News Article
InN looks to enhance nitride HEMTs
A high-mobility, single atomic layer of InN could improve the performance of nitride HEMTs
Inserting a singe atomic layer of InN in an AlGaN/GaN HEMT creates a highly localised two-dimensional electron gas. The surface barrier height is denoted by Φs.
Calculations from the University of California, Santa Barbara, suggest that inserting a single atomic layer of InN into the heart of an AlGaN/GaN HEMT improves device performance. According to Mao-sheng Miao and Chris Van de Walle, adding an ultra-thin InN layer increases the density, localisation and mobility of the two-dimensional electron gas, leading to significant performance improvements, such as superior switching.
These theorists decided to investigate this novel device because of the far higher electron mobility of InN than GaN.
To evaluate the benefits of inserting a single layer of InN into a HEMT, the West-coast duo carried out first-principles electronic structure calculations with macroscopic device simulations based on Schrödinger-Poisson solvers.
Van de Walle explains that the Schrödinger-Poisson solvers enable the accessing of features such as the localisation of the two-dimensional electron gas.
"But first principles calculations are needed in order to obtain microscopic parameters, such as the effective mass of electrons in the InN layer." They accomplished this with a form of density functional theory that employs an advanced hybrid functional and yields accurate results for the electronic structure.
"Such calculations are highly demanding, requiring up to twenty times more computing power than density functional calculations with traditional methods," says Van de Walle. "They can only be run on supercomputers."
One structure investigated by Miao and Van de Walle consists of 260 nm of GaN and 4 nm of Al0.34Ga0.66N, with a 0.36 nm-thick single atomic layer of InN inserted between. It is assumed that the InN is psuedomorphically grown on GaN, so it sustains a large compressive biaxial strain, accompanied by a corresponding relaxation of the lattice along the c-axis.
In contrast to a conventional GaN HEMT, where electron density is spread out over a width of at least 3 nm, the charge in the InN-containing structure is far more localised "“ it is spread over as little as 1 nm, judged by the value of the full-width at half maximum.
The nature of this electron gas has been investigated in more detail. Calculations suggest that for a fixed barrier height, the HEMT containing InN starts forming a two-dimensional electron gas at a smaller AlGaN thickness, and its density is much higher. And if the AlGaN thickness is held at 4 nm, the density of the two-dimensional electron gas is much higher when the atomic layer of InN is present, indicating that a transistor would be more sensitive to changes in gate voltage.
To switch off a HEMT, a negative voltage is applied that lowers the potential on the surface, and cuts charge transfer from surface states to the interface region.
Simulations by the team show that for a surface barrier height of up to 3.5 eV, there will be a small residual two-dimensional electron gas at the interface of a conventional device; but for the variant with an ultra-thin InN layer, the density of the two-dimensional electron gas goes sharply to zero at a surface barrier height of just 3.0 eV.
Further calculations determined the in-plane effective mass of the electron, finding it to be 94 percent of that for the bulk, and more than 30 times less than that of GaN. This indicates the high mobility for electrons in the two-dimensional electron gas.
Proving the promise of these novel HEMTs requires their fabrication and testing. This might takes place within the University "“ academic Stacai Keller, in the group headed by Umesh Mishra, has previously worked on InN/GaN heterostructures for potential use in HEMTs.
