Using Blacklight and other XSEDE resources, a computational chemist shows the feasibility of “tuning” carbon-based semiconductors, thereby mapping a path around the roadblocks of silicon-based electronics

Our ancestors had stone, bronze and iron. For us, silicon defines our age. This material’s unique properties as a semiconductor make it the foundation of modern electronics, and over a half century of innovation in smaller, faster circuitry — with silicon as the base material — has transformed the way we live. One of today’s smart phones has more computational power than Apollo 11. Our businesses run on computers and Internet trade. Our military conducts war via satellites and computerized drones. It’s not an exaggeration to say that America’s economic and physical security depend on silicon.

PHOTO:Tomlinson and

Aimée Tomlinson has trained undergraduates at North Georgia College & State University in using XSEDE resources that include Abe, Cobalt and Ember at NCSA and Trestles at SDSC along with Pople and Blacklight at PSC. This work has produced several student posters that have been presented at local, regional and national meetings

PHOTO:Marcela Madrid

Marcela Madrid, PSC, XSEDE Extended Collaborative Support Services

While ever more condensed circuitry — better performing silicon “chips”— has led us to where we are, roadblocks loom. Gordon Moore, for instance, of “Moore’s Law” — which holds that silicon devices will double their circuit density every two years — has predicted that this progress will eventually stall, as miniaturization is inherently limited by the size of atoms.

Awareness of these limitations drives research seeking workarounds to silicon. One of the most promising may be carbon — in the form of “conjugated polymers,” organic long-chain molecules that conduct electricity and can act as semiconductors. The 2000 Nobel prize in chemistry recognized three chemists — Alan Heeger, Alan MacDiarmid and Hideki Shirakawa — for the discovery and development of these molecule-sized wires.

“Conjugated polymers,” says Aimée Tomlinson, “have the potential to revolutionize the world of semiconductors.” Tomlinson is a computational chemist at North Georgia College & State University. She used Blacklight at PSC and other XSEDE resources in Illinois (NCSA) and the Trestles system at San Diego Supercomputer Center to study materials, called benzobisoxazoles (BBOs), that can be used as chemical building blocks to produce conjugated polymers (CPs). Her work is aimed particularly at solar cells.

“Organic materials provide several benefits over inorganics for solar cells,” says Tomlinson. They are “greener” to mass produce, less detrimental to the environment, she points out, and they have plastic-like properties that are advantageous in industrial processing, since they can be more easily molded than inorganics like silicon into various shapes.

Tomlinson worked with a team of her undergraduate students and collaborated with synthetic chemist colleagues at Iowa State University. Her quantum computations predicted behavior that closely matched experimental findings from lab-synthesized BBOs. Their results — reported in The Journal of Organic Chemistry (October 2011) — show that electronic properties of CPs can be varied in predictable ways, saving trial-and-error lab work, to reliably forecast desired electronic behavior, work that reinforces the promise of CPs.

Tuning the Orbitals

Compared to inorganic semiconductors, CPs are cheaper to manufacture, observes Tomlinson. Along with their plasticity, they are also greener — the production process doesn’t rely on toxic metallic inorganics. And in theory they offer a far greater versatility in electronic behavior.

In much the same way that silicon can be “doped” — by the addition of impurities — to produce specific electronic properties, CP organic “wires” can be doped by adding molecular side groups along the main chain. “From a theoretical point of view,” says Tomlinson, “if you add a chemical group that tends to donate electrons, that will make your polymer chain more of an electron donor, willing to transfer electrons to other molecules. If you add a group that withdraws electrons, that’s going to be more of an electron accepting material.”

But you need to “tune” a donor and an acceptor for them to work properly with each other, she adds. This requires knowing how side groups will affect the critical parameters — the “highest occupied molecular orbital,” or HOMO, and the “lowest unoccupied molecular orbital,” or LUMO. HOMO describes the highest-energy electrons in a polymer at rest; LUMO describes the higher energy level needed to start them moving in an electrical current. They’re both important for telling whether a polymer is likely to work for a particular device.

The ability to fine tune electronic properties by changing side groups may allow these organic materials to perform tricks that silicon could never match.

To capture light energy in a solar cell, which means to convert photons — quantum particles of light — into electricity, the energy of the photon, which is determined by its color (the wavelength of light involved) has to equal the difference between the HOMO and LUMO. For a pair of molecules, the energy difference between the HOMO of one and the LUMO of the other must be just right for the electrons to make the jump.

