Exotic Material Could Revolutionize Electronics

[phongn]

Menlo Park, Calif.—Move over, silicon—it may be time to give the Valley a new name. Physicists at the Department of Energy's (DOE) SLAC National Accelerator Laboratory and Stanford University have confirmed the existence of a type of material that could one day provide dramatically faster, more efficient computer chips.

Recently-predicted and much-sought, the material allows electrons on its surface to travel with no loss of energy at room temperatures and can be fabricated using existing semiconductor technologies. Such material could provide a leap in microchip speeds, and even become the bedrock of an entirely new kind of computing industry based on spintronics, the next evolution of electronics.

Physicists Yulin Chen, Zhi-Xun Shen and their colleagues tested the behavior of electrons in the compound bismuth telluride. The results, published online June 11 in Science Express, show a clear signature of what is called a topological insulator, a material that enables the free flow of electrons across its surface with no loss of energy.

The discovery was the result of teamwork between theoretical and experimental physicists at the Stanford Institute for Materials & Energy Science, a joint SLAC-Stanford institute. In recent months, SIMES theorist Shoucheng Zhang and colleagues predicted that several bismuth and antimony compounds would act as topological insulators at room-temperature. The new paper confirms that prediction in bismuth telluride. "The working style of SIMES is perfect," Chen said. "Theorists, experimentalists, and sample growers can collaborate in a broad sense."

The experimenters examined bismuth telluride samples using X-rays from the Stanford Synchrotron Radiation Lightsource at SLAC and the Advanced Light Source at Lawrence Berkeley National Laboratory. When Chen and his colleagues investigated the electrons' behavior, they saw the clear signature of a topological insulator. Not only that, the group discovered that the reality of bismuth telluride was even better than theory.

"The theorists were very close," Chen said, "but there was a quantitative difference." The experiments showed that bismuth telluride could tolerate even higher temperatures than theorists had predicted. "This means that the material is closer to application than we thought," Chen said.

This magic is possible thanks to surprisingly well-behaved electrons. The quantum spin of each electron is aligned with the electron's motion—a phenomenon called the quantum spin Hall effect. This alignment is a key component in creating spintronics devices, new kinds of devices that go beyond standard electronics. "When you hit something, there's usually scattering, some possibility of bouncing back," explained theorist Xiaoliang Qi. "But the quantum spin Hall effect means that you can't reflect to exactly the reverse path." As a dramatic consequence, electrons flow without resistance. Put a voltage on a topological insulator, and this special spin current will flow without heating the material or dissipating.

Topological insulators aren't conventional superconductors nor fodder for super-efficient power lines, as they can only carry small currents, but they could pave the way for a paradigm shift in microchip development. "This could lead to new applications of spintronics, or using the electron spin to carry information," Qi said. "Whether or not it can build better wires, I'm optimistic it can lead to new devices, transistors, and spintronics devices."

Fortunately for real-world applications, bismuth telluride is fairly simple to grow and work with. Chen said, "It's a three-dimensional material, so it's easy to fabricate with the current mature semiconductor technology. It's also easy to dope—you can tune the properties relatively easily."

"This is already a very exciting thing," he said, adding that the material "could let us make a device with new operating principles."

The high quality bismuth telluride samples were grown at SIMES by James Analytis, Ian Fisher and colleagues.

IMAGE: Surface electron band structure of bismuth telluride. (Credit: Image courtesy of Yulin Chen and Z. X. Shen)

Three-dimensional topological insulators are a new state of quantum matter with a bulk gap and odd number of relativistic Dirac fermions on the surface. By investigating the surface state of Bi2Te3 with angle resolved photoemission spectroscopy, we demonstrate that the surface state consists of a single nondegenerate Dirac cone. Furthermore, with appropriate hole-doping, the Fermi level can be tuned to intersect only the surface states, indicating a full energy gap for the bulk states. Our results establish that Bi2Te3 is a simple model system for the three-dimensional topological insulator with a single Dirac cone on the surface. The large bulk gap of Bi2Te3 also points to great potential for possible high-temperature spintronics applications.

