Merged Topics on Ceramic CPU Recovery with Yields

[cnbarr]

Cnbarr,

Your results support my theory that the gold is being trapped in the tungsten copper alloy top plate. Since normal pentium ceramic cpus do not have this integrated top plate heat spreader, the results are identical for crushed and uncrushed. Other ceramic cpus without integrated heat spreaders should not require crushing either: 286-486 ceramic types, motorolas, i960, AMD x86s, AMD K series, AMD athlon/duron ceramic series cpu, etc.

Your test leads me to believe the only cpus that require crushing (to approximately 1/4" sized pieces) are the ones with an integrated heat spreader on top (Cyrix x86, Pentium Pro, HP, etc.). A rule of thumb is that if it has a gold plated top, then it needs to be either crushed or processed in diluted nitric acid to remove the base metals (specifically the copper portion of the top plate) completely before processing for gold.

Nice work.

Steve

Thanks Steve,
I feel your theory to be correct as well, I will have to run a similar test with ones that have the integrated heat spreader and that should prove it.

As for the the other cpu's the only benefit I saw in my test between crushed and uncrushed, was the crushed fully processed in between 3-5hrs faster.

 

[lazersteve]

Why did the uncrushed cpus require more time in your opinion?

Steve

 

[cnbarr]

Hey Steve,
I think the uncrushed ones took longer due to slow acid penetration under the silicon chip, with the crushed ones the silicon chip had been virtually shattered allowing for a greater surface area for acid penetration thus allowing quicker dissolution of the gold underneath the chip. This is just my observation of the reaction, but I feel it is an accurate assessment.

 

[butcher]

I also think when crushed finer the fine gold wires inside the chip are exposed easier to the acids, gold salts are water soluble so once gold is dissolved settling can get most of the liquid water and HCl wash can get the rest of dissolved gold from the cumbersome ceramic powders.

I have not done CPU's for quite some time now, but when I did process them, I would just break to pieces and get as much gold as I could, then crush and reprocess the powder, I did not get much gold the second run, but I did get some.

That is a very interesting experiment.

 

[samuel-a]

cnbarr

I'm afraid your results are somewhat low for some reason.
It could be either your process, or, the fact that there's really only 4 grams in your 64 pieces.

In my videos about ceramic processors, i ran these exact cpu's, the video show a bit more then one lb (18 pieces), all in all i ran several kilos of them.
My avarage yield is 0.07g per one cpu's on the first run, that i only broke to 4-6 parts.
I still don't have exact numbers for the second crushing and leaching at the moment, but already mentioned that the second run was suprising.


p.s.
I see no reason to include gold plated metalic lids/heat spreader into the main lot, i do them separately.

 

[lazersteve]

Sam,

The cpus cnbarr ran had no gold top or bottom. I think you are talking about the bottom lids and not the top plate being removed since the Pentium model cpu he processed has no top plate. Additionally, the cpus (ceramic Pentium I) he ran typically do not have gold plated bottom lids in most case. Perhaps I misunderstood what type of cpus he processed.

This brings up a few extra points worth mentioning when it comes to cpu terminology in general.

• The top of the cpu is considered to be the side without the legs on it. This sounds obvious but is important as newer cpus are built using FC, or flip chip technology where the cpu die is actually flipped upside down. The point is that regardless of which way the cpu die is oriented to the overall cpu package with ceramic cpus, the top side has no contacts or legs that connect to directly to the cpu socket. The top of some ceramic cpus may have contact points that are external interconnects for flip chip dies, cache, decoupling capacitors, and voltage regulation surface mounted components.
  
• The top plate is an integrated heat spreader that is composed of a copper tungsten alloy in some ceramic cpu models (mentioned in post above) that is typically plated with gold externally. This heat plate is brazed to the top of the cpu via the top braze.
  
• The bottom of the cpu is considered the side of the cpu with the electrical contacts on it. These contacts may be legs or lands (aka pads) and are typically gold plated kovar or other base metal alloy.
  
• The top braze is a term for the weld that holds the top plate to the top of the ceramic housing. The jury is still out as to what the actual composition of this braze is and whether it contains any precious metals or not.
  
• The package of the cpu describes the physical arrangement of the cpu housing and it's external and internal connecting points. It also defines the number of external connection points (pads/legs) and the orientation of the die inside the cpu.
  
• The housing is the purple/gray ceramic piece that remains after the cpu is processed for values. It is composed of several layers of ceramic materials fused together with electrical traces connecting the legs/pads of the cpu to the die lands. These internal traces are most likely a Molybdenum paste that is applied to the separated layers of ceramic before the housing is fused into one piece.
  
