Silicon Replacement

[Cogman]

I got thinking about a post I made way back about the first creation of artificial diamonds. Now Ive been thinking. Diamonds have basicly no electrical conduction, but on the other side they have an extreamly high thermal conduction (very good for cooling). So is it possible that Artificial Diamonds Become the next Silicon (at least in the cpu core) I Think it would do wonders for the current computing. (starts thinking if my cpu core was entirly diamond and my radiator and water block where diamond, The overclocking would be grand)

 

[Falloutboy]

this has been discussed before and seems like a possibility but its 10 years away from aplication. but I agree its the way to go. chips made with diamonds could run at a much higher temp without issue (200-400C). granted a computer would be a space heater and would need some kind of exhaust outside the house or it would be a sauna inside

 

[iwantanewcomputer]

they need to be able to produce artificial diamond much cheaper(neaar the cost of silicon) and get it a lot purer. one of the biggest problems of current artificial diamons is their purety, it's not as good as real diamonds. and purity has to be extremely good for a microprocessor

the thermal conductivity would be very useful. diamonds have nearly twice the conductivity of copper and way more than silicon so this would allow the core (with the same heat output) to be much nearer to the actual temperature of the heatsink base, which would be closer to ambient

 

[cheesehead]

Although diamond does not conduct electricity, carbon nanotubes do so very well. I'm a little hazy on the chemistry and electronics, but in a few years we could feasibly be playing around with carbon-based processors on a 25nm or less process, built from trillions of nanotubes embedded in a thin peice of diamond. Because carbon has a higher resistance than silicon, the processor would run VERY hot, but it would take up so much less space it would put out less heat overall. Also, it would not be that difficult to put four our more cores in a single bit of diamond, which would make cooling a great deal easier.
Best of all, carbon is 100% nature-friendly!

 

[Falloutboy]

Originally posted by: iwantanewcomputer
they need to be able to produce artificial diamond much cheaper(neaar the cost of silicon) and get it a lot purer. one of the biggest problems of current artificial diamons is their purety, it's not as good as real diamonds. and purity has to be extremely good for a microprocessor

the thermal conductivity would be very useful. diamonds have nearly twice the conductivity of copper and way more than silicon so this would allow the core (with the same heat output) to be much nearer to the actual temperature of the heatsink base, which would be closer to ambient

from what I've read is the new artificial diamonds are identical to normal diamonds except they are more pure than mined ones

 

[TuxDave]

O_O Ok, back up everyone. If diamond does not conduct electricity, then how is it gonna replace silicon again? Although silicon is generally an insulator, it can be doped to make it conductive. You need SOMETHING to be conductive if you want the darn thing to do anything. If a diamond transistor doesn't turn on, it's useless.

 

[Cogman]

You just said it, Silicon is not conductive either. their is a problem with geting electrical conductivity in a diamond because as I remeber they use silicons ability to shrink when heated to place the materials needed (As I recall they do it this way, but times have changed and this is a long time ago). honestly I cant think of a good way they could get something conductive into the diamonds. But it would be very cool if they did.

Also, I thought Carbon-Nanotubes would give a higher Preformance/heat ratio? It would also be able to withstand the higher heats as Carbon-N.. are very durable.

 

[TuxDave]

Originally posted by: Cogman
You just said it, Silicon is not conductive either. their is a problem with geting electrical conductivity in a diamond because as I remeber they use silicons ability to shrink when heated to place the materials needed (As I recall they do it this way, but times have changed and this is a long time ago). honestly I cant think of a good way they could get something conductive into the diamonds. But it would be very cool if they did.

Also, I thought Carbon-Nanotubes would give a higher Preformance/heat ratio? It would also be able to withstand the higher heats as Carbon-N.. are very durable.

But I said the reason why Silicon is nice because you CAN make it conductive by doping it (injecting it with type III or V elements). Silicon by itself is not a very good insulator either until you oxidize it into SiO2. The good think is that SiO2 bonds well with Si since it has similiar lattice structures.

