NERSCPowering Scientific Discovery Since 1974

Modeling the Bonds of Iron and Water

February 28, 2011

Contact: Linda Vu,, +1 510 495 2402

Water lone pair orbitals in the first (left) and second (right) hydration shells of a high spin iron and aluminum. Image: © 2010 American Chemical Society.

Using supercomputers at the Department of Energy's National Energy Research Scientific Computing Center (NERSC), scientists have successfully modeled the atomic interactions between a high spin ferric iron ion and 64 water molecules for the first time.The model is based entirely on the fundamental equation of quantum mechanics, known as the Schrödinger equation; solving this equation is the only way to gain reliable insights into the behavior of atoms.

Researchers say this unprecedentedly accurate model will offer valuable insights into the process of underground mineral formation and extraction, as well as the transport of toxic materials in groundwater. By understanding these key reactions, scientists hope to develop effective strategies for carbon sequestration and other environmental remediation projects.

This research was featured on the cover of the Journal of Physical Chemistry A.¹   The study was carried out by the group of John Weare (Chemistry and Biochemistry Department, University of California San Diego) in collaboration with researchers at Pacific Northwest National Laboratory. The project was led by Stuart Bogatko as part of his doctoral research. The team is currently validating these computer models with experiments at DOE's Advanced Photon Light Source facility and the Stanford Linear Accelerator Center; a paper on this work is forthcoming.

Powerful Computers Increase Scientific Accuracy

"We can use quantum mechanic equations to gain valuable insights about Earth’s chemistry by accurately modeling how metal ions and other materials react in water at different temperatures and pressures. But until this study, accurate on the fly representations of these reactions were difficult to obtain," says Eric Bylaska, of the Pacific Northwest National Laboratory and co-author on the Journal of Physical Chemistry paper.

This figure illustrates the H-bond interaction between second shell waters and the first shell. The orange waters are trigonally H-bonded to the first shell (which is partially obscured to show the interactions more clearly). The light blue water is coordinated by a first shell acceptor bond.

He notes that in geochemistry, conditions like pressure, temperature and pH, change quickly and constantly. The problem is further complicated by the fact that atoms in the molecular system are moving. The accuracy of Bogatko's work comes from solving the Schrödinger equation for all 517 electrons in the iron ion-water solution system. This ensures a parameter free model, meaning that scientists can successfully predict how the molecular system will react in a wide range of conditions on the fly. When implemented in the NWChem chemistry code, this technique accurately predicted the structure of iron ions in a water solution—including the motion of water molecules in the hydration shells, the interactions of electrons of the solvating molecules with the metal ion, the formation of ion pairs and water exchange mechanisms.

Researchers have long known that the properties of a molecular system could be accurately predicted by solving the Schrödinger equation for all consisting atoms, but many choose to rely on approximate methods because they require less computing power. While this strategy works in many cases, Bylaska notes that scientific accuracy breaks down when surrounding variables like temperature or pressure change. Thus, it is not ideal for geochemistry research.

"If we ran our system of 64 water molecules and an iron ion on a desktop computer, it would take a couple of hours to solve one Schrödinger equation. But using supercomputers like NERSC's Franklin or Bassi, we can solve that same problem in less than one second," says Bylaska. "What we accomplished in a few hours on NERSC systems would have taken about 11 years to do on a desktop computer."

Using approximately 272 processors on NERSC’s Cray XT4 "Franklin" system, and about 112 processors on the facility's IBM POWER 5 "Bassi" system, Bogatko generated a simulation that lasted five times longer than any previous model. The system also contained twice as many water molecules as preceding studies. Bylaska likens these simulations to making a movie. "Each Schrödinger equation is equivalent to one frame of a film. We have to solve these equations in less than a second to get enough snapshots to see what is going on, and this would not be possible without supercomputers," he adds. "As technology becomes increasingly sophisticated, we will be able to accurately model even larger systems in better detail."

In addition to Bogatko and Bylaska, John Weare of the University of California, San Diego was also an author on the paper. The NERSC allocation was granted by DOE's Office of Basic Energy Sciences. Bassi was decommissioned in April 2010 and replaced with an IBM iDataPlex system called "Carver."

¹  Stuart A. Bogatko, Eric J. Bylaska, John H. Weare, "First Principles Simulation of the Bonding, Vibrational, and Electronic Properties of the Hydration Shells of the High-Spin Fe3+ Ion in Aqueous Solutions," The Journal of Physical Chemistry A, 2010 114 (5), 2189-220      ≫Back

About NERSC and Berkeley Lab
The National Energy Research Scientific Computing Center (NERSC) is a U.S. Department of Energy Office of Science User Facility that serves as the primary high-performance computing center for scientific research sponsored by the Office of Science. Located at Lawrence Berkeley National Laboratory, the NERSC Center serves more than 7,000 scientists at national laboratories and universities researching a wide range of problems in combustion, climate modeling, fusion energy, materials science, physics, chemistry, computational biology, and other disciplines. Berkeley Lab is a DOE national laboratory located in Berkeley, California. It conducts unclassified scientific research and is managed by the University of California for the U.S. DOE Office of Science. »Learn more about computing sciences at Berkeley Lab.