How to clean CPU pin corrosion?

[feipoa]

I have a few CPUs which came out of upgrade interposer modules. The gold pins on these CPUs have corrosion on them. One on particularly bad case, I bought the CPU upgrade kit new in a sealed the Gainbery upgrade box, so I know it hasn't been exposed to leaking motherboard batteries or numerous filthy fingers.

What causes this corrosion? Is there a chemical reaction between two dissimilar metals, e.g. from the mating interposer module? Is there some trace elements in the environment (air) which reacts with these dissimilar metals and causes corrosion?

How do I get rid of this corrosion?

The particular CPUs I am noticing this on are 1) IBM 5x86c-100HF, which was in a Gainbery VRM module, 2) Cyrix 5x86-133/4x.

The gold on the IBM 5x86c has some spots of blue and red all over the pins. Is this rust? How is it removed without damaging the chip's function?

The Cyrix 5x86-133 chip has more of a black film on some pins, which can be scrapped off with a toothpick and considerable effort, though there is always some I cannot remove. One of the pins on the Cyrix has this red corrosion to it, like on the IBM. The Cyrix, on the other hand, does not have any of the blue spotted corrosion. Please refer to the images.

I know on motherboards which had the leaking alkaline barell battery, that you are supposed to use something like acetic acid or citric acid to neautrilse the alkaline base of the battery's discharge. But in these cases, I do not believe they were ever in contact with leaking battery juices.

So any idea what is causing this and how to get rid of it? I am also enclosing a higher resolution photo from the IBM chip. And this photo is after I already tried to use a toothbrush with a) Isopropyl alcohol, b) electrical contact cleaner, and c) acetic acid (vinegar).

[bccwchan]

Corrosion particular happen to IBM 486 cpu, IBM made Nexgen cpu.

I therefore believe that IBM (OEM) has some problem in production process for them.

[CPUShack]

too thin of gold

[feipoa]

Is it speculation that IBM chips had too thin of gold, or is this known? And how would they make it thicker? When I worked at a semiconductor company, they would dip their parts into what looked like liquid gold, of course they also had a sputtering machine, but it didn't seem to get used much.

So did the CPU manufactures dip the pins in the "liquid gold"? And if so, did, perhaps, Cyrix dip theirs in something more viscus, or do double dips, whereas IBM did a single dip?

Even if the gold is too thin, something caused the corrosion, aside from a too thin layer of product (gold), I assume.

What is curious is that I bought two of these Gainbery modules, both were sealed. Its twin partner has no corrosion on the pins!

[Robev]

Most likely poor preparation of the pins, ie pins not clean, before gold coated.

[feipoa]

surface prep - that makes sense, though it seems like something of this nature would be tightly controlled. Was this issue more prevalent with IBM branded chips?

[CPUShack]

Most if not all of these packages are from a supplier, not IBM/Cyrix, IBM/Cyrix bough the packages from a package maker (such as NGK) and then install the dies etc

Very few semiconductor companies actually make their own packaging

I've seen the bad corrosion on Cyrix and ST branded chips too

[alexzu]

Ok, so the main question - how to clean it carefully without destroying anything ?
I have some prototypes with the same corrosion on gold problem, even some Pentiums with gold caps have corrosion around the cap (

[feipoa]

Does the supplier deposit the gold plating? Or would IBM have received non-deposited PGAs?

Indeed - how to get rid of the corrosion? I do have the feeling that, on the IBM chip posted above, that the corrosion is coming from below the gold. I say this because "scrubbing" with a toothpick on the IBM does not seem to help much like on the Cyrix chip. If I use metal to gently scrape, I easily reach the silver-coloured underneath of the pins, which kind of makes me think that the thickness of the gold itself has reacted and tarnished.

The Cyrix chip, on the other hand, does appear to have some black tarish substance "growing" on top of the gold. Washing with ISO, contact cleaner and vinegar helped some on the Cyrix chip, but not that much compared to using the toothpick. I think using acetone would be a bit too aggressive. I think there is epoxy on the die cap, which cold dissolve. Also, the printing on the die cap would likely dissolve.

[Robev]

I would just let the chip soak in methylated spirits for a day or 2 and the clean with a soft cloth of some kind

[feipoa]

A soft cloth through 168 pins? I doubt anyone on this forum has patience for that. Maybe a soft tooth brush though.

2 days? Will this soaking dissolve the surface printing? Loosen it? Cermaic is porous. Will the solvent penetrate the ceramic and either a) stain the ceramic, or b) damage the internals? Will the spirits get passed the sealant on the die cap? Has anyone tried this approach for the suggested interval?

[Robev]

Methylated Spirits is basically just alcohol with a bit of purple coloring in it.

It turns to a gas after a while and I have used it on a lot of Intel chips without any problems.
It is also great for softening hardened glue that was used to hold heatsinks on some Intel 486's etc.

It does not seem to have any affect on the print on the surface of the Intel chips I have used it on.

The other thing you could try is a Jewelry Cleaning Solution.
I have not tried it but I think it would have the same affect.

A Soft Toothbrush would be OK to clean the pins but I would not use it on any of the other chip surfaces.

[feipoa]

I have some jewellery cleaner. I will try this on the IBM. The Cyrix is too rare for such discovery.

[alexzu]

Please keep us updated ! )

[feipoa]

I scrubbed, soaked, scrubbed, and soaked with the jewelry cleaner, but the discolouration remains the same. I suspect the the corrosion as reacted with the gold such that there is no more gold remaining in these locations.