Molecular Modeling and Simulation of Surfaces

Stephen Warde

Molecular Simulations Inc.
240/250 The Quorum
Barnwell Road
Cambridge
UK CB5 8RE
http://www.msi.com



http://www.netsci.org/Science/Compchem/feature16.html

Introduction

The use of modeling and simulation to tackle research problems in the physical sciences has taken off during the 1990s. The technology grew from the combination of computer graphics and computational methods initially used to investigate the interactions of drug candidates with protein structure. As software and hardware improved and more general computational algorithms were developed the application of these techniques diversified - first to address problems in polymer science and then to deal with more general organic and inorganic materials. Today simulation has matured to the point where chemists, physicists, materials scientists and chemical engineers regularly apply it to practical industrial problems like optimizing polymer blend formulations, increasing the efficiency of organic crystallization processes, or improving the performance of catalysts. Many of these researchers are studying a set of problems which affect virtually every industrial product or process - those connected to behavior at surfaces and interfaces.

Modeling Surfaces

Graphical models of surface structure, often constructed by cleaving surfaces from 3D models of crystalline or amorphous solids, can provide valuable insight into surface chemistry. But simulation offers much more than this. Users can employ energy calculations in algorithms which minimize the energy of the surface - thus studying surface relaxation. They can place models of additive or solvent molecules on the surface and use such computations to study their binding or simulate their motion - providing valuable information about surface diffusion, adsorption, and stability.

Energy calculations are based either on molecular or quantum mechanics. Molecular mechanics approximates the energy of a system by summing a series of empirical functions representing components of the total energy like bond stretching, van der Waals forces, or electrostatic interactions. Although only semi-quantitative, it has the advantage of being very fast. Quantum mechanics methods use various degrees of approximation to solve the Schršdinger equation. Extremely accurate, although much more computationally expensive, these methods deal with electronic structure - enabling researchers to characterize chemical reactions.

Advances in Surface Simulation

Advances in surface modeling are based around improvements in both of these fundamental methodologies. More accurate and generic parameterizations, such as the Universal Force Field (1), now allow molecular mechanics simulations to cover a far wider range of compounds. In particular, the development of functional forms and parameters which describe with sufficient accuracy the energetics of metallic elements has allowed these methods to be applied to inorganic and organometallic materials. At the same time novel techniques, such as the CASTEP electronic structure code (2), have combined density functional approximations with advanced numerical algorithms to greatly increase the system sizes which be addressed by quantum mechanics.

Many viable applications of such techniques to commercially relevant systems are now emerging - I will briefly illustrate two: crystallization processes and surface catalysis.

Studying Crystallization Processes
Combining Modeling and Molecular Mechanics

The study of crystallization is a relatively new field for computational chemistry. Simulation is used to investigate growth mechanisms and to study the effects on these mechanisms of additives, impurities, and solvents. Applications include control of particle shape in products like drugs, pigments, and dyes and inhibition of unwanted inorganic phases such as the oxides resulting from corrosion.

Researchers at Schlumberger Research in Cambridge, UK, have recently published work examining the action of phosphonate retarders on hydrating cements (3). This provides an excellent illustration of a modeling application. Cement retarders are used to control the setting of slurries. The rate of setting and the microstructure of the set material can have profound effects on the mechanical behavior of the cement, yet the hydration process is poorly understood and retarders are generally identified by collecting large amounts of empirical data. The Schlumberger researchers set out to get an improved, semi-quantitative, insight into the action of retarders. They examined a process which they had identified as a key rate-controlling step in the hydration - the nucleation and growth of the inorganic crystalline phase Ettringite from its gelatinous precursor.

Phosphonate additive on a surface of ettringite - a key phase in the hydration of cement. This interaction has been successfully modeled at Schlumberger, UK.

Starting with a model of the known crystal structure of ettringite, they computed a theoretical morphology for the crystal. This was displayed on the computer screen and adapted until it matched the experimentally observed particle shape. In this way the key growth surfaces in the crystal were identified. Molecular models of these surfaces were constructed using the building tools of the Cerius2 molecular modeling program (4). Models of known phosphonate inhibitors were then "docked" on these surfaces and their geometrical fit, the match between their charge distributions and that of the surface, and their stability on the surface were studied using molecular mechanics minimization and molecular dynamics. Adsorption of the retarders on a particular surface inhibits further growth on that surface. Thus the effects of a retarder on crystal morphology can be established by modeling its binding to all of the significant growth faces. Results from these simulations matched experimentally observed changes in crystal morphology due to the presence of the retarder. Once such observations have been rationalized using modeling, it is possible to adapt the model of the retarder in order to find molecules which will be more effective or have alternative effects - novel cement retarders were proposed.

