Oxidation-proof copper could replace gold, meaning cheaper chips, says prof Scientists claim to have found a way to stop copper from oxidizing. If they're right, this could potentially allow copper to replace gold in electronics, leading to lower costs and, ultimately, cheaper components. The academics at Pusan National University in South Korea say they have developed a method to fabricate atomically …
That was my thought too. Instead of "damage" though, simply think "wear". Contact surfaces rub against each other on regular insertion and removal. Small amounts of wear are not generally an issue and indeed may even be desirable since they give the connection self-cleaning properties.
However here we are talking about surfaces smooth right down to the atomic scale. Surely a single insertion is going to scuff the surface sufficiently to completely destroy that fancy finish you're just applied?
we might - just might - wake up and fix the mess we made...
Cheer up! The ozone layer is on a path to recovery, acid rain has dropped across North America and Europe, forests are recovering in North America and Europe (North America is more forested than it has been in decades), waterways are recovering, China's air pollution is dropping, global extreme poverty is down, global literacy rates are up, whale populations are bouncing back, and, until late February of 2022, deaths due to war were at historical lows.
There are challenges ahead, but we've dealt with a lot of other long-term challenges in the last 50 years.
"Royal Mint wants to turn your old phone into gold"
But apparently only to turn the gold into useless bling - "gold, which they’re planning to use to make commemorative coins"
Actually, we have (world wide) such masses of gold that if it were all released on the market its value would collapse. The same is true for diamonds - the moment you buy one it's worth about half what you spent, as it's a purely artificial scarcity. Not really anyone's best friend.
Came here to say the same, interesting tech but conclusións are absolute nonsense.
Gold is used currently, but not such a thin layer, so if this tech were brought to market, the copper volume used for it would be far less than even current gold usage.
Perhaps that this tech will actually turn out not to prevent oxidisation completely and some cheap tat will get a shorter life? And I'm sure Apple with jump on any tech that shortens the life of its phones.
How much gold gets successfully recyled from circuit boards by recyclers playing with nasty chemicals in the third world. Maybe this could help prevent that?
I am sure other conclusions could be drawn that are at least plausible.
"And I'm sure Apple with jump on any tech that shortens the life of its phones."
Really? You pick on the one company that has had exceptionally good long term support for it's handsets. The last seven generations of their phones (not counting the SE models, since they're just a repackaging of one of other generations) all support the latest version of iOS.
Yes yes yes, software support is all well and good, but when it has a soldered and glued in battery that, with a regular charge / discharge cycle will begin to degrade after about 18 months, that doesn't count for very much.
Yes, I know battery technology has improved over recent years, but gains are incremental, and the same applies to pretty much all lithium ion or lithium polymer batteries (I'm not singling Apple out here, except on the poor repairability).
My old phone eventually had to be replaced due to the battery losing its ability to retain charge beyond a few minutes. The up-side here is that I didn't pay £1,000 for it, so paying for a third part to attempt to unseal it, replace the battery, and not break any part of it (or sending it to the manufacturer for a replacement for $wedge) was less economically viable than buying a new phone, which also didn't cost anywhere near as much as an iPhone (IIRC, about £190, and with capabilities equivalent to a "flagship" phone from approximately two years earlier).
Anyway, I've wandered from my point a little - Apple's software support might be "second to none", but their hardware support for older devices is more like "none". A colleague of mine once had to have an argument with one of their "geniuses" over a frayed charging cable, which they adamantly claimed was "water damaged" so wouldn't replace free of charge, and demanded £40 or so for a new one. When you fleece your customers in that way, you can afford to support software on devices that are approaching their economic end of life.
The conclusion notes that the price of copper is rising and that this use may further increase prices (*) undermining the economic case.
Except that, as the article already noted, "copper comes in at around $10k per metric tonne, while gold is over $62m per metric tonne"
So currently copper is just 1 / 60,000 the price of gold. Even if that went up ten times(!), the cost would still be a negligible percentage of the price of the gold currently used- i.e. close to nothing in comparison- and shouldn't affect the economics significantly.
(*) Only, as others have already pointed out, it probably won't.
