Graduate Student Coumba Ndoye holds a 12-inch silicon wafer being used as a substrate to grow compound materials that could help improve the next generation of semiconductor technology.
The increasing density and speed of today’s semiconductor devices has hit the limits of silicon’s capability — and a Virginia Tech microelectronics team is developing a technology to not just continue semiconductor miniaturization for logic and memory chips, but also to colocate photonics applications with silicon-based circuitry, including light generation by lasers and more efficient solar-cells. The team’s technology vision may open new possibilities of integrating optical and logic functions.
Mantu Hudait (left) and Marius Orlowski
“The half-a-century-old silicon semiconductor technology is running out of steam,” comments Marius Orlowski, the Virginia Microelectronics Consortium (VMEC) professor of ECE. “The current industry’s trajectory demands to continue the transistor miniaturization to render chips denser and faster. However, currently there is no consensus of how to accomplish this.”
Mantu Hudait, who joined ECE in 2009 after serving in Intel’s Advanced Transistor and Nanotechnology Group, agrees. “I don’t think anyone really knows how we will continue Moore’s Law [doubling the number of transistors on a chip every two years] in the foreseeable future where the device dimensions are 15nm or below. Therefore, it is an excellent opportunity to explore new materials and device architectures, innovation for low-power and high-speed logic, memory, photonics and interconnects, all monolithically integrated onto silicon substrate.”
Materials are controlled with single atomic layer precision. A key enabling tool for deposition of subnanometer dielectric layers is the so-called Atomic Layer Deposition chamber seen in the forefront.
Orlowski, Hudait, and Louis Guido are seeking to meet this challenge by synthesizing compound semiconductor materials on a silicon substrate that would take advantage of established silicon technology base as well as of the superior photo-electronic properties of other materials. “Silicon technology has 50 years under its belt and commands an immense infrastructure,” Orlowski says. “Circuit design methodology, manufacturing-and-material know-how — the whole industry is geared for silicon… Although we cannot claim we know all the properties of silicon, we know a whole lot about this element. It is probably the best researched element in the periodic table.”
To take advantage of silicon technology infrastructure, but add new functions, the team is growing crystals of other so-called III-V compound semiconductors — such as gallium arsenide — on a silicon substrate. Using a process called heterogeneous epitaxy, they are forcing the III-V material to grow in single-crystalline structure on large-area, low-cost silicon wafers.
The challenges of heterogeneous epitaxy involve defects, processes that need to be controlled on an atomic scale, co-integration of heterogeneous technology, and demonstration of manufacturing viability.
The crystalline defects arise mainly from mismatch between different crystal lattices and their thermal mismatch, according to Orlowski. When a material is forced onto the crystal template of another, “you are changing the natural distance between atoms, trying to squeeze or stretch a material’s structure into different atomic distances,” he explains. This puts the new material under compressive or tensile stress and allows this stress energy to acccumulate. “The more layers you grow, the more stress energy will be accumulated. In seeking equilibrium configuration, the composite material tries to release this energy in the form of defects. The question then becomes how to confine defects to certain areas and not let them propagate to the area where we want to form transistors that require highly defect-free crystals.”
Another difficulty is coming up with an optimal process flow for creating devices with these new materials — the process flow for creating devices in silicon can’t be transferred unaltered to this new world. “Our knowledge of these III-V semiconductors must yet catch up with our knowledge of silicon in the next five to 10 years,” says Orlowski.
Synthesizing compound semiconductor materials on silicon substrates is a big challenge, he says. “However, if this can succeed, then it opens wonderful new opportunities, such as allowing integration of silicon-based electronics with optical devices.” One major limitation of silicon is that it does not convert light well on its own. But with two or more materials on the same substrate, many new applications are possible, such as on-chip lasers for sensors and even efficient solar cells.
The team is well poised for success, according to Orlowski. “My experience is in the silicon world; Hudait is a specialist in growing III-V compound semiconductors and device structures heterogeneiously onto Si And Ge substrates, and Guido is a specialist in compound semiconductor-based laser and optoelectronic devices. It will take the three of us to realize this vision of bringing the two worlds together. Together we have all the expertise we need to meet this challenge.”