Researchers at Cornell University (Ithaca, NY) have discovered a new way to fabricate nanoscale structures on silicon that they believe could fuel the development of products as diverse as biological sensors and light-emitting silicon-based displays. The research, which was reported in a the April 9 Applied Physics Letters, was funded by a $2.8 million, three-year grant from Philip Morris Inc.
The nanofabrication process, called controlled etching of dislocations (CED), has produced an array of features on a silicon surface with tiny columns--the researchers call them nanobumps--just 25 nm across (about 75 atoms) six times smaller than the width of the most minuscule component of a commercial microprocessor. The technique has the potential to supplement conventional microelectronics manufacturing techniques such as optical lithography, which is currently limited, at least commercially, to production of features ranging near 150 nm in width.
Stephen Sass, Cornell professor of materials science and engineering, and Melissa Hines, associate professor of chemistry and chemical biology, believe that the etching process will enable them to produce silicon structures as small as 10 nm, which is the distance between binding sites on a human antibody. "We now have the tools to make very fine-scale surface features at the length scale of biologically important molecules, such as human antibodies" says Sass. "We would hope to develop properties and applications we can't even imagine today."
Sass and Hines note that a process based on CED can potentially produce tiny structures across an entire 6-in. silicon wafer, suggesting that the technique will scale up easily for industrial production. "I like to think of this as creating nano-Lego surface structures, which ultimately we can build on to make a variety of devices," says Sass.
"This work provides an interesting new approach to creating arrays of periodic nanostructures on a semiconductor surface," comments Harold Craighead, director of the Cornell Nanobiotechnology Center, where part of the research was performed. "If one can create large areas of precisely controlled objects, some of the problems associated with random sizes on depositions could be reduced."
Sass notes that the theory behind CED has been known for a long time. In their process, two identical silicon wafers are twisted, one against the other, at a precisely controlled angle, and then are bonded together to make a bicrystal. Because of this twist angle, rows of atoms in the two wafers don't line up correctly. Only where the rows cross each other do the atoms line up and form strong bonds. The result is a repeating atomic scale mismatch that can be simulated by a moire pattern. (See QuickTime demonstration at www.chem.cornell.edu/mah11/Nanofab.html.)
Sass has studied the grain boundaries of crystals for 30 years. In some regions (the squares of the moire checkerboard), the atoms line up correctly and, therefore, are strongly bonded. In other regions (the lines around the squares of the checkerboard), the atoms are poorly bonded, leading to dislocations. Using a solution of chromium trioxide and hydrofluoric acid, the Cornell researchers etched away any dislocations to produce an array of nanobumps. To prove their concept, they fabricated a test structure with nanobumps that were approximately 100 atoms (25 nm) in diameter and 160 atoms (38 nm) apart.
Sass and Hines calculate that if a silicon bicrystal were fabricated with a twist angle of 4 degrees, in principle, nanobumps just 5.5 nm in width, or about 20 atoms, could be created. In essence, though, the width of the nanobumps does not matter as much as the distance between them. Because this spacing decreases as the twist angle of one silicon crystal relative to the other increases, says Sass, an angle of 10 degrees would produce a distance between nanobumps of only two nm (about 6 silicon atoms), �although we really don't know what the limit of this technique is.�
Long term, the CED technique has the potential to open the door to manufacturing at biological dimensions because many molecules, such as human antibodies, have features on a similar scale. Other technology possible with CED, says Hines, might be light-emitting silicon devices. Normally silicon does not emit light, but in microscopic fragments of silicon, electrons are confined and travel in a way that allows them to emit light. "That might mean that you could make a flat-panel display for a computer out of the same stuff you use to make the computer itself," she says.