Realistic RTDs will have nonlinear electrostatic potential in their emitter. Typically a triangular well is formed in the emitter due to the applied bias and the emitter thus contains discrete quasi bound states. These states are typically very strongly bound with a very long lifetime of a very narrow resonance energy. However, the emitter also contains carrier sheet densities in excess of 1011/cm2 in quasi equilibrium conditions which characterized by strong scattering. Such strong scattering will broaden out the emitter quasi-bound states. NEMO introduced an empirical broadening model which accounts for the quasi-equilibrium conditions in the reservoirs. The overall device is partitioned into emitter & collector reservoirs and the central out-of equilibrium (NEGF) region. The broadening in the reservoir is very simple resulting in a non-Hermitian system – only charge is computed with Equilbrium Green Functions. The central RTD “feels” the surrounding reservoirs which contain broadened states. Carriers are injected from these reservoirs and the RTD.
For typical high performance InGaAs/InAlAs RTDs the relaxation is set to η=6.6meV which corresponds to a scattering time of about t=0.1ps. The relaxation rate in the reservoirs immediately affects the current in the device and the relaxation rate should not be used to match experimental data on a one-time basis.
- Realistic doping profiles
• triangular quantum wells in the emitter,
• confined states in the emitter
- very long lifetime / very narrow states in the mathematically ideal case
- High electron density in the emitter & Equilibirum conditions!
• strong equilibrating scattering,
• states are broadened
- NEMO introduced an empirical broadening model
- Partition the device into reservoirs and NEGF region Reservoirs are non-Hermitian – compute charge only Central NEGF region sees effects of thermalized states
- For typical high performance InGaAs/InAlAs RTDs:
• set the relaxation to h=6.6meV
• scattering time of about t=0.1ps.
- The relaxation rate should not be used to match experimental data on a one-time basis.
Università di Pisa, Pisa, Italy