These aren't superconductors in the traditional sense - but looks like a promising new way forward.

[General Zod]

So, how much do these cost to produce? Are they going to be cheaper than current methods?

[Fingolfin_Noldor]

So, how much do these cost to produce? Are they going to be cheaper than current methods?

And the usual rule of thumb for the semiconductor industry is that performance to wildly exceed that of what can be done with silicon or they won't budge a bit. Yields must also be good as well. Someone I talked to told me that it is very unlikely graphene would be ever used because no one has yet to produce the fabrication technology to produce high yields and quantity and performance does not exceed silicon by an order of magnitude.

[Starglider]

Someone I talked to told me that it is very unlikely graphene would be ever used because no one has yet to produce the fabrication technology to produce high yields and quantity and performance does not exceed silicon by an order of magnitude.

Historically, supercomputers were the route by which risky new low-volume high-cost fabrication techniques came into production. This trend died around 1990 with the Cray 3, which was a brave effort to make high-clocked processors based on gallium arsenide (that failed for debatable reasons). Since then the rise of clusters meant that most of the supercomputer research effort went into fast interconnects, which is arguably more useful for the rest of us anyway, since more and more applications became bandwidth and/or latency limited through the 90s and early 2000s (to some extent this is being rectified now by things like DDR3, SSDs, CSI/Hypertransport and on-chip memory controllers).

However it's possible that we're approaching diminishing returns in building huge clusters. Modern supercomputers have 10s to 100s of thousands of cores, and some problems just refuse to parallelise that far, and the ones that do are facing increasing overhead. The space and power demands of that much hardware are getting out of hand. At some point it would be cheaper to go for a tenth as many cores each ten times as fast, even if there's a lot of technical risk involved in making them. I was expecting something like Rapid Single Flux Quantum technology (based on low-temperature supeconductors, clocks up to 100 GHz or so), which is relatively mature, but concepts like this do hold more hope for eventual trickle-down to individual use.

[Fingolfin_Noldor]

Historically, supercomputers were the route by which risky new low-volume high-cost fabrication techniques came into production. This trend died around 1990 with the Cray 3, which was a brave effort to make high-clocked processors based on gallium arsenide (that failed for debatable reasons). Since then the rise of clusters meant that most of the supercomputer research effort went into fast interconnects, which is arguably more useful for the rest of us anyway, since more and more applications became bandwidth and/or latency limited through the 90s and early 2000s (to some extent this is being rectified now by things like DDR3, SSDs, CSI/Hypertransport and on-chip memory controllers).

However it's possible that we're approaching diminishing returns in building huge clusters. Modern supercomputers have 10s to 100s of thousands of cores, and some problems just refuse to parallelise that far, and the ones that do are facing increasing overhead. The space and power demands of that much hardware are getting out of hand. At some point it would be cheaper to go for a tenth as many cores each ten times as fast, even if there's a lot of technical risk involved in making them. I was expecting something like Rapid Single Flux Quantum technology (based on low-temperature supeconductors, clocks up to 100 GHz or so), which is relatively mature, but concepts like this do hold more hope for eventual trickle-down to individual use.

The semiconductor industry pretty much is a giant dinosaur that built up its entire infrastructure and technology around silicon. The same person I talked to (he's a material science person who did his PhD making laser diodes) told me that Quantum dot technology using silicon might hold promise enough to prevent use of other new materials. The same person told me that GaAs had very poor yields for what they want to do. GaAs is currently more used in making laser diodes, photodiodes etc. where the structure on average simpler. Remember for a fab to be profitable, it has to have 90% utilization. Low yields less than 90% prevent that.

