G C Crow

Self-consistent 2-D Monte Carlo Simulations of InSb APD

Self-consistent Monte Carlo simulations are used to study the low noise and high gain potential of InSb avalanche photodiodes. It is found that for an electron-initiated avalanche, excess noise factors well below the minimum McIntyre value persist up to gain values of around 60 for a 3.2 /spl mu/m avalanche region. For these very low noise values, it is found that multiplication has a very unusual voltage dependence which may be exploited for highly efficient novel low noise planar arrays operating at low voltage.

Monte Carlo simulations of high-speed InSb-InAlSb FETs

Self consistent Monte Carlo simulations which include impact ionization are used to study the high-speed potential of InSb field-effect transistors. It is found that the impact ionization has a strong influence on the performance of InSb for high speed. The ionization leads to a high electron drift velocity and substrate bias can be used to extract the holes which are generated in the channel. Residual hole density within the channel, however, limits the maximum speed. The substrate bias and buffer doping are critical for extracting holes from the channel without inducing excess ionization. Simulations yield a peak cutoff frequency of 820 GHz with a 0.03125-/spl mu/m gate, a source to drain voltage of 0.58, and a sheet doping density of 1.7/spl times/10/sup 12/ cm/sup -2/.

Monte Carlo simulations of AlGaN/GaN heterojunction field-effect transistors (HFETs)

Self-consistent Monte Carlo simulations are reported for AlGaN/GaN HFETs. Hot-carrier scattering rates are determined by fitting experimental ionization coefficients and the doping character of the GaN is obtained from substrate bias measurements. Preliminary simulations for a simple model of the AlGaN surface are described and results are found to be consistent with experimental data. The high-frequency response of short-gate-length transistors is found to be sensitive to the charge state of the free AlGaN surface and it is proposed that current-slump phenomena may also be related to deep levels at this surface. Breakdown calculations show interesting two-dimensional effects close to the drain contact.

Monte Carlo simulations of capture into quantum well structures

The purpose of this paper is to simulate capture events into quantum well structures of the kinds studied in luminescence experiments in order to understand the carrier dynamics, to extract local electron and hole capture times and to relate microscopic carrier processes to the luminescence signal. To do this, a quantum mechanical model for charge capture by a quantum well has been incorporated into a self-consistent Monte Carlo simulation of carrier transport in quantum well laser diodes. The capture model makes use of the established technique of modifying Fermis golden rule to calculate a dimensionless capture probability instead of a capture rate. This model has been used to simulate time-resolved photoluminescence experiments on a variety of InGaAsP-based structures which have unstrained quantum wells. Local electron and hole capture times have been extracted from the transport data for simulations performed at 300 K, and these times oscillate as a function of well width. Wells 85-90 ? wide permit fast electron capture into a second electronic subband (0.56 ps capture time) and hole capture into a second light hole band (0.44 ps). This may be contrasted with a 75 ? well for which electron capture into the single conduction subband exceeds 1.7 ps, and the peak hole capture time is 1.1 ps.

Monte Carlo simulations of hole transport in SiGe and Ge quantum wells

Monte Carlo simulations have been carried out to investigate factors which influence hole transport at 300 K for moderate electric fields (104 -106 V m-1 ) within compressively strained Si1-x Gex (x = 0.15 0.30) quantum wells deposited on Si. Drift mobilities in the range 900-1200 cm2 V-1 s-1 have been calculated for pseudomorphic structures with well widths 60 110 A, and trends in the mobility have been identified. SiGe alloy disorder and interface roughness are the main factors which limit the mobility; inelastic LO phonon scattering is less significant owing to the large phonon energy ~60 MeV and hence comparatively large threshold carrier wavevector for the onset of scattering. The mobility in SiGe is compared with that simulated for an experimental Ge/Ge0.7 Si0.3 structure. The simulated drift mobility ratio for Ge/GeSi versus SiGe/Si is 10:1, a result which is consistent with recent measurements. There are two main reasons for this - a comparatively low in-plane heavy hole effective mass and significantly reduced alloy scattering. This is despite greater surface roughness derived from the reported experimental data.

