Slashdot Mirror
NTT Verifies Diamond Semiconductor Operation At 81 GHz
Anonymous Coward writes "This story over at eetimes.com reports of a semiconductor made of diamond that is able to run at 81 GHz." Mmmm, foreshadowing.
← Back to Stories (view on slashdot.org)
Does anyone know how hot these things will get?
30 W/mm??
Don't they mean 30 W/mm^2?
SCO (noun.)- A Slimy Corporate Ogre. Often seeks free money.
This tech has some serious military applications.
Killing devices like the star drek phaser is not that far off. The high energy output potential because of the thermal characteristics is scarry! Just imagine if the output of a cell phone could have a signal db and directional capable antenna. Yipes you could get scrambled brains if the antenna was too close. The radar and remote sensor applications for this could kick current US stealth tech out the window as well.
OH THE SHAME I fell off the wagon and use sigs again!
I remember a Nova special about manufactured diamonds and how GM finally got the large ones made with no defects. A trip from a Debeers exec and the operation was shutdown and people were released. Back to the industrial diamond business!
Highly unlikely. See, what you don't realize is that this technology will likely be utilised in memory before processors. One of the first verification processes in semiconductor technologies is 'can we make memory with it'. They start off simple and let the circuits get more complex from there. We'll likely see very high speed memory before you see a Pentium 5 or Athlon Diamond XP. This is a great boon for computing. Memory has been a large bottleneck for a long time.
:)
Imagine 1GB of processor core clock speed memory. That would be friggin amazing for databases
You can't legislate goodness. Let each to his own destiny, by will of his freely made choices.
Anyone who's bothered to do the research into it knows that DeBeers is about as evil as a multinational can get. Somehow I doubt that they are going to play nice with another industry that wants to use thier bread and butter product for making something that doesn't cost $100,000 a gram.
As I see it, there are one of only two outcomes here:
#1) Someone finds a way to make cheap diamonds, and DeBeers goes after them (in more ways than just the legal route) to make sure that #2 happens.
which brings me to
#2) Nobody finds a way to make cheap diamonds, and DeBeers can triple their prices. Of course, the diamond supply is already kept artifically low to drive up prices, so meeting this new demand won't be a problem at all (it'll just cost you the price of a small car to buy a CPU.)
I don't like this one bit...nope...not one bit. As if Microsoft's monopoly wasn't bad enough.
One of the first verification processes in semiconductor technologies is 'can we make memory with it'. They start off simple and let the circuits get more complex from there. So why, then, is the gap between processing speed and memory access speed growing?
http://yetanotherpoliticalrant.blogspot.com
This month's issue of Wired Magazine has artificial diamonds as its cover story. Just finished reading it a few hours ago. Very interesting as to where this is going to take the diamond jewelry business (DeBeer's is in trouble) as well as the semiconductor industry.
It already is a consideration in many settings. Rackmount systems are the first to notice because of the density. As I understand it, any processor over 50 watts is pushing it for 1U applications. I think Sun is going to use mobile Athlon 64's for just that reason.
With Prescott set to top 100 watts, I think we've hit the limit of what desktop users are willing to tolerate. We're into "can't run it on summer afternoons" territory already. I've been using my laptop at home because of it.
When someone might yell at me, it has to be OpenBSD.
This is part of the reason why the fibre optic revolution has been more of a slow turn... fast pipes are great, but it helps if you know where to send them.
81GHz isn't going to solve the problem - but it will help.
Q.
Insert Signature Here
That inhuman pack of gunship flying, mercenary hiring, indigenous population exploiting *ssholes can suck it down and shut up.
Our obsession over "pretty sparklies" is disgusting, and what we are willing to ignore to ensure a steady flow is reprehensible. How many middle-class housewifes with a rock on a finger know the TRUE cost of that shiny bauble?
Lets wake up to ourselves and try to develop a modicum of common sense? Why are diamonds expensive? Because they are in demand. Why are they in demand - no it is not the industrial applications? Because they are expensive.
Q.
Insert Signature Here
That's exactly right. Even with all of the newest improvements in semiconductor technology and the resulting memory density(Remeber those old 512KB clunky SIMMS :) improvements, we are still placing the memory too far away from the processor. It should be closer(physically, logistically, electrically). With the new AMD Opteron, they got it right. Putting the memory controller on the processor is the first step in a long line of improvements that can be made. With a few more fundamental changes in design, memory might actually be able to keep up with processors in the future. One such design change would be getting the main memory bus off the motherboard PCB. With the memory controller on the processor itself, the compatability or portability of the memory modules between Opteron generations is no longer a viable excuse. There is no reason why the memory can't be a stacked silicon module that plugs into the side of the processor. That right there would solve quite a few problems as well as take full advantage of the Opteron's built-in memory controller and provide memory performance unchallenged by either DDR SDRAM or RDRAM technologies.
I'm betting we'll see 1Ghz memory(not effective via DDR or QDR, I'm talking actual bus frequency) within 1 year from this day.
Anyone wanna take my money?
You can't legislate goodness. Let each to his own destiny, by will of his freely made choices.
IBM has been testing SiGe chips at up to at least 350-400GHz last time I checked and producing and selling chips at up to at least 110GHz. Intel's made claims of tested transistors in the THz range.
