CONTRACTOR: PACIFIC NORTHWEST NATIONAL LABORATORY
MS K8-93
Richland, Washington 99352
CONTRACT: DE-AC06-76RLO 1830 (KC-04-03-02)
CATEGORY: Geochemistry
PERSONS IN CHARGE: Andrew R. Felmy
A. Surface Structure and Chemistry of Carbonate Minerals (D.R. Baer [509-376-1609; Fax 509-376-5165; E-mail dr_baer@pnl.gov], J.E. Amonette [509-376-5565, Fax 509-376-2303; E-mail je_amonette@pnl.gov], and J.P. LaFemina [509-375-6895; Fax 509-375-4486; E-mail jp_lafemina@pnl.gov])
Objectives: The purpose of this program is to develop a fundamental, microscopic understanding of the structure and chemistry of carbonate surfaces, including the interactions between adsorbates and mineral surfaces.
Project Description: This project involves an interdisciplinary theoretical and experimental effort designed to gain a fundamental, molecular level understanding of carbonate mineral surface structure and chemistry. Carbonate minerals are particularly important in the global carbon dioxide cycle and in subsurface contaminant migration processes. The availability of large single-crystals allows fundamental measurements to be made on well-defined surfaces. By linking experimental studies of geochemical reactions on single-crystal surfaces with first-principles quantum-mechanical model calculations to describe the surface and interfacial structure and chemistry, a systematic study of the factors controlling the surface chemistry of carbonate minerals can be made. In particular, the effects of substitutional impurities and other point chemical defects on the structure and geochemical reactivity of carbonate mineral surfaces and interfaces can be isolated and quantified. Moreover, this improved microscopic understanding will eventually provide insights into the behavior of these materials in natural systems.
The approach to meeting program goals involves three interdependent areas of effort: development of ab initio models for the structure and chemistry of the calcite cleavage surfaces; vacuum studies of the structure and chemistry of the cleavage surface; and comparison of surfaces in vacuum with those in model geochemical environments.
Results: Recent work has focused on experimental and theoretical studies of the effects of solution chemistry on the dissolution of calcite. The overall results suggest that the types of binding energy differences used to explain dissolution rates in solutions far from equilibrium (Surface Science 373 [1997] p. 275-287) be supported by observation made at equilibrium. Experimental measurements have included observations of pit structure at equilibrium and the effects of carbon dioxide, Mn and Sr on the rates of pit growth while dissolution is occurring. Theoretical work includes the addition of back reactions to our Kinetic Monte Carlo model of dissolution and initial molecular level calculations of solution flow surface topography effects on dissolution. The shape of pits formed in water at equilibrium with calcite provides information about the energies of steps on the surfaces exposed. Previous work has demonstrated that for nearly pure water, steps on the cleavage surface with slightly different geometries dissolve at different rates. (Geochimica et Cosmochimica Acta 60 [1996] 4883) Both terrace ledge kink (TLK) and Kinetic Monte Carlo (KMC) models can reproduce the results by assuming different binding energies which depend upon the detailed geometry of different binding sites.
Although the slowest dissolving site was assumed to have the largest binding energy, steric/entropic effects may have caused these differences in dissolution rates. New observations of pits at equilibrium confirm the different binding energy assumption and provide an indication of two different energies for the two types of surface steps. Since surface energies per unit area are equivalent at equilibrium, steps on the side of an equilibrium pit with the shallowest slope have the highest energy. Under far from equilibrium condition, the steps having higher energy dissolve more slowly than the steps having lower energy and consequently, the slopes of the sides of pits reverse accordingly.
Various divalent cations are well known to influence the dissolution and growth of calcite. Earlier results have shown that Mn significantly slows the dissolution rates of calcite and that the rapidly dissolving steps were the most affected. Additional work demonstrates that this effect depends upon ion size. Although Mn influenced the fast and slow dissolving directions to different degrees, Sr (as might be expected due to its larger size) primarily influences the open sites of the normally fast dissolving direction. For both metals, the fastest dissolving steps were the most influenced, but the smaller ion also decreased the dissolution of the steps with the more closed geometry.
A Kinetic Monte Carlo (KMC) model has been used within a solid-on-solid model of surfaces to determine the reaction or dissolution rates of five different surface sites. (Surface Science 373 [1997] 288-299) A back reaction has been added to the KMC model to allow study of dissolution nearer to equilibrium. This model has been used to fit the rates obtained experimentally for fast and slow directions at different concentrations of bicarbonate in solution. The results provide information about not only the change of the dissolution rates, but also the change of the pit shape.
