Monte Carlo modelling of noise in advanced III–V HEMTs

Abstract

One of the main objectives of modern Microelectronics is the fabrication of devices with increased cutoff frequency and decreased level of noise. At this moment, the best devices for high-frequency, low-noise behavior are High electron mobility transistors (HEMTs) based on InGaAs and InAs channels. In this work, a complete analysis of ultrashort-gate HEMTs has been carried out by using a semiclassical Monte Carlo simulator, paying special attention to the noise performance. The validity of the model has been checked through the comparison of the simulated results with static, dynamic and noise measurements in real HEMTs. In order to reproduce the experimental results, we have included in the model some important real effects such as degeneracy, surface charges, presence of dielectrics and contact parasitics. The cryogenic performance of the HEMTs has also been analyzed. The influence of the parasitic resistances, width of the devices, value of the δ-doping and recess length has been analyzed when scaling down the gate length of the transistors to 50 nm aiming at achieving higher cutoff frequencies and better noise performance. The important effect of the impact ionization mechanisms and the consequent kink effect on the noise in both InGaAs and InAs based HEMTs have also been studied. Finally the advantages of the use of a double gate topology are quantified.

Noise and charge discreteness as ultimate limit for the THz operation of ultra-small electronic devices

“…Firstly, we present semiclassical Monte Carlo simulations for a two terminal device 43 – 45 . Inside the device, transport is assumed to be ballistic without electron-phonon collisions.…”

Comparative study on noise characteristics of As and Sb‐based high electron mobility transistors

“…From the viewpoint of the circuit design, NF min (i.e., a ratio of signal to noise at the gate divided by a corresponding ratio at the drain) is the most important parameter related to the noise. Therefore, we calculate NF min by using the noise spectral densities and the Y ‐parameters, according to the following formulas : where R n is the noise resistance, Y cor is the correlation admittance (associated to the correlation between the gate and drain noise sources), and Y opt is the parallel optimum matching admittance for satisfying the minimum noise conditions. The associated power gain G ass corresponding to the minimum noise condition is also calculated as follo...…”

“…Figure 7 shows NF min and G ass as a function of I ds at 100 GHz for the devices A (a), B (b), and C (c). V ds is varied as 0.2, 0.4, and 0.6 V. NF min shows the typical U-shape, which originates basically from the balance between the dependence of g m and <DI 2 ds > (i.e., f T and P) on I ds [26]. The device C shows the smaller NF min at the low V ds of 0.2 V; this indicates that the lower V ds operation and the larger g m (i. e., the higher f T ) beat the larger <DI 2 ds >, eventually.…”

“…From the viewpoint of the circuit design, NF min (i.e., a ratio of signal to noise at the gate divided by a corresponding ratio at the drain) is the most important parameter related to the noise. Therefore, we calculate NF min by using the noise spectral densities and the Y-parameters, according to the following formulas [18,20,[24][25][26]:…”

High-field transport and hot electron noise in GaAs from first principles: role of two-phonon scattering

“…Later, Monte Carlo calculations [34,39,40] and analytical models [41][42][43] of transport in GaAs used PSD as an additional observable to study intervalley processes and their role in the Gunn effect and PSD peak. Subsequent works focused on the effect of short channels [44] or impurities [45] on PSD at cryogenic temperatures, Monte Carlo modeling of noise in modern heterostructure devices [46,47],…”