Diamond Transistors-Any Being Made?

[kelticwizard]

Read an article some time ago about the use of diamond as a substrate for transistors, and I presume chips as well.

Any models coming close to the production line?

Also, does anyone have an idea what the forward voltage would be on a diamond transistor-like 0.6 V for silicon or 0.3 V for germanium? What would the voltage for a diamond transistor or diode be?

[I_Forgot]

I think it would be silicon on diamond substrate, so the transistors wouldn't change much. The big advantage to using a diamond substrate instead of silicon is that diamond is a much better thermal conductor. The problem is that silicon and diamond don't have matching crystal lattices so such pairings tend to be weak mechanically. Stresses due to differing thermal coefficients can cause the thing to fracture at the boundary.

I_F

[SY]

You can't use diamond as the actual semiconductor- the bandgap is waaaay too large. I-F has outlined the advantages as a substrate.

[kelticwizard]

Thanks. The article only touched on diamond's possible use in transistors. The main part of the article was about diamondlike coatings, and it even gave a project where a diamondlike coating could be cooked up in your home oven. It was the NY Times from way back.

Since then I was on the lookout for diamond transistors which didn't seem to be coming anytime soon. Now I know why.

[JohnG]

I would not hold your breath. Diamond is a wide-bandgap semiconductor, and in theory one can make transistors. I think that currently, only experimental Schottky diodes have been made, although maybe a p-n junction diode has been fabricated by now. A BJT would probably have a BE drop about 3-4V. However, a MOSFET would not have such a voltage drop. The advantages of diamond semiconductors would be high voltage parts with low on-state voltage drop compared to silicon, and high temperature operation. Perfect for your direct-drive ESL...

Note that a silicon transistor on a diamond substrate may have some advantages, i.e. a highly thermally conductive insulating substrate, but ultimately it's still a silicon transistor.

Note that wide-bandgap devices are coming. SiC (silicon carbide) Schottky diodes are available from Digikey, and a number of researchers have fabricated SiC BJTs, JFETs and MOSFETs. I'm not a semiconductor designer, but my understanding is that the wafer quality and size is where silicon was decades ago, and the same properties that make wide band-gap materials attractive electronically also make them the devil to process. SiC has all kinds of crazy defects, and doping is by ion implantation or epi layer growth, since diffusion apparently doesn't work very well (temps to high?).

John

[kelticwizard]

John:

Thanks for your informative answer.

Why is the work being done with silicon carbide? Does it contain properties similar to diamond but is more workable?

[JohnG]

I'm really an end-user, but I'll answer as best I can. If anyone knows more, let me know.

SiC ( silicon carbide), like diamond, has a wide bandgap, which is good for high temperature and high voltage operation. It also has high thermal conductivity, although not as good as diamond. However, silicon carbide has a lot more process development behind it than diamond, and wafer sizes are increasing with decreasing defect density, meaning more devices per wafer and higher yield, both of which are driving the cost down. Like I mentioned before, SiC Schottky diodes are available from Digikey. These are great for switching power supplies, but I'm not too sure there are major benefits for anything under 200V.

Another big potential benefit of SiC is the ability to grow a silicon dioxide insulating layer. This is a huge natural benefit that is one of the main reasons standard Si technology has progressed so far. However, because of the carbon in SiC, and many other reasons I can BS about but don't actually understand, the oxide and the interface between the oxide and the SiC has a lot of room for improvement. That's why SiC MOSFETs are not commercial yet.

Diamond may be on the way, but my guess is it will be a while before anything is commercially available, and good transistors are probably a long ways off.

While this may not be of immediate help for audio, it is pretty interesting.

John