The AFM examination shows that solution flow strongly affects step dissolution rates and the shape of "deep" pits. We have developed a 2D cellular automata (CA) model to simulate the effects of dissolution, readsorption, diffusion, and fluid flow on calcite surface crystallographic step evolution. A unique feature of the CA model is the ability to treat fluid flow effects on the step kink growth kinetics and morphologies. Our AFM data in the reaction-controlled regime (fast flow) have been reproduced quantitatively. Initial calculations for slower flows (diffusion controlled) have qualitatively reproduced the observed AFM step velocities and morphologies, and have revealed a dependence on fluid flow direction. Efforts are currently focused on eliminating a computational bottleneck caused by large differences in time constants for the fluid versus the solid and on developing methods to determine the back-reaction rate constants from the fluid flow dependencies. Future work will focus on extensions to 3D using the lattice gas automata (LGA) approach, which will permit simulation of calcite pit growth for reactive boundary layers in the flowing fluid.
B. Structure and Reactivity of Fe/Al/Si Oxide and Oxyhydroxide Surfaces and Aqueous Interfaces: Molecular to Thermodynamic Scale (James R. Rustad [509-376-3979; Fax 509-376-3650; E-mail jr_rustad@pnl.gov] and Andrew R. Felmy [E-mail ar_felmy@pnl.gov])
Objectives: The objectives of this program are to (1) develop molecular models of hydroxylated mineral surfaces, (2) use these models to better understand the relationship between surface structure and reactivity, and (3) use this information to improve thermodynamic calculations based on surface complexation models.
Project Description: Ferric oxides have high specific surface areas, high affinities for oxyanions and heavy metals, and under some conditions can actively respond to changes in redox conditions in natural environments. These minerals are therefore important in a variety of low-temperature geochemical processes, particularly those in which adsorption and dissolution couple with fluctuations in redox potential. For many solutes, measurements of sorption density versus aqueous concentrations suggests the presence of a heterogeneous array of surface sites having a range of affinities for the probing solute. Crystallographic differences in oxide site coordination numbers and their geometrical arrangement are a fundamental aspect of this heterogeneity. In this project, the effects of crystallographic heterogeneity on adsorption are evaluated using computational molecular models. These results are then used to produce a more robust thermodynamic description of adsorption at the mineral water interface. Silica surfaces are important in both low-temperature geochemistry as well as in higher-temperature processes involved in rock deformation. Surface hydration of quartz and the behavior of water in grain boundaries can be an important factor in determining rheological properties of clastic sediments.
Results: Molecular statics calculations of proton binding at the hydroxylated faces of goethite are used to guide the development of a thermodynamic model which describes the surface charging properties of goethite in electrolyte solutions. We find that pair formation between surface sites and bulk electrolyte species (Na+,Cl-, NO3-) are required to describe the surface charging of goethite. If these species are included, the final thermodynamic model is shown to be consistent with the surface charging properties of goethite over a range of pH values and ionic strengths. This work is written up in a paper entitled "Molecular statics calculations of proton binding to goethite surfaces: Thermodynamic modeling of the surface charging and protonation of goethite in aqueous solution" which is in review in Geochimica et Cosmochimica Acta.
We applied our methods for potential development from ab initio calculations to the Al(III)-water system. This was done because (1) Aluminol surface sites are an important in adsorption to aluminum oxides and clays, (2) Al(III) can be done with very large basis sets using correlated methods, and (3) Thermodynamic data for Al(III) hydrolysis in solution are more extensive than that for Fe(III). We found that very large basis sets were required for convergence of the quantum mechanical calculations to the experimental values. The force field fit to the ab initio calculations was then used to calculate the heat of solution of Al(III), where we were within 1 percent of the experimental results. The results of this work were reported in a paper entitled "Interaction of Al3+ in water from first principles calculations," Journal of Chemical Physics 106, 9769, 1997.
Molecular dynamics simulations using the transferable/polarizable model by Rustad et al.[J. Chem. Phys. 102, 427, 1995] were applied to study the surface relaxation of the non-hydroxylated, hydroxylated, and solvated surfaces of (hematite). "Hydroxylation and Hydrogen Bonding on the (012) and (001) surfaces of alpha-Fe2O3" is in press in Surface Science (vol. 343). We have also shown that consideration of possible proton arrangements on the hematite (012) surface must consider tautomer-tautomer interactions beyond one unit cell. For example, in agreement with temperature -programmed desorption measurements, we predict about 75 percent dissociation on the (012) surface. However, to find this minimum required searching over tautomeric mixtures of unit cells within a 2 x 2 super-cell. The results of the calculations on the longer wavelength fluctuations in proton distribution are now being written up. Molecular statics calculations on gas-phase and solvated clusters representing aqueous species, and gas-phase and solvated slabs representing surfaces were applied to investigate acid/base reactions involving silica. Our gas-phase approach, which was previously applied to alpha-FeOOH, predicts a surface pKa of 8.5 for the reaction >SiOH -> >SiO- + H+ which is in good agreement with estimates based on potentiometric titration. However, the model gives an unrealistically large pKa for the reaction >SiOH2+ ->SiOH+H+. The model-dependence of this result was checked by using two different types of interaction potentials: the first is based on quantum mechanical calculations on H4SiO4 clusters, whereas the second is an empirical model fitted to the structure and elastic constants of alpha quartz. Because these models gave similar results, we hypothesize that the failure of the gas-phase model is due to intrinsic solvation effects not accounted for by our previously-developed correlations. We test this idea by carrying out energy minimization calculations on gas-phase clusters with one hydration shell, as well as molecular dynamics simulations on fully-solvated H5SiO4+ and a fully-solvated (0001) surface of beta quartz. The solvated systems do indicate that SiOH groups do not protonate in any of our solvated models. A paper entitled "Molecular Dynamics Study of Proton Binding to Silica Surfaces" was submitted to Journal of Colloid and Interface Science.
