My results of specific types of IC chips, flatpacks and BGA

[Tzoax]

The name of this package is CHIPFET. It was never mentioned here on the forum before - when i tried to search it by this name.

It was and still is manufacturing by ON Semiconductor. Here on this link you can see obsolete package (that contained gold bonding wires) and a newer package with copper bonding wires.

http://www.onsemi.com/PowerSolutions/MaterialComposition.do?searchParts=NTHC5513



The transition from gold bonding wires to copper bonding wires happened at 2010. Also on the bottom of this PDF document you can see several affected models:

NSS35200CF8T1G
NSS12601CF8T1G
NSS20601CF8T1G
NSS40601CF8T1G
NSS40600CF8T1G

Link for this PDF this:

https://www.onsemi.com/pub/docs/pcn/FPCN16501.pdf



This is a link for the first model - NSS35200CF8T1G
Here you can see more details about where it is used, schemes, diagrams and so on.

https://www.onsemi.com/pub/Collateral/NSS35200CF8T1G-D.PDF



On this link you can see today's price of chipfets (with copper wires).It is about 0.5 euros for 1 piece.


https://eu.mouser.com/Semiconductors/Discrete-Semiconductors/Transistors/MOSFET/_/N-ax1sf?P=1z0xullZ1z0z7ptZ1yzxao2Z1yrzaht&pop=3r74r



Here you can see that Package/Case is ChipFET-8

https://eu.mouser.com/ProductDetail/ON-Semiconductor/NTHC5513T1/?qs=ZXBb0xZ9WeBEPMIgTlVWZw==

On the first link where is data about yields - you will see obsolete model that contained gold - NTHC5513T1
This is a link about the same model:

https://www.mouser.com/ds/2/308/nthc5513-d-1194434.pdf

So, when you see something like this, save it, if it is made before 2010 there is a chance it contains 193g of gold per kilogram.




Alex

 

[shmandi]

I don't have exactly the same package transistor. The one on photo is 6 pin, 2mmx1mm package.
Not really focused, but you can see bond wire in middle of photo. Just for size comparison with other parts.

 

[shmandi]

This one is the same size package but 5 pin. Bond wires are whole lenght.
I am not saying that certain ChipFets don't contain that much gold, I just wont belive it until I see it. Maybe there is mistake in document and bond wire weight should be in µg instead mg.

 

[shmandi]

This is photo of LED I mentioned in one of previous posts. Size is 1mmx1.5mm, so weight is probably around 1mg.

 

[Tzoax]

This one is the same size package but 5 pin. Bond wires are whole lenght.
I am not saying that certain ChipFets don't contain that much gold, I just wont belive it until I see it. Maybe there is mistake in document and bond wire weight should be in µg instead mg.

Great pictures. I thought that it was mistake too...but i found several documents about chipfets from different years and the numbers are same. This is from 2003.



I compared those results by several years and numbers are still the same, i think it is not mistake... if it is - the mistake is present 15+ years.

Another thing is - when i compare my results from specific type of components i processed with those data - it is always close to those data (80-90%).

I think i located couple of them on this picture - it is a cellphone board.

 

[cosmetal]

All,

It appears that similar results have been reported before on GRF back on Aug 14, 2014 within an OP by ahmadbayoumi http://goldrefiningforum.com/~goldrefi/phpBB3/viewtopic.php?t=21061

Specifically, within solar_plasma's Aug 19, 2014 response, solar_plasma refers to a document entitled "Testing black chips and SMD Packages on their Raw Gold Content", by Roy Eugster, Switzerland, April 2013. A copy is attached and can be found online at https://www.scribd.com/document/248694489/Testing-Black-Chips-and-SMD-Packages-on-Their-Raw-Gold-Content Eugster shows a yield of 229 gms/kg on his tested SOIC and SOP packages. :shock:

I couldn't find any information on Roy Eugster. Has anyone heard of him before? If solar_plasma is still active on GRF, perhaps he could add some more information.

