Page 112 - Wire Bonding in Microelectronics

W ir e Bond Testing 91

increased elongation. Therefore, the pull force tends to be indepen-
dent of the specific wire-breaking load for many common device
geometries. These results may be scaled down for integrated circuits,
except that small 25 µm (1 mil) diameter annealed Al wire elongates less
than approximately 10%, and then only after high-temperature exposure
or significant annealing. Otherwise it’s typically ~2%.
There will be special cases where the effect of large wire elonga-
tions can change the bond-pull geometry and, hence, the measured
pull force, even more than indicated in Figs. 4-7 and 4-8. This occurs
when the pulling probe (the hook and arm) is misplaced, is flexible, or is
free to pivot where it is joined to the force gauge or load cell.
The effect may occur even if the hook does not slip in the case where a
high package bond pad (post) is involved. Here, the wire span will be
considerably longer on the chip side of the hook than on the package
side. The relatively greater increase in length (elongation) of the chip-
side span during the test will result in moving (swinging) the hook
nearer the package pad and in pulling on the wire at some angle, ϕ,
from the vertical. The effect will be enhanced if the pulling hook was
initially placed nearer to the package bond than to the chip bond.
These changes in the bond-pull geometry, which can result in lower
measured values of pull force, must be taken into account in any pull-test
calculations involving wires with high elongations.
Stress-strain type measurements have been made during pull
testing on a number of large-diameter power-device wire bonds to
determine any unique characteristics that could influence the pull test.
Both the measurement and its interpretation are much more difficult for
pulling a typical wire-bond loop than for measuring the stress-strain rela-
tionship of a long piece of wire. In pulling a standard 250 mm (10 in)
length of wire, the elongation is normally read directly from a recorder
(see Sec. 3.2, Fig. 3-1 for examples of stress-strain curves.). However, in
pulling a large-diameter wire-bond loop, the total length of wire is gen-
erally less than 6.25 mm (0.25 in), and, in addition, the measurement
indicated by the apparatus is in reality the increase in loop height (which
is nonlinear with wire elongation) and is very small compared to the
elongation of the standard length of wire. Thus, when determining wire-
bond-loop elongation, the sensitivity of the measurement apparatus
must be increased to its maximum, and any system nonlinearities, such as
a slight irregularity of the screw-thread pitch on the stress-strain machine
or bending of the pulling hook, will have a greater effect and must be
corrected for in each curve.
A typical, corrected force versus rise-in-pulling-hook curve for a
200 µm (8 mil) diameter emitter wire bond from a power device is
shown in Fig. 4-9. There are three distinct regions in this curve.
Region 1 is the triangular loop formation and elastic wire-
tensioning region. Although the curve increased linearly for this
bond, other bonds often showed variations as the loop formed into a
triangle, generally within the dotted curves. Point 2 denotes the elastic