Diamond CMOS at last?

Author: EIS Release Date: Apr 2, 2024

P-channel diamond mosfets have been made many times because the carbon bulk can be doped with boron, or surface hydrogen can pull appropriate carriers into the channel.

NIMS n-channel diamond fet
N-channel diamond mosfets have not proved so easy, blocking the route to diamond CMOS and all the high-voltage (bandgap = 5.5eV), high temperature and high thermal conductivity advantages that it might bring.

Now researchers at Japan’s National Institute for Materials Science (NIMS) have made n-channel diamond mosfets, by finding a way to phosphorus-dope the carbon lattice without: too much phosphorus, the large atoms doing too much damage, or the phosphorus atoms bringing dopant-neutralising hydrogen with them.

The result is a mosfet with ~150cm2/V/s field-effect mobility at 300°C – the graph shows curves from Vg = -20V (black line) to Vg = +10V (yellow).

Part of the key to the device is having a low-doped (~1017/cm3 ) n-channel layer on the synthetic diamond substrate, over-laid by highly-doped epitaxy. “The use of the latter diamond layer significantly reduced source and drain contact resistance,” according to the Institute.

The n+ layer was etched away to expose the channel, but left under the source and drain contacts. aluminium oxide was used as a gate dielectric.

Epitaxy was grown using microwave plasma chemical vapor deposition equipment made at the institute, which avoided over-doping and excessive hydrogen, and the resulting layers were atomically flat, with only a slight angle in the cut of the underlying substrate causing steps in the epitaxy surfaces.

The “only one point we can say is: we are using two identical CVD machines adapted to different doping levels for channel layer and contact layer”, NIMS researcher Satoshi Koizumi told Electronics Weekly.