No doubt about that, but it was the economics of this specific case I had in mind. As far as that's concerned, barring all but the most ludicrous increases in the price of copper (i.e. several orders of magnitude) this technique would- assuming it works and everything else remaining equal- still be economically worthwhile regardless.
This is true, but an annual increase in demand of 66.4 tonnes (well, actually more like ~22 tonnes, because the atomic mass of copper is roughly a third that of gold) is about the amount you'd need to do the plumbing in a new housing estate of 100 homes. In other words, so close to negligible on the global scale (currently about 28 million tonnes annually) that it could safely be considered to be zero (an increase in demand of less than 0.00001%).
That would depend on whether you are talking about the thickness of the coatings in atoms, or in nanometres. As someone with a (distant) background in chemistry, I tend to think in terms of the former, but from an engineering standpoint, one may well be thinking in terms of the latter. Either way, the point is that you wouldn't be replacing one tonne of gold with one tonne of copper, you would, at most, be replacing it with ~460 kg (using the relative densities) or ~323 kg (using the relative atomic mass).
Of course, the whole point is moot, if the cost of depositing the coating, atom by atom, is prohibitive. Bear in mind, that a layer only 1 micrometre thick is around 4500 layers of atoms.
First of all, this is nice. The surface science guy in me very much appreaciates the article. However...
"Oxidation-resistant Cu (copper) could potentially replace gold in semiconductor devices[...]"
What does this refer to? To gold plated copper pads (in surface finish with ENIG/ENEPIG)? I hope not... because the copper pads are many (10+) microns thick, and getting to that thickness with sputtering (the technique used in the work) will take forever.
Wire-bonding? Copper is already in use, so the application would then be related to augmenting the shelf-life of copper wires. Would this mean using the single-crystal copper wire mentioned in the article? Because what with wire-bonding being mostly used for lower-end applications, I am not sure any increase in material costs is justifiable.
Moreover, as is mentioned in the article "The flatness of a surface is decisively influenced by the interface structure between the film and the substrate"
The substrate is a double-side polished aluminum oxide. So how are you going to get THAT into the picture when it comes to making the process practical for electronics?
I can sympathize with the need to make bombastic claims in today's engineering ruled world, but they might have chosen the wrong application.
What does this refer to? To gold plated copper pads (in surface finish with ENIG/ENEPIG)? I hope not... because the copper pads are many (10+) microns thick, and getting to that thickness with sputtering (the technique used in the work) will take forever.
I wonder if the technique could be applied to the surface of pads made with conventional techniques, or whether the sputtering (as far as I know, similar to thin film vapour deposition techniques) would just preserve the surface features, but a couple of atoms higher? I guess it depends on whether this can be applied to a non-smooth surface to smooth it by "filling in the cracks".
Let’s take for example the RAM sticks I’m due to take delivery of later today.
The contacts on the side of the board that I shove into the motherboard will most likely be made of gold.
When I shove it into the motherboard, it will get a bit scratched, which is intended behaviour by design to give it a better connection. What would happen if single crystal super-smooth copper gets scratched, would it still have anti-oxidation properties.
Also, out of the £380 I paid per stick, I suspect very little of it is for the purchase of the gold in those contacts. This copper thing sounds like it would be much more difficult to apply, so would there actually be any cost saving?
"most likely be made of gold"
Actually they're most likely to be coated with gold (often over nickel) to a median thickness of about 15 microns (range around 3-50 microns depending on the application).
When a connection is mated, the gold will be scratched to some extent. Depending on the connector type, the gold may either only be partially scratched through, resulting corrosion resisting endurance for multiple insertions, or fully scratched through (for thinner coatings) at high insertion force, resulting a gas tight contact with the underlying metal. In this case, the remaining gold prevents corrosion attacking the edges of the gas tight contact, but the downside is that insertion endurance can be low or once-only.
I do electroplating and gold is a common thing to plate for the obvious shiny factor. (Even at a hobby scale it works out as perhaps £0.50 per square inch to do to a reasonable thickness; it's not exactly expensive!)
When plating a thin layer of gold would be between 0.10 and 0.25 microns. A normalish amount to plate would be 0.5 to 1 micron, depending on application. Jewellery such as rings gets a thicker 3 to 5 micron finish to allow for wear with pulling it on and off and rubbing it against things. The electoplating process limit is around 5 microns depending on the composition of the chemicals that your using. I don't think anybody can put 50 microns down with modern methods.
