ajay kumar dharmireddy
@sircrrengg.ac.in.
Associate Professor, Department of E.C.E
ajay kumar dharmireddy
@sircrrengg.ac.in.
Associate Professor, Department of E.C.E
ajay kumar dharmireddy
In the realm of electronics, device modeling is an essential endeavor that aids in forecasting how devices will behave in different scenarios. In this brief, a neural network-based machine learning framework for modeling the MOSFET transient characteristics curve is presented. Gate length (L), width (W) and oxide thickness (Tox), channel doping concentration, and source/drain doping concentration are input variables for our machine learning (ML)-based MOSFET device model. The model is trained on TCAD-generated datasets, capturing complex relationships and ensuring high accuracy while significantly reducing computational costs. Compared to conventional methods, this approach accelerates modeling, enhances scalability, and adapts to future semiconductor device designs. By bridging the gap between physics-based simulations and machine learning, this work contributes to efficient transistor modeling and advances semiconductor research.
In traditional MOSFETs, below 45nm technology, device scaling is a big challenging task because it results severe short channel effects (SCEs). The causes of this are, the ratio between operating voltage and the thermal voltage shrinks with scaling down of MOSFET. This leads to higher leakage currents stemming from the thermal diffusion of electrons, and poor electrostatic control of gate over the channel, and In planar devices drive strength is dependent on the channel width. Deep research on this issue results many non-classical devices which have good electrostatic control over the channel, thereby greatly minimizes SCEs and allows the progress of device scaling . Double gate FET, multi gate FinFET, and Tunnel FET are the resulting devices. Sub-threshold slope(SS), on state current and off state currents are the design parameters to measure the device performance. In Low power -High speed electronic circuits, the devices with more steeper characteristics are required. The speed o
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