C. Theoretical Characterization of the Physics and Chemistry of Soil Minerals (A.C. Hess [509-375-2052; Fax 509-375-6631; E-mail ac_hess@pnl.gov] and M.I. McCarthy [509-375-6824; Fax 509-375-6631; E-mail mi_mccarthy@pnl.gov])
Objectives: This program is designed to provide the theoretical basis for understanding a broad range of complex subsurface processes. It compliments many of the experimental efforts presently supported by OBES/Geosciences and will serve as a broad-base link between several experimental and theoretical projects. The goal of this program is to investigate the microscopic properties of minerals and mineral interfaces that directly affect the macroscopic transport of contaminants through the subsurface. The knowledge obtained from in-depth studies on several classes of mineral-contaminant systems are used to construct a general understanding of soil physics. This is, in turn, used to predict the behavior of a broad range of subsurface minerals and contaminants. The knowledge gained from these studies may also be directly transferred to large-scale subsurface transport models and remediation design models.
Project Description: An integrated theoretical approach is used which combines methods from ab-initio quantum mechanics and classical mechanics. This involves the development and use of ab-initio time-independent periodic Gaussian basis density functional theory and the use of Hartree-Fock theories to study the electronic structure and physical properties of bulk solids, clean surfaces and interfaces. Molecular dynamics and molecular mechanics techniques are also employed, in conjunction with the quantum mechanical calculations, to study interfacial dynamics. A principle focus our activity is the continued development of fully self-consistent, all electron, low order scaling periodic density functional theory. The approach we have developed is implemented in the program GAPSS (Gaussian approach to Polymers, Surfaces and Solids) and has been targeted toward a number of massively parallel computer architectures. This state-of-the-art software, combined with the capabilities of MPP computer systems, is allowing us to address the complexity of geochemical systems using predictive theoretical methods. This research program is jointly supported by OBES/Geosciences and OBES/Chemical Sciences.
Selected Highlights: (A.C. Hess, M.I. MCCarthy, J.E. Jaffe, G.K. Schenter, Zijing Lin, Maciej Gutowski, Peter Zapol) A characterization of selected water/mineral interfaces is being undertaken to determine the reactivity of minerals in the saturated and unsaturated zones. The adsorption and chemidissociation behavior of water on metal oxide surfaces and silicates constitute some of the most fundamental processes in geochemistry. The presence of molecular water overlayers and adsorbed chemidissociation products on such surfaces strongly influence the transport properties and chemical transformation of contaminants in the subsurface environment. Geometric and Electronic Structure of the 260 K Ordered Overlayer of H2O on MgO (001): A molecular level understanding of the interaction of water with the external surfaces of metal and transition metal oxides is key to describing a range of phenomena including the dissolution and growth of materials, the uptake and retention of contaminants in the subsurface environment, and the promotion or inhibition of many chemical reactions of commercial importance. Using fundamental experimental techniques from surface science, in conjunction with high level theoretical methods, a picture of the geometric and electronic structure of the water-oxide interface has emerged in recent years. Such studies have begun to unravel some of the complex interrelationships that exist between adsorbate concentration, system temperature, surface structure and composition, and the effect of topological and morphological surface defects on the observed chemistry. Recent experimental results on one system of interest, MgO (001) with H2O, have demonstrated that at temperatures below room temperature (where the motion of water is slowed down) at least three distinct phases of water (Tphase1 < 180K, Tphase2 = 220K and Tphase3 = 260 K) exist as unique ordered overlayers on cleaved and thin film MgO (001) samples. Although the experimental studies have been able to demonstrate the presence of these distinct phases, a molecular level picture of the geometric structure of the surface-adsorbate complex remained elusive. This highlight briefly describes the progress being made by a team of theoretical researchers from the Materials and Chemical Sciences Department and the Environmental Molecular Sciences Laboratory to determine the geometric and electronic structure of the water overlayer in the high temperture (260 K) phase. The approach adopted in this study was to utilize a combination of classical molecular dynamics and solid state quantum mechanical methods to obtain the minimum energy orientation of the water molecules in the presence of the a surface and the electronic structure of the adsorbate-surface complex. Beginning with the experimental coverage of 2/3 of a monolayer and the knowledge of where the water oxygen atoms were located relative to the MgO surface cell, molecular dynamics calculations that rapidly quenched the water molecules in the presence of the surface from thousands of random orientations were performed by G.K. Schentner. The atomic potentials used in these simulations were previously developed by M.I. McCarthy based on solid state quantum mechanical studies of the MgO (001) water interface at a different adsorbate coverage. The orientation of water molecules which yielded the lowest total energy was then used as the initial atomic configuration for subsequent studies using periodic Gaussian basis density functional theory as implemented in the program GAPSS (A.C.Hess and J.E. Jaffe). Those calculations provided an assessment of the electronic structure of the water-surface complex and the fine details of the equilibrium orientations of the water molecules. These latter calculations, however, are of grand challenge proportions and were performed using GAPSS in parallel mode by M. Gutowski on 128 nodes of the newly installed EMSL IBM-SP.
