Basic Research Needs for Geosciences - Energetics Meetings and ...

GRAND CHALLENGE: COMPUTATIONAL THERMODYNAMICS OF COMPLEX FLUIDS AND SOLIDS(UO 2 ) 2 (μ-OH) 2 (OH 2 ) 6 2+ ; see Figure 22), which have been shown to be the dominant Al(III) andU(VI) species for intermediate pH and concentration conditions (Richens 1997; Baes andMesmer 1986; Soderholm et al. 2005; Clark et al. 1995). Their stability and chemistry influencetransport, the kinetics of reactions in the solution phase, and interactions with solid phases, suchas adsorption and other reactions in the interface region (Figure 23). The measurement of thestructural and reactive properties of these species is difficult because of their amorphous andfluxional nature (Soderholm et al. 2005; Clark et al. 1995).Fortunately, in recent years there has been a dramatic advance in both the theory available tocalculate the properties of complex many-body materials and in the performance ofcomputational platforms available for simulation. This was made possible by the DOE-sponsoredhigh performance light sources and new neutron sources (e.g., X-ray, EXAFS, HEXspectroscopy, neutron scattering) and by the development of high resolution measurements(Brown 2001; Soderholm et al. 2005). The challenge is to continue the remarkable developmentof these technologies which will lead to a new level of understanding and control of processes asdiverse as self-organization in colloids, the catalytic biochemistry of signal transmission ineukaryote cells, and the prediction of the chemistry that may occur under the extremetemperature and pressure conditions encountered in advanced nuclear power systems.The complexity of the processes of interest (e.g., accurate treatment of the highly correlated andpoorly screened f and d electrons in actinide and transition metal elements, complex bondbreaking and formation, changes in bond valance and polarization, etc.) requires reliablepredictions with simulation methods based on first-principles evaluation of forces (thecalculation of forces directly from the electronic Schrödinger equation) (Marx and Hutter 2000;Valiev et al. 2002). The problem is made more difficult by the amorphous structure of thespecies of interest (e.g., the solvated uranyl dimer illustrated in Figure 22; the absorbed speciesin Figure 23), thus requiring many atoms to provide a complete description of chemicalprocesses. Together these problems lead to a grand-challenge-level computational problem.RESEARCH APPROACHESDynamical methods at the atomic time scale and upscalingCurrently the most efficient approximations to the solution of the electronic Schrödingerequation are based on the Density Functional Theory (DFT) approximation (Marx and Hutter2000; Valiev et al. 2002). While simulations with this approximation are very efficient and theresults adequate to provide reliable interpretations for many systems, there are limitations to themuch wider applications of these simulations to more realistic models of technology problems.The most important of these are:1. The DFT level of approximation to the electronic Schrödinger equation is not sufficientlyaccurate to treat many materials problems (energies and bonding in transition and actinideelements and long range (van der Waals) forces).2. The scaling of existing solution methods is not sufficient to exploit the performance of thenext generation parallel computers (tens to hundreds of thousands of processors).3. Even with the improvements in scaling of these methods, the particle and time scales ofmany interesting processes are orders of magnitude larger than can be directly simulated.70 Basic Research Needs for Geosciences: Facilitating 21 st Century Energy Systems