'Off-The-Wall' Chemical Reactions Could Help Scale Up Production of Gallium Arsenide Chips

BUFFALO, N.Y. -- When the byproducts of chemical reactions stick to the walls of chemical reactors, it's more than a nuisance. By reacting with the walls, chemicals undergo changes that may alter the outcomes of experiments.

That sticky issue is one reason why gallium arsenide (GaAs), a material long expected to succeed silicon as the primary component of computer chips, hasn't progressed beyond specialized applications.

A new type of chemical reactor designed and built by chemical engineers at the University at Buffalo appears to have solved that problem. It is described in a paper, "A New Counterflow Jet Reactor for Purely Homogeneous Kinetic Studies of Endothermic Reactions," published this month in the American Institute of Chemical Engineers Journal.

Still a research tool, the new reactor could help bring mass manufacturing of gallium arsenide (GaAs) computer chips a step closer to reality.

"In order to obtain thin films with special properties, we need to understand their chemistry better," said T.J. Mountziaris, Ph.D., UB assistant professor of chemical engineering and developer of the reactor.

"The main obstacle in using chemical-vapor deposition to grow these GaAs films is that we haven't been able to understand the effects of chemical kinetics. We didn't have a way to tell precisely what reactions were happening while these films were being grown."

To prepare thin films of electronic materials with special properties, Mountziaris explained, scientists need to understand precisely the underlying chemical and physical phenomena.

"The complex chemical reactions involved in chemical-vapor deposition are not very well understood," he said. "This is the major obstacle in commercial development of CVD."

To better understand the chemical reactions, he added, it is necessary to measure the kinetics at atmospheric pressures, since those are the pressures at which the films are grown. But until now, the only way to measure gas-phase kinetics without interference from reactor walls has been to measure them at very low pressures and then to extrapolate their values to higher pressures.

Based in part on supercomputer simulations they conducted at the Pittsburgh Supercomputing Center, the UB researchers have developed a chemical reactor in which the gas-phase reactions are isolated from the reactor walls and occur at nearly atmospheric pressures.

"It's a matter of breaking down the process, one step at a time," said Mountziaris.

"Because this reactor allows us to control the amount of time during which the initial reaction occurs, we can eliminate the secondary reactions. Once we understand the kinetics of the very first reaction step, then we can increase the residence time of the chemical species and try to understand the secondary reactions, too."

To isolate the reaction zone, the researchers used two counterflowing jets of gas: one heated, the other cooled. The cool jet carries a reactant, such as tert-butyl-arsine, diluted in hydrogen, which then collides in the reactor with a hot jet of hydrogen. When the gases collide, the hot gas transfers heat to the cool gas, driving the reaction.

"We had to find out which conditions would give us a suspended-reaction zone away from the walls," said Mountziaris. "There was a narrow range of jet velocities and temperatures that would allow us to precisely confine the reaction zone."

By carefully controlling the flow rates of the jets and the temperatures of the reactants, the researchers found that they were able to create a reaction zone that was suspended between the walls of the reactors.

"This reactor allows us to find out what chemical species are being produced by the gas-phase reactions and at what rates," explained Mountziaris. "This is very important because these are the species that will stick to the substrate in a chemical-vapor deposition reactor, leading to growth of a thin film."

He emphasized that the new reactor is a research tool. However, in providing scientists with a fuller understanding of kinetics in the gas phase, it could lead to construction of predictive models of the chemical-vapor deposition process.

Such models are necessary before scaled-up versions of CVD reactors producing GaAs and other compound semiconductor films chips can be developed.