James

 

[Tzoax]

All,

It appears that similar results have been reported before on GRF back on Aug 14, 2014 within an OP by ahmadbayoumi http://goldrefiningforum.com/~goldrefi/phpBB3/viewtopic.php?t=21061

Specifically, within solar_plasma's Aug 19, 2014 response, solar_plasma refers to a document entitled "Testing black chips and SMD Packages on their Raw Gold Content", by Roy Eugster, Switzerland, April 2013. A copy is attached and can be found online at https://www.scribd.com/document/248694489/Testing-Black-Chips-and-SMD-Packages-on-Their-Raw-Gold-Content Eugster shows a yield of 229 gms/kg on his tested SOIC and SOP packages. :shock:

I couldn't find any information on Roy Eugster. Has anyone heard of him before? If solar_plasma is still active on GRF, perhaps he could add some more information.

James

James, it is actually 2.29g of gold not 229g because 0.229% of 1000g is:
0.229/100=x/1000
x=0.229*1000/100=2.29

And between those packages there is not chipfet package i mentioned.

 

[shmandi]

I found few that look similar. I will look at them tomorrow under microscope.

 

[cosmetal]

All,

It appears that similar results have been reported before on GRF back on Aug 14, 2014 within an OP by ahmadbayoumi http://goldrefiningforum.com/~goldrefi/phpBB3/viewtopic.php?t=21061

Specifically, within solar_plasma's Aug 19, 2014 response, solar_plasma refers to a document entitled "Testing black chips and SMD Packages on their Raw Gold Content", by Roy Eugster, Switzerland, April 2013. A copy is attached and can be found online at https://www.scribd.com/document/248694489/Testing-Black-Chips-and-SMD-Packages-on-Their-Raw-Gold-Content Eugster shows a yield of 229 gms/kg on his tested SOIC and SOP packages. :shock:

I couldn't find any information on Roy Eugster. Has anyone heard of him before? If solar_plasma is still active on GRF, perhaps he could add some more information.

James

James, it is actually 2.29g of gold not 229g because 0.229% of 1000g is:
0.229/100=x/1000
x=0.229*1000/100=2.29

And between those packages there is not chipfet package i mentioned.

Boy, that's embarrassing! I was always better with geometry.

Any info on when (historically) the ChipFETS entered the scene? The oldest doc I found was from EE Times dated 07/18/2007 (attached)

James

 

[Tzoax]

I found few that look similar. I will look at them tomorrow under microscope.

That's great, i can't wait to see pictures. Here is how internal components and bonding wires should look like:

https://www.vishay.com/docs/71127/71127.pdf

Any info on when (historically) the ChipFETS entered the scene? The oldest doc I found was from EE Times dated 07/18/2007 (attached)

The above PDF is from 2003. I will try to find some earlier documents.

 

[kernels]

Those pictures are not internal components / bonding wires, they are the pad layout on the PCB (top / bottom layer) where you solder the component.

 

[cosmetal]

All,

This is (to me) an absolutely fascinating section of Tzoax’s original post. Thank you, Tzoax, for continuing your work.

Of course, I have many, many questions. And I tried to find those answers on my own. But, again, I need experienced help in that quest.

So, here goes:

•I knew about MOSFETs and MISFETs (did you know that a MOSFET is also a MISFET but a MISFET is not a MOSFET? :shock: ). But I did not know about ChipFETs. In my search for their history, I also discovered that there are BioFETs and FlowFETs and, I am sure, there could be more.

•Has anyone researched these other FETs?

Peace,
James

(I will never post a simple arithmetic calculation again when I have the flu! :roll: )

 

[shmandi]

Here are some candidates that I found. According to marking code I don't think any of them is exactly the same as in your datasheets. First 4 are 8 pin, last is 6 pin. All are cca. 3x2x1.2mm

 

[shmandi]

Here are in the same order as on previous photo.
No. 2 has thinner and les bond wires.
No. 1 and 3 could have good yield to weight ratio as they have shorter (copper) leggs.
On last 3 photos there is (my) hair for thickness comparison.

 

[Tzoax]

Those pictures are not internal components / bonding wires, they are the pad layout on the PCB (top / bottom layer) where you solder the component.