• The die is the silicon wafer (aka chip) that is mounted inside the cpu housing. This wafer is mounted to the ceramic housing using a mixture of epoxy and gold alloy braze. If the wafer is not completely removed, broken, or chemically etched free of the housing the layer of gold alloy braze under it will not be recovered in the refining process. This is a primary area of lowered yields when processing this type of cpu.
  
• The lands or pads are the points where the bonding wires that connect the cpu die to the cpu housing attach. The same term is used for the external connections on the cpu where no legs are present. In all cases lands are typically gold plated.
  
• The bonding wires are the short jumper wires that extend from the cpu die to the cpu housing die lands. They can be exposed by removing the bottom plate (aka lid) from the bottom of the cpu housing. The purpose of these wires is to conduct electricity from the legs or pads of the cpu to the appropriate contact point on the cpu die. These wires are typically made of one of two metals. The gold colored bonding wires are high purity gold. The silver colored wires are almost always high purity aluminum. There are two main theories as to whether or not these wires extend into the internal layers of the ceramic package housing:
  a) that the wires do not extend through the housing structure to the legs. For this to hold true there must be a secondary method of conducting the electricity to/from the legs and die. Goldsilverpro has stated that he believes this connection within the ceramic to be some form of Molybdenum trace.
  b) that the bonding wires extend through the ceramic housing all the way to the legs. This is most likely not the case as no one (to my knowledge) has ever been able to produce a photograph at any magnification of the ceramic when broken or crushed that exposes these 'phantom' wires.
  
• The lid of the cpu is typically a cover on bottom of the cpu. This lid is typically gold plated kovar, high iron base metal alloy, or a thin ceramic material. The lid is also soldered to the bottom of the cpu housing with a silver colored solder. This solder alloy is believed to contain some precious metals, most likely silver if any.

You can search the forum for more discussions on the construction of cpus and check out the documents on my website.

Steve

 

[cnbarr]

I also think when crushed finer thefine gold wires inside the chip are exposed easier to the acids, gold salts are water soluble so once gold is dissolved settling can get most of the liquid water and HCl wash can get the rest of dissolved gold from the cumbersome ceramic powders.

Hey butcher, I am still in mental debate over there being fine wires in the ceramic body of the chip. I've broken and looked at several under a 600x microscope and haven't found any indication of fine wires, that's not to say there isn't any fine gold wires, I just haven't found any in the cpu body, yet.

cnbarr

I'm afraid your results are somewhat low for some reason.
It could be either your process, or, the fact that there's really only 4 grams in your 64 pieces.

In my videos about ceramic processors, i ran these exact cpu's, the video show a bit more then one lb (18 pieces), all in all i ran several kilos of them.
My avarage yield is 0.07g per one cpu's on the first run, that i only broke to 4-6 parts.
I still don't have exact numbers for the second crushing and leaching at the moment, but already mentioned that the second run was surprising.

Hey Sam, I felt like my yields were a little low as well, but my average per/cpu was not much different than yours. It worked out to be 0.0625g per/cpu, getting identical results from two concurrent batches makes feel like I did extract all values, but I will re-run them just for confirmation.

Sam,
The cpus cnbarr ran had no gold top or bottom. I think you are talking about the bottom lids and not the top plate being removed since the Pentium model cpu he processed has no top plate. Additionally, the cpus (ceramic Pentium I) he ran typically do not have gold plated bottom lids in most case. Perhaps I misunderstood what type of cpus he processed.

This brings up a few extra points worth mentioning when it comes to cpu terminology in general.

Hey steve, you are correct in the type of cpu I processed, I posted a picture of them in the contest I just ran, it is the last picture in the set http://www.goldrefiningforum.com/phpBB3/viewtopic.php?f=67&t=12921#p129261

Thank you for the terminology list, I feel at times there is much confusion in some of the differences in the terminology used.

 

[lazersteve]

The accuracy, battery life (if applicable), and/or calibration of either Sam's or Cnbarr's scales can be to blame for the slight variation in yields.

Steve

 

[samuel-a]

Hi Steve

This is great list, i really liked your explanation for each term, especially the phantom wires enigma, lol :mrgreen:
I'm sure many will benefit from it.

As to the terminology, I'd like to think i'm quite versed with it, so i think there wasn't any confusion by me. I just mentioned as a side note, to process gold plated lids OR heat spreaders (that are metallic based obviously) at seperate reaction vessel as i sensed some confusion in previous posts, sorry if i offended you somehow.


Me and cnbarr, generaly speaking, processed the same type of cpu's, now, i know that no batch is ever same (different; sub-models, year, manufacturer etc'...), but just like cnbarr "felt" the yield is low, i do too. My guts (and experience) tells me he have more gold to extract, but i could be wrong of course. I only seek his best interest.

This is the type of processors that i have processed and my yield data relates to, i believe they are similar to cnbarr's batch.

