Phil,
     I agree with the "possible but unreliable" view. I did try gold ball
bonding to nickel/immersion gold a few years ago. I couldn't get a consistant
process and eventually redesigned the substrate for thick electrolytic gold.
Fortunately the board vendor had that capability. The only people I have spoken
to who have claimed to succesfully do it had yields and second bond strength
variability that I would find unacceptable for a reliable high volume, or high
wire count product and would certainly be for a 700 wire MCM. The situation may
have changed since. Wire bonders have evolved since I last worked on automatic
machines. In particular 120KHz transducers are in use. This may have changed
things. Talk to bonder manufacturers about the application. I have found this to
be worthwhile in the past. These days the machine manufacturers sell a process
as well as equipment.  They have a wealth of experience with solving customers
bonding problems and know what works and what doesn't.

     One possibility; could the board vendor subcontract the wire bondable gold
deposition? I have done this in the past with succcess, but the companies I know
are in the UK. Electrolytic gold plating is quite aggresive, so a suitable
photoimagable etch resist is required. This is laminated to the copper outer
board surface, exposed with the pattern and developed to expose the copper where
the tracks and pads will be. It is then nickel gold plated. The resist is
stripped and the board is passed through copper etch. The nickel and gold act as
an etch resist, keeping the copper in the regions where it is desired. If
regions of thick gold and immersion gold, or other finish is needed for
soldering, then that also is possible, but with more process steps and expense.

regards,  Jeremy






Phil Hersey <[log in to unmask]> on 29/01/99 21:42:47

Please respond to Phil Hersey <[log in to unmask]>

To:   [log in to unmask]
cc:    (bcc: Jeremy J Drake/HQ/CUK/Celestica)

Subject:  Re: [TN] seeking aluminum wedge bonding advice




>Phil,
>     For aluminium wedge bonders try K&S, Palomar, or Delvotek. Wedge bonding
to
>nickel, immersion gold is a reasonable process. I have not tried it over three
>layers of photo imagable dielectric though, but can see no fundermental reason
>why it shouldn't work. Bonding takes place at room temperature, so the Tg of
the
>material isn't so important for bonding.

Have you tried gold ball bonding to copper/nickel/flash gold?  One vendor is
telling me that gold ball bonding to copper/flash silver (4u") works just fine.
I'm skeptical.  I only know what I run across in traderags etc... but it seems
to me that the consensus is that while gold ball to immersion flash gold is
possible, it is somehow unreliable or inconsistent for MCM purposes (my MCM has
700 wires in a 1.5x1.5" area to all kinds of different devices).  If it isn't
reliable for gold, I can't see how flash silver would be any better!?  Shoot, I
have enough problems consistently bonding to thickfilm gold!

A three layer built up board is going
>to limit your choice of suppliers at the moment, but you may be able to find
one
>who can give you a gold bondable surface.

MicroVia in Florida can do 3 microvia layers, but they can't put on soft gold
right now, only thin immersion flash.

Jeremy, thanks for the bonder edu!

Resp.
Phil Hersey, Hytek