5

As a partial reparation for my previous question today.
Please allow this rather esoteric/ but on topic question.
I was sending out a pcb that will have gold wire bonding pads.
It was a option at our board house. I got a call when I was requesting a quote.

"How thick do you want the soft gold bonding layer?"
"I have no idea." I said, which is my normal response.
"What's your "standard" thickness.?", I asked.
I was shifted to engineering. 30-50 micro inches was the answer, after a nice discussion.
(It's always a pleasure to talk with the engineer's.)
So that's what I ordered.
I could order a thicker layer.
Has anyone done this?
I'll be wire bonding soon, so I can let you know if it works.

Edit: Well I got the quote back and my jaw dropped. The price is more than $2k for a few pcb's. For that cost I could evaporate my own Ni and gold layers.

Edit 2: (Adding more to the question.)
So the reason for the high price is that the board house needed to send it outside for processing. I asked for quotes on their other types of Gold finishes.

The first is called Deep Gold. This is an electrolytic process and puts down 30 u inches of gold. (I don't know if there is any underlying Nickel.) This is also called hard gold and from limited reading I don't think it works for wire bonding.

The second is Immersion Gold. or ENIG and puts down 3-10 u in (75-250 nm) of gold over a thicker nickel layer. Again from limited reading it seems that one can not wire bond to this with gold. But maybe with aluminum. (Though there can problems with the Al/Au interface.)

And my final crazy idea is to make a mask for my pcb and evaporate my own Ni/Au layer.

Community
  • 1
George Herold
  • 4,754
  • 1
  • 18
  • 25
  • @mkeith, thanks, I think I'm getting gold over copper. I've got an expert showing me how to do the wire bonding... hey I should ask him how thick! – George Herold Nov 12 '14 at 01:22
  • 2
    Gold over copper will corrode. The gold will peel right off the copper after a couple weeks. Guess how I know. – The Photon Nov 12 '14 at 02:33
  • @ThePhoton, Interesting. Did you have a board house make it for you? I'm still waiting for a firm quote from advanced circuits, I'll ask them if they do a nickle layer in between. (Certainly they must know how to do it.) – George Herold Nov 12 '14 at 13:19
  • Are you ball bonding? (Au) or wedge bonding (Al)? – placeholder Nov 12 '14 at 17:00
  • 1
    @GeorgeHerold There's a place in PA that offers training in wire bonding, next course is in early Dec. I think all gold over copper *requires* a barrier layer of some kind (usually Ni, but I guess other things would work). Unfortunately Ni is a bit magnetic so that that doesn't always work in sensitive applications. – Spehro Pefhany Nov 12 '14 at 17:01
  • @placeholder, (as is usual) my answer is I have no idea. I was hoping to just do this and not have to learn all about it. (but now that looks like a fools dream.) I just ordered the 2nd ed,. of "Wire Bonding in Microelectronics" – George Herold Nov 12 '14 at 18:02
  • @SpehroPefhany, Thanks, I'll be doing this at the CNF (Cornell nano-fab) and there is a wire bonding expert there. I talked with him a bit a few months ago, and just sent him an email asking about options. – George Herold Nov 12 '14 at 18:04
  • $2K? Wow. Hard to imagine how the makers of LCD modules, toys etc. can do COB if they are paying prices like that. ;-) – Spehro Pefhany Nov 12 '14 at 18:14
  • @SpehroPefhany, well Adv. Circuits has to send it outside for processing, (which explains the price.) As the Photon said below there seems to be a big push to reduce the amount of gold (for obvious cost reasons.) And there are newer processes using palladium. – George Herold Nov 12 '14 at 19:52
  • @GeorgeHerold, If you were asking for a quick turnaround (like 5 days or less), that will interact with the external process to bump the price up. Say you requested 5-day turn with gold plating. If they need 2 days to turn the plating around with the external vendor, then they have to treat the rest of the build like a 3-day turn. – The Photon Nov 12 '14 at 23:12
  • @ThePhoton, It was still over $2k at 4 weeks. – George Herold Nov 13 '14 at 01:28

2 Answers2

2

According to documents I have from a PCB vendor, a typical spec for wire-bondable electrolytic soft gold is and 1.97 \$\mu\$in (min) gold over 188 \$\mu\$in (min) nickel. These are called out as IPC minimum values, though I don't have the IPC documents in front of me.

ENEPIG (electroless nickel, electroless palladium, immersion gold) plating can also produce a wire-bondable surface. A typical spec for that is 197 \$\mu\$in nickel, 12 \$\mu\$in palladium and 1.1 \$\mu\$in gold. Again these are all minimum values.

Recently I had success with a design that spec'ed the gold thickness as 25-30 \$\mu\$in, but really you don't want a very thick gold layer because excessive gold does bad things to solder joint reliability.

The Photon
  • 126,425
  • 3
  • 159
  • 304
  • Thanks, I've spent the morning reading all about ENEPIG and all sorts of other wire bonding stuff. The soft gold thickness was 30 u in. (1 u in = 25nm) the reason for the large price in the quote was because they needed to send it out for processing. Does the place you used do small proto-type quantities? OK I'm stopping here and just editing my question. – George Herold Nov 12 '14 at 18:12
  • Most PCB shops, afaik, will send boards out to a plating shop for gold plating. This might be related to the environmental/safety issues with some of the chemicals involved in plating processes. Certain processes might be easier to get done in-house than others. If you visit chat I could drop names of a couple of shops I've used. – The Photon Nov 12 '14 at 20:15
0

Wow, I want to answer my own question.
Well, it's not my answer at all. Credit goes to JL, the first response here.
And then while chatting 'The Photon' found the following link. Which looks just about perfect, for prototype wire bonding. (I hope someone will give him another +1 for me.)

George Herold
  • 4,754
  • 1
  • 18
  • 25