Selective grinding

[KreAture]

I just got my industrial diamond grinding wheel (100mm) to try out some selective grinding on ram chips that I wanted to test. I did this once already with a 180 grit wheel but found the resulting pulp to have very fine particles.

This wheel is 45grit so very rough. This also makes it very fast.
The reason for doing this is to selectively grind off the top middle of every chip on some ram sticks with the TSOP style packages and see if they hold much more than the BGA ones without having to do a lot of depopulating. The idea is really to later do a grinder using one or two round wheels and grind the packages to powder with the module as carrier. No depopulation needed. The process can also take off the top of the PCB on both sides to grind off all the gold fingers as well. Just to eliminate cutting of fingers.

Plan is to do a quick sift of the sludge and remove the largest chunks and any capacitors/resistors that may end up in there before doing a quick gravity separation. I can remove quite a lot of the copper and most of the molded packaging without loosing gold. May have to use 2 stages of mesh to get this really good, but it may not be necessary.

Then a quick (since it is dust) [stt]AR[/stt]AP process to remove base metals before a HCl/Cl gas process takes all the gold out so it can be dropped cleanly.

The manual grinding to test this was really quick on some sticks I had but a automated machine would be a big help. I'd hate to do this on kilos of sticks manually.
The 45grit wheel left nice chunky gold vs the 180grit and it was easy to pan out to have a look.

I've been searching the forum but not found much on this.
Seems like burning and ball milling is more preferred?

 

[kernels]

You can't be serious that the 45 grit wheel left you with "nice chunky Gold" ?!? All the Gold inside ICs are in the bond wires, which are practically invisible to the naked eye ? Also, you don't want AR to get rid of the base metals, it will also dissolve Gold. Also, how are you working with Chlorine gas, I have been in some pretty advanced labs where they decided it was too dangerous and difficult to work with Chlorine gas directly.

 

[KreAture]

Woops misspelled. Meant AP.
I am testing different setups for Cl gas with scrubber after reaction vessel to avoid unnecessary venting of gas.
The gas is created anyways if you use any Cl method instead of proper AR but atleast with the gas method I am not creating unwanted salts.

When I say nice and chunky I mean compared to the 180 grit. Everything was -200 with the 180 grit. Very hard to gravity separate. With the 45grit it seems a lot of the bond wires are ripped loose in larger pieces at least 5-10x as long as their diameter. Will take some pictures in the microscope tomorrow. They definitely separated easier and the grinding was very quick.

 

[KreAture]

Here is a quick picture of what the gold from the selective grind looks like after a quick panning.

I added my favorite piece of 0.4mm enameled copper wire as scale.
Compared to the dust I got with the 180grit wheel this is more shreds.

You can also see some plastics and resin pieces in there as well as a lot of copper.
Idea is to "acid wash" the copper and any other metals away before dissolving the gold.
I won't be doing this fine panning for larger runs though. This was mostly to see what was in there without all the black encapsulation material.

 

[kernels]

Cool, so ordinarily I would say that panning bond wires is impossible and you lose about as many as you retain, but in this case, because your grinding breaks them up into quite small bits, they probably settle better with panning than the long spindly ones do.

When you pyrolize before mechanically breaking up the chips, you get very long spindly bond wires that definitely do not settle in a pan. Have you thought about how you intend to gravity separate them in bulk ?

 

[FrugalRefiner]

Now, to test your methodology, you need to examine and test all the other byproducts of your process. I'm not being critical. Just saying you need to check all the other "stuff" you pan off to make sure you're capturing a satisfactorily high enough percentage of the available gold to justify the process.

There could be some gold still tied up with the plastics and other metals. From a given sample, refine the gold from what you're panning. Then process all the other materials and compare the results.

It's an interesting approach! Keep us updated as you "refine" your process.

Dave

 

[jarlowski1]

I am curious to see how this pans out (pun intended) :lol: I currently have about 30 pounds of chips I have been kicking around. I am concerned though 2 things happen when you grind on material 1 you create dust in the air so adequate ventilation and a respirator would be required. 2 with all that dust goes your gold too. So how are you preventing these from happening?

 

[Yggdrasil]

Just to clearify this a bit.
If I understand it correctly, this is done on the fingers only, not the chips itself?
So the spindly long bond wires do not apply here.
Correct me if I'm wrong.

Still, it may be better just to make sure one has clean tight cut fingers, or at least just split off the layer attached to the fingers.
It seems to be a high risk of loosing a significant part of your gold this way.

 

[g_axelsson]

Are you sure it is gold and not brass? A drop of nitric would remove any copper or brass from the concentrate and answer that question.
What type of chips are you grinding?

Göran

 

[Yggdrasil]

Oops.
I reread the original post, I'm the one that misunderstood the post.
I stand corrected

Edited for misspelling.

 

[KreAture]

I have only visually inspected and re-panned the stuff I pan off so far but the plan is to run it in a metal leech first and then do the HCl+Cl method on that as well as the concentrate just to verify my method.
I did find a few very small specs when repanning so some gold turns into so fine powder that it is difficult to pan off. I still think keeping a bit more of the waste will allow a mechanical gravity separation to work "easily".

The grinding is done wet, not dry so as to reduce or eliminate dust. It also cools the grinding wheels allowing minimal wear. (Not that it is needed for a manual test, but for industrial approach I think it would be mandatory. It would create a lot more heat than beating it in a mill.)

