(c) Dr Paul Kinsler. [Acknowledgements & Feedback]

Logo (c) Paul Kinsler Logo (c) Paul Kinsler

Optically Pumped Terahertz (far-infrared) Emitters

This work was done within the Institute of Microwaves and Photonics at the University of Leeds, and involves theoretical collaboration with Dr P. Harrison and Dr R.W. Kelsall.

Is this just Greek to you? Why not try looking at the IRC #physics channel information maze for some simple explanations of semiconductor physics.

Project Summary.

The race to make a far--infrared laser in semiconductors is happening right now. This THz frequency region has potential for use in short range communications such as wireless computer networks, and in micro- and medical imaging technologies -- but device applications are more practical when based on solid-state technology, thus providing the motivation for this project. Current solid-state microwave oscillators can be used as radiation sources for frequencies up to approximately 150 GHz (2mm); optical sources in the infra-red are available for frequencies down to typically 10 THz (30 microns). In between these two lies the technologically and commercially vital terahertz frequency (far-infrared) range, so the aim of this project is to design a convenient radiation source capable of bridging this gap. To do this we will explore ways of extending the operating wavelengths from the mid infrared (8-14 microns) to the far infrared (beyond 30 microns).

Our approach is to obtain terahertz emission from asymmetric quantum wells in GaAs/AlGaAs, aiming to suceed at designing devices that have the room-temperature population inversion necessary for laser emission. For simplicity, we will initially investigate optically (CO2 laser) driven quantum wells. This follows the recent success of Capasso's `Quantum Cascade Laser', which utilizes transitions between subbands in quantum well structures.

diagram of a quantum well

However, getting such a device to work is tricky. In order to get a population inversion between THz-separated two subbands in a prototype quantum well device, we have to know a lot about how the electrons move around the structure, how they get in, get out, and how they hop between their quantum energy levels. Unlike the mid-infrared region, here the useful radiative process are completely swamped by electrons interacting with the sea of acoustic and LO phonons that make up the vibrations in the crystal lattice of the semiconductor. Worse, the electrons repel and scatter off each other, a problem that gets worse as we increase the number of them when tring to increase the laser output. The only way around this is to get clever, and study in detail this myriad of scattering processes. By tailoring the design, we can alter the wavefunctions, the subband spacings, and the phonons -- thus altering the scattering processes. Here at the IMP we are studying sets of prototype designs using theory and comprehensive computer simulations to understand how to make a THz laser work. These designs are for simple stepped quantum wells, grown with AlGaAs barriers, with a GaAs deep well and a AlGaAs step. In the same group, biased triple-well structures are also being studied by Kate Donvan.

The initial work focussed on averaged scattering rates, and assumed a range of "reasonable" electron distributions in the subbands. These early studies showed limited potential at low temperatures for the simple three-subband devices, and, more excitingly, the four-subband structures shows promise at room temperature -- a must for anything to have real commercial prospects.

However, there were some complications. Initially the modelling assumed that bulk-like phonons permeated the structures as an approximation to the real phonon modes. Subsequently I extended the modelling to include a more sophisticated phonon model known as the Dielectric-Continuum-Model (DCM). The enabled me to understand better the electron dynamics within the multi-level quantum system defined by the well. The effects of phonons on the dynamics is crucial. Room temperature operation is a challenging target, because although the emitter subband spacings are less than the minimum LO phonon energy, the extra thermal energy of the electrons can allow LO phonons to be created. This can disrupt any population inversion that has been created. The DCM caculations gave results that were significantly different in detail to the original calculations, with resulting changes to our projected device performance. However, from a more general perspective the changes were not too great.

Diagram: Inter-subband electron-phonon scattering. (12k gif)

Note also that the electron-electron scattering is of similar strength to the phonon processes, and cannot be ignored. The scattering rates change with electron density, and are significantly affected by the Pauli exclusion effect. And since our designs rely on asymmetric quantum wells, we have additional Auger scattering processes. In some cases these can dominate the normal "symmetric" scattering processes. I added in these electron--electron scattering processes, and used these more complete calculations to estimate the potential of sets of three-subband and four-subband prototype structures. It seems the three-subband prototypes have little potential for terahertz emission due to the strong electron--electron scattering -- the only exception being at lower electron densities. However, recent results show that our four-subband prototypes are very promising, even to the extent of there being the possibility of room temperature operation.

Diagram: Inter-subband electron-electron scattering. (11k gif)

We also needed to treat the electron dynamics properly, as these earlier calculations had all made assumptions about the electron distributions in the subbands. I am doing this by implementing a Monte Carlo program that can track an ensemble of electrons as they scatter off phonons and other electrons. So far, the Monte Carlo results highlight the fact that the combination of optical pumping and LO phonon emission cause significant electron heating, effectively killing off any potential for three-subband devices at low temperature. As I type, there are four-subband calculations in progress -- the electron heating is likely to be less of a problem for these, but nevertheless the rapid LO phonon emission process make the outcome uncertain.

Other References:

( Research homepage; )
Date=20000223 0224 19990121 1203 19970702 Author=P.Kinsler