Physics based compact modeling of symmetric double gate MOS transistors with high mobility III-V channel material

https://doi.org/10.1016/j.spmi.2021.106925Get rights and content

Highlights

  • A core compact model of surface potential, charge and drain current of symmetric III-V Double gate MOSFET is reported.

  • Inversion charge and potential is derived from explicit solution of Schrödinger-Poisson equations.

  • Effect of conduction band non-parabolicity and velocity overshoot is modeled.

  • Model predicted results are in a reasonable agreement with numerical simulation results.

Abstract

In this work, we report a physics-based core model of surface potential, inversion charge, and drain current of a symmetric double-gate MOSFET with high mobility III-V channel material. The model efficiently captures all essential device physics of III-V MOSFETs. The carrier quantization effect inside the quantum well formed between oxide and channel region, is modeled by solving time-independent Schrödinger wave equation. The surface potential and inversion charge are obtained from the explicit solution of the Poisson and Schrödinger equation. A first-order correction has been applied to incorporate the modifications in sub-band energy due to applied gate bias voltages. The conduction band non-parabolicity effect is also included in our proposed model. The drain current has been derived using drift-diffusion transport including the velocity overshoot effect. The core model is free from any empirical parameters. The model predicted results have been validated with self-consistent Schrödinger-Poisson solver and commercial device simulator (TCAD) for different channel thicknesses, effective masses, a wide range of gate and drain bias voltages, and different non-parabolicity factor. The model predicted results are found to be in reasonable agreement with numerical simulation data.

Introduction

Due to the aggressive scaling, power density inside the scaled device increases substantially which demands a proper thermal management scheme thereby increasing the production cost [1,2]. Power dissipation by a logic chip can be minimized by reducing supply voltage but that leads to degradation in switching characteristics. To mitigate this problem, researchers are looking for high mobility channel material where inversion charges can travel at a much higher speed than silicon thereby resulting in higher drive current at reasonably low supply voltages. The III-V compound semiconductors such as InAs, InGaAs, InSb, etc. are a promising candidate as an alternate channel material in the next generation scaled MOS devices due to their superior electron transportation such as low electron effective mass, high electron mobility, and high saturation velocity [3].

Although III-V channel MOSFETs show attractive transport properties, there are some inherent problems. Due to the higher permittivity of III-V materials, III-V MOSFET suffers from poor electrostatic integrity (EI) which results in higher short channel effects. By introducing ultra-thin body, multiple gate architecture and channel engineering, gate controllability and therefore EI can be improved [2]. Achieving a good quality interface is another one issue for III-V MOSFET. Literature survey reveals that Al2O3 with InGaAs shows good interface quality with less trap density [4].

For circuit simulation purposes, a compact model of the device is essentially required. Also, compact models are an integral part of a process design kit (PDK). Due to the low effective mass and ultra-thin channel structure, III-V MOSFET shows stronger quantum confinement. As a result, electron energy split into multiple sub-bands [[5], [6], [7]]. In Silicon channel based MOSFET, due to comparatively higher effective mass, the quantum effect is insignificant and often that effect is introduced into the model in terms of correction scheme (charge centroid and modification in gate oxide thickness). However, those strategies are not applicable for III-V MOSFETs where Schrödinger and Poisson equation needs to be solved explicitly to obtain inversion charge and surface potential. This makes the model development task much harder than Silicon MOS devices. Moreover, the conduction band of III-V materials is strongly non-parabolic in nature and parabolic relation between energy-wave vector are not fully valid. Additionally, the mean free path of III-V materials is significantly higher than Silicon, and charge transport property is semi-ballistic in nature at shorter channel length. As a result, the velocity overshoot phenomenon takes place. A compact model of III-V MOSFET demands all these features that need to be incorporated accurately without increasing the computational burden.

Literature survey reveals that there are few attempts made on the development of a compact model of III-V MOSFET. Modeling of inversion charge and capacitance of III-V double gate MOSFET reported in Refs. [[8], [9], [10], [11]]. The modeling of surface potential and charge inside a potential well is reported in Ref. [12]. In that work, the authors have used a triangular potential well to model the quantum well. A first order perturbation method has been applied to correct sub-band energies and wavefunctions in presence of applied electric field. The surface potential and inversion charge is derived using numerical approach. In Ref. [13], a compact model of double-gate MOSFET is reported where the drain current and terminal charges are derived using polynomial based interpolation scheme. In Ref. [14], a surface potential-based compact drain current model of III-V on Insulator MOS transistor is reported where the relation between Fermi level and inversion charge is represented using one interpolation function. The conduction band non-parabolicity and velocity overshoot effect are not included in all these above-mentioned models. In another work, a compact model of inversion charge with quantum capacitance and drain current is reported in Ref. [15], where the contribution of 2nd sub-band and band non-parabolicity has been incorporated in form of a correction factor. In Ref. [16], a charge based drain current model of buffered InAs-on-insulator MOS transistor is reported. The velocity overshoot effect is included by modifying the saturation velocity by a channel length-dependent saturation velocity. This model also includes the band non-parabolicity effect. However, this model is primarily applicable to buffered III-V-on-insulator transistors with single gate operation.

