Bruce,
This all depends on what the end user is using that surface for.
For example on thin PCBs that will be soldered to the nickel thknss
could be much thinner than if something would require intimate contact
electrically. Generally, the immersion gold surface is used for
soldering to. However, some fabricators are driving the gold thknss
above typical (from 2-10 u inches to 5-20 u inches) and are having
somewhat reasonable success in wire bonding. While many fabricators
shoot for EL Ni thickness' of 200 to 250 u inches, some RF and
microwave designers are finding the nickel to be too lossy (DB loss)
and have tested and liked the electrical performance of 50-100 u
inches of EL Ni better. I think that the PCB application will probably
dictate what that thickness will be.
dhh
______________________________ Reply Separator _________________________________
Subject: Re: Nickel Diffusion Barrier
Author: [log in to unmask] at SMTPLINK-HADCO
Date: 9/28/95 1:21 PM
Anybody has any idea of what should be the minimum nickel thickness & gold
thickness requirements for electroless nickel & immersion gold process?
B.H.Lee
Process Engineer
---------- Forwarded message ----------
Date: Wed, 27 Sep 1995 14:57:00 -0500
From: Jerry Cupples <[log in to unmask]>
To: Seppanen Gordy <[log in to unmask]>
Cc: [log in to unmask]
Subject: Re: Nickel Diffusion Barrier
Gordy, you said:
>Are you familiar with the AESF report from approximately 1976 titled
>"Interdiffusion of Electrodeposits with Basis Metals"? It is a thesis from
>Pensylvania State University. My copy has four pages missing, although I am
>not sure why.
Nope, missed that one. I didn't mean to say I had done any exhaustive
literature search, although I have patents on codeposition (Ti/W, Cu, Sn)
with sputtering to make metal pads on IC metallization (from my IC
packaging days at Mostek), and am familiar with some related literature
dating back about 10 years. We are talking about interstitial solid phase
solutions and diffusion, just generally.
>The gist of the report is that nickel is an ideal Cu/Au diffusion barrier, but
>the thickness should be at least one micron and in some cases two to prevent
>diffusion.
That would be 40-80 microinches, right? So the IPC-RB-276 requirement is
200 microinces, or 5 microns. My point is that I've read authoritative
papers (sorry I can't cite any) saying 50 uin of Ni is enough. I certainly
agree that the metallurgy of Cu/Ni/Au coatings is quite sound.
>The testing was at 100#161#C and 175#161#C so it is reletively low. At the
>extreme of the study at 100#161#C after 500 days they observed Cu on the
>surface of
>the Au.
My mail reader shows 100(poundsign)161(poundsign)C here. I guess you mean
100 to 161 degrees Celsius. I'd guess at that temperature for 500 days the
diffusion would be maybe 100X accelerated. We need to see what Mr. Ahrenius
has to say.
>Honeywell in the mid 70's experienced this the hard way on some
>connectors of torpedos that were stored on sea water craft and began to note
>the signs of green Cu on the Au connector surfaces. We also have evidence that
>suggests the diffusion occurs in a hemetic atmosphere and that if the
>difference between the Au and Cu surface resistance is crucial to operation,
>failures can occur without a Ni barrier.
Yeah. I have heard stories about Harpoon missiles on ship deck launchers
which came back to TI looking like they had been to the moon and stuck in
the cheese. But I am concerned only with meeting IPC Class 2 conditions. My
own specs here call for 100 microinches (2.5 microns) of nickel under 30
microinches (0.75 microns) of gold. This is a commercial computer network
interface controller product, which should normally live in an office
environment.
It's interesting to hear what has happened in those extreme environments,
though. The Bellcore standards are usually designed around CO or vault
conditions, which can be warm and have battery acid in the vicinity...
cheers,
Jerry Cupples
Interphase Corporation
Dallas, TX
http://www.iphase.com
|
