Slashdot Mirror
Molybdenite As an Alternative To Silicon
An anonymous reader writes "Molybdenite (MoS2) can be used to make transistors that consume 100,000 times less energy in standby state. This mineral, which is abundant in nature, is often used as an element in steel alloys or as an additive in lubricants. Research carried out in Switzerland at the Ecole Polytechnique Fédérale de Lausanne's Laboratory of Nanoscale Electronics and Structures (LANES) has revealed that is a very effective semiconductor. Molybdenite's 1.8 electron-volt gap is ideal for transistors and gives it an advantage over graphene (which does not have a gap)."
Isn't this just Moly disulphide, the lubricant in Molykote? http://www.dowcorning.com/content/molykote/anniversary.aspx?bhcp=1
Sorry, but gray text on gray background is making my eyes bleed.
In the latest technologies a lot of current is wasted to subthreshold conduction . Current that flows then the transistors should be "off".
A material with a higher bandgap 1.8ev to silicons 1.1ev will naturally have less leakage. As it is an exponential thing the leakage should not just be a reduction of 1.1 to 1.8 thing but much more significant.
Molybdenum is a CRITICAL trace element in the development of any food crop we have.
This reeks of the dumbest thing one could do, EVER.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
Oh dear. This means they might have to rename Silicon Valley to Molybdenite Valley, but that doesn't sound nearly as nice.
There are plenty of materials out there that make good semiconductors, the question is: can we make them?
Moly disulfide is a material a couple of different graphene groups have been looking at (hey, we know there's an issue with graphene). What this paper really means is that the Ecole group has figured out how to *make* MoS2 better than other people, and that's really the hard part. Of course, they're still making devices using scotch tape exfoliation...
It's really hard to mass produce 2D materials.
You want to save energy by replacing as many of the the currently installed systems in the world? Why do I get the feeling that trashing perfectly good equipment, and manufacturing replacements is not the best use of our energy resources.
We hope your rules and wisdom choke you / Now we are one in everlasting peace
in reality, you will just get more features out of the same die consuming the same amount of power than today. We did great with small CPU, the software we run on them just became full of bloat (not to speak about all the HD crap). That said, Intel's business is to sell you a new CPU every few years, not make it last 15 years.
This mineral, which is abundant in nature, is often used as an element in steel alloys or as an additive in lubricants. That is a joke, isn't it? Or is it just /.?
From Wikipedia:
Molybdenum is the 54th most abundant element in the Earth's crust and the 25th most abundant element in the oceans, with an average of 10 parts per billion; it is the 42nd most abundant element in the Universe.
That is not abundant that is pretty rare. Considering 35% of the planet is silicon ... or is it more?
Regards,
Angel
Cost free eBook I read (by iBook/Kobo/Amazon/ObookO/Gutenberg etc.): "The Green Odyssey" by Philip Jose Farmer.
"...Molybdenite's 1.8 electron-volt gap is ideal for transistors and gives it an advantage over graphene (which does not have a gap)..."
With Graphene, I can scribble it on Scotch Tape and get a Nobel Prize; can I do that with Molybdenite?
Plus, can you image a goddamn smart phone with a week long battery life?? Or a laptop that runs for days without needing to recharge? A server farm that could be powered by solar power and a few large battery power storage units?
You have misunderstood the article. It clearly says molybdenite transistors consume 100.000 times less energy than silicon ones in STANDBY. Not when operational. Sure, it would increase efficiency of mobile devices where you turn unneeded transistors off to save energy, but it would do nothing for when the system is operational and in use. Thus your idea of a server farm being solar powered is completely without basis.
Molybdenite's strength is in mobile applications: when the device is in standby mode it consumes a lot less energy than traditional silicon-based ones. But it has another strength here: silicon is a 3-layer material, whereas molybdenite is monolayer. This means that you can make smaller chips, or cram more stuff in a chip of the same size.
With all of these other materials available for computer circuitry , are their computers manufactured with materials that are far more efficient but cost more (probably enormously more) to produce that are only available to a select few ?
Google just announced their new browser update, "Chrome Moly".
It is not a quirk of mathematics. It is a quirk of language.
While it can be parsed the way you say, most would parse it to mean "1/100000 of previous consumption".
It might not be the "right" way, but it is the way most people read it.
So on one hand what you state is correct from a mathematical standpoint but on the other hand irrelevant. :p
It is technically incorrect but the phrase "xxx times less" has become the way people express that something is 1/xxx of what it used to be.
