High-speed semiconductor device models compared
Researchers at the University of Illinois Urbana-Champaign in collaboration with Air Force Research Laboratory engineers have studied two semiconductor simulation tools using different energy-transport models for RF gallium nitride (GaN) high-electron-mobility transistors (HEMTs). Accurate simulation of these devices – which promise to revolutionize wireless communications – must capture the important physical processes involved and the carrier transport mechanisms in order to realize the material’s full potential, say the researchers.
The two energy-transport models studied included the Fermi kinetics transport model and a hydrodynamics transport model as it is implemented in the device simulator Sentaurus from Synopsys. The researchers say they found that the Fermi kinetics solver has mathematical properties that allow it to better handle the extreme conditions under which gallium nitride devices will operate.
“This is the first time a direct comparison has been made between the state-of-the-art commercial program and a custom-developed research code,” says Shaloo Rakheja, a professor of electrical and computer engineering at the University of Illinois Urbana-Champaign. “It is important for the semiconductor community to understand the strengths and limitations of each.”
According to the researchers, the most important difference between the two programs is how they model the electronic heat flow. The commercial package uses Fourier’s law, an empirical model that does not necessarily work well for semiconductors, while the Fermi kinetics transport solver uses more fundamental thermodynamic principles for this purpose.
This accounts for the different predictions each program makes, say the researchers.
“There is a strong connection between the underlying physics and the behavior of each program,” says Rakheja, “and we wanted to explore that in the context of a device technology that’s highly relevant today: gallium nitride.”
To compare the two codes, the researchers simulated an elementary gallium nitride transistor with each. They found that the two programs gave similar results under modest operating conditions.
However, when they introduced large, transient signals of the kind expected in high-speed applications, they obtained unexpected results for electron temperature from the commercial package. It predicted that at short time scales the electron temperature would dip below the ambient temperature, while the Fermi kinetics solver gave more consistent temperature profiles.
In addition, say the researchers, when they examined the rate of convergence – a mathematical indicator of simulation self-consistency – of each, the Fermi kinetics solver converged faster, suggesting that the Fermi kinetics solver is more computationally robust. The researchers are now using the solver’s robustness to simulate more gallium nitride devices, with an aim at understanding how the material heats up as it operates at high speeds and use this information to design devices that fully take advantage of the material’s properties.
“Gallium nitride has really been a game changer,” says Nicholas Miller, an Air Force Research Laboratory engineer. “As the technology continues to evolve into more sophisticated forms, a critical component of the development cycle is modeling and simulation of the transistors.”
For more, see “A comparison of a commercial hydrodynamics TCAD solver and Fermi kinetics transport convergence for GaN HEMTs.”