Monte Carlo simulations of charge transport in high-speed lasers

A self-consistent ensemble Monte Carlo calculation of carrier transport in multiple-quantum-well lasers has been developed in an effort to understand the impact of picosecond carrier dynamics upon the modulation bandwidth. The model has been applied to InGaAsP-based devices which are designed to emit at 1550 nm. Results are discussed for structures with ungraded confinement layers of varied width and different numbers of QWs. Simulations have been carried out at fixed and also modulated bias, from which the intrinsic frequency response can be derived.

Monte Carlo simulations of electron transport in coupled Si quantum wells

Coupled wide and narrow tensile strained Si X2-valley quantum wells could provide the basis for an SiGe transistor operating in a velocity modulation mode. Ideally, charge would be rapidly switched between high- (wide) and low- (narrow) mobility channels under the action of an applied gate bias, with little or no modulation of the total channel charge density. With this application in mind, the Monte Carlo technique has been used to simulate in-plane electron transport for a back-doped Si/Si0.85Ge0.15 double-well structure. The equilibrium band profile from the Schottky or oxide top gate to the SiGe virtual substrate is such that electrons are confined to the narrow well when the gate is unbiased. The calculated in-plane mobility is sensitive to the transverse field applied across the quantum well structure - the maximum:minimum mobility ratio is 13:1 at 77 K. At 77 K, a lower mobility in the narrow channel (4000 cm2 V s-1) is mainly due to surface roughness and Coulomb scattering by supply layer impurities. The simulations also predict that mobility modulation would be far less effective at 300 K; scattering by acoustic and inelastic g phonons (LO, TA, LA modes), and consequent distribution across the available subbands (9) of the entire well structure reduces confinement to either the wide or narrow well, and hence the predicted maximum:minimum mobility ratio is less than 2:1.

Performance predictions for a silicon velocity modulation transistor

A Monte Carlo simulation has been devised and used to model submicron Si velocity modulation transistors with the intention of designing a picosecond switch. The simulated devices have nominal top and back gate lengths of 0.1 μm, and the conduction channels have similar thickness. Mobility modulation has so far been achieved by heavily compensated doping and interface roughness at one side of the channel. The simulated devices have a high intrinsic speed; simulations performed for T=77 K suggest that current can be switched between the low and high mobility regions of the channel within 1.5 ps. However, in unstrained Si devices the main obstacle to practical device operation is the rather small current modulation factor (the ratio of the steady drain currents for the device operating in the high and low mobility regimes), which decreases towards unity with increasing drain–source bias. Such a device should work best for small electric fields along the channel (∼105 V m−1), the regime where impurity scatte...

A k.p model for carrier capture by a quantum well

A k.p model for electron and hole capture by a quantum well has been applied to the calculation of the rates for capture via phonon and alloy scattering by a 30 AA In0.7Ga0.3As-InGaAs quantum well, a system which is suitable for lasers operating at a wavelength of 1.55 mu m. The standard approach to carrier capture by a quantum well, namely the application of Fermis Golden Rule, has been adapted in order to derive the dimensionless local capture efficiency per unit incident particle current density. This has the advantage that the capture data can be more readily incorporated into many-particle transport simulations of quantum well lasers. Structure is observed in the capture efficiency as a function of the energy and in-plane wavevector of the incident carrier. This is partly attributed to instances where incident electrons or holes in barrier states undergo strong transmission into the well region, but mixing between the quantum well subbands is also important through the effect upon both the density of final states and the scattering matrix elements for capture transitions.

The transient signal response of submicron vertical silicon field effect transistors

A Monte Carlo simulation has been used to model steady state and transient electron transport in a vertical geometry n-type silicon metal-oxide field effect transistor (Si n-MOSFET). Detailed time-dependent voltage signal analysis has been carried out to demonstrate the modulation response of the device, which has been compared with the result of an identical analysis performed on a more conventional planar geometry silicon-on-insulator n-MOSFET of similar dimensions and doping. The effective source-drain separation of the modelled devices is 90 nm, while the polysilicon gate length and channel width are 50 nm. The calculations suggest that provided the contact resistance and capacitance associated with the drain of similar experimental structures can be limited, the upper limit to the current gain cut-off frequency of this vertical MOSFET is 160±10 GHz, a 50% improvement over the planar device.