Non-Chemical What is on the inside of a CPU?

[Paige]

Is it worth crushing to get something out?

Paige

 

[goldsilverpro]

Much of the gold value will be under the chip, if it's attached with gold/silicon solder. I believe the solder makeup is 96Au/4Si. About the only thing that will dissolve it is true hot aqua regia. The acid can only dissolve the solder starting at the edges of the chip. It can literally take take days for the hot acid to completely undermine a large chip.

The main reason for breaking up the CPU is to break up the chip. This provides much more chip edge area for the acid to penetrate. It will still go slow. When the gold is gone, you'll be able to slide the pieces around on the pad.

We used to speed this up on all-gold side braze packages by heating the part up to the melting point of the solder, on a heating block, and removing the chip with a vacuum probe. A sharp pointed knife, like an Z-Acto will work. We then put chip and package in the aqua regia. I think the solder melts around 700-800 F.

There are several possible sources of gold on these parts. The percentages are educated guesses and only apply to all-gold parts - gold in every category below.

High value: 60%
- Gold(80%)/Tin(20%) eutectic braze used to attach the gold plated metal lid to the ring
- Gold(~96%)/Silicon(~4%) eutectic braze used to attach the chip to the pad

Medium value: 30%
- Gold plating on the legs
- Gold plating on the metal lid

Medium Low value: 10%
- Gold plating on the inside fingers
- Gold plating on the inside pad - the flat square area where the chip is is mounted
- Gold plating on the bottom of the chip
- Gold plating on the ring where the lid is attached

Very Low value: 2%
- Pure gold bonding wires going from the chip to the fingers - about a mile (1 mil dia.), or two (.7 mil dia.), of wire to the troy oz. This figures out to $.005 to $.01 per inch of wire.

 

[catfish]

GPS:

This is very informative, for just last night a newby posted questions about the gold in the ceramic of CPU's.

I, too have always wanted to know just what kind and how much gold was in a straight 386/486/pentium ceramic chips.

I wish we had some one to do a good test experment on just what the amount of gold was in a CPU chips of different types.

I hope tom341 sees this post.

Thanks again for sharring some of your outstanding research and knowledge.


Have a good day,

Catfish

 

[goldsilverpro]

Here are two great links on CPU's

http://www.cpu-collection.de/
http://cpu-museum.de/

 

[goldsilverpro]

When I ran CPU's, I received them mixed in 30 gallon barrels. The average value for the all gold ones, at about a $300 market, was $28/#. If I remember right, there were about 20 pieces/#.

The all gold side braze packages are quite valuable. They come in purple or white. With all gold lids, they can run $250/#, for the smaller ones. The common 40 leads can run $125/#.

At the bottom - unsealed and untrimmed:
http://www.chelseatech.com/packages.htm

40 lead sealed and leads trimmed - about $125/#
http://www.cpu-collection.de/?tn=0&l0=md&l1=1974&l2=Motorola#XC6800B,EngineeringSample
http://www.cpu-collection.de/?tn=0&l0=md&l1=1975&l2=Intel#C8080A

18 lead - about $225/#
http://www.cpu-collection.de/?tn=0&l0=md&l1=1977&l2=Intel#C8008

If you run into these soldered on a board, you won't lose much by prying or cutting them off.

I knew a guy in Houston that had 30-40 CPU's, of different types, assayed. It cost him $100 per assay from a lab in California.

It's very difficult, but possible, to get all of the gold off of CPU's that have Au/Si under the chip. It takes real aqua regia, hot.