The accuracy, battery life (if applicable), and/or calibration of either Sam's or Cnbarr's scales can be to blame for the slight variation in yields.

I don't know what cnbarr is doing, i can only speak for my self...
I'm not using some fancy analytical laboratory scale, I have one that is one hundredth of a gram accurate (xxx.x1).
I'm calibrating my scacle every other day and always before doing 'analytical' work.
My scale is always allowed 30 minutes (at least) to adjust with ambient temp' before use and always located on a leveled plain.
I own a standart 100g weight, purchased at a reputable local laboratory supplies vendor.

Some will say it's an over kill, some will say it's not enough...
I took this picture for you steve:




p.s 1 -

The lid is also soldered to the bottom of the cpu housing with a silver colored solder. This solder alloy is believed to contain some precious metals, most likely silver if any.

I have confirmed that in many cases it is indeed silver based solder.

p.s 2 -

The top braze is a term for the weld that holds the top plate to the top of the ceramic housing. The jury is still out as to what the actual composition of this braze is and whether it contains any precious metals or not.

I think in either cases this solder should be considered as gold bearing, even if plain tin/lead solder was used, it most likely dissolved some gold form the surrounding plating while applied.

 

[lazersteve]

Sam,

Of course, I am not offended by your comments. We are all working towards the same goal here, so I would be a fool to take your remarks as inflammatory in any way.

You are being very through in your scale calibration and accuracy, but it had to be mentioned as no two scales are exactly in 'sync' with one another. There are many other sources of errors in weighing yields as you well know, some of which I mentioned in the earlier posts of this thread. The same scale in perfect calibration can give errors based upon external anomalies (eq: gravity, magnetic fields, gold purity, slag/flux, etc.)

I'm very confident in stating that the braze that holds the top plate on is not normal solder (tin/lead) as it's melting point appears to be way above that of conventional solders.

Cnbarr's test, regardless of the actual yields, show the cpus he tested yields the same amount of gold crushed or not. I feel his results will be very different when he attempts the same test on ceramic cpus with integrated gold plated heat spreaders. I do feel if he processes everything in dilute nitric to completion first, he may get nearly the same results for crushed and uncrushed even with the gold topped cpus.

Steve

 

[butcher]

the pins and the silicone chip have to be connected by the bonding wire, and as far as I can tell it is inside the ceramic case.

I guess I have been trying to leach Phantoms.

 

[lazersteve]

I agree with you Butcher, but is it a requirement that the electrical connection is made with gold wire?

Steve

 

[cnbarr]

Hi Steve

This is great list, i really liked your explanation for each term, especially the phantom wires enigma, lol :mrgreen:
I'm sure many will benefit from it.

As to the terminology, I'd like to think i'm quite versed with it, so i think there wasn't any confusion by me. I just mentioned as a side note, to process gold plated lids OR heat spreaders (that are metallic based obviously) at seperate reaction vessel as i sensed some confusion in previous posts, sorry if i offended you somehow.


Me and cnbarr, generaly speaking, processed the same type of cpu's, now, i know that no batch is ever same (different; sub-models, year, manufacturer etc'...), but just like cnbarr "felt" the yield is low, i do too. My guts (and experience) tells me he have more gold to extract, but i could be wrong of course. I only seek his best interest.

This is the type of processors that i have processed and my yield data relates to, i believe they are similar to cnbarr's batch.

Hey Sam,
I just wanted say that I had no derogatory intent with reference to terminology, I feel as though we were all on the same page so to speak.

As for the cpu's they are the same as the ones in your post, I am re-running them right now to see if I missed any gold, but I think the discrepancies in yields may be, as Steve pointed out, my scales. They are old, I do not have calibration weights for them, so the likelihood of them being off by a 0.1g is possible. Your yield data says that I should have got 4.48g from 64pcs, so if my scales are off by -0.1g, over 4g of weight would be a -0.4g discrepancy. Until I get new scales I will the 2nd run of the cpu's be the deciding factor.

Cnbarr's test, regardless of the actual yields, show the cpus he tested yields the same amount of gold crushed or not. I feel his results will be very different when he attempts the same test on ceramic cpus with integrated gold plated heat spreaders. I do feel if he processes everything in dilute nitric to completion first, he may get nearly the same results for crushed and uncrushed even with the gold topped cpus. Steve

Hey Steve,
I will be getting to this test by the weekend, I have about 3lbs of cyrix MII and cyrix 6x86 I am going to use for this. I quite excited for the results actually, I'll post more details as I have them.

And thank you both Steve and Sam for the healthy interaction, it these type of discussions that show how much meticulous detail must be payed attention too in this craft.

Chris

 

[vegaswinner]

Very fast settling white precipitate in hcl-cl reaction? What is it? (lead? silver? copper 1) Am I missing a step out?

