XRF analysis gold plated SPARC VI chips

[nlvince]

Hi everyone,

Since I was curious on the gold content of the fully gold plated SPARC VI chips I decided to have them analyzed.
The gold content doesn't show in the analysis as can be seen in the attachment.

The result was far from satisfying for me as well as for the guy who owns the XRF, so I gave him four pieces to let a professional refiner have a look at them and determine the contents.

When I have those results I will update the post.

Vincent

 

[cosmetal]

ENIG plate? All the gold is hiding in the wire bonds.

The color doesn't look right for hard plate (your photo off-color?). Also. a lot of scratches.

The old prospector told the new greenhorn "If you thinks it's gold . . . it ain't."

 

[glorycloud]

The old prospector told the new greenhorn "If you thinks it's gold . . . it ain't."

:lol: :lol: :lol: :lol:

Indeed!! The e-scrap dreamers like I used to be now know this all to well.
Well, at least in the vast amounts of gold that we thought it should be as
we drooled over our "golden" e-scrap delights. 8)

 

[nlvince]

Here's a better picture without use of the flash

 

[anachronism]

ENIG plate? All the gold is hiding in the wire bonds.

The color doesn't look right for hard plate (your photo off-color?). Also. a lot of scratches.

The old prospector told the new greenhorn "If you thinks it's gold . . . it ain't."

Which wire bonds are you referring to?

 

[cosmetal]

http://spectrum.ieee.org/tech-history/silicon-revolution/chip-hall-of-fame-sun-microsystems-sparc-processor

Internal gold wires attaching the silicon die to the CPU leadframe.

James

 

[anachronism]

Do you have a photo of one of YOUR chips with the heatsink removed?

 

[nlvince]

I will remove one of the heatsinks tomorrow and then post a picture. I have a total of around 280 pieces so I hope there's some gold hiding underneath the heatsink :G

Here's a video that also shows the chips still attached to the board:

https://youtu.be/EoznUpL8p5M?t=14m43s

 

[g_axelsson]

Nice scrap, is it your video?

Looking at the pictures and the movie it looks like there is a fiber board below the heat spreader. Then it probably is a flip-chip design, like any P4 cpu. From the XRF results the heatspreader looks like it is copper base with heavy nickel plating and gold flash.
With a flip-chip there are no gold bond wires.

As has been stated before, thin plating is often mirror finish while thick gold plate has a more dull appearance.

Best way to get gold from the CPU:s is to sell them to collectors and buy gold for the money. :wink:

Göran

 

[nlvince]

Here's a picture with the heat sink removed.

@ Göran: Yeah that's also what I thought from the results. It's not my video but I noticed that those boards have the same chips on them.

I'll let the experts here decide wether it's a flip-chip or not.

 

[g_axelsson]

It looks like the base is made from some ceramic material, but the fact that the chip is soldered (probably indium solder) to the heat spreader means there are no bond wires.

Sell them to collectors, at $5 each you only need to sell two to get more worth from them than the total gold is worth. CPU-world.com is a good site for that or ebay.
Either that or selling them to boardsort or similar company.

Göran

 

[cosmetal]

nlvince,

Just curious, what prompted you to identify these as SPARC CPUs in the first place?

James

 

[goldsilverpro]

Any gold XRF result on gold plated materials is meaningless.

 

[nlvince]

nlvince,

Just curious, what prompted you to identify these as SPARC CPUs in the first place?

James

They are described as SPARC in this thread: http://www.theregister.co.uk/2010/12/03/oracle_sparct4_fujitsu_sparc64/

And I couldn't find any other info on them so I presumed that they were indeed SPARC CPUs.

Any gold XRF result on gold plated materials is meaningless.

Is this because the peaks from the base metals are so large that the smaller ones from the plating are measured as noise in the baseline?

It looks like the base is made from some ceramic material, but the fact that the chip is soldered (probably indium solder) to the heat spreader means there are no bond wires.

Sell them to collectors, at $5 each you only need to sell two to get more worth from them than the total gold is worth. CPU-world.com is a good site for that or ebay.
Either that or selling them to boardsort or similar company.

Göran

Thanks for the detailed information. I've decided to throw 10 in HCl and leave them there for a while. Should this result in extremely low yield as you mentioned I'll sell them to collectors. But this is just as a confirmation for myself. When I have the results I will post the results here.

Vincent

 

[g_axelsson]

XRF works by exiting atoms with X-rays and then analyses the characteristic X-ray spectra it gives off. Every atom has it's own spectra.
X-rays penetrate a bit into an object so you see signals from both deep down and the surface (it's called interaction volume). Most will come from the surface and then less from deeper down. Different alloys absorbs X-rays differently so there are quite a lot of calculations to get it correct.
One of the assumptions is that the sample is homogeneous, an alloy for example or a well mixed powder. Making measurements on a thin plating will mess up the calculations but usually it gives too large numbers for the plating and too little for the underlying metals. In any case it is messing up the results. To get a proper measurement you need to melt the heat spreader, mix it well and then grind a flat fresh surface without any oxidation on it.

Come to think about it, did he polish or grind the surface before measuring it? Maybe the thin gold was rubbed off in the process?

Göran

 

[g_axelsson]

Digging through old posts on CPU-world and googling makes it quite clear that this isn't a CPU but a chipset, probably for the SPARC 64VII, so the collector interest is probably not very high.

http://www.cpu-world.com/forum/viewtopic.php?t=28307

Göran

 

[nlvince]

XRF works by exiting atoms with X-rays and then analyses the characteristic X-ray spectra it gives off. Every atom has it's own spectra.
X-rays penetrate a bit into an object so you see signals from both deep down and the surface (it's called interaction volume). Most will come from the surface and then less from deeper down. Different alloys absorbs X-rays differently so there are quite a lot of calculations to get it correct.
One of the assumptions is that the sample is homogeneous, an alloy for example or a well mixed powder. Making measurements on a thin plating will mess up the calculations but usually it gives too large numbers for the plating and too little for the underlying metals. In any case it is messing up the results. To get a proper measurement you need to melt the heat spreader, mix it well and then grind a flat fresh surface without any oxidation on it.

Come to think about it, did he polish or grind the surface before measuring it? Maybe the thin gold was rubbed off in the process?

Göran

The chip wasn't melted or grinded before the measurement. So the sample was in no way homogeneous.
Also, one night in about 10% HCl gave a complete removal of the gold color leaving a grey surface. The weird thing about it is that the letters are still present. The liquid didn't change color and stannous is negative.
This tells me that refining has no use for these chips (as I was told but I still had to check).

Thanks Göran for the information!

Vincent