Some of the MMICs are cool. Simple, cheap, fast, unconditionally stable. They are all spec'd in RF terms but most make great time-domain amps.
InP is about the fastest semiconductor out there--it's about the only one where you can get over 1 THz f_T.
The most interesting part to me is whether you can make a small device exhibit gain above its plasma frequency.
In a plasma, electromagnetic waves can't propagate above a frequency proportional to the square root of the free charge density. (The math is a bit like a metal waveguide below cutoff--the Z component of the propagation vector k_z becomes imaginary, so there are only real-exponential solutions.)
In a macroscopic plasma, the dropoff is very striking, but I don't know how much difference it makes in a 10-nm device, say.
BITD I had an idea for making tunnelling varactors based on my Ni-NiO-Ni tunnel junction technology. The idea was to find a metal that formed a negative-height Schottky barrier with NiO. (Negative-height barriers exist--that's one common way of making ohmic contacts to silicon.)
Theory predicted that the result would be electrons spilling out into the oxide, just as in a normal PN junction, _but at metallic carrier densities_, so that the plasma frequency would be in the hundreds of terahertz.
Since the oxide was so thin (30 angstroms or so), the result ought to be a varactor that works up to the near IR. I was hoping to import all kinds of classical microwave devices into the photonic/plasmonic space--parametric amps and dividers especially.
I tried several candidate metals, mostly rare earths, but didn't find a suitable one that didn't react with the NiO. :(
Cheers
Phil Hobbs
2-terminal oscillator devices, like tunnel diodes and impatt sorts of things, seem to have fallen out of favor. A terahertz tunnel diode would be cool.
My senior paper at Tulane was "The tunnel diode slideback sampling oscilloscope" which won some award and required me to present it at some IEEE sessions, a real bore. My wife accompanied me when I presented, and the old farts pulled me aside and told me "We don't allow women here."
Do you know why InP is used only in some niches, tiny parts? We use some SOT89 mmics, still classic darlington types.
If anyone is interested in using InP MMICS (or the newer GaN parts) as pulse amps, beware that newer parts often have an internal bias servo loop that wrecks low frequency behavior. These tend to be the kind you can power from a stiff source, typically a 5-volt power supply and a series inductor.
The data sheets and s-params of course don't go below 100 MHz to better hide the LF behavior.
Also note that a "50 ohm" part rarely is. We experiment and hack our own Spice models.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.