Materials Advance May Help
the Semiconductor Industry

This micrograph shows a mock field-effect transistor with a layer of crystalline strontium titanate instead of silicon dioxide as the gate electrode. The layer was grown in registry with the silicon template making up the transistor's base. ORNL tests show that the transistor works.

A barrier to future increases in computing power is a restriction set by a compound of silicon itself. As transistors are downsized, the use of silicon dioxide to control electron flow will limit transistor performance.

After 10 years of research, Rodney McKee of ORNL's Metals and Ceramics (M&C) Division, Fred Walker of the University of Tennessee, and Matt Chisholm of the Solid State Division, have found a solution. They demonstrated that amorphous silicon dioxide conventionally used on silicon chips can be replaced with a crystalline oxide whose superior electrical properties will allow reduction in transistor size without loss of performance. In March 1999 the team built a field-effect transistor (FET) using crystalline strontium titanate and demonstrated that it performs as well as conventional transistors.

A FET is a common switching device used in modern electronic equipment. This tiny semiconducting device consists of three metal electrodes and a silicon base. When a conventional FET is turned on, electrons injected by a source electrode flow as a current through the silicon base for collection at a drain electrode. To turn the transistor off, a gate electrode between the other electrodes applies an electrical voltage to a dielectric film, causing it to "pinch off" the current by raising the silicon base's resistance. In this way, a transistor can function as an on-and-off switch. It can also store bits of information (a "1" if switched on, a "0" if switched off).

As transistor size is reduced, the silicon dioxide layer needed for the dielectric film will eventually become so thin (<3 nm) that it will be useless. The reason: it will leak electrons through quantum mechanical tunneling.

For decades, crystalline oxides have been considered a promising solution to this problem for the semiconductor industry. These materials have the physical thickness to support an electric field yet are able to store electrical charges more effectively. But no one had been able to produce the high-quality layer needed to support shrinking semiconductors. Using molecular beam epitaxy, a precisely controlled process for growing thin films under ultrahigh vacuum, and a $400 video camera to tape the deposition of oxide films on silicon, McKee and Walker learned which conditions allow film crystals to grow in registry with the silicon crystal template beneath, producing a perfect film.

Because strontium titanate exerts a stronger influence on the transistor's conductivity than silicon dioxide, the gate electrode can take up less space, compressing the area between the source and drain electrodes and shortening the distance the electrons would travel. The benefits? Transistors with strontium titanate are likely to be smaller and faster.