Historically you could have done it; Mercury fire gilding could probably have done it for instance. Mercury is a liquid at room temp, and it dissolves gold. You therefore mixed mercury with gold to form a form of liquid gold, paint it on, and then heat the item to >400c. The mercury then boils off, and you have a really, really thick layer of gold deposited. However, Mercury fumes are not that great for your health, and as a result the Health & Safety people have more to say about people using that process this they do with the modern "safe" version which dissolves gold using potassium cyanide.
And with that little bit of chemistry that most people won't care too much about over, something more interesting is that copper basically "eats" gold atoms. Nobody plates copper over gold. (more than once)
Also, electroplating suffers from similar problems as with painting in that the preparation amounts to 80% of the work, with only 20% going to slapping the top layer of paint on. The same applies to electroplating gold; the layer your putting on is so thin that it shows through the layer underneath, so everybody puts down a thick layer of nickel which is nice and shiny first. You then put the (thin) gold on top of the nickel so you get the gold colour from the gold combined with the shinyness of the nickel.
If I was plating RAM chips which are probably going to be moved between systems perhaps twice in their lifetime then i'd probably only have a .10 micron thickness of gold over the top of about 20 microns worth of nickel. The gold layer will get scratched in places on an atomic scale, but the nickel layer will prevent any oxygen from penetrating to the copper.
Mercury fire gilding, of the traditional form, is extremely hazardous. I think this is the basis or ormolu gold plating, that could be applied to non-metallic items. You make an amalgam of gold in mercury, and paint that on the object, The object is then baked, which evaporates the mercury, leaving the gold behind on the surface. I think some burnishing is needed after that.
The big problem is mercury vapour. Workers in this trade were lucky if they lived into their forties.
Mercury-gold amalgam is also used in gold mining, to extract very fine gold particles. The amalgam clumps together, making it easier to extract the gold physically. This can cause massive pollution problems.
I just read that there is about 0.1g of gold in a CPU chip and I expect they have more than most. So at $62 million per tonne, that works out to about $6.20. If the copper was absolutely zero cost, your stick would come down to about 377 pounds. Might make more difference on cheaper things but they probably have less gold in them too
"...about 66.4 tonnes of gold was used in electronic components last year – which is likely to be much lower than the figure for copper."
Given the average thickness of gold, the areas using gold and comparing that to the average area of a PCB, perhaps 70% covered in copper, have multiple copper layers and many wire/film interconnects, that has to be the biggest understatement of the century.
"semi-permanent resistance to oxidation"
Isn't "permanent" one of those absolute words? How can something be semi-permanent? Or is it something almost exactly the same but different?
Now, apart from all the excellent chemical and materials explanations given above, for which I thank the contributors for expanding education, does this not strike people as being similar to the lead-free solder whiskers issue? ie, look for "cheaper" or "safer" and end up with "shorter lived" because the corrosion resistance os only "semi-permanent", ie not permanent so has a shorter finite life than using gold?
After Papua New Guinea became independent from Australia in 1975,[5] Bougainville was given provincial status in 1976.[6]
In 1988, tension erupted into a civil war between the Bougainville Revolutionary Army and Papua New Guinea government forces.[5] One key issue of conflict was the Panguna mine, which was closed in 1989.[6] The civil war ended with a ceasefire in 1998, that was followed with the Bougainville Peace Agreement from 2001.[7] The agreement established the Autonomous Bougainville Government,[8] and mandated a referendum on the independence of Bougainville to be held 10-15 years later than the election of the first Autonomous Bougainville Government, which was June 2020 at the latest.[5] The referendum would be non-binding, and the final say would rest with the Papua New Guinean government.[5][9]
In November 2019, Raymond Masono, Vice-President of the Autonomous Region of Bougainville, campaigned that he would plan to reopen the Panguna mine if the referendum resulted in a vote for independence. Panguna closed in 1989 due to the civil war and is now estimated to hold copper worth up to $60 billion. With independence, all of Papua New Guinea's interests in the mine would transfer to Bougainville, giving it a 60% share in all projects and retaining all mining licences. The remaining 40% would be left for investors to bid on.[10]
> "Copper comes in at around $10k per metric tonne, while gold is over $62m per metric tonne"
No electrical engineers in the house??