Tomlinson used the quantum chemistry software GAUSSIAN09 to compute the electronic properties of BBOs. Carnegie Mellon chemist John Pople received the 1998 Nobel prize in chemistry for his work in the development of GAUSSIAN and its underlying theory, which provides a way — with the computational power of modern high-speed processors — to solve the quantum equations that govern electrons and provide accurate predictions of their behavior.

Blacklight’s shared memory was essential for these GAUSSIAN calculations, says Tomlinson, because the volume of data involved in the computation goes up with the square of the number of atomic orbitals in each atom, multiplying between atoms. “The shared memory helped tremendously,” she says. “The more memory you can put on these calculations, the faster they go.”

PSC scientist and XSEDE consultant Marcela Madrid helped to coordinate with Tomlinson and notes that her computations made efficient use of Blacklight. “She used approximately 90 percent of the 128 gigabytes of memory available on 16 cores.”

The calculations showed deviation between predicted HOMO values for BBOs and those of the synthesized BBOs ranging between 2.4 and 12.8 percent. Corrections based on more accurate starting molecular structures brought all the errors below 3.5 percent, a reliable range of predictive accuracy to identify compounds to test in a device. The full value of the result, notes Tomlinson, may be best gauged in time. The results give confidence that Tomlinson’s computations can predict the HOMO-LUMO parameters for dozens of molecules concurrently in one to three weeks. In contrast, to synthesize and test the same molecules could take one to six months for each molecule.

A Simple Twist
For this BBO compound, one of several that Tomlinson modeled, her calculations showed that the 90° twisted geometry (left) gave closer correlations with experimental findings than the planar geometry. “Nonplanarity is supported by other research,” explains Tomlinson, “and we know that the experiments are in solution where these compounds have flexibility to move around.” The associated “electrostatic potential map” (right) shows polarity of charge, positive (blue) to negative (red). “The red region is electron rich and becomes increasingly electron poor as we go from red to blue.”

Electronic Paper and Molecular Cookbooks

Tomlinson’s goal is to identify photoelectronically active CPs that can yield efficient solar cells. The ability to predict CP electronic properties — and to fine tune those properties by changing side groups — may allow these organic materials to perform some tricks, says Tomlinson, that doped silicon could never match.

In particular, because they are physically flexible carbon chains, CPs may offer the ability to imbed computers in everyday devices. And they can be transparent. “Smart” windows could change optical properties in response to incoming light or user commands. “Electronic paper” that includes these smart windows could combine the functions of ordinary paper with those of a computer terminal.

One avenue that the researchers are exploring arises because the side groups of semiconducting polymer chains can themselves be polymer chains. Longer chains festooning the main chain could make the polymer more soluble in organic solvents, and thus more accessible to be chemically manipulated. Long-chain side groups may also broaden the extent to which those groups could alter the properties of the entire polymer, yielding new components.

Possibly the most exciting outcome of this research could be a molecular cookbook. Tomlinson envisions a database that would allow device designers to stipulate the physical, electronic and photonic behavior they need, and from that generate a list of compounds that would meet the specifications. It’s a level of programmed innovation that would revolutionize device design.

“That’s a long way off,” says Tomlinson, and would inevitably take the contributions of many researchers. But her work is an important step that helps the field make the crucial transition from random testing into targeted development.

How Polymers Conjugate
The ability of organic molecules to act as conductors — and semiconductors — stems from the physics that pulls atoms together into molecules. When two atoms share one of their electrons with each other, they form a single bond; if two atoms share two electrons each, they form a double bond — indicated in the bonding diagram by two bars.

When a chain of carbons alternates between single and double bonds, as in the bonding diagram, the clouds can overlap, and the system becomes “conjugated” — bonded in a manner in which an electron from each carbon atom gains the ability to “come loose.” These electrons then are no longer tied to the atoms they came from — they can move freely along the chain of carbons. In this sense, alternating single and double bonds in an organic polymer can act as a molecule-sized wire.

© Pittsburgh Supercomputing Center, Carnegie Mellon University, University of Pittsburgh
300 S. Craig Street, Pittsburgh, PA 15213 Phone: 412.268.4960 Fax: 412.268.5832

This page last updated: September 26, 2012