Page 147 - Wire Bonding in Microelectronics

W ir e Bond Testing 125

forces below the chosen force level at the time the test is performed.
There is no assurance against later bond strength degradation due to
Au-Al intermetallic and subsequent void formation, ultrasonic-
cleaner-induced wire-bond vibration fatigue, or wire-bond flexure fatigue
due to temperature or power cycling, etc. Such possible failure mecha-
nisms are described elsewhere in this book (see Chaps. 5, 7, and 8).
The effect of post-NDP-test screens and environments on bonds
should be thoroughly understood by the device user. These are dis-
cussed elsewhere in this book. In some cases, the devices could be
chosen or designed to minimize post-NDP-test degradation, such as
using monometallic wire and bond-pad systems and using high bond
loops. The NDP test, as well as the destructive pull test, is not appro-
priate for screening the quality of ball bonds. The welded area of a
ball bond would have to be less than the cross section of the wire to
fail during a standard NDP test [pulled at 3 gf for a 25 µm (1 mil)
diameter wire]. This could only happen if the bonding process were
completely out of control.

4B.5 The Current Status of the NDPT (2008)
for Critical Space Applications
The NDPT is expensive and costs about as much as manually making
the bonds in the first place—and much more if the bonds were made
on an autobonder and NDP tested manually. Therefore, the NDPT is
performed only on critical high-reliability military or class K for space
(equivalent) and some implanted medical devices. Currently, there is
pressure to reduce costs of all systems, military and commercial, and
the NDPT is an area of concern. In addition, with the introduction of
fine-pitch, costs and time as well as mechanical problems in imple-
menting the NDPT tend to reduce its use. The chosen approach for
critical military and space applications (with limitations) is to use sta-
tistical process control for devices with more than 84 terminations and
package-pad-pitch of ≤305 µm (12 mils) (see MIL-STD-883G/H,
Method 2023.5, paragraph 3.2). This specification is adequate to serve
as an alternative to the NDPT. There is no mention of multitiered
packages, which are a primary NDPT problem, as well as the chip pad
fine pitch. The entire procedure depends upon the bond pull test for
both Al wedge and Au ball bonds; thus, there is no ball-shear test eval-
uation required for SPC of ball bonds. [Note, the JEDEC Shear test is
now (2008) called out in MIL-STD-883G/H, [4-7] and hopefully, this
will also be added to any SPC requirements.]
The difficulty of performing the NDPT under multitiered and/or
fine-pitch (<150 µm on the package and ~60 µm on the chip) condi-
tions is real, and the above (Method 2023) is helpful. However, the
problem with any SPC approach is that one must choose which param-
eters to control and how to measure them. There are few specific details
given in this test method. Cleaning with UV-ozone, plasma, or with