Tuesday, September 26, 2017

The terahertz gap

At a thesis proposal talk yesterday, I realized that I hadn't ever written anything specifically about terahertz radiation (THz, or if you're trying to market something, t-rays).   Terahertz (1012 Hz) is the frequency of electromagnetic radiation higher than microwaves, but lower than what is traditionally labeled the far infrared.  Sometimes called "mm wave" radiation (1 THz would be a free-space wavelength of about 0.3 mm or 300 microns), THz is potentially very useful for communications (pdf, from here), imaging (here, here, here), and range detection (see here for an impressive google project; or here for an article about THz for self-driving cars), among other things.  It's also right around the frequency range of a lot of vibrations in molecules and solids, so it can be used for spectroscopy, though it's also around the energy range where water vapor in the atmosphere can be an efficient absorber.

This frequency region is an awkward middle ground, however.  That's sometimes why it's referred to as the "terahertz gap".

We tend to produce electromagnetic radiation by one of two approaches.  Classically, accelerating charges radiate electromagnetic waves.  In the low frequency limit, there are various ways to generate voltages that oscillate - we can in turn use those to drive oscillating currents and thus generate radio waves, for example.  See here for a very old school discussion.  It is not trivial to shake charges back and forth at THz frequencies, however.  It can be done, but it's very challenging.  One approach to generating a pulse of THz radiation is to use a photoconductive antenna.  Take two electrodes close together on a semiconductor substrate, with a voltage applied between them.  Smack the semiconductor with an ultrafast optical pulse that has a frequency high enough to photoexcite a bunch of charge carriers - those then accelerate from the electric field between the electrodes and emit a pulse of radiation, including THz frequencies.

The other limit we often take in generating light is to work with some quantum system that has a difference in energy levels that is the same energy as the photons we want to generate.  This is the limit of atomic emission (say, having an electron drop from the 2p orbital to the 1s orbital of a hydrogen atom, and emitting an ultraviolet photon of energy around 10 eV) and also the way many solid state devices work (say, having an electron drop from the bottom of the conduction band to the top of the valence band in InGaAsP to produce a red photon of energy around 1.6 eV in a red LED).  The problem with this approach for THz is that the energy scale in question is very small - 1 THz is about 4 milli-electron volts (!).  As far as I know, there aren't naturally occurring solids with energy level splittings that small, so the approach from this direction has been to create artificial systems with such electronic energy gaps - see here.   (Ironically, there are some molecular systems with transitions considerably lower in energy than the THz that can be used to generate microwaves, as in this famous example.)

It looks like THz is starting to take off for technologies, particularly as more devices are being developed for its generation and detection.  SiGe-based transistors, for example, can operate at very high intrinsic speeds, and like in the thesis proposal I heard yesterday, these devices are readily made now and can be integrated into custom chips for exactly the generation and detection of radiation approaching a terahertz.  Exciting times.


xykademiqz said...

Hi, I'd like to add that the THz gap usually refers not to the total absence of sources in the THz but to the absence of coherent sources with high enough power. Also, I'd like to mention that the broad concept of difference frequency generation (e.g., Cherenkov nonlinearity, frequency combs, four-wave mixing) is a very popular indirect way to generate THz radiation. Tunability of the frequency and the output powers differ among different incarnations.

Douglas Natelson said...

Thanks. I was going to talk about frequency mixing and nonlinear optics in there, too, but I was struggling to come up with a really accessible, intuitive way to explain it. I guess I could talk about overtones in musical instruments, and for things like optical rectification it's easy to look at a classic anharmonic spring and see that greater excitation = shift in average oscillator position. I need to think more about accessible ways to explain such concepts if I want to be better able to write for a very broad audience.

DanM said...


To be fair, terahertz gas lasers have been around since the late 1960's, and can produce upwards of a watt, continuous-wave at specific frequencies. Very coherent and very high power (for a terahertz source). They just happen to be unpopular because they're, well, gas lasers.

But anyway.

DanM said...

Also, just so you know, at terahertz conferences you are no longer allowed to use the phrase "terahertz gap" or to show the spectrum, or else you are at risk of getting mocked. Ask a certain UC San Diego professor about the "spectrum shower" t-shirt he was required to wear, once.

Douglas Natelson said...

I knew I'd get a response from you, Dan :-) Maybe that's why I waited so long to write something on this.
I like the public mocking approach. Perhaps there needs to be more of that when people talk about other over-used graphs, like Moore's Law or Tc vs. time for superconductors.

Anonymous said...

Gas lasers also lack broad tuning ability... The other funny thing is the overlap spectrally of the nomenclature "Terahertz", "Far infrared", and "sub-mm waves".... You find almost non-overlapping sets of research papers depending on which descriptor you use but the actual wavelengths overlap heavily.

DanM said...

I saw Rick give a talk yesterday. He showed the spectrum, and then looked straight at me and apologized. Seeing as the audience was mostly not terahertz people, I forgave him. Think globally, act locally.