 

[Matthias99]

Originally posted by: TuxDave

Originally posted by: Cogman
You just said it, Silicon is not conductive either. their is a problem with geting electrical conductivity in a diamond because as I remeber they use silicons ability to shrink when heated to place the materials needed (As I recall they do it this way, but times have changed and this is a long time ago). honestly I cant think of a good way they could get something conductive into the diamonds. But it would be very cool if they did.

Also, I thought Carbon-Nanotubes would give a higher Preformance/heat ratio? It would also be able to withstand the higher heats as Carbon-N.. are very durable.

But I said the reason why Silicon is nice because you CAN make it conductive by doping it (injecting it with type III or V elements). Silicon by itself is not a very good insulator either until you oxidize it into SiO2. The good think is that SiO2 bonds well with Si since it has similiar lattice structures.

I think (although I haven't been following this research very closely at all) that the idea would be to use diamond in a 'silicon-on-insulator' (SOI) type configuration (since it conducts heat very well and is an excellent insulator), or to use it for cooling purposes, NOT to use it as a 100% replacement for silicon. The diamond/carbon nanotube stuff sounds interesting, but is years and years from going anywhere (again, AFAIK).

 

[TuxDave]

Originally posted by: Matthias99

Originally posted by: TuxDave

Originally posted by: Cogman
You just said it, Silicon is not conductive either. their is a problem with geting electrical conductivity in a diamond because as I remeber they use silicons ability to shrink when heated to place the materials needed (As I recall they do it this way, but times have changed and this is a long time ago). honestly I cant think of a good way they could get something conductive into the diamonds. But it would be very cool if they did.

Also, I thought Carbon-Nanotubes would give a higher Preformance/heat ratio? It would also be able to withstand the higher heats as Carbon-N.. are very durable.

But I said the reason why Silicon is nice because you CAN make it conductive by doping it (injecting it with type III or V elements). Silicon by itself is not a very good insulator either until you oxidize it into SiO2. The good think is that SiO2 bonds well with Si since it has similiar lattice structures.

I think (although I haven't been following this research very closely at all) that the idea would be to use diamond in a 'silicon-on-insulator' (SOI) type configuration (since it conducts heat very well and is an excellent insulator), or to use it for cooling purposes, NOT to use it as a 100% replacement for silicon. The diamond/carbon nanotube stuff sounds interesting, but is years and years from going anywhere (again, AFAIK).

Ok... that's still a little strange but I guess the idea is mostly science fiction right now.

 

[Cogman]

hey it was Just a thought. SciFi is a little strong as I think it is possible.

 

[TuxDave]

Originally posted by: Cogman
hey it was Just a thought. SciFi is a little strong as I think it is possible.

Science fiction doesn't mean it'll never become feasible. It just means that there's many obstacles and no solutions in sight.

 

[jhu]

Originally posted by: iwantanewcomputer
they need to be able to produce artificial diamond much cheaper(neaar the cost of silicon) and get it a lot purer. one of the biggest problems of current artificial diamons is their purety, it's not as good as real diamonds. and purity has to be extremely good for a microprocessor

the thermal conductivity would be very useful. diamonds have nearly twice the conductivity of copper and way more than silicon so this would allow the core (with the same heat output) to be much nearer to the actual temperature of the heatsink base, which would be closer to ambient

diamonds can be made for cheap! take a look here. hopefully debeers will go away in a few yeas.

 

[Falloutboy]

cost is still a bit too high I think the number is around 100 bucks for a 2 carate (still really cheap comaperd to natural diamonds but too much and too small for mass produced chips)

 

[darfur]

I wrote a paper about the use of artificial diamonds as a silicon replacement last fall for a class.

If I can remember from my paper correctly (I don't have it anymore), diamonds can withstand 1000°C until they start to have problems. Also, I believe researchers inject electrically charged boron into the diamonds to give them conductivity.

