Atomera, Game Changer, Oxygen in Semis

[Arthur Hanson]

Atomera has come up with a method of improving semis by inserting an Oxygen molecule between the source and the drain on transistors on semis, improving the performance and power consumption significantly. It claims to make semis faster, more efficient and lower manufacturing costs. This might be the beginning of a whole series of architectural changes, rather than node shrink, the improve performance in many ways, while lowering costs. This could have significant impact, by being another factor that could significantly grow the market for semis. It will be interesting to see if this can be applied to solar and display. My main curiosity is if this could be applied to solar as a major breakthrough? Does anyone know of any similar improvements in chemistry/materials coming to the semi sector?

ATOMERA – Quantum Engineered Materials

 

[IanD]

Only technical reference on their website is a paper from 2012 -- if this really was a revolutionary technology I'd expect a lot more than this...

 

[Arthur Hanson]

With rampant theft or misappropriation of technologies, I have seen more walls put up and less information being released in many cases. One super major company's engineers I have run into at technical gatherings always answer they KNOW NOTHING, and I don't blame them. I would never tip my hand to much or to early, until I had the technology locked down in as many ways as possible.

 

[IanD]

Agreed, but no news or partners or path to product in over 5 years suggests it's another non-starter looking for money...

 

[U235]

Interesting Arthur

These quantum confinement effects -- and potential mobility enhancements -- will naturally be accessible in nanowire transistor structures.

But whether it's worth introducing these fabrication steps on older nodes I have no idea.

The "series of architectural changes, rather than node shrink" is already happening: Planar MOSFET -> FinFET -> Gate-all-around/ nanowire -- performance improvement, without (much) gate length shrink.

 

[Daniel Payne]

No named customers, yet they are a public company, just how in the financial world does this scheming actually work?

11 engagements.

The technology sounds like it may only apply to bulk CMOS, and not FinFET process nodes, further limiting its potential usefulness.

 

[Arthur Hanson]

Dan, I agree it's unusual and the NASDAQ can be fooled as in the case of Bernie Madoff being its CEO for a time. Only time will tell and with the speed things are moving, it won't be long. I having an expert in this area of chemistry/physics look at it if his time permits. It will only need a few applications to be applicable to generate large returns. Nanotech physics and chemistry is the future.

 

[name99]

It's not clear to me EXACTLY what they have that they believe to be new, ie what they are selling. This masters thesis, for example
Simulation-based Study of Super-steep Retrograde Doped Bulk FinFET Technology and 6T-SRAM Yield | EECS at UC Berkeley
seems to consider the point to be well-known.

The issue is not any fancy quantum effects, it is (as I understand it) that by placing a layer of oxygen at the base of a finFET, you prevent diffusion of dopants between the fin and the bulk semiconductor, and this allows you to better control two distinct dopant levels in these two geometrically distinct areas, without leakage from one to the other.

This is nice, and gives you a few (up to maybe 10%) performance boost, but it's not a "game changer", it's simply one in a menu of possible options for future improvements. It gets you much the same advantages as finFET on SOI, so it all really boils down to which costs more, and which delivers slightly more bang.

I don't know exactly how patents work in the semiconductor space. Presumably you can't patent the knowledge that an oxygen layer in silicon has the properties it has. You likely can patent a specific way of embedding that oxygen layer in silicon, and that's likely the content of Atomera's patents. The question, then, is whether there are alternative (perhaps better, perhaps adequate) ways to create that same layer.
And as far as everyone growing rich is concerned, regardless of how solid the patent is, there's an absolute ceiling on how much it is worth based on the fact that no-ones going to pay more for this tech than the cost of building a finFET on SOI.

Mears & Atomera seems to have a lot of interest in superlattices, ie essentially artificial semiconductors. At some point their time will come, but again it's not clear to me EXACTLY what Atomera is offering here that's special. Superlattices (in vastly more complex configurations than are imagined for logic and memory) are being constructed everyday right now, either through MOCVD or e-beam, primarily for the purposes of high end optics, like quantum cascade lasers. I could imagine superlattice wafers being manufactured (like SOI wafers are manufactured) as a better substrate for all the rest of the complexity of manufacturing a CPU, but that's not a novel idea, it's just one that's waiting around for the economics to force it to become necessary.

 

[Daniel Payne]

One day in the 1970's I went to the library at Wang Laboratories and started browsing through the list of semiconductor patents only to be shocked that the concept of a CMOS inverter, NAND gate and NOR gate were all patented. So yes, you can patent anything. Technically the idea has to be new and unique, however in practice the folks working at the patent office are swamped with too many applications and too little time to do an adequate "prior art" search in the literature, and therefore ridiculous ideas that are already in widespread use are in fact prior art, therefore should stop the vast majority of patent applications from being granted.

In our high tech industry it was absurd that Apple was granted design patents for several prior art ideas:

1) A rectangle shape with rounded corners
2) A green icon for making a phone call
3) Rows and columns of icons on a Home screen

 

[name99]

Daniel, a DESIGN patent is not the same thing as a "standard" patent, and you destroy your credibility by complaining about the Apple issues you list.

Design patent - Wikipedia

You may or may not like the concept of IP as applied to ornamental flourishes and suchlike, but it has nothing to do with the concept of patents per se.

 

[Daniel Payne]

Daniel, a DESIGN patent is not the same thing as a "standard" patent, and you destroy your credibility by complaining about the Apple issues you list.

Design patent - Wikipedia

You may or may not like the concept of IP as applied to ornamental flourishes and suchlike, but it has nothing to do with the concept of patents per se.

Both design patents and standard patents are a very broken system, because neither system does a sufficient "prior art" search. Our patent system today makes lawyers rich, and is not a system based on justice.

Do you honestly think that a rectangle with rounded corners should be allowed as a design patent?

 

[Arthur Hanson]

I just had a conversation with a retired fab researcher and he said they have the scientists and sales people, but not the production process people. He said it does have potential, but someone has to be able to put it in a viable productions process and that's the person they don't have or haven't listed.

PS I agree with you on the patent process and this is a world wide problem. It's an absolute jungle of a mess designed as a full employment program for lawyers. We need a standard world wide patent process with research filed in a block chain process so the person who is really first gets the credit.

 

[Arthur Hanson]

I finally got ahold of my contact in SV and he said Atomera has theoretical people, but not manufacturing applications people and this is the problem. They only have the science and sales people and need to convince a manufacturer to put money, talent and time on the line to prove if it works on a mass production level.