Page 2, 3/22/12 (PRW) There are many other practical applications of quantum chemistry. Our graduates will be using the results of such calculations, whether or not they perform them. Heat of combustion is a simple, valuable, and instructive problem for introducing theory and use. MODULE CONTENT In four sessions of 1 hr 15 min, the following topics were covered: Review of molecular structure using visualization software; Black-box introduction to quantum-chemistry calculations; Theoretical background on quantum chemistry; Properties calculable by computational chemistry and molecular simulations.
The design course at UMass is taught in an amphitheater-shaped computer classroom containing 26 student computers with 52 seats, an instructor computer, computer projection, a sound system, and a second screen for transparency projection. Classroom monitoring software (NetOp) also allows display of and interaction with students computers. Lecture notes were presented on the board and with PowerPoint slides. Students used Gaussview and Gaussian for Windows software (Gaussian Inc.) for computational quantum chemistry and MathCAD or Excel to calculate ideal-gas thermochemistry. Review of molecular structure. Students recall of molecular nomenclature and three-dimensional structure was first refreshed. These topics are covered in organic chemistry courses but are seldom reinforced sufficiently in the curriculum. Students may become hazy about structures like aldehydes, carboxylic acids, esters, proteins, and sugars. Modern molecule-drawing codes such as Gaussview (Gaussian Inc.), Molecular Studio (Accelrys Inc.), Spartan (Wavefunction Inc.), Hyperchem (Hypercube Inc.), and CACHE (Fujitsu) have proved very helpful. In these codes, 3-D-rotatable drawings can be created by dragging and dropping base molecules and fragments. We began with methane, created by choosing a tetravalent (sp3) carbon from the menu and clicking on the drawing; hydrogens are added automatically. Ethane was created by next clicking on one of the hydrogens, thus replacing it by a second tetrahedral carbon. Propane and higher alkanes were generated as easily. It is useful to rotate molecules with the mouse, appreciating the spatial relationships among the atoms. By measuring the bond lengths, bond angles, and dihedral angles that can be used to quantify structural positions, students develop some intuition about the size scales and geometries. Ethyl alcohol is created by choosing divalent (sp3) oxygen from a menu and clicking the carbon, substituting an OH for a CH3. Amines, rings, pi-bonded atoms, and amino acids are added similarly. As an in-class example, the class created glucose and then added a fructose to make sucrose. With this visual aid, students are easily encouraged to play with and thus self-review molecular structure. Black-box quantum chemistry. Performing quantum-chemistry calculations with menu-driven software is an effective way to introduce some of the jargon before explaining it. Given a simple, well-characterized molecule like ethane, students could run in-class calculations. Any software companys code may require some specialized commands, but they all require choices of task, quantum-chemistry method, and basis set. We performed Geom+Freq (geometry optimization with frequencies), a Hartree-Fock level of theory, and a 3-21G basis set. Methods and basis sets were explained later. Simple geometry optimization is a good task to start with. Ethane was used to give a speedy first result and to illustrate vibrations and internal rotations. Students easily accepted that the optimum
Page 3, 3/22/12 (PRW) geometry would have minimum energy (Fig. 1). Geometry is adjusted by adjusting atomic (i.e., nuclear) positions typically by Newton-Raphson techniques, which they have seen before. By choosing optimization from a menu, the other task options can be noted briefly. For example, energy can be calculated for an arbitrary geometry. Harmonic frequencies can be computed, given an optimized geometry. A transition state (and thus a rate constant) may be calculated by minimizing energy for all coordinates except the reaction coordinate, for which energy is maximized. Next, theoretical method and basis set were selected from menus containing an alphabet soup of choices. Hartree-Fock is the most basic ab initio method, fine for this first calculation. At this stage, it was sufficient to explain that other methods like B3LYP, MP2, CCSD, and so on make up a hierarchy of higher levels, more accuracy, and more expense. A basis set was simply explained as a set of parameters and functions for describing electron orbitals. Just as with levels of theory, 3-21G can be escalated to complex, more accurate, and more expensive basis sets like 6-31G, 6-31G(d), 6311G(3df,3pd), cc-pVQZ, and