National Center for Computational Sciences

By: Leo Williams

Simulation explores mechanism for proton transfer in water

Magic happens at the nanoscale. From antibacterial coatings to fuel-cell-powered cars, tomorrow’s inspiring new technologies depend on today’s researchers learning how to control the novel behavior of small-scale systems in a predictable way.

A team of materials scientists led by Jorge Sofo of Penn State University and Thomas Schulthess of Oak Ridge National Laboratory (ORNL) has used ORNL’s Jaguar supercomputer to successfully simulate such a system. Working from fundamental principles of quantum mechanics, the team accurately simulated the behavior of water in the presence of the common catalyst titanium dioxide, a material routinely used in solar cells and hydrolysis. The work not only improves our understanding of a process that is already important in areas such as fuel cells and the geosciences; it also prepares the way for simulations of ever more complex systems.

"This whole simulation sets the stage for a lot more work on more complicated systems," explained Paul Kent, a member of the team who worked extensively on the computer application used in the research. "This is much more than a proof of concept because we’ve got a lot of science out of this, but the idea is obviously to move on to more complicated materials."

Specifically, the team simulated the process by which water passes protons from one molecule to another. While a molecule of water—H2O—contains two hydrogen atoms and one oxygen atom, one of the hydrogens will occasionally break off; this leaves an unconnected hydrogen—the nucleus of which is a single proton—and a hydroxyl molecule (consisting of the leftover oxygen and single hydrogen). This proton can then be exchanged with other hydrogens in other water molecules, transporting the proton through the water.

The mechanism observed in the simulation, known as the Grotthuss mechanism, governs the way in which a proton is passed between hydroxyl atoms. More generally, the simulation shows that first-principles molecular dynamics—in this case performed using the Vienna Ab-Initio Simulation Package (VASP)—can be used to explore this process.

The process is very fast and localized. To capture the behavior of water molecules, the simulation tackled a system of 700 atoms in steps of a half femtosecond each. Equivalent to a half-quadrillionth of a second, a half femtosecond is to a second as a second is to 63 million years. To get enough useful information, the simulation proceeded through more than 20,000 time steps to get to 10 picoseconds, or 10 trillionths of a second of data, making it one of the largest such simulations undertaken to date.

The simulation put the water in contact with titanium dioxide for two primary reasons. First, the catalyst spurred the process enough for a 10-picosecond simulation to yield sufficient results. Second, technologies such as fuel cells typically use catalysts, albeit more expensive catalysts such as platinum, and the group aims to improve our understanding of the processes and thereby improve the technologies.

"We wouldn’t be able to study the Grothuss mechanism in these simulations without the titanium dioxide," Kent explained. "The water is swimming around at room temperature, and every so often it will adsorb on the surface of the titanium dioxide briefly and a hydrogen will break off. And it will do a little interchange with other water molecules in the area, so they can swap protons. Without the titanium dioxide, the process is too infrequent for these simulations."

Titanium dioxide, also known as titania, is far less expensive than platinum. Used in a wide range of industrial applications, this common catalyst also makes white paint white, protects skin from ultraviolet radiation, and activates oxygen sensors.

The VASP application used to simulate the process solves the Schrodinger equation, which is, among other things, able to describe the chemical bonding between individual atoms. While VASP is a mature code, Kent has optimized it for use with Cray supercomputers such as Jaguar, improving the robustness of the application and its ability to scale to ever more processors.

Kent said the team sticks with the VASP code because it has proven very robust and effective over the years.

"We haven’t changed the fundamental method used in the code at all," he explained. "In fact, we don’t want to, because one of the reasons that people use it so much is there’s a sense of trust and experience." Eventually, he said, more efficient linear scaling applications are likely to take over, but these are not yet ready for use in large science runs.

The computer specialists are able to compare their results with results from a team of experimentalists led by Dave Wesolowski of ORNL. Wesolowski’s team evaluated the same system of water and titania molecules using neutron-scattering techniques at the Intense Pulsed Neutron Source at Argonne National Laboratory. It has looked at other oxide materials as well.

The collaboration illustrates the benefits of experiment and computer simulation working together. On the one hand, computational scientists need to know that their work matches systems in the real world, and Wesolowski’s neutron scattering provided validation to Sofo’s team. On the other hand, even the most advanced experimental techniques are limited in the amount of information they can provide, especially when looking at the scale of atoms and molecules.

"The concept that we’re proving is that we can study this system with sufficient accuracy to get good agreement with experiment, where we’ve got data," Kent explained. "One of the cool things about this work is that the neutron scattering gives us a fingerprint for the dynamics of the water. And we can, in our simulations, go off and compute that fingerprint as well.

"But experimentalists can’t see the individual atoms move around. They only get this fingerprint, which is an average of what’s going on. Now maybe there will be some more sophisticated technique in the future. But the idea here is to have this close collaboration so that if we can verify that we’ve got the same dynamics that they’re seeing, we can then go in and see the whys."

In the case of water and titania, neutron scattering is able to measure the vibrational modes of the water—in other words, how the hydrogen and oxygen atoms in a water molecule wiggle. The simulations, however, show in detail not only how individual molecules wiggle, for example the bending and stretching of the oxygen-hydrogen bonds, but also the collective response of all the waters in the system. This enables researchers to understand the necessary conditions for different processes such as local changes in the acidity of the water.

Because the simulations were able to match the neutron scattering results for vibrational modes at varying temperatures and densities, the researchers have reason to believe that other aspects of the system may be accurate as well, aspects that cannot be measured by neutron scattering.

"So we’ve got this cross reference," Kent noted, "and as closely as we’re looking it seems to be accurate. The fact that the modes line up and are shifted appropriately because of the titania, it’s confidence building. The cross reference is by no means everything, but it gives a hint that the dynamics in the simulation are pretty good and that we’re justified in looking in more detail."

Kent stressed that these simulations are unlikely to lead to great technological innovation in the short term, but they are a necessary step along the way.

"For example, proton transport is critical for fuel cells," he noted. "This is clearly a major motivation for learning what we can about proton transport in water—on a surface that is well characterized and clean and where the neutron-scattering people can do measurements. Looking at a fuel cell membrane is much more complicated, so this is a prerequisite."