I have lots of experience with thermosonic gold ball bonding. You can read the literature and you will probably find some papers that say you can do it. I have never been able to do it. We gold ball bond all day long with little problems at all but this is to soft gold over nickel. The gold is 25 micro inches and the nickel is about 200 micro inches. I can tell you that if I have any problems with stitch lifts or nonsticking stitches, I XRF (xray fluorescence)the PCB and the gold is thin at about 3 to 6 micro inches. Why? My PCB supplier forgot to selectively plate up the wire bond areas. I have never been able to thermosonic gold ball bond to ENIG. Have tried this with Palomar, Hughes, and Panasonic bonders. No luck!
Here are a few suggestions.
For prototypes, you might be able to do make a ball bond to the substrate. This will leave a little gold ball on the stitch pad. Then you can go back and ball bond to the die and stitch bond to the ball stuck to the substrate pad previously. The idea is that the ball bond will stick to ENIG better than a stitch bond and you are building up the substrate gold thickness using a ball bond placed on the stitch pad.
Reverse your bonding. Ball bond to the substrate and stitch bond to the die.
Use MacDermid Sterling Silver plating instead of ENIG. It works for me. You can gold ball bond to it. I can anyway. Not nearly as good as thick gold but it works well with little rework. One caution here. The PCB has to be clean and stored properly so it does not oxidize. The advantage is that you can solder to it. This is not the case with thick gold due to the gold embrittlement issue.
The process that works the best is to use selectively plated PCB. Adds about 15% to the cost of a PCB plated with thick gold. This is what my supplier charges me. The soldered areas are 3 to 6 micro inches of flash gold over 200 micro inches of nickel and the wire bonded areas are 25 micro inches of soft gold over 200 micro inches of nickel.
Good luck,
Chris
reply »