Dielectric films grown on wide bandgap semiconductors are needed for next generation high power and high frequency microelectronic devices. Methods for selectively depositing materials onto these wide bandgap semiconductors simplify device fabrication processes, reducing time and costs.

Now Mark Losego of Georgia Institute of Technology, in collaboration with Elizabeth Paisley from Sandia National Laboratories and the team of J.P. Maria from North Carolina State University, have recently reported the development of a new selective area epitaxy (SAE) method for growth of dielectric MgO thin films on gallium nitride (GaN) surfaces using surface chemistry modification.

GaN-based materials are excellent candidates for use in advanced microelectronic devices because of their ability to operate at high frequencies, high powers, and high temperatures. The growth of atomically smooth MgO epitaxial layers onto GaN is one route to forming gate dielectrics for these devices.  However the selective growth of oxides on semiconductors is uncommon.  “Our demonstration of using MBE [molecular beam epitaxy] to selectively grow MgO on GaN surfaces is one of the few examples of selective area epitaxy of an oxide by physical vapor deposition [PVD],” Losego says.

Selectivity is generally easier to achieve with a chemical vapor deposition (CVD) process, than with PVD. “The reason for this is the use of elemental precursors in MBE. Atomic precursors ‘stick better’ to the underlying surface than molecular species, so we had to modify the surface chemistry to alter the sticking coefficient,” Losego explains.

Treatment with hydrochloric acid had been previously shown to chlorinate GaN surfaces, but the effect of this surface chlorination on atomic adsorption had not been previously studied. Losego and colleagues found that treating GaN surfaces with hydrochloric acid impeded MgO film growth under certain conditions, and regions of “growth” or “no growth” could be readily patterned on GaN surfaces. A monolayer of Cl adatoms–the only detectable difference in surface structure–was identified as the source for adsorption impediment (a macroscale example of the results of this process are shown in the photograph).

The inset of this figure demonstrates that higher resolution features can also be replicated when photolithography is used to pattern the surface chlorination.

The group’s work is described as “a nice example of scientific creativity” by Raffaella Calarco, senior scientist at Paul Drude Institute for Solid State Electronics in Berlin. Calarco was not involved in the study. “Instead of the most common approach using a hard mask, here a simple change of surface chemistry due to an appropriate Cl-termination is employed,” she says.

Stephen J. Pearton at the University of Florida, who was also not involved in the study, adds: “It is a clever approach that avoids the need for more complex masking steps in order to achieve selective area growth. The next steps forward will need to expand the range of oxides that can be deposited in this fashion.”

According to Losego, the researchers are now exploring extending this understanding of surface chemistry to other materials and to CVD processes.