Search

Understanding and mitigation of substrate RF losses and signal distortion are critical to enable high-performance GaN-on-Si front-end-modules. While the origin of RF losses and consequently a decreased effective substrate resistivity $({\rho }_{eff})$ in GaN-on-Si substrates is now understood to be diffusion of Al and Ga atoms into the silicon substrate during III-N growth, the effect of upper III-N buffer layers on the ${\rho }_{eff}$ degradation under stressed conditions remains unclear. In this paper, we show that up to 50% variation in ${\rho }_{eff}$ at 2 GHz can take place over more than 1,000 s when the substrate is stressed at 50 V. Additionally, Coplanar Wave Guide (CPW) large-signal measurements under the same experimental conditions show a variation of $2^{\mathrm{ nd}}$ harmonic power of up to 5dB. A thermally activated stress and relaxation behavior shows the signature of traps which are present in the C-doped layers. With the help of a simplified TCAD model of the GaN-on-Si stack, we link this behavior to slow charge redistribution in the C-doped buffer continuously modifying the flat-band voltage ( $\text{V}_{\text {FB}}$ ) of the Metal-Insulator-Semiconductor (MIS) structure. Free carrier transport across the buffer is shown to have the greatest contribution on the large time constants, highlighting the importance of vertical transport paths in GaN-on-Si stacks.

For operation as power amplifiers in RF applications, high electron mobility transistor (HEMT) structures are subjected to a range of bias conditions, applied at both the gate and drain terminals, as the device is biased from the OFF- to ON-state conditions. The stability of the device threshold voltage (Vt) condition is imperative from a circuit-design perspective and is the focus of this study, where stresses in both the ON and OFF states are explored. We see rapid positive threshold voltage increases under negative bias stress and subsequent recovery (i.e., Vt reduces), whereas conversely, we see a negative Vt shift under positive stress and Vt increase during the subsequent relaxation phase. These effects are correlated with the thickness of the GaN layer and ultimately result from the deep carbon-acceptor levels in the C-GaN back barrier incorporated to screen the buffer between the silicon substrate and the epitaxially grown GaN layer. Methods to mitigate this effect are explored, and the consequences are discussed.

This article presents a physical model of the threshold voltage in MOSFETs valid down to 4.2 K. Interface traps close to the band edge modify the saturating temperature behavior of the threshold voltage observed in cryogenic measurements. Dopant freezeout, bandgap widening, and uniformly distributed traps in the bandgap do not change the qualitative behavior of the threshold voltage over temperature. Care should be taken because dopant freezeout results in a different physical definition of the threshold voltage. Using different definitions changes significantly the threshold current level. The proposed model is experimentally validated with measurements in large-area nMOS and pMOS devices of a commercial 28-nm bulk CMOS process down to 4.2 K. Our modeling results suggest that a pMOS-specific phenomenon in the gate stack is responsible for the non-saturating temperature behavior of the threshold voltage in pMOS devices.

Dopant pockets in combination with a III-V heterostructure have become a staple in simulations of tunnel field-effect transistors (TFET) to achieve acceptable on-currents (ION) and to break the ION-subthreshold swing trade-off in pTFETs. Questions on the scalability and variability of these dopant pockets remain, however. We therefore propose an alternative concept using two opposite sheet charges at the tunnel junction that are realized by exploiting different bonding types. With fully quantum mechanical 30-band k.p-based simulations, we investigate two configurations that exhibit such sheet charges both at the device level and in a ring-oscillator: a lattice-matched GaAs-Ge-GaAs heterostructure TFET and a homostructure InGaAs TFET. By comparing to pocketed broken gap GaSb-InAs TFETs, we show that built-in sheet charge is a valid alternative for both nTFET and pTFET, with the prospect of lower variability and better scalability.

Nano-ridge engineering (NRE) is a novel method to monolithically integrate III–V devices on a 300 mm Si platform. In this work, NRE is applied to InGaP/GaAs heterojunction bipolar transistors (HBTs), enabling hybrid III-V/CMOS technology for RF applications. The NRE HBT stacks were grown by metal-organic vapor-phase epitaxy on 300 mm Si (001) wafers with a double trench-patterned oxide template, in an industrial deposition chamber. Aspect ratio trapping in the narrow bottom part of a trench results in a threading dislocation density below 106∙cm−2 in the device layers in the wide upper part of that trench. NRE is used to create larger area NRs with a flat (001) surface, suitable for HBT device fabrication. Transmission electron microscopy inspection of the HBT stacks revealed restricted twin formation after the InGaP emitter layer contacts the oxide sidewall. Several structures, with varying InGaP growth conditions, were made, to further study this phenomenon. HBT devices—consisting of several nano-ridges in parallel—were processed for DC and RF characterization. A maximum DC gain of 112 was obtained and a cut-off frequency ft of ~17 GHz was achieved. These results show the potential of NRE III–V devices for hybrid III–V/CMOS technology for emerging RF applications.

The temperature dependent carrier transport characteristics of n-type gate-all-around nanowire field effect transistors (GAA NW-FET) on bulk silicon are experimentally compared to bulk fin field effect transistors (FinFET) over a wide range of temperatures (25–125 °C). A similar temperature dependence of threshold voltage (VTH) and subthreshold swing (SS) is observed for both devices. However, effective mobility (μeff) shows significant differences of temperature dependence between GAA NW-FET and FinFET at a high gate effective field. At weak Ninv (= 5 × 1012 cm2/V∙s), both GAA NW-FET and FinFET are mainly limited by phonon scattering in μeff. On the other hand, at strong Ninv (= 1.5 × 1013 cm2/V∙s), GAA NW-FET shows 10 times higher dμeff/dT and 1.6 times smaller mobility degradation coefficient (α) than FinFET. GAA NW-FET is less limited by surface roughness scattering, but FinFET is relatively more limited by surface roughness scattering in carrier transport.