News

A "nanopeapod" in which carbon or boron-nitride nanotubes have the intriguing ability to encapsulate other chemical species. Simulations using SDSC's DataStar supercomputer showed that encapsulating Buckyball "peas" added new features to the electronic structure of the host nanotube, which may prove useful in molecular-scale electronic devices.

by Paul Tooby, SDSC Senior Science Writer

In the tiny world of nanomaterials--smaller than the wavelength of visible light--strange new materials can exhibit unexpected behavior and properties, not at all what we would expect from our experience in the larger scale of everyday life. And because they behave in novel ways, researchers are excited about the potential applications of these new materials to make dramatic improvements in technologies from electronics to medical diagnostics.

For example, nanoelectronic devices may offer vastly-expanded data storage capabilities with reduced power consumption, and a newly-discovered nanoscale device, described below, emits body-penetrating infrared light that could one day improve medical diagnosis.

The research groups of physics professors Marvin Cohen and Steven Louie at UC Berkeley collaborate on investigations in which they use supercomputers at the San Diego Supercomputer Center (SDSC) at UC San Diego as a key part of their computational physics research. They combine this with theoretical approaches in which they aim both to explain the properties and behavior of nanoscale materials and also to predict entirely new materials and phenomena never seen before.

Nanomaterials are a very dynamic area of research, with exciting discoveries from buckeyballs to nanotubes emerging in recent years through collaborative efforts among theoretical, computational, and experimental researchers. In this creative research environment, the Berkeley scientists find that computational science is an essential tool, sometimes responding to experimental and theoretical advances, and at times leading and identifying novel materials and phenomena. "Our ab initio simulations model materials from first principles," said Professor Cohen of UC Berkeley. "These realistic simulations, which can yield discoveries of materials and properties never seen before, are very computationally intensive, requiring the large, data-oriented resources that SDSC provides."

Nanopeapods and Buckeyballs
To explore nanomaterials in a number of areas, the Cohen and Louie groups received a combined allocation of some 1.3 million processor hours on DataStar. Processor-hours refers to how time is allocated on a supercomputer. A project with a one million hour allocation could run on 1,000 processors of a parallel machine for 1,000 hours, or about 42 days. Running the same project on a single-processor computer would take more than 115 years!

One class of inhabitants of the exotic nanoworld the researchers are exploring is known as "nanopeapods," in which carbon or boron-nitride nanotubes have the intriguing ability to encapsulate other chemical species. In nanopeapods, or functionalized nanotubes, the details of the position of guest atoms within the cage-like structure are of great interest to materials scientists because they affect electronic and other properties of these systems.

The researchers studied the electronic and geometrical structure of boron nitride and carbon nanotubes in encapsulating the C60 molecules known as Buckeyballs. Using ab initio calculations with the PARATEC and SIESTA codes, they performed self-consistent density functional theory computations. The structural complexity of nanopeapods means that a very large number of atoms had to be included in the simulation cell, requiring major computational resources. In the nanopeapods, the researchers found that the Buckyball "peas" added new features to the electronic structure of the host nanotube, resulting in changes in the electronic transport properties that may prove useful in molecular-scale electronic devices.

The Cohen and Louie groups have also studied other nanoscale electronic devices using first-principles calculations of electron transport. These simulations used a scattering-state code that the researchers developed to calculate the electrical current in these molecules and nanostructures as a function of bias voltage. One motivation for this research, in addition to learning more about the fundamental science, is eventually to be able to extend Moore's law for ever-faster computing devices by developing devices that are smaller, more energy-efficient, and faster to overcome the limitations to further shrinking silicon-based semiconductors.

Simulating electron transport through single-molecule devices, the researchers explored behavior suggested by experiments in which resistance in the nanoscale devices can sometimes unexpectedly decrease. This is in striking contrast to "normal" behavior in bulk metal wires, where resistance increases as bias voltage increases. This nonlinear behavior is exciting to physicists and electrical engineers because it holds out the promise of new electronics applications.

Light-emitting Nanotubes
Yet another area of research involves the optical properties of nanotubes. It turns out that carbon nanotubes can emit light in the infrared range. The researchers have simulated this behavior in the one-dimensional world of nanotube semiconductors, where the excitons, or electron-hole pairs, are very strongly bound. "With the help of SDSC resources, we were able to make the exciting discovery of a giant excitonic effect in the optical response of the nanotubes," said Professor Steven Louie of UC Berkeley. "This effect is orders of magnitude larger than what is found in standard semiconductors such as silicon." The ability to emit infrared light with strikingly different optical properties than in bulk material has potential applications in such areas as medical diagnosis, since infrared light can penetrate the human body.

The researchers computed the optical properties from first principles in three steps using three codes, two running on 64 processors in runs of up to 1,500 hours and the final step computing on as many as 640 processors in runs of 24 hours. The computationally most burdensome parts of these codes involve the algebraic manipulation of very large matrices, as well as Fast Fourier Transforms. The data-oriented resources of SDSC were vital to their research, the scientists explained, because of the very large computational requirements, the large memory requirements, and especially the need for very fast data communication between nodes that is provided by DataStar.

Nanoscale Superconductivity and Friction
Another area the researchers explored is superconductivity in novel materials. By building new structures not found in nature such as boron nitride nanotubes containing C60 molecules and potassium, new electrical properties including superconductivity can emerge. By simulating these structures, the researchers are gaining insights into the basic physics taking place, which they hope can eventually lead to higher-temperature superconducting materials with major technological applications such as more-efficient electricity transmission.

Finally, the wide-ranging research of the Cohen and Louie groups used SDSC resources to explore friction at the nanoscale. This is important in the development of nanomachines such as tiny motors. The simulations are helping answer questions related to how mechanical energy is dissipated at these small scales, with properties that differ greatly from what is observed at the scale of bulk materials we experience in everyday life. For these simulations, the researchers used classical molecular dynamics simulations of tens of thousands of atoms. --Paul Tooby.