I did two separate tests, and the picture is of the "assumed gold" from the SSOP package covered old style RAM sticks. The chips are ground off the pcb without grinding any of the PCB surface (as I had already cut off gold fingers and just wanted to test the process on the chips). The solder and pads as well as most of the legs are left on the PCB's. It looks quite funky. These chips have the leadframe on top of the die so the grind has to be a bit deep. Some systems have the chip ontop of leadframe and can be ground less. The latter leaves entire chip on ram stick including legs and solder but I have only seen one like that a year ago and none in the sticks I tested now.
The second test was of some 2-sided sticks with BGA's on, and the gold from that is very fine dust.
I guess the SSOP packages have coarser bond wires.

For large scale separation I think the best way would be a shaker table. The fine pulp would, after sifting, be well suited to very good separation on one of those tables demonstrated by MBMM in their videos.

My reasoning for doing the grind and rough separation is really to give the CuCl2 some area to work on instead of the slow leech. The gold dust will also dissolve more easily in the Cl bubbler setup as there is more contact area and the bubbles will easily agitate the solution keeping particles in suspension. (At least in my head.)

 

[kernels]

I would love some pics of the grinding setup, sounds very cool.

How long would it take to ground one ram module down ?

If you are already wet-grinding, then it doesn't sound like a bad plan to pipe the outlet into a sluice box, you then get initial concentration basically free. You would probably just have to run some surfactant, I suspect the ground epoxy will be hard to wet.

 

[jarlowski1]

It does sound very cool! I'm sure it will be a great system once all the kinks are worked out. It may have been mentioned already but I was thinking that some wire may still be stuck in the epoxy bits. What your thoughts on recovering that?

 

[KreAture]

From what I can see the epoxy bits are just mud. The gold seems to brake loose from it quite easily.
I suspect the wetting energy in epoxy does not extend to noble metals.

I'll have a look and see if I can get some separate pics of the different "fractions" that separate off.
In essence there is the epoxy and internal green lacquer of the BGA chips that turn into black and green "mist" in the water. Then there is the copper flakes from the different copper parts, leadframes and the internal pcb's of the BGA's that become this beautiful copper mosaic mess. The gold is separated from the copper mostly as all the green and black just washes right off. To recover the finest gold however, it looks like one does best to sift with a fine mesh like 150 or even 200+ in order to create a larger window of separation. (Density vs particle size and area. Just like normal goldpanning.)

Also, I see no reason for there to be brass inside copper legged SSOP chips or the BGA's. Brass or Phosphor bronse is more used for connection pins that don't or do need to flex a bit and where copper would not have the abrasion resistance required. The color balance of my camera does make it harder to see though. In real life it is very gold-y. I'll see if my little stunt-button will partake in the next photoshoot to give a color reference.

 

[KreAture]

Here's a picture of the grinder I will be using. (For now I have used a flat plate to test but the machine that would pull in the sticks and process would be better with this type grinder.)

It's 25 or 30mm wide so would be well suited for processing one side of a memory stick.
These are really cheap brazed grinders and I have used them for rock processing before. They take a lot of abuse. They would also be quite easy to replace given the right design. Also, such a small rotating mass will be safer than huge flywheels and large motor-systems. I happen to do a lot of brushless control (and I wrote my thesis on it way back when) so these are the scales of "motion" I feel comfortable with.

I know ppl may think I am nuts but specialty built systems is my passion.
It's no different than a pastamaker really. A single machine to do one job really well. (Except if you get gold from your pasta you are in the wrong business.)

 

[johnny309]

OK.....
The only downside is mathematics.
Aka.... platting and wiring inside chips are in size of um(micrometers),and giving the additional friction involved in your job..,..soon you will be putting your gear in AR.....to recover the gold.

Sell the boards to company that have the knowledge and technology and you have more gain.

 

[cosmetal]

Here's a picture of the grinder I will be using. (For now I have used a flat plate to test but the machine that would pull in the sticks and process would be better with this type grinder.)
grinder_30mm.png
It's 25 or 30mm wide so would be well suited for processing one side of a memory stick.
These are really cheap brazed grinders and I have used them for rock processing before. They take a lot of abuse. They would also be quite easy to replace given the right design. Also, such a small rotating mass will be safer than huge flywheels and large motor-systems. I happen to do a lot of brushless control (and I wrote my thesis on it way back when) so these are the scales of "motion" I feel comfortable with.

I know ppl may think I am nuts but specialty built systems is my passion.
It's no different than a pastamaker really. A single machine to do one job really well. (Except if you get gold from your pasta you are in the wrong business.)

Are you currently using a commercially available high-speed, flexible shaft rotary tool like a Foredom as your grinding wheel power source?

Thanks,
James

 

[KreAture]

I am using a low rpm high torque motor to be able to control the grind and avoid flinging boards all over the place.
The grinding discs are 100mm dia, but for the mechanical design I want to build I need drum shaped grinders.
They do not have to be large diameter so I want to try with the 25-40mm ones as they are cheaply available.

The panned out gold from a tiny test of the sticks mentioned earlier (after removing all visible copper to see if I am getting most of what I believe is in the ram) weighs in at a tad less than 25 mg.
There was one 2-sided stick with SSOP chips, one 2-sided stick with BGA and one single sided stick with BGA. All were registered memory so 9 instead of 8 chips pr set. 18 SSOP's and 45 BGA's.
I think the older SSOP has more gold pr IC but yield data online works in kg and they are much heavier thus probably not higher yield pr kg.
Still, about 5mg/side for the sticks seems not to far off considering there will be missing gold in the finer dust that would be caught by a chemical process.