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

[niks neims]

i am planning to test them.

Great! I have wondered about them myself! Can`t wait to see your results!

Too bad they are so tiny, TO-220 is much larger!

Any idea about TO-263 package https://en.wikipedia.org/wiki/TO-263
or
MOSFET https://en.wikipedia.org/wiki/MOSFET,

They seem to be common on MB as well... Sadly I suspect they also do not have any Au :/

-Artūrs

 

[patnor1011]

I have always collected them and processed with regular IC. I think I mentioned that few times in the past.

 

[Tzoax]

i am planning to test them.

Any idea about TO-263 package https://en.wikipedia.org/wiki/TO-263
or
MOSFET https://en.wikipedia.org/wiki/MOSFET,

They seem to be common on MB as well... Sadly I suspect they also do not have any Au :/

-Artūrs

I checked several sources for TO-263 (D2PAK) package and unfortunately nowhere is mentioned gold bonding wires - they use aluminum bonding wires for that package.



Also, i collected some of the transistors mainly from motherboards. When i checked couple of them (by model) i found that they are DPAK package - which also does not contain gold bonding wires. There are also D2PAK packages.



But - anyway will process these transistors from the picture just to be sure.

Alex

 

[g_axelsson]

As there are so many different components from countless manufacturers and over decades, there is always the odd exception in electronic components. But in general, high power applications uses thicker aluminium bond wires to handle the larger currents. I wouldn't be surprised to see both gold, silver and palladium in trace amounts even in the power transistors batch.

Göran

 

[niks neims]

writing on a hunch here, correct me guys, if you think I am wrong:

If transistor in question have some kind of heat-managment (bolted to the heatsink like TO-220 or soldered to the main PCB like TO-263 for some form of heat management) there is likely no gold bond wires - aluminium is used instead and the heat management is used to counter the much higher than gold electric resistance of aluminum (and thus much higher heat expenditure)?

-Artūrs

 

[geedigity]

Postby niks neims » February 16th, 2018, 4:26 pm
writing on a hunch here, correct me guys, if you think I am wrong:

If transistor in question have some kind of heat-managment (bolted to the heatsink like TO-220 or soldered to the main PCB like TO-263 for some form of heat management) there is likely no gold bond wires - aluminium is used instead and the heat management is used to counter the much higher than gold electric resistance of aluminum (and thus much higher heat expenditure)?

I have processed large transistors consisting of TO3, TO66, and other odd ball packages with metal caps and heat sinks. More than a few different types had gold bonding wires, and these were high current transistors.

 

[g_axelsson]

The heat is mostly generated in the die, a transistor could be viewed as a variable resistor. If the bond wires adds too much to the energy dissipation it's easy to either increase the diameter or add more wires in parallel.
The resistance of the bond wires are usually in the milli-ohm range.

The resistivity of gold is 2.4 x 10^-8 Ohm metres while aluminium have 2.8 x 10^-8 Ohm metres, just 15% higher resistance.

Göran

 

[niks neims]

huh,

OK, I learned something today

Still those transistors are fastened so tight to the motherboards, that removing them, even the golden ones, is usually not an option for my cherry-picking operation. I think, I`ll stick with BGA & larger IC, let the end refiner of boards get transistors together with MLCC & pins

-Artūrs

 

[Tzoax]

I would love to see the result of these Tzoax, unlikely to bother collecting them, but that seems like a surprisingly high yield.

Hein, i just found several documents about even better yields - 193.2g of gold per kilogam for this type of component. I thought that it was mistake, so i found several documents that confirms that, and even on a manufacturer's page it can be checked. Bad news is that they stopped using gold bonding wires that were replaced with copper bonding wires at 2010.

That's an interesting find!

But if I did the math right, it'll take over 100,000 to make a kilo.

Still, it's one more source of gold.

Dave

Dave, does 193.2g of gold per kg of a certain component type sounds better?


This is so far highest yield about any component i know about. And the strange thing is that it was not mentioned here on GRF - not even once. Any guesses?

 

[shmandi]

Dave, does 193.2g of gold per kg of a certain component type sounds better?

It is hard for me to belive. I crushed several of transistors like that and tried to measure bonding wire. I think my (very) rough calculation was around 5-6 micro grams per wire. So it would take 100.000 transistors for just 1 gram of gold. Also visually it looks 1 copper leg tenths of times bigger than gold bonding wire.
For some comparison you can print photo of transistor you posted any draw thin line with sharp pencil from one leg to middle and another leg to middle. I don't think that you will be able to imagine 20% gold in those 2 lines.
But I am aware that there are many different types, so maybe some have more gold. But 20%?
Small SMD LEDs you can find in cellular phones look like much better yielding than SMD transistors. Somewhere I read yield could be around 5%.
Similar types than on link: https://www.dhgate.com/store/product/4000pcs-lot-0805-smd-super-bright-led-white/98788152.html

 

[g_axelsson]

I would say... because they are just tiny and most would not contain that much gold. That's 20% by weight in bond wires and doesn't sound plausible.

