Upconverting Mid IR(THz) Radiation into the Visible

There are lots of resonator-enhanced nonlinear optics things. Room-temperature mid-IR detectors are mostly crappy (HgCdTe and suchlike), so a better one is welcome.

This one is going to be a bit of a schlepp to get working at any reasonable efficiency over any reasonable detection area, but who knows--somebody might figure out a way to scale it.

Cheers

Phil

I'm betting that the Astronomers will be pretty interested. Opens up a new EM window; never know what you'll find.

As for the underlying mechanism, are there any tutorial articles you could suggest?

Thanks,

Joe Gwinn

Nah, astronomers can afford cryogenically-cooled detectors. HgCdTe arrays are fairly heartbreaking, but at least you get many pixels, and a given pixel can have any etendue you like.

A single resonator gets you at most one optical mode, i.e. an etendue of lambda**2/2 per polarization. With thermal light, the mode volume is only very sparsely filled, so that's not a lot of photons per second.

I'm actually not a big nonlinear optics guy--I took one course on it in grad school, and have never built a parametric converter or optical harmonic generator. I used to have one (or at least IBM bought one for me), but I haven't gone through the math in 35 years or so.

From a phenomenological POV, the magic-goo guys give you this nice material with some nice large second- and third-order dielectric susceptibilities. (For a solid, these are both tensors.)

For given incident E fields, you use the susceptibilities to calculate the nonlinear dielectric polarization, and apply the Helmholtz propagator to figure out what the output power will be for the given geometry. (I forget exactly how that part of the calculation goes, but it isn't super complicated IIRC.)

Then you moan about how tiny the output power is, and look for ways to make it bigger. With a crystal, you can sometimes find a geometry and a choice of incident k vectors so that the nonlinear polarization phase-matches to a propagating wave.

At that point, the output signal builds up and builds up with distance, so you can sometimes make a pretty efficient frequency converter. (For parametric converters, "pretty efficient" is a few percent in general.)

For a resonator, you probably have to use FDTD to calculate the output, though you can certainly do some decent first-order estimates analytically. It'll look like a single dipole absorber, and probably a single dipole emitter as well. (That's a clue that the efficiency is going to be poor when the input and output wavelengths are very different.)

Fun stuff. I'd like a chance to build a parametric converter one of these times, but it's well below making ECDLs on the priority list.

Cheers

Phil Hobbs

Ahh. Way more complicated than I realized.

Maybe the THz imaging folk then. Or just bored physicists.

Joe Gwinn

It's worth trying this sort of stuff out--new technologies typically don't appear overnight. However, for it to be more than a pretext for writing papers and dissertations, you have to think fairly deeply about how to scale it to technologically-useful levels of efficiency and cost.

Coming up with a new scheme that looks like it could scale well is very exciting.

Cheers

Phil Hobbs