[Eleas]

The semiconductor industry pretty much is a giant dinosaur that built up its entire infrastructure and technology around silicon. The same person I talked to (he's a material science person who did his PhD making laser diodes) told me that Quantum dot technology using silicon might hold promise enough to prevent use of other new materials. The same person told me that GaAs had very poor yields for what they want to do. GaAs is currently more used in making laser diodes, photodiodes etc. where the structure on average simpler. Remember for a fab to be profitable, it has to have 90% utilization. Low yields less than 90% prevent that.

Sounds about right. When silicon finally does face serious competition, I suspect it will come from a material used widely in other applications of the kind you mention (laser diodes, photodiodes), so that it is already perceived as a mature technology, and most importantly doesn't require an entire new industry to be built from scratch for that purpose.

[Simon_Jester]

Remember that the silicon industry has a good point here. It really doesn't make sense to replace the very expensive silicon fabricators with even more expensive graphene or gallium arsenide fabricators unless you get a performance increase so dramatic that it's good for everyone. For a lot of people, the performance advantage of graphene is irrelevant, but the cost increase is very relevant indeed.

For a more mundane analogy, silver is a better electrical conductor than copper. It's more efficient, and if we replaced all our copper wires with silver, the electrical grid would run with less waste. But except for applications where you absolutely need the best conductivity you can get and money is no object, no one uses it... with reason. I might get slightly cheaper kilowatt-hours from a silver electrical grid, but the price of construction would swamp any savings. So no one even thinks about building silver transmission lines.

The same will most likely go for superconductors.

[Memnon]

I think a good question here would be whether using this material would enable the semiconductor industry to raise its margins...
According to Late 2008 Article

Basically, the news isn’t that good. iSuppli noted that the quarterly profits produced are dwindling to the degree that PC manufacturers (known for low margins) now have better profit margins than semiconductor manufacturers.

“Semiconductor profitability has eroded steadily since mid 2004, with quarterly net profits having fallen into the single-digit range in 2008, down from the 17 percent to 19 percent range in 2004,” said Derek Lidow, president and chief executive officer of iSuppli. “The semiconductor industry now is less profitable as a percentage of revenue than the notoriously low-margin PC business, something that hasn't occurred before, except during a short period of the severe market downturn in 2001.”

So it would seem that the conditions are ripe for some kind of margin-raising mechanism.

[Starglider]

For a more mundane analogy, silver is a better electrical conductor than copper. It's more efficient, and if we replaced all our copper wires with silver, the electrical grid would run with less waste.

That would be completely pointless, because you could get the same decrease in resistance by increasing the diameter of your copper cables by 5% or so. Obviously this would be far cheaper than using silver (incidentally this is why silver hi-fi interconnects are nonsensical). Superconductivity is fundamentally different, in that it allows economical transmission of power coast-to-coast, which no plausible copper grid could do.

[Simon_Jester]

That would be completely pointless, because you could get the same decrease in resistance by increasing the diameter of your copper cables by 5% or so. Obviously this would be far cheaper than using silver (incidentally this is why silver hi-fi interconnects are nonsensical). Superconductivity is fundamentally different, in that it allows economical transmission of power coast-to-coast, which no plausible copper grid could do.

You have missed my point, which was entirely about cost-effectiveness. It is true that replacing copper with silver cables would be a waste of money, and that this is why we don't do it. And that was the very thing I was trying to get people to take away from my post.

If silver were as common and cheap as copper we'd use silver and not even think of copper cables, any more than we think of using electrical cables made of iron. But unless someone finds a bunch of gigantic undiscovered silver lodes lying around, that's never going to happen.

Likewise, if superconducting cable were even within shouting distance of copper in terms of price and resilience, it would probably replace copper for nearly all applications... which it won't, for the foreseeable future. At the moment, the refrigeration costs make superconducting cable a waste of money except for specialized applications; if we ever invent room-temperature superconductors, they're likely to be very exotic materials that no foreseeable manufacturing technique can make cheaply enough to justify the switch from copper.