Such methods have been successfully applied to other inorganic materials - examples include the work of scientists at Zeneca (5) and Calgon (6) on scale inhibiting chemicals - and to organic crystals - researchers at Bayer have studied the effects of additives on one of their products, a disulfonic acid (7), and numerous examples have been reported by crystal engineering researchers at Brooklyn Polytechnic University (8).

Surface Catalysis
Applying Novel QM Techniques.

Heterogeneous catalysis has been a focus for simulation effort for a relatively long time, with numerous studies of zeolites, metal oxide, and metal catalysts available in the literature. But this is an area in which the possibilities offered by new quantum mechanics techniques are particularly exciting, allowing users to study larger, metal-containing systems at a previously impossible accuracy and to simulate the chemical reactions being catalyzed.

One such method is CASTEP (2) - the Cambridge Sequential Total Energy Package. There is a considerable body of work based around this program, ranging from studies of surface diffusion in semiconductors (9) to predictions of the surface structure of glasses (10). An example is the successful modeling of the chemisorption of carbon monoxide on the surface of a typical palladium catalyst (11). The CASTEP-computed chemisorption energy agrees closely with the experimentally obtained value. Calculation and display of properties like charge density and electrostatic potential allowed researchers to understand the likely mechanism of attack during any reaction with the chemisorbed molecules.

Chemisorption of CO molecules on a typical catalytic surface - as simulated by the CASTEP program. The charge density difference is shown mapped onto the charge density isosurface. This shows valence charge redistribution relative to the superposition of free atoms.

In a recent paper in Science, the same Cambridge research group reported its work on the behavior of an entire zeolite system during a reaction at its internal surface (12). This work explained the effectiveness of the zeolite as a catalyst for the reaction studied. Deeper understanding of such processes could have huge commercial consequences, since zeolite catalysis is responsible for the production of billions of dollars worth of chemicals annually.

Integrating Methods

The examples outlined above represent a minute fraction of current surface modeling work. Other applications include predicting the adhesion of polymer coatings to metals, understanding the nature of chemical vapor deposition processes, and determining surface structure by simulating analytical data such as EXAFS spectra. This last example illustrates a key role of modeling - integrating the entire range of approaches to surface problems. Modeling work can be used to draw together information from theoretical predictions, analytical studies, and structural databases. Combining all of these with graphical visualization and the user's chemical intuition offers a uniquely powerful vehicle for understanding some of today's toughest research challenges.

Modeling the surface chemistry of an organic crystal. This is a surface of the drug Chloramphenicol Palmitate. Visualization of the surface chemistry rationalizes the poor bioavailability of this particular polymorph - the active groups are not accessible to hydrolysis at the relevant surface.

References

(1) Rappé A.K., Casewit C.J., Colwell K.S., Goddard W.A., Skiff W.M., J. Amer. Chem. Soc., 114, 10024, 1992

(2) Payne M.C., Teter M.P., Allan D.C., Arias T.A., Joannopolous J.D., Rev. Mod. Phys., 64, 1045, 1992

(3) Coveney P.V., Humphries W., J. Chem. Soc., Faraday Trans., 92(5), 831, 1996

(4) Cerius2, Software Environment for Chemical Computing, Molecular Simulations Inc.

(5) Black S.N. et al, J. Chem. Soc., Faraday Trans., 87, 3409, 1991

(6) Bendikson B., Molecular Simulations International Science Symposium, Philadelphia, September 1995

(7) Karbach A., Reichel F., Scharschmidt J., Wirges H.P., mc2 Magazine, Vol 1 No 3, Molecular Simulations Inc., 1994

(8) Myerson A.S., Jang S.M., J. Cryst. Growth, 156, 459, 1995

(9) Milman V. et al, Phys. Rev. B50, 2663, 1994

(10) Milman V. et al, mc2 Magazine, Vol 2 No 2, Molecular Simulations Inc., 1995

(11) Hu P., King D.A., Crampin S., Lee M.H., Payne M.C., Chem. Phys. Lett., 230, 501, 1993



NetSci, ISSN 1092-7360, is published by Network Science Corporation. Except where expressly stated, content at this site is copyright (© 1995 - 2010) by Network Science Corporation and is for your personal use only. No redistribution is allowed without written permission from Network Science Corporation. This web site is managed by:

Network Science Corporation
4411 Connecticut Avenue NW, STE 514
Washington, DC 20008
Tel: (828) 817-9811
E-mail: TheEditors@netsci.org
Website Hosted by Total Choice