Every transistor contains 2 or maximum 3 bond wires, one per connection. I rather spend my time on picking one BGA or flat pack with hundreds to thousands of connections, each one connected with a bond wire.

A similar proportion goes for connectors... the smallest connectors with the most pins contains the highest amount of gold per weight. As pins gets smaller the thickness of the plating has to stay the same to work, therefore the relative amount of gold goes up. But in grams the smaller connectors contain less gold as the surface area is smaller.

By the way, do you have a link to the document describing 20% gold by weight?

Göran

 

[Tzoax]

Dave, does 193.2g of gold per kg of a certain component type sounds better?

It is hard for me to belive. I crushed several of transistors like that and tried to measure bonding wire. I think my (very) rough calculation was around 5-6 micro grams per wire. So it would take 100.000 transistors for just 1 gram of gold. Also visually it looks 1 copper leg tenths of times bigger than gold bonding wire.
For some comparison you can print photo of transistor you posted any draw thin line with sharp pencil from one leg to middle and another leg to middle. I don't think that you will be able to imagine 20% gold in those 2 lines.
But I am aware that there are many different types, so maybe some have more gold. But 20%?
Small SMD LEDs you can find in cellular phones look like much better yielding than SMD transistors. Somewhere I read yield could be around 5%.
Similar types than on link: https://www.dhgate.com/store/product/4000pcs-lot-0805-smd-super-bright-led-white/98788152.html

I know - it was hard to believe for me either - now i am preparing pictures, documents with http links and you will see that this component have almost 20% of gold by weight. Sad thing is that i never collected them, i always thought that there is no value so it all ended soldered on the boards when i sold them to scrapyard (before i scrap IC chips, fingers, pins and mlcc's.). You will believe when you see official data about this on manufacturer's website... i need some time to prepare the links and pictures.

It is also possible that some/many of people know about this and keep this as "secret", i don't know...the name of this component is not ever mentioned here. It is a small component - but few pieces will payoff for example a whole PCI card.

Alex

 

[kernels]

I'm guessing that those parts will be very small package SMD LEDs ?

 

[g_axelsson]

Sorry, I thought you talked about SMD transistors still. My bad.

Eagerly waiting for the post and the links.

Göran

 

[Tzoax]

I would say... because they are just tiny and most would not contain that much gold. By the way, do you have a link to the document describing 20% gold by weight?

Göran

Göran, while i am preparing pictures and links please do the math yourself in case i am wrong, here is the picture:

I'm guessing that those parts will be very small package SMD LEDs ?

No, it isn't. But it is a very small package.

 

[kernels]

Wow, in that case I'm looking forward to hearing the answer,

One more guess . . .

Very small TVS (transient voltage suppression diodes), they come in small packages and have to handle very large current spikes.

 

[Tzoax]

Wow, in that case I'm looking forward to hearing the answer,

One more guess . . .

Very small TVS (transient voltage suppression diodes), they come in small packages and have to handle very large current spikes.

Nope, it's not. I remember well that some of the telecom boards i had contained hundreds of them by board - unfortunately i gave those too to a scrapyard. Maybe i have some more of these boards at home, i will try to find them tomorrow. And of course i will start collecting these for testing.

 

[g_axelsson]

The math checks out... but man, 13 mg components, don't sneeze, you will blow away a fortune.

75 components per gram, 75.000 to get a kilo, but what a kilo to refine. They spent 10 cents per component in just gold, no wonder they developed a cheaper process.

The question then becomes... how common are the high gold models?

Göran

 

[g_axelsson]

Looking closer at those numbers, there's something iffy with that data sheet. What is the chance that the bond wires of gold weighs the same as the copper ones? Down to 0.4% precision. I would like to see hard data, in other words what comes out of the beaker in the end. I suspect someone faked (or copied) some numbers in the rush to finish a document needed before going to market.

Göran

 

[Tzoax]

The question then becomes... how common are the high gold models?

Göran

It is used in cellphones, PDA's, computers, printers, digital cameras, disk drives and so on... -of course according to them. I am preparing link for those infos either, don't worry.... just don't sneeze anymore, ok?

And data is real, i watched 3 different documents (made different year) about this data and numbers are same. That was the first thing i thought - it was a mistake. But it isn't. Soon you will see.