Page 92 - Wire Bonding in Microelectronics

Bonding W ir e Metallur gy and Characteristics 71

The bonding method, as well as bond quality, can also affect the
burn-out current. Shorter lengths of gold ball-bonded wire will burn
out at a higher current than an equivalent length of gold wedge-wedge
bonded wire. This results from the wedge-bond neck constricting the
wire and limiting the thermal flow, whereas the large ball serves as
a heat spreader, making good thermal contact with the chip. As an
example, everything else being equal, a 25-µm diameter gold wedge-
wedge bond burned out at 0.6 A DC, but a similar ball-wedge bond
burned out at 1 A DC. The burn-out point would be displaced (from
the center) toward the wedge bond since there is better heat conductiv-
ity out through the ball, cooling that side of the wire. If heat flow was
symmetrical, then burn out would occur at the center of the wire.
Some wire manufacturers give tables of burnout data for their
product, usually identified only by letters or other code, so it may not
be applicable to an other manufacturer’s product. Nevertheless it can
be useful and points out that different wire dopants can influence
both the burn-out and resistivity. See Web sites in further reading.
Plastic-encapsulated devices comprise over 95% of integrated cir-
cuits. Thus, it is surprising that there have been minimal studies of
(gold) wire burn-out in such conditions. Some organizations have
made limited studies for internal use, but not published them. One
such study contracted to a university, is available as an unpublished
report [3-33], and some of its conclusions are used in this section.
Encapsulated wires will carry considerably more current than open-
air wires due to the increased thermal conductivity of the surround-
ing plastic compound (with respect to air). However, at some point,
as the current increases, it heats the wire sufficiently to affect the
thermal characteristics of the adjacent encapsulant (glass transi-
tions, melting, charring, volatilization, etc.). Ultimately this leaves
an air gap between the wire and the plastic, or, the plastic otherwise
become thermally insulating, and the wire burns out quickly. The
report [3-33] finds that the voltage drop across the wire (an indica-
tion of temperature) increases erratically as if there were a series of
plastic (thermal-characteristics) transitions. Some encapsulated 30-µm
(1.2-mil) diameter gold wires sustained currents of several amperes
for over an hour before failure. As with bare wires, longer encapsu-
lated ones failed at lower current levels.
As stated above, gold wire burns out in air by forming neat balls
on each side of the open wire. However, when encapsulated, interac-
tion with the plastic and the filler results in complex failures, gener-
ally of the type shown in Fig. 3-11 [3-34]. Here, no definable ball was
formed (or else any ball fragments fell off during decapsulation) and
particles of the inorganic filler adhere to the wire. The burn-out cur-
rent for the 25-µm diameter gold wire in this particular case was 1.1
A (data obtained by slowly, manually, increasing the current) which
was about twice as high as expected for a 25-µm diameter Au wire of
equivalent length in air. Another difference between open cavity and