Monte Carlo transport analysis to assess intensity dependent response of a carbon-doped GaN photoconductor

Abstract

Evaluation of the photoresponse in wurtzite GaN photoconductive switches is presented based on kinetic Monte Carlo simulations. The focus is on electron transport physics and assessment of high frequency operation. The roles of GaN band structure, Pauli exclusion, and treatment of internal fields based on the fast multipole method are all comprehensively included. The implementation was validated through comparisons of velocity-field characteristics for GaN with computational results in the literature. Photocurrent widths of less than ∼7 ps for the 1 μm device can be expected, which translates into a 100 GHz upper bound. Photocurrent pulse compression below the laser full width at half maxima at high applied fields are predicted based on the interplay of space-charge effects and the negative differential velocity characteristics of GaN.

Density-Dependent Effects on Pulse Compression in GaN Photodetectors Probed by Monte Carlo Studies

“…The scope of this work is to further probe of the potential applications of GaN PCSS devices for pulse compression, which could then lead to increased frequency operation. Previous work has been performed to show that when operating in the negative differential mobility (NDM) regime which is present in the GaN field versus velocity characteristic, one can achieve output current pulses whose temporal widths match or are lower than the input laser pulse [34]. Similar trends have been shown experimentally in materials such as GaAs with pulse compression of up to 66% based on a similar negative differential mobility material property [35,36].…”

“…The use of a laser with sub-bandgap energy would only excite electrons from the traps to the conduction band in the carbon-doped GaN section, while having negligible carrier generation throughout the rest of the p-i-n device, and avoiding any hole creation. The p-i-n GaN structure was taken to be 12 µm thick, with other parameters in line with a previous report [34]. The carbon-doped GaN was assumed to be 1 µm thick with each transverse dimension taken to be 50 µm.…”

“…The utilization of the NDM regime requires biasing the PCSS device at a field that is just around the maximum point on the drift-velocity curve. For GaN, this field has been predicted to be around 180-200 kV/cm [34,41,42]. When modeling high densities of particles, it becomes necessary to utilize a computational method to effectively manage larger densities without increasing the computational weighting significantly.…”

“…This section will briefly touch upon these aspects and the transport modeling methodology used for charge carrier transport in the simulation. A more rigorous account of this methodology can be obtained in a previous work [34].…”

“…Collision times were determined based on the energy of the particle at the different instants. The model included interactions with acoustic modes via the deformation potential, in addition to polar optical phonon interactions, and intervalley deformation potential scattering [34]. The piezoelectric scattering was excluded since it is known to be negligible at room temperatures with negligible influence on high field transport [46].…”