You can yell at people until you're blue in the face but it is pointless to try to change the language back to what it used to be
You're doing some freaky ass computing if all the transistors in your CPU are active at the same time.
I may agree with what you say, but I will defend to the death your right to face the consequences of saying it.
Actually, I could see a use for it, if it can be deposited in parts of a silicon chip: Cutting power to parts of the chip.
Though there are a lot of problems with the few solutions I've come up with in about 2 minutes of reading this. Mostly involving using different materials on the wafer, but if it is that much more efficient, I could see the relatively expensive process of cutting a few parts to be replaced with this out with lasers eventually panning out.
Then again, most of the actual silicon I've looked at personally with any degree of knowledge was 1980s/early 1990s level stuff. Now they are more or less black boxes described by whitepapers.
English is not mathematics. And no, the rest of us aren't changing our language to match your obsessive needs for it to be so.
let me know when you have I-V curves for a moly disulpide FET. Both p and n types please.
I learned many moons ago, that one of the most important things about Si is the fact that it's so easy to grow an oxide. It's EXTREMELY useful when processing integrated circuits. Otherwise everything electronic would use III-V's.
Any new material which aims to replace Si is going to need an equivalent process capability.
Personally I'm hoping for a breakthrough in organic semiconductors. I want to be able to screen print transistors at home.
Absolute statements are never true
Molly's Revenge are one of the local Irish bands seen here in the Bay Area. (Apparently they were a follow-on to an earlier band called Dance Around Molly, but with a name like "Molly's Revenge" they eventually had to wrote a song involving someone named Molly and some revenge...)
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
That was true up to a point. Over the last 5 or so years, we hit the 'good enough' point on computers. Power efficiency is where it is at now. With the last round of upgrades in my home, I went from an average power usage of 180kw on my computers to an average of 40kw. That doesn't even include the fact that most of my computers can actually go into stand by now.
Unless we create a magic battery, power consumption will always be a huge thing for laptops and cell phones. Data centers too certainly measure performance/watt. But I agree, for the regular desktop it's no longer a big deal, if it ever was.
Live today, because you never know what tomorrow brings
I'll take the Mexican divorce, thanks.
Sent from my ASR33 using ASCII
More and less are opposites, so is multiplication and division. Most people take ten times more to mean x * 10 and ten times less x / 10. Neither is mathematically correct, but it has a certain logical consistency.
Live today, because you never know what tomorrow brings
Leakage current has been the dominant issue since the days of the P4 and the first nVidia vacuum cleaner. As the devices have gotten smaller, the leakage has gone up significantly. To combat this, they've stopped increasing clock speed and started to use a lot of clock-gating and power gating where parts of the chip are inactive or even turned off. They are at the point where a higher gate voltage to turn on should not offset the reduced power dissipation due to leakage current.
Town name was Molybdenum?
Sorry, but gray text on gray background is making my eyes bleed.
40 kW average? Home? That would be 83 amp service @480V, for a single computer. How many of these supercomputers do you have?
"National Security is the chief cause of national insecurity." - Celine's First Law
You know...deserts have sand, too.
What else can happen when an unstoppable force collides with an immovable object?
That's a horrid lead-in that tells us exactly nothing. "Less energy than" what?
Some of the possibilities:
Molybdenite (MoS2) can be used to make transistors that consume 100,000 times less energy in standby state than when they are not in standby state.
Molybdenite (MoS2) can be used to make transistors that consume 100,000 times less energy in standby state than silicon-based transistors do in standby state.
Molybdenite (MoS2) can be used to make transistors that consume 100,000 times less energy in standby state than Cowboy Neal does to zip up his pants.
Seriously, don't make us guess what you mean.
AKA "Molly be damned"
the preceding comment is my own and in no way reflects the opinion of the Joint Chiefs of Staff
Replacing a gigawatt server farm with one that uses watts is such a substantial energy savings, that it is difficult to imagine. It would be like replacing a factory with the power usage of a lightbulb and producing just as much. This is a difference of 100,000 times less energy. That is worth it.
Where is the mod rating for "scary"? Also,
Well, many people have already addressed the flaws with your argument (such as the energy costs of manufacturing replacements for all this equipment), but more critically, you've failed to realize that this is at such an early stage of discovery that it's still highly likely to fizzle without going anywhere.
Just because you can make a few samples that perform well in a lab doesn't mean you can produce products with it in a consistent and energy-efficient manner. Just look at gallium arsenide - Two decades ago everyone was predicting GaAs to be the future of the semiconductor industry, two decades later it is only seen in a few niche RF devices (and those manufacturers are STILL having yield problems), much of it due to the simple fact that GaAs doesn't form a nice insulating layer when it oxidizes and other manufacturability issues.
retrorocket.o not found, launch anyway?