The attached figure depicts the theoretically predicted geometric structure of the ordered overlayer of water molecules that constitutes the 260 K phase. The structure depicted in that figure is currently the most accurate picture of what this water-oxide interface looks like at the molecular level. Based upon more specialized results obtained from the quantum mechanics, information has also been obtained on the chemical nature of the water-surface interaction, water-water interaction in the overlayer and the energetics associated with theadsorption process itself. Studies of this type, although highly fundamental in nature, are the basis for our understanding of interfacial processes.
Adsorption of CO and H2 on the ZnO (1010) Surface: The geometry of carbon monoxide molecules and dissociated hydrogen on the nonpolar (1010) surface of ZnO was determined by quantum-mechanical energy minimization. CO was found to physisorb in a tilted geometry in good agreement with experiment, and the binding energy also agreed with experiment when electron correlation was taken into account. Little change in the surface relaxation of the ZnO in response to the CO was found. In contrast, at monolayer coverage H2 was found to dissociate and strongly chemisorb to the surface, and to cause substantial changes in the ZnO top layer geometry. Our results are consistent with the limited experimental data available and show that of changes in surface geometry in response to reactive adsorbates can be important for the correct prediction of adsorption energetics. ZnO is commercially important in methanol synthesis, where reactive H atoms at the Zn sites play a key role.
Molecular Strucute of MgO (001) + H2O 260 K Phase
Top View
Side View
Side View
(Rotated 90º)
Atomic Relaxation of the BeO (1010) Surface: To continue our general studies of the atomic structure of oxide surfaces an atomic level relaxation of the nonpolar (1010) surface of BeO was calculated by minimizing the surface energy within the framework of the ab initio Hartree-Fock method. A six-layer two-dimensionally periodic slab model was used, permitting a full symmetry-conserving relaxation of the two outer layers. The Be-O surface bonds showed a small rotation angle of about 4 accompanied by a large (about 10%) reduction in surface bond length. Significant contraction of backbonds and a small rotation of second layer bonds were also found. The relaxed BeO (1010) surface is thus predicted to be similar to the ZnO (1010) surface but different from the corresponding surfaces of all other II-VI compounds. Evidence from a bond population analysis suggests that this behavior can be described in terms of partial double bond character in the surface bonds. Since multiple bonding is related to small atomic radii, it follows that the small radius of the oxygen atom is the ultimate cause of the type of surface relaxation we predict.
Bulk and Surface Properties of the Magnetic Insulators MnO and NiO: Properties of the antiferromagnetic oxides MnO and NiO are presently under study due to the important role that they play in the redox chemistry of the subsurface. Our recent studies of the bulk materials using PG-DFT have confirmed the high-spin antiferromagnetic state as the ground state of these materials and have accurately predicted the ordered magnetic moment.
In addition, a number of other electronic structures which lead to other ordered magnetic states have also been investigated. We find that the next lowest energy state corresponds to a ferromagetic state that is predicted to be 0.32 eV per formula unit higher in energy than the antiferrromagnetic state. A similar result is also found using periodic Hartree-Fock theory although our PDFT results predict that the ferromagnetic state is metallic. Our current work focuses on determining what the ground state electronic configuration is for the (001) surfaces. This includes both out theoretical work and experimental collaborations which are designed to better understand and determine the magnetic order parameter at the (001) surface. The second phase of this study will focus on understanding the interaction of molecular adsorbates with the (001) surface.