Here's my process so far.

Broke 20 cpu's into 4/6 pieces exposing the silicon wafer and gold wires.
Dropped in AP and left to bubble for 2 weeks (all pins had been digested leaving just foils, some sunk to the bottom of beaker, some still attached to the ceramic but came off with a rub)
Poured off AP and rinsed with hcl until rinse was clear, 5 rinses.
Covered cpus, foils, everything in the beaker with 36% hcl and added 15% CL (Sodium hypochlorite from reagent supplier) in small increments.
Almost immediately after adding the first bit of CL the solution went very cloudy, this white stuff settled very quickly.
Continued to add CL until no more visible gold.
Left to sit for an hour then poured off AUCL3 through 3 filters, being careful not to pour off any white stuff into filter.
Nice clear aucl3 solution put to one side.
Added some water to the broke cpus and white stuff to see what would happen, this turned into a very milky solution and the white stuff was not settling (will check it in the morning)

What is it and have I missed something.

Thanks

 

[butcher]

White chloride salts

NaCl (table salt), dissolves in water (clear solution).

Cu(I)Cl (copper I chloride), dissolves in Hydrochloric acid (HCl), (blue to green to black solution depending on concentration and dilution).

PbCl2 (lead chloride), not very soluble in cold water, but is very soluble in boiling hot water; PbCl2 in hot water is clear, will precipitate when water is cold.

AgCl (silver chloride), insoluble in hot water, basically is insoluble in HCl, can be separated from the other chloride salt's above because of this properties, it is fluffy and takes time to settle in solution so if using techniques above to separate it from the other chloride salts, give it time to settle in these solutions before decanting solutions, silver chloride is white in acidic chloride solution, but if salts are washed well in water and let sit in sunlight they will change darker violet to gray black, light is reducing them to silver metal (principle used in photography).


NaClO sodium hypo-chlorite (bleach), is made chlorine gas and sodium hydroxide (NaOH) usually in a electrolytic cell operations of brine salt (forming chlorine and sodium hydroxide).
Cl2 gas + 2NaOH --> NaClO + NaCl + H2O
Notice salt (NaCl) and sodium hydroxide (NaOH) in the formula, these can be in our bleach also besides the NaClO, also when sodium hypo-chlorite is mixed with our hydrochloric acid solution we make by products such as listed below, (note also it is possible bleach can also lower acidity of our solution) (basically making salt out of our HCl acid),
NaClO + 2HCl --> Cl2 gas + NaCl + H2O
Notice salt is a byproduct.
If heated NaClO decomposes to sodium chlorate and salt,
NaClO --> NaClO3 + 2NaCl


So you will have or can have several byproducts from your reaction, salt some sodium chlorate (depending on condition) and also insoluble chloride salts of metals from your materials you processed (depending on your process methods it is possible to have un-dissolved gold hidden in these salts also).

Now you have the tools to figure out what that white salt is, table salt or some other metal chloride salts, (well actually table salt is also a metal chloride salt we eat).

Hope all your white salts are silver.

 

[vegaswinner]

Hope all your white salts are silver.

Me too... :lol:

I will get on on to separating these salts this week to see what I have.

Thanks butcher for the detailed response.

 

[johnnybreaux]

I'm not even a "newbie" yet (I haven't started). Just after reading your input; it would seem a bit more beneficial to recover the gold and make jewelry, trinkets, etc. and sell them for more. What do you think?

 

[butcher]

johnnybreaux,

Welcome to the forum.
Yes I think if a person could make nice jewelry the gold would be worth much more, the thing is a jeweler is not only an artist but someone who is a skilled mechanic in his craft, it can take years to learn recovery and refining, and years to become a jeweler, not that I am saying it cannot be done, as if a person worked hard he could learn both skills, the artist in us I think we are born with not something easily learned.

 

[glondor]

If you have the skills, Johnny, go for it.

 

[cnbarr]

I'm not even a "newbie" yet (I haven't started). Just after reading your input; it would seem a bit more beneficial to recover the gold and make jewelry, trinkets, etc. and sell them for more. What do you think?

First off Johnny I will say welcome to the forum, second, this topic is geared mostly towards the processes and recovery from various ceramic cpu's and the yields that can be acquired from them. Please do not take what I am saying in the wrong, I have no intent to be audacious or derogatory in any way.

That aside, I will say I have absolutely no experience in jewelry making, but if a jeweler has to pay little or no money for the metals they use I would say they would fair out better then most. Your intent in refining would all depend on your goals toward refining, many of us here refine for the pure passion of it, and many to stash away a little gold for a rainy day, or both. But if you can find a niche with jewelry making and refining of PM's, I'd say you would probably do quite well.