Not to rain on the "OMG look how many GHz or THz that is!" parade, but there are even higher numbers to "OMG" at =)
Put InP (indium phosphide) and SiGE (silicon germanium) in to google for more max Ghz fun...
There is probably even faster stuff than that out there!
I am sick and tired of people claiming tubes sound better,
Well, since "sounding better" is a subjective judgement, anyone who says so is right. They'd be wrong if they claimed that tube amps were more accurate than transistor amps.
What tubes have going for them is that they don't clip the way that transistors do when their power limit is exceeded. Transistors clip, tubes go non-linear and approach their power limit asymptotically.
Of course, anyone who routinely overdrives their amp like that has probably lost the ability to hear much more than the ringing in their ears.
-jcr
The only title of honor that a tyrant can grant is "Enemy of the State."
Sure, they just say "FET device". Possibly just a single FET. Early days yet, it is a long way to get from a single transistor to a microprocessor. The more immediate applications for this are high power high frequency devices which currently still use vacuum tubes.
Actually, according to an article in a recent (this month?) Wired magazine, there is a corporation in Boston which is developing ultra-pure diamonds using a vapor disposition techinque. While the initial generations of diamonds produced in this way will be expensive, if they prove useful, mass production will ultimately drive the price of diamonds through the floor. Haha! take that DeBeers! (seriously, DeBeers's corporate executives cannot come to the US without being arrested, and they are single handly responsible for keeping the price of diamonds so high that wars can be financed via "blood diamonds" even though the mineral is not actually rare)
Galium Arsenide is the material of the future, and always will be.
The reasons that we can't with current generation hardware are:
Hardware, software, and blinking lights!
Interesting - I would have thought that was not true, but see this breakdown of the differences between SRAM and DRAM. Static RAM is indeed static - I would have thought the flip-flops would pass current all the time (making them comparable to DRAM), but they actually don't.
Many other properties are required to make transistors useful for implementing microprocessors, memory, or other complex peripherals used in computers. Complementary devices are needed (high speed in both N and P channel) to implement high density logic without excessive static power consumption. Withness the sucess of GaAs semiconductors for CPUs, using only N-channel devices! (but GaAs is well established for RF amplifiers of course).
High quality insulators are needed to fabricate many layers of metal interconnect (SiO2 happens to be glass... but obtaining good insulators on top of most materials is quite difficult). A way to produce a good ohmic contact in a tiny area between the semiconductor and metal is also needed (for a single transistor used in an amplifier, only 3 contacts are made and large area can be devoted to them). Vertical fabrication techniques used for high power single transistors obviously can't be used for complex circuits with many transistors, so all these requirements must be able to be met with many fabrication steps.... and the fact that they're growing the film at such a low temperature will make all the subsequent steps quite difficult. Even after acheiving all these difficult properties, a new material like diamond must support extreemly small geometries over very large areas to achieve the transistor counts required to be competitive with modern silicon CPUs.
Perhaps someday diamond semiconductors may be useful for complex circuitry.... but to think that now is quite silly. The targeted application is analog signal gain at these extreemly high frequencies, which will open up a massive portion of spectrum that's previously been unavailable. That is arguably even more important than continuing Moore's "law" for increase in CPU speed and memory size.
PJRC: Electronic Projects, 8051 Microcontroller Tools
Yes, they are. Or rather, it is: De Beers is the only one of any significance. This article about the "Diamond Invention" has been posted to Slashdot before, and despite being written in 1982 is just as true today. The myth that diamonds are valuable was created by De Beers early last century, and they have been able to maintain that myth through the virtue of being a thuggish monopoly.
Don't buy diamonds. For our wedding bands, we chose unadorned titanium. :)
Given that the final shuttle accident report was released today, I'm surprised that no-one else has touched on this topic.
The Reinforced-Carbon-Carbon panels have been noted to get very pitted and pot-marked over time. Indeed there has always been serious concerns over this component.
Given the chemical process for synthesizing diamond wafers, isn't it reasonable to deposit a single sheet part super heat conductive material that would replace the reinforced-carbon-carbon on the space shuttle wings. Diamond is the hardest substance known to man. Isn't it reasonable that such a macro-application would be reasonable and logical.
Other near term application could be heat sinks in other industrial super-heated applications. I could even imagine sythesizing the linings of cannon barrels out of sheet diamond. How about aircraft "black boxes" made out of sythesized diamond so that they absoluetly CANNOT be destroyed.
On more application could be to organically grow the hull of a small submarine capable of diving to tremendous depths. A sufficiently polished application could be optically transparent!!!! That is no portholes required. Remember "transparent aluminum" from star trek. A chemically deposited transparent diamond panel could probably kick it's ass in strenth.
How about armor for tanks, helicopters and planes???? A thin panel may be stronger then the most exotic alloy.
A sufficiently advaned systhesizing process may be capable of produce "machine grade" parts that will effectively NEVER wear.
The 20th century was the century of steel. With a reliable diamond production process, and technology that generates carbon nanon-tube threads (as well as bucky ball "bearings"), this could be the century of carbon!!!!!!
-------- -------- Support Wesley Clark for president!!!