In this work, we propose a compact model of the surface potential, inversion charge and drain current of a symmetric double-gate MOS transistor with high mobility III-V channel material. The unique attribute of our work is as follows. We solve 1-D Schrödinger and Poisson equation explicitly along confinement direction to find sub-band energy, inversion charge density, and potential. A 1st order correction to the sub-band energy, effective mass and 2-D density of states (DOS) due to band non-parabolicity effect has been incorporated. Velocity overshoot effect has been incorporated by modifying saturation velocity (νsat) using a channel length-dependent saturation velocity. The core model is free from any empirical parameters. The model predicted result has been validated against numerical simulation data for various channel thicknesses, different effective mass, and a wide range of bias voltages.

This article is organized in different sections. Section-1 describes the introduction part. Section-2 represents the model development framework. Section-3 contains the model implementation and validation with a professional numerical simulator. Finally, the conclusion is summarized in Section-4.

Section snippets

Device structure and model development framework

Fig. 1 (a) shows the device structure considered for analytical model development. In this work, we consider Tch is the thickness of the device, Lg is the physical channel length, and Tox is the oxide thickness of front and back gates of the semiconductor device. In addition, we assume x=0 is at the center position of the channel thickness, and y=0 is at the start position of the channel. The potential well profile is shown in Fig. 1(b). All energy levels are measured with respect to the bottom

Device simulation framework

For III-V compound semiconductor materials, all physical parameters such as electron effective mass, band-gap energy, electron affinity and dielectric constant are estimated using Veggard's interpolation formula [25]. The device simulation is carried out with the help of Schrödinger-Poisson solver [17] and a commercial 2-D device simulator, SDEVICE from Synopsys Inc [26]. All simulations have been carried out at room temperature (T = 300K). To evaluate the electron density and surface potential

Conclusion

In this article, we have reported a core model for surface potential, inversion charge, and drain current of a double gate MOSFET with low effective mass III-V channel materials. The model correctly captures the variation in surface potential, sub-band energy, and inversion layer charge. The model predicted results have been validated with self-consistent Schrödinger-Poisson solver data along with commercial device simulator (TCAD) and a reasonable agreement has been achieved. Terminal charges

Author statement

Ho Le Minh Toan: Conceptualization, Methodology, Visualization, Investigation, Software, Validation, Writing- Original draft preparation. Subir Kumar Maity: Supervision, Writing- Reviewing and Editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References (30)

  • D.J. Frank

    Power-constrained CMOS scaling limits

    IBM J. Res. Dev.

    (2002)
  • J. del Alamo

    Nanometre-scale electronics with III-V compound semiconductors

    Nature

    (2011)
  • K.K. Bhuwalka et al.

    In0.53Ga0.47As-Based nMOSFET design for low standby power applications

    IEEE Trans. Electron. Dev.

    (2015)
  • S. Takagi et al.

    III–V/Ge channel MOS device technologies in nano CMOS era

    Jpn. J. Appl. Phys.

    (2015)
  • S. Luryi

    Quantum capacitance devices

    Appl. Phys. Lett.

    (1988)
  • V.P. Trivedi et al.

    Quantum-mechanical effects on the threshold voltage of undoped double-gate mosfets

    IEEE Electron. Device Lett.

    (2005)
  • M. Balaguer et al.

    In-depth study of quantum effects in SOI DGMOSFETs for different crystallographic orientations

    IEEE Trans. Electron. Dev.

    (2011)
  • S. Mudanai et al.

    Capacitance compact model for ultrathin low-electron-effective-mass materials

    IEEE Trans. Electron. Dev.

    (2011)
  • S. Oh et al.

    Physics-based compact model of iii-v heterostructure fets for digital logic applications

  • C. Yadav et al.

    Capacitance modeling in iii–v finfets

    IEEE Trans. Electron. Dev.

    (2015)
  • M.D. Ganeriwala et al.

    Modeling of charge and quantum capacitance in low effective mass iii-v finfets

    IEEE Journal of the Electron Devices Society

    (2016)
  • E. Marin et al.

    Analytic potential and charge model of semiconductor quantum wells

    IEEE Trans. Electron. Dev.

    (2015)
  • A.S. Roy et al.

    Compact model for ultrathin low electron effective mass double gate mosfet

    IEEE Trans. Electron. Dev.

    (2014)
  • S.K. Maity et al.

    Compact drain current modeling of InAs-OI-Si MOS transistor including quantum confinement

  • C. Yadav et al.

    Compact modeling of charge, capacitance, and drain current in iii–v channel double gate fets

    IEEE Trans. Nanotechnol.

    (2017)
  • Cited by (0)

    View full text