For carrying electricity you DON'T go by weight. You want Mho per dollar. (Mho is the inverse of Ohm, so Conductivity.)
Aluminums are best. Gold is terrible (but has no Oxide).
Weight is a minor factor in overhead spans, and there is a zig-zag zone where hi-test Aluminum and Al-clad Steel compete with plain steel. (Yes, steel is excellent for the price and for the strength, but must be awkwardly large for its conductivity and good Oxide (rust) resistant joints are difficult.)
I remember the OLD DAYS (1980's/1990's) when ALUMINUM (or Aluminium fer ya Brits!) was used for ALL chip interconnects and traces. My suggestion to reduce COST is forget about copper and use U-shaped Micro-Channels for ALL interconnects using a very thin layer of aluminum on the surface of that U-channel using thin film sputtering or vapourization.
If your surface layers is a U-shaped microchannel, it will perform as a surface conductor using the "Skin Effect" for conductance. It actually outperforms SILVER which is the most conductive metal. Aluminum does oxidize BUT some simple thin film silica (i.e. pure glass) works wonders as a protective layer on top of the Aluminum thin film. The skin effect STILL works even with the thin film of glass on top of the aluminum U-channel surfaces!
Aluminum is $1.6647 per pound while Copper is $4.7640 per pound as of March 23, 2022 so the price difference is significant!
ANOTHER OPTION is Intrinsically stretchable and highly conductive polymers (ISHCP) using doped poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) which has 95% of the conductivity of pure Copper at 1/20th the price! Plus, these polymers can be used to make ALL-PLASTIC CHIPS that are flexible and very, very inexpensive!
See this link:
https://www.science.org/doi/10.1126/sciadv.1602076
The gold nanowires used to connect chips to outside connectors can be CHANGED OVER with the aluminum U-shaped microchannels that act as surface effect conductors. Now chip prices can drop even further.
And if you REALLY want to be bold, even the chip-traces can be converted to U-channels surface filmed with aluminum and silica to remove the need for all those dopants. A P-N junction could be made out of U-channel within a U-channel USING A CHEAP POLYMER as the semiconductor!
I have to think further but I think an ALL surface-effect P-N junction diode COULD be made out of aluminum, pure silica (glass) and a polymer to form gates/transistors/etc. for modern CPU, GPU and DSP chip making needs at 1/20th the price of today's processes.
I will get back to y'all on this!
Everything in this post IS NOW FREE AND OPEN SOURCE under GPL-3 licence terms!
V
With modern DIRECT electron beam etching using multiple beam heads (hundreds+!), you can definitely create 100 nanometres and less U-shaped non-channels with sputtered or vapour deposited thin aluminum films to create surface-effect connections and traces. You can still use UV photolithography processes but for accuracy I would probably go with multi-beam electron beam etching of nanoscale U-channels!
For prevention of oxidation, using plain pure silica or borosilicate glass can ALSO be thin-filmed on the surface of an aluminum U-channel to STILL allow surface effect electrical current transmission. P-N junctions and Diodes which can turned into the gates/logic circuits of a microprocessor and memory storage system can ALSO be crated out of nanoscale U-channels for costs that are 25% and less of current processor manufacturing costs. Polymer-based U-channels can bring that down even further to 10% of today's manufacturing costs!
Surface effect current transmission using thin film Aluminum U-channels is ONE future of future CPU, GPU, DSP and ASIC production! I think it's a good one to explore for mass market chip micro production!
V
This was a while back now, I know, but I've been trawling your posts, and this one had to have a reply.
If you use very thin film aluminium interconnects, you will have to keep it totally oxygen free for the entire lifetime of the device, because although metallic aluminium conducts well, aluminium oxide is a good insulator.
If you use a thin film in the presence of oxygen, it could not be thinner than the depth of penetration of oxygen into the surface, otherwise the trace would become totally non-conductive!
Are you another instance of the amanfrommars AI, or AI simulation?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
