
those curves are for Al wire, equivalent data for Au wire would be sim-
ilar. Note that Au wire for use in thermosonic and thermocompression
bonding is annealed and would generally have stress-strain character-
istics nearer to those of curve A in Fig. 3-1. To avoid metallurgical change
or damage to wires during pulling, the wire must not be stressed beyond
its elastic limits, region 1 of the stress-strain curves.

4B.2 Metallurgical and Statistical Interpretation
of the NDP Test
The metallurgical and statistical interpretation of the NDPT was
given by Harman [4-56] and the following treatment is taken from