Categories: LabsLawrence Berkeley

Nanostructural problem-solvers

The preeminent physicist-futurist Richard Feynman famously declared in a 1959 address to the American Physical Society that “there’s plenty of room at the bottom.” He then invited them to enter the strange new world of nanoscale materials, none of which had actually been invented, except in Feynman’s fantastical imagination.

It took another generation of scientists before nanotechnology emerged, but Feynman’s assertion still rings true. There’s plenty of room at the nanoscale and scientists at Lawrence Berkeley National Laboratory (LBNL) in California are at the forefront in constructing new materials there.

Paul Alivisatos, director of LBNL’s Materials Science Division, is a world leader in nanostructures and inventor of many technologies using quantum dots — special kinds of semiconductor nanocrystals. Quantum dots, which are one ten-millionth of an inch in diameter, fluoresce brightly, are exceedingly stable and don’t interfere with biological processes because they are made of inert minerals. Alivisatos and his colleagues have constructed dozens of variations in which the fluorescent color changes with the dot’s size. Today life-science researchers use quantum dots as markers, allowing them to visualize with extreme accuracy individual genes, proteins and other small molecules inside living cells and fulfilling a prediction Feynman made in his famous lecture.

LBNL physicist Lin-Wang Wang likes to say that some day we will view the 21st century as the “nanostructure age.” Wang and LBNL colleague Andrew Canning, a computational physicist who helped pioneer the application of parallel computing to material science, want to use computational methods to understand the emergent behaviors of novel materials built from exceedingly small blocks.

“There are a lot of challenges and there are still many mysteries to be solved,” Wang says. “For example, we still don’t quite understand the dynamics of the electron inside a quantum dot or a quantum rod. There is a lot of surface area in a quantum structure, much more than the same material in bulk. So how the surface is coupled with the interior states and how this affects the nanostructure properties is not well understood.

The research team is not starting from scratch, of course. There are established equations that predict the behavior of the electron wave function in these materials. The devil lies in the size of the problem.

“In terms of computation the nanostructure is challenging. For example, if you have a bulk material the crystal structure is a very small unit cell, just a few atoms, that repeats itself many, many times,” Wang says. “So computationally, you can treat bulk structures by calculating one unit cell — you only deal with a few atoms.

With only a few atoms, you can represent the whole, much larger structure of the material. However, for a quantum dot or a quantum wire you have to treat the whole system together. These systems usually contain a few thousand to tens of thousands of atoms, and that makes the computation challenging.”

To solve a problem containing thousands of atoms requires new algorithms that handle the physics differently without compromising accuracy and parallel computing on a massive scale.

Says Canning, “We know we need to solve the Schrödinger equation for these problems, but to do so fully is exceedingly computationally expensive. What we did was make advances to approximate, solve the problem, and still get the physics right.”

Canning collaborated with Steven Louie’s group at the University of California-Berkeley, to improve the Parallel Total Energy Code (Paratec), an ab initio, quantum- mechanical, total energy program. The program runs on Franklin, the Cray XT4 at LBNL’s National Energy Research Scientific Computing Center (NERSC). The massively parallel system has 9,660 compute nodes, but is due to receive an upgrade, increasing its processing capability to a theoretical peak of about 360 teraflops.

“Paratec enables us to calculate thousand-atom nanosystems,” Canning says. “The calculation is fast and scales to the cube of the system, rather than exponentially, as a true solution of the many-body Schrödinger equation.”
Besides massive parallelization of the codes, the researchers also developed many new algorithms for nanostructure calculations. For example, Wang devised a linear scaling method, called the folded spectrum method, for use on large-scale electronic structure calculations.

The conventional methods in Paratec must calculate thousands of electron wave functions, but the Escan code uses the folded spectrum method to calculate only a few states near the nanostructure energy band gap. That means the computation scales linearly to the size of the problem — a critical requirement for efficient nanoscience computation.

Wang and Canning recently worked with Osni Marques at LBNL and Jack Dongarra’s group at the University of Tennessee, Knoxville, to reinvestigate and significantly improve the Escan code by adding more advanced algorithms.colleagues also have recently invented a linear scaling three-dimensional fragment (LS3DF) method, which can be hundreds of times faster than a conventional method in calculating the total energy of a given nanostructure.

The code has run at 107 teraflops on 137,072 processors of Intrepid, Argonne National Laboratory’s IBM Blue Gene/P. The researchers have in essence designed a new algorithm to solve an existing physical problem with petascale computation. Wang says the LS3DF program is designed for materials science applications such as studying material defects, metal alloys and large organic molecules.

Within a nanostructure, the physicists are interested mainly in the location and energy level of electrons in the system because that determines the properties of a nanomaterial. For example, Wang says, electrons within a quantum rod or dot can occupy a series of quantum energy states or levels as they orbit the atomic nucleus and interact with each other. The color emitted by the material typically depends on these energy states.

Specifically, the scientists focus on two quantum energy levels: the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO), which the Escan code can calculate. The energy difference between these two levels determines the material’s color.

The color also changes with the quantum dot’s size, providing one way to engineer its properties. In principle, knowing the electronic properties of a given material lets the researchers predict how a new nanostructure will behave before actually spending the time and money to make it. It’s a potentially less expensive way to experiment with new nanomaterials, Wang says.

This article originally appeared in Deixis: The CSGF Annual, 2008-09.

Share
Published by

Recent Posts

Genomic field work

A UC Davis fellow combines computing, the corn genome and growing data to help farmers… Read More

March, 2024

Tracking space debris

 A Computational Science Graduate Fellowship recipient monitors threats to satellites in Earth’s ionosphere by modeling… Read More

February, 2024

Decisive achievement

A computational sciences fellow models COVID-19 virus variants and examines how people weigh complex decisions. Read More

October, 2023

Learning climate

A Colorado State fellow employs machine learning for climate modeling, putting provenance behind predictions. Read More

August, 2023

Statistically significant

A LANL statistician helps cosmologists and epidemiologists grasp their data and answer vital questions. Read More

July, 2023

Cutting carbon, blocking blooms

Besides bioplastics research, the LANL Biofuels and Bioproducts team is studying carbon neutrality and applying… Read More

June, 2023