Yes, superconducting cables have a capability that no normal conductor can match. But even with that capability, they don't pay and won't in the near future. Transmitting power coast-to-coast just isn't worth it, unless for some bizarre reason you absolutely need to put all your power plants in one place and use them to run houses and factories three thousand miles away.
______

And to reiterate, my point is that replacing a basic (but cheap) material with a vastly more expensive material that offers only a marginal increase in performance doesn't pay. And it isn't obstructionism on the part of industry to refuse to make the switch.

[Starglider]

You have missed my point, which was entirely about cost-effectiveness. It is true that replacing copper with silver cables would be a waste of money, and that this is why we don't do it.

Even if silver cost the same as copper, it would still be entirely pointless. Silver is more dense than copper; the difference in density is greater than the difference in resistivity, so silver cables would be heavier and require more expensive pylons. Silver is also less flexible, considerably weaker, more susceptible to thermal stress (because it expands more) and almost as vulnerable to corrosion. The whole 'silver wiring is better than copper wiring' meme is an idiotic brainbug perpetuated by idiot 'audiophiles' (and the companies who like to rip them off). The only time that silver conductors would be preferable to copper is in a volume-constrained application*; on-chip interconnects perhaps, if anyone could come up with a workable chemistry for mass production.

* There may be some exotic (probably high-frequency) applications where skin effects make a silver-plated copper wire worthwhile, but if there are, I haven't encountered them (hi-fi is not one!).

If silver were as common and cheap as copper we'd use silver and not even think of copper cables,

No, we wouldn't, and you obviously don't know what you're talking about. Hint; critical applications such as satellites and military aircraft could easily afford silver wire, but they use copper or aluminium, because those are technically better materials.

Likewise, if superconducting cable were even within shouting distance of copper in terms of price and resilience, it would probably replace copper for nearly all applications... which it won't, for the foreseeable future.

Silver is about 5% better at transmitting power than copper on a per-volume basis and considerably worse on a per-weight basis. Superconductors are infinitely better, in resistivity terms; they are in an entirely different class. Your analogy between silver wire and superconducting wire is worse than useless because superconducting wire is worthwhile in principle and silver wire is not.

[Simon_Jester]

Even if silver cost the same as copper, it would still be entirely pointless. Silver is more dense than copper; the difference in density is greater than the difference in resistivity, so silver cables would be heavier and require more expensive pylons. Silver is also less flexible, considerably weaker, more susceptible to thermal stress (because it expands more) and almost as vulnerable to corrosion.

OK. You're right; I forgot to take those factors into account.

So my example of "copper versus silver" is profoundly flawed, because of other material properties of silver besides conductivity.

However, the actual point my example was meant to illustrate is not profoundly flawed. Imagine, instead of using silver, that we invent some technomagical way of "treating" copper that increases its conductivity by 10% without changing any of its other physical properties. However, the cost of "treated copper" is ten times greater than the cost of the copper itself.

No one is going to replace the power grid of normal copper with treated copper any time soon under those conditions; it won't pay to do so. The improvement in performance* doesn't justify the cost increase. Industry is not being irrational in refusing to tear down the many thousands (millions?) of miles of copper wire in the country and replace them with "treated copper." Nor is it somehow wrong for makers of electrical cable to continue making untreated copper cable for most applications. Except in very unusual cases, the superior conductivity of "treated copper" is completely irrelevant.

*Which, in this purely hypothetical scenario, would actually exist, unlike the case of "silver versus copper..."

The whole 'silver wiring is better than copper wiring' meme is an idiotic brainbug perpetuated by idiot 'audiophiles' (and the companies who like to rip them off).

In my case, it comes from reading a table of conductivity values for various metals; I am neither an audiophile nor a manufacturer of audio equipment. I you are specifically angry at those guys, please point your anger elsewhere, because they're not here.
_______

No, we wouldn't, and you obviously don't know what you're talking about. Hint; critical applications such as satellites and military aircraft could easily afford silver wire, but they use copper or aluminium, because those are technically better materials.

