Science @NERSC: Silicon Nanowires

Ball and stick representation of the atomic
structure of a silicon nanowire. The image
highlights the different arrangement of atoms
on the surface of the wires (orange) and in the
crystalline core (green).

As the second most abundant element in the earth's crust, silicon is vital as a principal constituent of all kinds of materials, from semiconductor wafers to cement. But now researchers are trying to broaden its utilization, wondering if it could be used as an efficient thermoelectric material, one that turns heat into electricity.

Professor Giulia Galli and co-workers from the University of California, Davis are using the Bassi and Franklin supercomputers at NERSC to explore this question. And the answer is important because nanostructures made of silicon may find application in sustainable energy areas such as power generation and heat dissipation.

Although silicon is rather inefficient in its bulk form, scientists suspect that silicon nanowires may exhibit high efficiency in converting heat into electric current. However, the fundamental reasons for this are not understood and different interpretations have appeared in the literature. Professor Galli's calculations explain that the thermal conductivity strongly depends on the surface structure of the wires. Wires with rough (amorphous-like) surfaces have much lower thermal conductivity (up to 2 orders of magnitude) than bulk silicon.

Galli's studies depend on NERSC resources and the group uses a variety of methods and computer codes to solve their research problems, a common and critical characteristic of much research carried out at NERSC. For example, the silicon nanowire studies involve a combination of classical molecular dynamics, density functional theory (DFT), and Boltzmann transport equation calculations to provide the necessary microscopic description of the thermoelectric properties.

"Access to novel resources such as Franklin has allowed us to run new projects on integrated nanostructures," said Galli.

The work has resulted in three publications so far:

• D. Donadio and .G. Galli, "Atomistic simulations of heat transport in silicon nanowires," Phys. Rev. Lett. 102, 1195901 (2009)
• T. Vo, A. Williamson, V. Lordi and G. Galli, "Atomistic design of thermoelectric properties of silicon nanowires", Nano Lett., 8, 111(2008).
• J-H. Lee, G. Galli and J. Grossman, "Nanoporous Si as an efficient thermoelectric material", Nano Lett. 11, 3750 (2008), featured in Nature Materials.

The "holy grail" of materials research involves not only explaining the properties of existing materials but also predicting properties of yet-to-be-synthesized ones. Galli's work suggests that creating a small amount of disorder in the surface of a square silicon nanowire could lead to a ten-fold improvement in thermal conductivity, and if some of the silicon atoms in an alloy nanowire were replaced by silicon's close relative, germanium, the conductivity would improve even more. Now it's up to the synthetic chemists to follow her clue.

Visit Professor Galli's Ab initio Nanoscience Group for Simulation and Theoretical Research on Materials (ANGSTROM) web pages for additional information.