Hein, thanks you for explanation, i was thinking that those 4 wires in the middle represents gold bonding wires from inside of package.

All,

This is (to me) an absolutely fascinating section of Tzoax’s original post. Thank you, Tzoax, for continuing your work.

Of course, I have many, many questions. And I tried to find those answers on my own. But, again, I need experienced help in that quest.

So, here goes:

•I knew about MOSFETs and MISFETs (did you know that a MOSFET is also a MISFET but a MISFET is not a MOSFET? :shock: ). But I did not know about ChipFETs. In my search for their history, I also discovered that there are BioFETs and FlowFETs and, I am sure, there could be more.

•Has anyone researched these other FETs?

Peace,
James

(I will never post a simple arithmetic calculation again when I have the flu! :roll: )

Thank you James, it took me couple of seconds to realize that you made a mistake in calculation...but for that few first seconds i felt amazing - that really "charged my batteries", so thank you for that.

All of the time i am trying to confirm to myself that i can rely on data from manufacturers. Knowing that there are many manufacturers and many changes through years i needed to test the most of common IC's to see how it matches with those data - and to be sure i can approximately predict gold content before i buy something, or when i am cherry picking. So far i am satisfied - all of the tests i made is lower than data from manufacturers by 10-20%, but i think it is normal since i am still learning and improving my skills. At the end i am aware how much gold i can extract for most common types of IC's and i think it is one of essential knowledge that anyone interested about refining should have. Based on that knowledge i can save time in the future by avoiding low gold content IC's, i can buy boards from telecom, medical and military equipment knowing what to expect etc.

I recently started to pay attention about those smallest IC chips and transistors and now i see it is a whole new world. There are so many of packages, and the yields are surprising.

Here are in the same order as on previous photo.
No. 2 has thinner and les bond wires.
No. 1 and 3 could have good yield to weight ratio as they have shorter (copper) leggs.
On last 3 photos there is (my) hair for thickness comparison.

Amazing pictures, thank you shmandi.

Those packages are "SOT23-8 thin".





I have data for the same package - with 3 legs. When you calculate it - it is about 1.7g ofgold per kg. So this package with 6 legs should have even better yields. And i think it will have to be even better yields than that because it is "thin" package.

 

[shmandi]

Two of them (1 and 3) have almost no legs and they are very close to package of ChipFets.
I will keep on looking for similar package transistors, maybe I find one that is full of gold

 

[Tzoax]

Two of them (1 and 3) have almost no legs and they are very close to package of ChipFets.
I will keep on looking for similar package transistors, maybe I find one that is full of gold

Chipfet packages does not have legs at all, they have 6 pads on bottom-side. And they are smaller than SOT23 package. I think they should look like these i found on cellphone board. When i scrapped them off with a screwdriver - 4 pads (from one side) left behind on a board, but it is obvious that there were 8 pads, and it looks much more like chipfet.





I don't have such a good microscope like you have. I will try to burn them and see whats inside.

 

[Tzoax]

This is best shots i made with my phone and microscope.

Gold bonding wires were still connected to a dice.



On this picture one gold bonding wire was still connected to a dice and a pad.



Does this first images look like those bonding wires will have 19% of total weight?

 

[kernels]

Awesome work mate, love this thread!

 

[shmandi]

Chipfet packages does not have legs at all, they have 6 pads on bottom-side. And they are smaller than SOT23 package. I think they should look like these i found on cellphone board. When i scrapped them off with a screwdriver - 4 pads (from one side) left behind on a board, but it is obvious that there were 8 pads, and it looks much more like chipfet.

Those two (no.1 and 3) have only pads. And also dimensions same as in datasheet of chipfets.
Another one I found is probably the most gold so far. But is not chipfet as I see only one transistor junction inside. As it is probably power mosfet, the wires inside are much thicker than in ordinary transistors or ICs. You can also see 3 parallel wires between dye and pin.
I don't think is close to 20%, but could be few % of weight. Just a guess of course.
If you have any normal digital camera photos are much better through microscope than phone camera. I use some old cheap Nikon and does the job.