Latest News


Good Day Sunshine

Published on January 15, 2010 by Aaron Dubrow

Photovoltaic cells produce electricity directly from sunlight. [Dept. of Energy]

In just one hour, the sun provides the Earth with enough energy to power the planet for a year. So an obvious way to solve the world's energy crisis would be to develop solar panels capable of capturing the sun's rays and turning them into usable electricity. However, solar panels (also known as solar photovoltaics, or PVs) are currently the least used alternative energy power source—and the reason is simple.

"It's much more expensive at present than wind or solar-thermal conversion methods," said Jeffrey Grossman, professor of Materials Science and Engineering at the Massachusetts Institute of Technology (MIT). "Cost and efficiency are the big problems."

Grossman believes both problems can be overcome with the addition of some critical knowledge. Using the high-performance computers (HPC) at the Texas Advanced Computing Center (TACC), his group at MIT has performed atomic-scale computational simulations that reveal the fundamental mechanisms behind energy conversion and storage.

"We ask questions about how energy conversion works—why different materials and different interfaces do what they do," Grossman said. "Once you understand these mechanisms, then the next step is to use that understanding to predict some improvement in the material."

Grossman's nanoscale virtual experiments help predict which materials, structures, and dopants are most likely to produce cheap, efficient, and viable solar PVs in the near future.

Improving Solar Cells

Humans have been studying and experimenting with important materials for thousands of years—refining iron, growing silicon—yet much remains unknown about the fundamental nature of the material world. This is particularly true at the nanoscale (less than 100 nanometers, or 400 times thinner than the width of a human hair), where materials often have radically different characteristics than they do as bulk substances.

It is at the nanoscale—and on the level of individual atoms—that solar photovoltaics work. But the behavior of such minute systems cannot be fully explored by physical experiments or microscopy. In order to ‘see' the quantum interactions of thousands of atoms and understand how they act in concert to convert photons into electricity, it requires HPC.

Over the course of the last year, Grossman has used Ranger, one of the most powerful supercomputers in the world, to make a number of important discoveries about the nature of promising new materials for solar energy conversion.

Predicting the electronic, structural, and optical properties of hybrid amorphous systems (right), and towards a microscopic understanding of light-induced degradation in amorphous silicon (left).

To convert a ray of sunlight into electricity, a solar photovoltaic cell first absorbs the sun's photons. The photons knock the electrons in the material into a higher state of energy, creating a positive and negative charge in the system. The next step is to pull those charges out of the material. Currently, the best solar PVs made out of a single material get around 25 percent efficiency, while those made of cheaper materials get only six percent or less.

To solve the cost-benefit problem that solar PVs face, Grossman has been exploring the mechanics of fullerene–polymer-based organic solar cells, which are among the most promising alternative materials for conversion.

"It's very hard to measure, understand and characterize experimentally how a charge separates at the interface between one type of material and another, in this case fullerenes and a polymer," Grossman said. "We provided some insights into that separation process, and our calculation showed that it's not a simple one-step separation. It actually takes two steps."

The hybrid material possesses two electronic states that are favorable to the electron, according to Grossman's simulations. One state is shared across the interface and allows the electron to easily move across the boundary from the polymer to the fullerene, while the other state is fully localized on the fullerene.

"The fact that these two states are next to each other energetically allows the electron and the hole to separate so efficiently," Grossman said. His findings will be explored further by experimental collaborators, possibly leading to breakthrough improvements in solar PVs.

Searching for Better Materials

Clarifying the behavior of existing solar PV technologies is important, but Grossman is interested in taking his HPC-driven explorations into virgin territory. "We want to use this computational framework, and the field expertise we've gained, to develop new materials and new ideas," he said.

To that end, Grossman explored new ways to efficiently separate an electron from a hole in a material. One promising discovery involves nanowires, in particular the tapering edges that are a by-product of their production. These edges had been viewed as undesirable by many experimentalists, but Grossman suspected they might have unexpected benefits for solar energy extraction.

"We thought that the tapering might change the electronic and optical character of the nanowires, so we studied the feature and found that, in fact, the tapering in this quantum confinement region, by itself, can make an electron go to one side and a hole to the other," Grossman explained. "Now we have evidence that the shape of the material at this size scale influences its charge."

Another set of simulations by Grossman's group explored the nature of defects in amorphous silicon, a cheap material whose large number of defects and traps make it an inefficient converter of solar electricity—for now. Grossman believes that by studying the fundamental electronic, structural and optical nature of the material, he will discover a way of improving the efficiency of this less expensive material.

Today, computational materials science is able to do something the field was not previously able to do: simulate virtual systems accurately and quantum mechanically on a scale that can interact with the experimentalists' synthesized materials.

"There's a real drive for these simulations to represent actual interfaces that would be in a real solar cell," Grossman said. "We're trying to meet the experimentalists at scales that are relevant, using first-principle methods. In order to do that, we needed a lot of electrons."

Jeffrey C. Grossman, Department of Materials Science and Engineering, MIT

And a lot of electrons means a lot of computing power. "The only way we can do these kinds of calculations, at these sizes, with the complexity that we're looking for, is on a machine likeRanger," Grossman said. "It's essentially the difference between being able to solve the problem and not."

Since there are millions of potentially useful materials, each with countless configurations, lab-bench production may never be able to characterize all of them. But computational methods can come close.

"We can predict how something might behave before it's made," Grossman said. "This allows us to explore different phase spaces in what could be potentially game-changing advances, thereby guiding synthesis efforts."

It's difficult to fathom the impact that new materials and new frameworks for discovering and producing these materials, could have on our world, not only for solar photovoltaics, but also for thermoelectrics, hydrogen storage, and solar fuels. However, the continued growth of these alternative energy sources depends on finding cheaper and more efficient materials, and that task can be dramatically accelerated with high-performance computers and materials scientists like Grossman.

"The behavior of materials is going to be at the heart of advances that are needed at the global scale today, and that includes energy, but also health and information technology," explained Grossman. "I'm excited to be able to apply theoretical physics and computational science to find disruptive advances in new materials."

Story Highlights

Solar photovoltaic cells are one of the most promising alternative energy sources, however their adoption has been impeded by high costs and low efficiency.

To better understand the fundamentals of solar energy conversion and to identify potential new materials for cheaper and more efficient cells, researchers from MIT have been simulating the atomic behavior of solar cells using the Ranger supercomputer at TACC.

These simulations have led to insights about the mechanics of nanowires, amorphous silicon, and fullerene–polymer-based organic solar cells that may enable next-generation energy solutions.


Faith Singer-Villalobos

Communications Manager | 512-232-5771

Aaron Dubrow

Science And Technology Writer

Jorge Salazar

Technical Writer/Editor | 512-475-9411