so on. It was intentional not to give more details than a sense of hierarchy at this point. These calculations run very quickly. Examination of the resulting geometry showed that bond lengths and angles changed slightly, and dihedrals adjusted so that methyl hydrogens were offset. These quantities can be compared with spectroscopic measurements for small molecules. With more time, the class can compare results of different methods and basis sets. Instead, we looked at NISTs online Computational Chemistry Comparison Benchmark Database [3], which usefully provides extensive comparisons for many simple molecules. Animation of the resulting frequencies was especially informative. Animation of the individual frequencies showed the real physical motions that occur, such as stretches and bends. The lowest frequency shows what would be methyl internal rotation along the C-C axis if the calculation were not at zero K. Animation showed the methyls moving harmonically against each other, hindered from being able to rotate. Theoretical background on quantum chemistry. Using quantum-chemistry calculation has become quite easy, although using it wisely requires some understanding of its principles. Ken Cox of Rice University once commented about molecular-modeling software, We used to worry about making it foolproof and user-friendly. Now it almost always converges, and anyone can use it. The challenge now is to make it fool-friendly and user-proof! This is important here because everyone is subject to naive mistakes when new to a subject. A one-hour lecture briefly described three key features of theory required to use ab initio calculations effectively: Calculating energy through the Schrdinger equation Basis sets and how to choose them Levels of theory and how to choose them
Briefly, the simplest form of the Schrdinger equation for energy: EY = H Y (1)
has a quantum Hamiltonian operator H acting on the function (the system wavefunction, an eigenfunction of the equation). Analogously to the classical Hamiltonian, H divides energy into kinetic-energy and potential-energy components. The full form is simplified by the BornOppenheimer approximation, which treats the positions of the nuclei as moving very slowly
Page 4, 3/22/12 (PRW) (relatively unmoving) compared to the fast motion of the electrons, and a time-independent solution is found. a COURSE EVALUATION Student feedback was generally very positive. The only negative comments came from two students who had covered similar material in an advanced physical chemistry course. Understandably, they found much of the material redundant. The main source of error in the students homeworks was calculation of frequencies and thermochemistry from a non-optimized geometry. Harmonic frequencies are based on a parabolic representation 2E/r2 at the bottom of the energy well (Fig. 1) through the force constant k in the Hookian spring equation. If the energy is not at a minimum, then 2E/r2 does not represent a proper frequency. REFERENCES 1. Westmoreland, P.R., Chemistry and life sciences in a new vision of chemical engineering, Chemical Engineering Education, 35(4), 248-255 (2001). 2. P. R. Westmoreland, P.A. Kollman, A.M. Chaka, P.T. Cummings, K. Morokuma, M. Neurock, E.B. Stechel, and P. Vashishta, Applying Molecular and Materials Modeling: An International Comparative Study, Kluwer Academic, New York (2002); also available as a PDF file at http://itri.loyola.edu/molmodel/ . 3. National Institute of Standards and Technology, Computational Chemistry Comparison Benchmark Database, Version 1, http://srdata.nist.gov/cccbdb . 4.
5. 6. 7. 8.
Reid, R. C., Prausnitz, J. M., Poling, B. E., and OConnell, J. The Properties of Gases and Liquids, Fifth Edition, McGraw-Hill, 1987. Benson, S. W. Thermochemical Kinetics, 2nd Ed. Wiley-Interscience, 1976. Fredenslund, A., R. L. Jones, and J. M. Prausnitz (1975). Group-Contribution Estimation of Activity Coefficients in Nonideal Liquid Mixtures. AIChE J., 21, 1086-1099 (1975). Fredenslund, A., J. Gmehling, and P. Rasmussen. Vapor-Liquid Equilibrium Using UNIFAC, Elsevier Scientific Publishing Company, Amsterdam (1977).
Transition
state
Energy E
from
H =
Minimum w.r.t. all but
reaction coordinate
Figure 1. Representation of geometry optimizations to the ground state and the transition state.
Biosketch Phil Westmoreland is Professor of Chemical Engineering at the University of Massachusetts Amherst. His research interests include gas kinetics by molecularbeam mass spectrometry, experimental polymer decomposition and flammability kinetics, applications of computational quantum chemistry and reactive dynamics, and reaction theory to these problems. He is a graduate of N.C. State (BS 73), LSU (MS 75) and MIT (PhD 86) and won ASEEs Corcoran Award in 2002.