4.4 Materials .1 Bonding Wire
4.4.4 Pad Cleaning
In order to make high quality, reliable wirebonds, the bonding pads must be clean. Many techniques have been tried over the years, but of all the methods, UV-ozone [86] and oxygen plasma [50] have proved to be the most effective in removing organic contamination. They are also effective against certain inor- ganic materials that form either a volatile oxide or, if not volatile, one that can be easily removed. While these techniques have been shown to remove a wide variety of contamination types, care must be exercised in their use. Because of the strong oxidizing environments present in O2plasma and UV-ozone reac- tors, metals such as silver, copper, and nickel may oxidize, and thus reduce their bondability. To reduce such effects in plasma reactors, argon is sometimes mixed with the oxygen. These oxygen-argon plasma cleaners are quite effective, Table 4.5 Wirebond pull strength for various thicknesses of autocatalytic gold plating over a nickel barrier (2.5mm thick) on a copper metallized printed wiring board
Pull Strengthb, grams (force) Gold Plating
Thickness,mm
Number of Bonds
NDPTa Failures
As Bonded
After 1508C Agingc
0.40 129 1 10.6 9.8
0.65 149 0 10.0 10.1
0.90 138 0 9.4 10.6
aNDPT¼non destructive pull test (at a 2.5 g (force) limit)
bSample sizes approximately 70 bonds. Standard deviations within10%.
c160 h (polymide-glass board material)
combining reactive ion cleaning with physical sputter etching. With any kind of plasma environment, there is a possibility of active circuit radiation damage.
Based on this author’s experience, this probability is extremely low for oxygen- argon plasma cleaners and should not be viewed as a deterrent to their use.
Similarly, because UV radiation can excite impurity states (color centers) in alumina-based ceramics, there is a tendency for white alumina ceramic sub- strates to appear yellow after UV-ozone treatment. The induced color change can be reversed by a subsequent thermal treatment. Table 4.6 and Fig. 4.14 show the effectiveness of UV-ozone cleaning (over solvent cleaning) in remov- ing intentional surface contamination.
Before leaving cleaning, a few comments about ultrasonic cleaning should be made. Historically, there have been several published reports (e.g., [71]), and much anecdotal conversation describing wirebond degradation or failure due to ultrasonic cleaning. Most of the reported incidents center on wirebonds in cavity type packages, such as those used for hybrids or hermetic single chip applications.
As with all mechanical structures, a wirebond has a resonant frequency which if excited will cause the wire to vibrate and in turn may cause fatigue and ultimate failure. The resonant frequency of a given diameter bonded wire is dependent on the length and height of the loop. For reasonable geometries and relatively short lengths (<2.5 mm) the resonant frequency of a typical wirebond is quite high (>30 kHz). Historically ultrasonic cleaners operated in the 20 kHz regime, and most of the reported damage occurred with long wire bonds (>2.5 mm) placed in large industrial cleaners (high energy). Thus, the ultrasonic cleaning of cavity type devices with short wires should be safe. Today, ultra- sonic cleaners span a broad frequency range from 20 kHz to over 100 kHz.
According to Harman ([38], p. 230), it is unlikely that high frequency ultrasonic cleaners (>50–60 kHz) will damage wirebonds.
Table 4.6 Average ball bond shear strength (grams (force)) for various cleaning treatments and thermal aging conditions for thermosonically bonded 25.4mm gold wire on 1mm thickness aluminum (on silicon). Average ball diameter was 90mm (3mm)
Sample Set Cleaning Conditions As Bonded Thermally Aged
A No cleana 50.9 (7.1) 47.8 (7.9)c
Plasma cleanb 52.2 (6.5) 52.1 (6.7)
No Clean 50.0 (6.2) 48.6 (7.1)d
B Contaminatede 38.9 (4.1) 40.3 (5.8)
Solvent clean 37.3 (6.1) 37.9 (7.3) Plasma cleanf 47.5 (6.0) 47.9 (6.7) UV-Ozone clean 53.0 (5.1) 54.2 (5.8)
aNo clean as received from substrate fabrication
bArgon-oxygen plasma (90% Ar, 10% O2)
cSample set A aged fro 96 h at 150oC
dSample set B aged for 168 h at 125oC
eContamination agents were photoresist and outgassing products of epoxy cure
fArgon-oxygen plasma (50% Ar, 50% O2)
With pin or ribbon leaded packages in which the pin or ribbon feeds directly inside the package to form the wirebond attachment point, special care needs to be taken to ensure that the external lead structure does not resonant. Resonance in these external leads can set up vibration on the pin or ribbon end inside the package and can cause wire or wirebond failure, especially if the wire is rela- tively stiff. This would be especially important when parts in hermetic quad flat packages are cleaned prior to board attachment.
With today’s fully encapsulated microcircuits, the cleaning of parts ultra- sonically poses little risk, especially for leadless or short leaded components.
The potential danger occurs when cleaning exposed wirebonds in open packages or in chip-on-board or flex applications. Another potential danger could be associated with microelectromechanical systems (MEMS) where ultrasonic resonance could cause mechanical failure of the MEMS struc- tures themselves in addition to the potential damage to wirebonds. Again, it is a question of the resonance frequency of the structure compared to the ultrasonic agitation frequency. In all cases with exposed wires and struc- tures, if ultrasonic cleaning methods are employed, cavitation should be avoided [38].
Fig. 4.14 Auger electron spectra of aluminum metallized silicon substrates both pre- and post cleaning with solvents and UV-ozone: (a) as processed substrate (uncontained); (b) substrate contaminated with photo resist; (c) substrate cleaned with solvent; and (d) substrate cleaned with UV-ozone