NCSA NEWS

How the fuel cells will be won

By J. William Bell, NCSA

Story posted October 26, 2006

With calculations of their most basic reactions' activation energies, fuel cells may someday move off the frontier and into widespread use.

Hydrogen often wears the black hat when we talk about the prospect of everyday fuel cell use. (In other words, when we talk about fuel cells as a remedy to our hunger for oil.) Hydrogen is difficult to transport and store. Alternative means of getting it -- creating it on the fly from oil- or alcohol-based fuels -- generate their own problems. The hydrogen they produce is less than pure and thus less efficient.

But oxygen, which is the other necessary reactant in many of today's fuel cell designs, is something of a villain in its own right. Splitting oxygen molecules into oxygen atoms and the subsequent formation of water are currently the rate limiting step in the process of getting energy from fuel cells, the reaction that restricts overall power production. According to the University of Wisconsin at Madison's Manos Mavrikakis, this oxygen reduction reaction, as it's known, is responsible for about 60 percent of common fuel cell designs' overall efficiency loss.

Until this part of the process becomes more efficient, low-temperature fuel cells will not be commercially viable. Current approximates by the Department of Energy's Office of Energy Efficiency and Renewable Energy peg the costs of a set of fuel cells capable of running a car at more than $100 per kilowatt. By comparison, according to the office, a car's internal combustion engine produces energy at about $30 per kilowatt. Building the set, meanwhile, would cost more than $8,000.

"There are two goals for people studying these fuel cells, improved cost and more power," says Mavrikakis, an associate professor in the chemical and biological engineering department. He and his collaborators at Wisconsin and a group of experimental chemists at Brookhaven National Laboratory are addressing both of these concerns, exploring the oxygen reduction reaction and the catalyst that provokes it. Namely, they're looking to make the reaction more efficient and to reduce the cost of the expensive all-platinum catalyst. The Wisconsin team includes Ye Xu and Anand Nilekar; the Brookhaven team is led by Radoslav Adzic.

"This is the new role theoretical chemists are starting to play. Guiding experiment to synthesize new materials. Designing the right catalytic materials to steer reaction selectivity in the desired direction. It's mind boggling to be able to do this from first principles on a computing system like NCSA's Tungsten," he says.

"The theoretical elements elucidate what's going on and make the experimental elements much stronger," says Adzic, a senior chemist at Brookhaven. "They allow us to think further. We can always do a little better if we understand a little more. That's why [working with people like Mavrikakis] helps."

The cell, the reactions, and the energies

A hydrogen-oxygen fuel cell, like a battery, has two poles. At the anode, hydrogen is exposed to a catalyst and breaks into protons and electrons. Beyond the anode is a polymer membrane that allows protons through but does not permit electrons. The electrons are instead shunted through an external channel. As they pass through this channel, electricity is generated. The channel returns the electrons to the cathode side of the fuel cell, which is where the protons also end up after their trip through the membrane.

The cathode is a catalyst as well. Traditionally it is pure platinum, and it is fed oxygen rather than hydrogen. Upon interacting with the catalyst, oxygen molecules are broken into atomic oxygen. The ions meet up with the freed protons and electrons from the anode side. This reaction forms water, the fuel cell's predominant by-product.

Mavrikakis and his team focus on the oxygen reduction reaction that breaks the bonds of the oxygen molecules and the formation of a hydroxyl intermedate, which consists of a single hydrogen bound to a single oxygen. Adding a second hydrogen to ultimately form water is "pretty easy," according to Mavrikakis. It is not thought to play a significant role in improving catalysis at the cathode.

They treat each step as a separate, intense calculation. The only inputs are the types of atoms and their original positions in space -- thus "first principles." Each of the atoms naturally interacts electrostatically and electronically through the catalyst's surface with the rest of the atoms in its vicinity, yielding forces acting on each atom. A density functional theory simulation code called DACAPO calculates these forces and moves the hydrogen, oxygen, and catalyst atoms based on the forces acting upon them, until the minimum energy, equilibrium structure is determined. The team is looking for each reactions' activation energy, which represents the minimum amount of energy needed to transform reactants to products.

Who was that masked metal?

