Neutron Science, Nanoscience, & New Simulations

ORNL theorists simulated 6 to 8 layers of molecules of dodecane (a lubricant) sandwiched between mica layers. The narrowly confined liquid-like dodecane was transformed into a solid-like structure as a result of interfacial forces. This effect may explain the orders-of-magnitude higher viscosities observed in confined-fluid experiments compared with those for bulk fluid. (Schematic redrawn by Judy Neeley.)

Peter Cummings, director-designate of the Nanomaterials Theory Institute of the Department of Energy’s Center for Nanophase Materials Sciences (CNMS), to be located at ORNL, says “Theory, modeling, and simulation will be essential tools for both the interpretation of data produced by neutron scattering and the guidance of experimental measurements of phenomena at the nanoscale.” In collaboration with experimental programs at the CNMS (on which construction is expected to begin in February 2003), this institute’s personnel will develop or improve computer models that describe material features, from the single electron, through atoms and molecules, to functioning nanodevices.

MODELING AND NEUTRON SCIENCE

Cummings, now at Vanderbilt University, gives an example of the importance of coupling simulation with neutron science. In 1991 Cummings (then on the University of Virginia faculty) collaborated with ORNL researchers Hank Cochran, Mike Simonson, and Bob Mesmer (all of the Chemical Sciences Division) to perform molecular simulations of supercritical water on ORNL’s Cray supercomputer. Supercritical water is a non-liquid, non-gaseous state of water produced at high temperature and pressure. The simulations predicted distances between oxygen atoms and between oxygen and hydrogen atoms and indicated the presence of hydrogen bonds—transitory bonds between hydrogen atoms and the negative parts of other water molecules.

In December 1993 the prestigious scientific journal Nature published a paper that claimed that hydrogen bonding is essentially absent in supercritical water, disputing the Oak Ridge results. The paper’s conclusions were based on neutron scattering experiments conducted at the Rutherford-Appleton Laboratory in the United Kingdom. The paper’s authors asserted that all existing water models, including the model used in the ORNL simulations, were flawed. Subsequent publications based on molecular simulation calculations by the ORNL group and by other research groups challenged the results of the U.K. laboratory scientists and their collaborators from the University of Rome. These papers showed that the scattering results reflected a possible error in the correction for the results of inelastic neutron scattering.

In response to these challenges, in 1996 the British-Italian group re-examined its neutron scattering data, improving the correction for the inelastic scattering results. From their re-analysis, group members found that hydrogen bonding is indeed present in supercritical water, as predicted by the ORNL simulations. They also re-analyzed the data they had obtained a decade earlier for ambient water—data that had been the basis for understanding and modeling water structure and used by researchers worldwide.

As a result of the new analysis, the British-Italian group, led by Alan Soper, revised the description of water under ambient conditions (i.e., at room temperature and atmospheric pressure). The structure of water at ambient conditions is extremely important because of water’s role in many biological and chemical processes at ambient conditions.

“This story illustrates an unusual interplay between neutron scattering experiments and molecular simulation calculations,” Cummings says, noting that this observation was published by the researchers in 1998. “It shows that both the interpretation of scattering results and the models used in molecular simulations have been improved by this interaction. Modelers can help determine the validity of experimental measurements and confirm findings on how atoms are arranged in a sample material. That’s why having the CNMS with its strong theory component so closely allied with the Spallation Neutron Source at ORNL should result in higher-quality science.”

MODELING AND NANOSCIENCE

Researchers are creating microelectromechanical systems (MEMS)—motors, sensors, and actuators on the micron scale—that can be combined with integrated circuits in a thin film or on a doped silicon chip. But micromotors, which may include tiny rotors, pumps, and valves, will likely require special lubricants to overcome friction between moving parts sliding by each other. Enter the field of nanotribology, an area of research in which Cummings and other modelers have been involved.

Cummings, Cochran, Shengting Cui of UT, and Clare McCabe, now at the Colorado School of Mines, have simulated the viscosity of lubricant-like molecules, such as dodecane, between two parallel sheets of mica 2 to 3 nanometers (nm) apart. They have predicted the viscosity (a measure of resistance to flow) of the fluid confined between the two surfaces of a surface-force apparatus, representative of parts of a motor sliding by each other.

In their simulation, 3 to 8 layers of dodecane molecules are sandwiched between the mica layers. A key element in understanding the predicted phenomena is that the attraction between molecules in the mica surfaces and the lubricant atoms is greater than the attraction between atoms in the lubricant.

The theorists predicted that the viscosity would be 6 to 7 orders of magnitude (or 1 million to 10 million times) higher than the viscosity of the same liquid in bulk (i.e., not confined to regions of the order of nanometers), in agreement with experiments by Steve Granick, a researcher at the University of Illinois. They also predicted that when the liquid is sheared by the surfaces moving in opposite directions, the liquid becomes less viscous the faster the shearing is—that is, the liquid flows more easily. “That’s because the faster the shearing, the more the clusters of atoms in the lubricant stretch out and align themselves, allowing them to flow over each other more easily,” says Cummings.

“We can model dodecane between mica sheets to a high degree of accuracy,” he continues. “Through simulation we confirmed that dodecane between mica sheets 2.36 nm or less apart becomes solid-like, confirming Jacob Klein’s experiments at the Weissmann Institute in Israel. Steve Granick’s experiments led him to conclude that nanoconfined dodecane is a glass—a state intermediate between liquid and solid—rather than a solid. Neutron scattering would help resolve whether a lubricant between surfaces a few nanometers apart is glass or solid.”

A solid lubricant would not be a lubricant, so what’s the solution? “On the basis of the simulations, we have concluded that branched lubricants should be used instead of linear ones because branched lubricants don’t form solids as easily as linear ones,” Cummings says. In linear organic compounds, the carbon atoms are linked in a straight line. In branched organic compounds, one or more carbon atoms branch off from the carbon-atom backbone to which they are attached.

Another possible course of action suggested by the simulations is to reduce the attraction between a lubricant and the two surfaces sandwiching it. This approach will decrease the confined lubricant’s density, reducing its viscosity. “This is the key concept behind developing new surface-phobic lubricants or lubri-phobic surface materials,” says Cummings, explaining that “phobic” as used here means the tendency to avoid a nearby material. “Modeling should help materials scientists synthesize new materials with the desired properties.”

Beginning of Article

Related Web sites

• ORNL's Chemical Sciences Division
• Nature
• Rutherford Appleton Laboratory
• Spallation Neutron Source

Search Magazine