Brown fiber CPUs - how to proceed?

[resabed01]

I have about 25 (600g) of these brown fiber CPUs that were pulled from Sun servers. They have a gold plated copper heat spreader and, of course, gold bond wires. Lots of gold plating inside where the chip hides as you can see where I pulled off the ceramic lid. Wondering what would be the best way to proceed with these. Note they were desoldered from the motherboards so there is plenty of tin on those legs.

Thanks.

 

[denim]

Perhaps you could soak them in HCL (dilute 3:1) with H20 for a few days to get rid of the tin. Then rinse well and give them a nitric leach hoping the wires would come off with the leg foils, then go on to dissolving with AR or HCL/CL. I'm sure you what to do after that. I could be wrong on the pin type, they may have tin in them.

Best of luck to ya-

Dennis

 

[g_axelsson]

If there were ever any gold on those legs I think it is long gone by now. Any gold plate on the pins would have been dissolved in the solder bath and is mixed up in any solder joint on the boards that you got the CPU:s from. The gold is dissolved by the wave solder machine and mixed with the tin in the machine.
The only gold that is practically recoverable is possible gold braze between the chip and the heat spreader, gold plate on the lid and heatspreader and in gold bonding wires.

I would try to cut out the central cavity without any pins and then go from there.

Göran

 

[resabed01]

Forgot to mention, I checked the legs with a magnet, they are Kovar or some kind of magnetic alloy. The legs themselves were never plated with gold although, the holes the legs are inserted into are plated through with gold. You can't really see it in the first picture but it's there visible on the top. I thought about using the shear to cut off all 4 sides to eliminate the majority of the base metals but I would be sacrificing some gold and possibly quite a bit of silver in the solder.
I'm thinking a series of HCL baths to try and eliminate as much of the tin and base metals first.
Thanks for the suggestions guys.

 

[Alentia]

You may try to simmer it in NaOH for couple of weeks to convert all tin into Stannous Hydroxide. Use Stainless Steel bowl, not glass vessel.

If pins are magnetic and there is visible gold plating on them, simmer in HCL for few days or more. If no gold visible on pins, cut out centers as Goran recommended.

If pins are non-magnetic you can go straight to HNO3 treatment.

Complete with AR and precipitate.

 

[g_axelsson]

I still think you should cut the center parts out if you can.

You have two different sorts of scrap here, gold plated + bond wires and flash plated gold dissolved in tin with a few percent silver. To mix it in a rather small batch is just to make things complicated.

To try to get to the silver and gold with the pins I would treat it in HCl, it will dissolve the tin and kovar, the silver will be converted to silver chloride (I think) and the gold that is mixed up in the tin will drop as a very fine sediment or being locked in with the silver chloride.
Then I would decant the dissolved kovar / tin mixture and also removing the left over circuit boards, it will still contain copper buried in the PCB and would affect the process later. You have to check that all holes are free of any sediment or left over metals from the pins.
Now you will have a sludge of silver chloride and microscopic gold particles and probably a lot of other metal oxides. Next step is to turn the silver chloride into metal and digest it with nitric, decant / filter and cement on copper. If some gold goes into solution it will follow the silver and cement with it. Any remaining material could be treated in AR to get gold out of it.

... but that is how I would do it if I had to, I'm not totally convinced it is economically to do it on such a small batch. The silver is just a few percent of the tin and gold is only flash plated, it is only there to protect the surface before the pins are soldered, not a mechanical protection for wear as in a contact surface. It will be a lot of work for very little return.

Disclaimer : I have never processed these type of chips so my procedure is purely theoretical and could contain fatal errors and could be improved in many ways. Read and use at your own risk, think it through for your self. :mrgreen:

Göran