University of Rochester

Light-Emitting Silicon Coupled with Conventional Circuitry

November 28, 1996

Engineers have taken an important step toward the long- sought marriage of electronics and optics by integrating a porous silicon light-emitting diode (LED) into conventional microelectronic circuitry for the first time, making possible an all-silicon system that can process light as well as electricity. A paper on the work appears in the Nov. 28 issue of Nature.

The team of engineers from the University of Rochester and the Rochester Institute of Technology (RIT) accomplished the task by creating a sturdier form of porous silicon that can withstand the rigors of today's manufacturing environment. The work is the first to use silicon for both the electronic and optical components on a single chip, eliminating the need for other materials or special processes during fabrication.

"This is the first time that a silicon light-emitting diode has been integrated with microelectronics," says Philippe Fauchet, leader of the group and professor of electrical engineering, optics, and physics and astronomy at the University of Rochester, and a senior scientist at the University's Laboratory for Laser Energetics. "This is a necessary rite of passage if porous silicon is to truly become a technology."

Silicon's robustness and low cost have made it the workhorse of the electronics industry. Silicon chips power everything from greeting cards to space shuttles and nearly all electronic appliances. Indeed, "Silicon Valley" is synonymous with technical innovation. The silicon-based computer chip industry is worth hundreds of billions of dollars a year.

But silicon has one major flaw: It emits light only feebly. This flaw has grown in significance in recent years as scientists turn to light to carry the increasing amount of digital data zipping around the globe. To meet the demand, companies are laying miles and miles of fiber optic cable, whose beams of light transmit more information more cleanly and much faster than electricity.

Currently engineers must turn to other materials, such as gallium arsenide or organic polymers, to provide optical capabilities. However, these materials pose problems of their own: They're either much more costly than silicon or extraordinarily fragile, and integrating them with silicon circuits demands large and costly changes in fabrication lines where chips are manufactured.

That's why scientists were jubilant in 1990 when they discovered that a specially prepared type of silicon, known as porous silicon because it's permeated with pores, can emit light. However, its porosity makes it so fragile that the material has not withstood conventional manufacturing techniques and has been almost useless to the electronics industry. The number of groups studying the material in the United States has dwindled from hundreds to a handful.

The Rochester team strengthened the material by chemically modifying it. Engineers removed hydrogen atoms from the outer layer of tiny silicon nano-particles less than 100 angstroms wide, about one-thousandth the width of a human hair. They replaced the hydrogen with a double layer of silicon oxide to create a modified form of porous silicon known as silicon-rich silicon oxide. These steps enabled the material, which is about three-quarters air and only one-quarter silicon, to withstand temperatures of 900 degrees Celsius typically reached in the fabrication process, as well as other processing steps such as the deposition of various layers and etching with photoresist.

"A silicon wafer travels four or five miles on the factory floor in the transition from a nearly useless material to a high- quality computer processor," says Fauchet. "The fabrication line can cost several billions dollars to set up and typically involves hundreds of steps. Because of the enormous investment, it's important to adapt any new technology to the fabrication lines already established."

The chip was manufactured at RIT by a team led by Karl Hirschman, a member of the faculty in RIT's Department of Microelectronic Engineering and a graduate student in Fauchet's laboratory. It was Hirschman's job to actually integrate the LED with conventional electronics -- in this case, a transistor, which controlled the current that modulated the output of the LED.

Also working on the project were research associate Leo Tsybeskov, who did much of the device design and materials science research; and graduate student Sid Duttagupta, who helped with the materials processing.

Fauchet's group holds the record for the most stable porous silicon LED -- they powered the device for 11 straight days before stopping the experiment. The LED is 10,000 times more efficient than the first light-emitting porous silicon in 1990, and its brightness is about 1 milliwatt per square centimeter. It can flash up to 10 million times per second.

Nevertheless, the engineers say that many more improvements must be made before porous silicon becomes a standard material on the shop floor. They'd like to boost the efficiency 10-fold, to 1 percent, and they'd like to increase its frequency 100-fold, to one billion flashes per second.

"This is really the first time that porous silicon has lived up to its promise," says Fauchet, who has the largest team in the United States studying the material. His work is currently funded by the U.S. Army and the National Science Foundation.

Fauchet is organizing a symposium as part of next week's Materials Research Society meeting in Boston. There he will present his latest results showing that porous silicon is about as efficient as the widely used amorphous silicon in converting sunlight to electricity.