Yes, you're right. I neglected to look up properties of the two metals besides their conductivity, which was a mistake. I was wrong to assert that silver would replace copper if the two had equal costs.

However, even so, I think that my original point (which the "copper versus silver" thing was merely a bad example of) is sound. Marginally greater quality at vastly higher cost does not justify an industry-wide switch, which is why we may not be seeing chip manufacturers switch away from silicon any time soon, even if this new material is better.
________

Silver is about 5% better at transmitting power than copper on a per-volume basis and considerably worse on a per-weight basis. Superconductors are infinitely better, in resistivity terms; they are in an entirely different class. Your analogy between silver wire and superconducting wire is worse than useless because superconducting wire is worthwhile in principle and silver wire is not.

And yet, superconducting wire is still so much more expensive that its "infinitely better" resistivity is useless except for specialized scientific applications where you absolutely positively need the highest possible currents.

So to hell with silver wire; we still have the more fundamental issue of cost-benefit analysis. Nobody builds refrigerated power lines of liquid-nitrogen superconductors for very good reasons, even though those lines could carry electrical power with vastly less waste than the existing copper ones.

[Starglider]

However, the actual point my example was meant to illustrate is not profoundly flawed. Imagine, instead of using silver, that we invent some technomagical way of "treating" copper that increases its conductivity by 10% without changing any of its other physical properties.

Your 'basic point' is blatantly obvious, if not tautological. Your inability to get basic physical facts right is what made it worthless, and the fact that a load of irritating so-called 'audiophiles' constantly make the same mistake is what made it annoying.

However, the cost of "treated copper" is ten times greater than the cost of the copper itself.

This does in fact exist, deoxygenated copper is a slightly better conductor, but it is only actually worth using in a few specialist applicatons (and of course audiophiles, again)

[Simon_Jester]

Your 'basic point' is blatantly obvious, if not tautological. Your inability to get basic physical facts right is what made it worthless, and the fact that a load of irritating so-called 'audiophiles' constantly make the same mistake is what made it annoying.

I was attempting a justification of the industry's "it doesn't pay even though the performance is better" stance in response to this:

The semiconductor industry pretty much is a giant dinosaur that built up its entire infrastructure and technology around silicon...

I did not intend this to trigger a huge argument; after all, my basic point was fairly obvious.

Then I chose a truly bad example to illustrate it with, based on a table of conductivity values I read a few years back in a freshman physics textbook. That was a mistake, because I should have checked other relevant physical parameters such as density.

I compounded my mistake by inadvertently offending you. Unlike you, I haven't experienced the gullible audiophile arguments, so I didn't realize I would be annoying you by bringing it up. Since the matter seems to be rather serious to you, I apologize for my unintended insult.

[sketerpot]

Most power lines are actually made out of aluminum, often wound around a core of steel wire. The key thing is to get a balance of high conductivity, low weight, and high tensile strength, for a reasonable price. Copper just isn't the right material for this, most of the time.

Oh, and to further muddy the waters with real-world facts, there are some ways to improve the strength of these wires that make them several times more expensive. This can be worth it, if you want to carry a huge amount of current at a very high voltage. Remember, you also have to pay for towers and the land to build them on. Money saved here can make it economical to (for example) up the voltage some more, and get lower losses that way.

Can we stop nitpicking the damn analogy now? It's dead, Jim.

[Simon_Jester]

Cool. Didn't know they used aluminum for high-tension power lines. I'm learning a lot of metallurgy as applied to electrical engineering here.

More seriously, I sincerely hope that the bismuth compound described is the next wave in electronics. I'm just not clear on whether it will be or not, and a lot of stuff gets promised as the wave of the future without panning out. I'm still waiting for my flying car, after all.