I toured a facility that produced micro-chips with a 70 nm process a couple years ago, it was a student facility, hence the out of date machinery, but it could successfully deposit layers of all kinds of different materials on the chip dyes with relative ease. I'm sure it won't take a few engineers more than a couple months to figure out how, and then put it into an industrial machine.
Where is the mod rating for "scary"? Also,
Until humans find a use for it.
If Google really cared they would fix Android Chrome to reflow text, instead of discriminating
I think you are probably overestimating the savings from this change. Is the vast amount of energy used in a data center really from leakage in idle transistors? I'm a bit skeptical of that.
We hope your rules and wisdom choke you / Now we are one in everlasting peace
silicon is a 3-layer material, whereas molybdenite is monolayer
Huh? A MOS transistor is three-layer: Metal, Oxide, Semiconductor, no matter what is the semiconductor.
You will probably never see this in practice in any large scale semiconductors because of two reasons: 1. There is already an established process and infrastructure to refine silicon. 2. Mo costs more due to rarity and does not have a well established industrial process to refine it. As it stands, it is incredibly expensive and requires large amounts of energy to refine: http://en.wikipedia.org/wiki/Silicon#Purification Mo being rarer to begin with will have a higher material cost, and will also probably be that much more difficult to refine. I am not sure what processes there are to distill/refine Mo into the superpure state needed for semiconductors, but I cannot imagine that there are established processes that can handle the worldwide demand for semiconductors.
Yes, yes... I did mean what. I was going to put the monthly kwh usage but decided I didn't want to go through the math considering the times of use changes better computers brought and so forth. I didn't switch back to watts. Lets just chalk it up to Verizon Math, and move forward. ;)
Most transistors are idle most of the time, and any electron gap below 1.6 eV is going to leak like a an old ladies bladder. Silicon has a gap of 1.1, allowing electrons to cross it with relative ease. Molybdenum disulfide has a gap of 1.8 eV, higher than the charge of a single electron, making it orders of magnitude more difficult to leak any power.
Think about some basic gates, like the nand, nor, not, and, xor, and, or, the different kinds of flip flops, and think about what percentage of the transistors is idle even when the circuit is active. In many circuits, you will have less than half the transistors active, and even then it is likely that a large portion of your circuitry is dormant at any given time. So figure if a totally idle processor uses about 50 watts, while active as possible is only like 130, it is a vast amount of power being consumed. You could cut the power consumption of server farms by more than 50%, maybe even 75-85% on the far side of peak usage hours.
Where is the mod rating for "scary"? Also,
Clearly you didn't read the article, and your condescending tone is not just obnoxious, it is pathetic. Come off it.
The importance of this discovery is held a lot higher than other discoveries precisely because it lacks the various problems of certain materials. First of all, it is just another layer added to silicon chips, the silicon is going to do much work that it already does. Second, it has a high voltage gap that makes it much more efficient and avoids the issues that graphene has in this area, but still has a very, very tight structure very conducive of smaller manufacturing processes.
Also, unlike materials like gallium arsenide, molybdenum disulfide has a long history of use in other industries, its properties and methods for handling it properly are quite developed, significantly reducing the turnaround time. I would venture to wager that this material will be integrated into existing manufacturing processes within the next few years. Not to mention the on chip fiber optic converters, which are in significant development and could seriously blast through many existing bottlenecks for speed.
But go ahead, sit their with your unrealistically excessive pessimism and ride around on your high horse, everyone loves people like that.
Where is the mod rating for "scary"? Also,
Lifespan of most computers being, what, 5 years? 10?
Wouldn't take long for a decent swathe of the word's install base to move to a new technology, assuming it's worth it.
Well yeah, but the OP was saying to take your brand new devices and replace them immediately with this new technology. My point is what you seem to be talking about as well, that it would probably be a better use of resources to replace them when they reach EOL and not sooner.
We hope your rules and wisdom choke you / Now we are one in everlasting peace
But they all pronounced it Molly Be Damned (except the kids, who had to say "Molly Be Durned").
I see even classic Slashdot is now pretty much unusable on dial up anymore.
Hrm, and for power control even if it was in a separate layer it wouldn't be that bad, as it'd be a relatively small number, meaning it wouldn't even need the relatively hard work I was thinking it would.
Nifty.