I could be very wrong, but this is how i remember it.

EDIT: falloutboy is correct. Artifical diamonds can now be produced more purely and perfectly than nature produces them. This is how you can distinguish between the two. The artificial ones are simply too perfect to be natural.

 

[TuxDave]

Originally posted by: darfur
I wrote a paper about the use of artificial diamonds as a silicon replacement last fall for a class.

If I can remember from my paper correctly (I don't have it anymore), diamonds can withstand 1000°C until they start to have problems. Also, I believe researchers inject electrically charged boron into the diamonds to give them conductivity.

I could be very wrong, but this is how i remember it.

EDIT: falloutboy is correct. Artifical diamonds can now be produced more purely and perfectly than nature produces them. This is how you can distinguish between the two. The artificial ones are simply too perfect to be natural.

What class was this for and what are the issues that you discovered that prevents diamonds from use in today's technology? Etching issues? Metallization issues? Performance issues?

 

[darfur]

It was just a computer literacy course, but it's required before i can take other computer science courses.

1) Intel and AMD have a very large investment in silicon, and there is no reason for them to start using diamonds.

2) They can't produce large enough diamond wafers to make a sufficient amount of chips from.

3) Like fallout boy said, they are still too expensive to mass produce, especially for consumer grade products.

4) Artificial diamond creation is still being worked on. They aren't at a stage where they can use them for that kind of application. IIRC they are still 10-15 years out. (There have been people working on creating them for at least 15 years already).

I don't believe there are any performance issues to speak of, because I don't think a diamond processor has been created yet that actually does something.

I don't recall off hand any specific problems they are having with etching or metallization, but the four reasons above I think are the big ones.

BTW TuxDave, I like your sig

 

[TuxDave]

Originally posted by: darfur
It was just a computer literacy course, but it's required before i can take other computer science courses.

1) Intel and AMD have a very large investment in silicon, and there is no reason for them to start using diamonds.

2) They can't produce large enough diamond wafers to make a sufficient amount of chips from.

3) Like fallout boy said, they are still too expensive to mass produce, especially for consumer grade products.

4) Artificial diamond creation is still being worked on. They aren't at a stage where they can use them for that kind of application. IIRC they are still 10-15 years out. (There have been people working on creating them for at least 15 years already).

I don't believe there are any performance issues to speak of, because I don't think a diamond processor has been created yet that actually does something.

I don't recall off hand any specific problems they are having with etching or metallization, but the four reasons above I think are the big ones.

BTW TuxDave, I like your sig

Haha... thanks.

Well, if they manage to bring down cost and have the basic processes down, I think we'll get some very interesting things happening in computer technology. *cross fingers*

 

[darfur]

Something else I seem to remember. Silicon will reach it's limit as far as heat goes around 2015-2020 if current trends continue.

Don't quote me on that one though.

 

[FrustratedUser]

Originally posted by: Cogman
I got thinking about a post I made way back about the first creation of artificial diamonds. Now Ive been thinking. Diamonds have basicly no electrical conduction, but on the other side they have an extreamly high thermal conduction (very good for cooling). So is it possible that Artificial Diamonds Become the next Silicon (at least in the cpu core) I Think it would do wonders for the current computing. (starts thinking if my cpu core was entirly diamond and my radiator and water block where diamond, The overclocking would be grand)

It is not as easy this.
The reason Si is used is not the thermal conductivity. It is a semiconductor and it can be either P or N doped. You can also grow tiny layers of other materials used in semiconductors.
Other materials that are used now for very high frequency components are the so called III-V materials, found in group III to V in the periodic system. Two examples are GaAS and InP (Gallium Arsenid and Indium Phosfide).

Diamond on the other hand is not a semiconductor.

A material that is being used now for high power and high frequency transistors is SiC (Silicon Carbide).