Page 103 - Wire Bonding in Microelectronics

82 Cha pte r F o u r

If both bonds are on the same level (H = 0), and the loop is pulled
vertically (ϕ = 0), in the center (ε = 0), and (θ = θ ), then the more
t d
familiar equation is obtained:

f = F (4-4)
θ
wt sinθ + cos tanθ
t t d
where θ = θ = θ. Note that, in general, for bonds of a given strength,
t d
larger values of h/d will result in higher pull force, F, values. Equivalent
equations using angles θ , θ , and F are:
t d
f = F (4-5)
wd sinθ + cosθ tanθ
d d t
Note that all of the above equations are solved for the force or ten-
sion in the wire, f and f (usually at break). If a reader wants to calcu-
wt wd
late the actual pull force, then the equations must be solved for F. A
wire will break when either f or f first reaches its breaking strength.
wt wd
This entails assigning a breaking strength (value) to each side of the
wire. Typically, this is about 60 to 75% of the manufacturer-specified
breaking load of the wire for Al wedge bonds (due to heel deformation
and metallurgical overworking), but is nearer 90% for Au bonds (either
ball or crescent bond break). The wire normally breaks just above the ball
in the heat-affected zone (see Chap. 3).
A plot of the calculated pull force (F) at wire rupture for wedge
bonds is given for a typical two-level semiconductor device-bond
configuration in Fig. 4-2, pulled straight up (ϕ = 0) at the center of the
loop. With everything else being equal, it is apparent that the higher
the loop height, the higher the bond pull force will be. For a given
bond-to-bond spacing, d, lowering the loop will result in a force multi-
plier, increasing the values of f and f for a given force at the hook, F,
wt wd
and thus yielding a lower force at wire rupture.
The position of the hook (indicated as εd in Fig. 4-1) and the pull
angle, 4, will significantly affect the distribution of forces at the bonds.
One can choose a ε or ϕ value that will give equal forces on each bond,
and it will result in a more equal test of both bonds. This is possible
with some automated pull testers. However, manual pull-test opera-
tors would be significantly slowed by such a procedure. In addition,
most specifications (such as ASTM F459-06 [4-1] and MIL-STD-883
G/H, Method 2011) [4-7], and most in-house requirements specify
that the hook be placed in the center between the bonds. So this is con-
sidered the standard hook-placement position for normal testing of
wedge bonds, but not for fine pitch ball bonds which may peel the
bond pad (see Chap. 9 and its references). We note that major accepted
specifications must be changed to allow such hook placement for
Cu/Lo-k devices and other fine pitch pull tests! (MIL-STD-883G/H
now allows non-center hook placement.) Whenever the hook is moved
close to a wedge bond, a higher proportion of the wire force is applied