The reactions' activation energies are important because they determine the efficiency of the fuel cell overall. The lower the activation energies, the more readily the reactions occur. Fuel cell designers use different catalysts to manipulate the activation energy of the oxygen reduction reaction, which, remember, is the rate limiting step of the overall fuel cell process. Platinum is currently the state of the art, but Mavrikakis' calculations and Adzic's experiments are showing that there are likely better options available. The team previously computed at NCSA's sister site, the San Diego Supercomputer Center. Work continues at NCSA and at sites supported by the Department of Energy.

In 2005, articles by the team in The Journal of the American Chemical Society and Angewandte Chemie, International Edition showed the value of what are called platinum-monolayer catalysts. The bulk of these catalysts are a cheaper material with a layer of platinum that is a single atom thick covering them. They say that nature can't be fooled, but the team has proved that a good mask goes a long way.

In their studies, the group compared the oxygen dissociation and hydroxyl formation steps when exposed to a variety of catalysts, including pure platinum as a control and platinum-covered ruthenium, iridium, rhodium, gold, or palladium. Ultimately they found that palladium with the platinum monolayer, which is markedly less expensive than pure platinum, offered the best overall performance characteristics. It improved the overall efficiency of the oxygen reduction reaction by 33 percent.

"The catalyst needs to do both steps in a nice way. The goal is to strike a balance between facile oxygen dissociation and facile hydroxyl formation," Mavrikakis says. If, for example, the oxygen breaks apart too easily, designers risk covering the catalyst's entire surface with atomic oxygen that's strongly bound. "New oxygen won't find a spot. Imagine it is like an airplane with no runway at the airport. It takes off again." And it takes off in it's molecular form, hindering the efficiency of the overall reaction.

These findings assumed that oxygen dissociation occurred prior to the addition of protons. However, that assumption has not been proven to reflect reality. Currently, the teams' calculations focus on other competitive reaction paths for the oxygen reduction reaction. These would include the addition of one, if not two, protons before the molecular oxygen bond breaks. The team also avoids accounting for any already-formed water molecules on the cathode side. In the future, they hope to add such molecules, develop the reactivity trends, and explore how those molecules might influence the oxygen reduction reaction as well.

'Everything is proven in reality'

Only a very limited number of people in the world can even make the custom platinum-monolayer catalysts. The team at Brookhaven uses a specially devised technique. Copper is first allowed to absorb on the metal, whether palladium or one of the other candidates. It is then placed in a platinum solution wherein that one-atom layer of platinum replaces the copper. "It's very elegant," says Adzic. "There's not enough platinum in the world [to supply all the fuel cells that might someday be in demand]. By working with a monolayer, you can reduce the platinum load in a fuel cell by 10 times."

With knowledge from the Mavrikakis team on the basic chemistry of the catalytic reaction, the folks at Brookhaven are further refining the candidate catalysts, looking for ways to reduce not only that platinum load but also the load of the other exotic elements like palladium. Currently, they're experimenting with substrates that have a nickel core, palladium on top of that, and then a final platinum monolayer.

"Everything is proven in reality," says Mavrikakis. "We've proven that [platinum monolayers can make the necessary oxygen dissociation and hydroxyl reactions] go faster and cheaper. The question that remains is how long do they go faster and cheaper. Is it only for the first second? Or will it last for the typical lifetime of your car?"

To that end, Adzic teams with a group, led by Francisco Uribe, at Los Alamos National Laboratory. They test the performance of prototype fuel cells. Already they've seen some last thousands of hours without diminished performance, according to Adzic. This approach -- from theoretical calculations on systems at NCSA, to experiment with the candidate catalysts in the lab, to experiment with real-world prototypes -- may someday clear oxygen's good name. And bring fuel cells into the mainstream and onto the streets.

This research is supported by the National Science Foundation (ENG/CTS), the Department of Energy (Basic Energy Sciences and Energy Efficiency and Renewable Energy), and S.C. Johnson.

Team members:
Radoslav Adzic
Manos Mavrikakis
Anand Nilekar
Kotaro Sasaki
Francisco Uribe
Miomir Vukmirovic
Ye Xu
Junliang Zhang