An Introduction to Molecular Orbital Theory Levels of Calculation Classical (Molecular) Mechanics (previous 2 lectures) • quick, simple; accuracy depends on parameterization; no consideration of orbital interaction; not MO theory) Molecular Orbital Theory (Quantum Mechanics) • • Ab initio molecular orbital methods...much more demanding computationally, generally more accurate. Semi-empirical molecular orbital methods ...computationally less demanding than ab initio, possible on a pc for moderate sized molecules, but generally less accurate than ab initio, especially for energies. Relative Computation “Cost” Molecular mechanics...cpu time scales as square of the number of atoms... Calculations can be performed on a compound of ~MW 300 in a minute on a Pentium computer, or in a few seconds on a high performance computer. This means that larger molecules (even peptides) can be modeled readily by MM methods. Relative Computation “Cost” Semi-empirical and ab initio molecular orbital methods...cpu time scales as the cube (or fourth power) of the number of orbitals (called basis functions) in the basis set. Semi-empirical calculations on ~MW 300 compound take a few minutes on a Pentium pc, or several seconds on a high performance computer. Ab initio calculations (to be discussed later) of such molecules can take hours. Semi-Empirical Molecular Orbital Theory Uses simplifications of the Schrödinger equation E = H to estimate the energy of a system (molecule) as a function of the geometry and electron distribution. The simplifications require empirically derived (not theoretical) parameters (“fudge factors”) to bring calculated values in agreement with observed values, hence the term semi-empirical. Properties Calculated by Molecular Orbital Theory Geometry (bond lengths, angles, dihedrals) Energy (enthalpy of formation, free energy) Vibrational frequencies, UV-Vis spectra NMR chemical shifts IP, Electron affinity (Koopman’s theorem) Atomic charge distribution (...but charge is poorly defined) Electrostatic potential (interaction w/ point +) Dipole moment. History of Semi-Empirical Molecular Orbital Theory 1930’s 1952 Hückel Dewar 1960’s Hoffmann 1965 Pople 1967 Pople treated systems only PMO; first semiquantitative application Extended Huckel; included bonds CNDO; first useful MO program INDO History... 1975 1977 1985 Dewar Dewar Dewar 1989 1970’s Stewart Zerner MINDO/3; was widely used MNDO AM1; added vdW attraction & H-bonding PM3; larger training set ZINDO; includes transition metals, parameterized for calculating UV-Vis spectra Basis of Molecular Orbital Theory Schrödinger equation: E = H (can be solved exactly ONLY for the Hydrogen atom, but nothing larger!!) P.A.M. Dirac, 1929: “The underlying physical laws necessary for the mathematical theory of a large part of physics and the whole of chemistry are thus completely known.” Basis of M.O. Theory... Three Simplifying assumptions are employed to ‘solve’ the Schrödinger equation approximately: • Born-Oppenheimer approximation allows separate • • treatment of nuclei and electrons Hartree-Fock independent electron approximation allows each electron to be considered as being affected by the sum (field) of all other electrons. LCAO Approximation Variational Principle Born-Oppenheimer Approx. States that electron motion is independent of nuclear motion, thus the energies of the two are uncoupled and can be calculated separately. Derives from the large difference in the mass of nuclei and electrons, and the assumption that the motion of nuclei can be ignored because they move very slowly compared to electrons Htot a (Tn) + Te + Vne + Vn + Ve Kinetic energy Potential energy (Tn is omitted; this ignores relativistic effects, yielding the electronic Schrödinger equation.) Hartree-Fock Approximation Assumes that each electron experiences all the others only as a whole (field of charge) rather than individual electron-electron interactions. Introduces a Fock operator F: F which is the sum of the kinetic energy of an electron, a potential that one electron would experience for a fixed nucleus, and an average of the effects of the other electrons. LCAO Approximation Electron positions in molecular orbitals can be approximated by a Linear Combination of Atomic Orbitals. This reduces the problem of finding the best functional form for the molecular orbitals to the much simpler one of optimizing a set of coefficients (cn) in a linear equation: = c1 f1 +c2 f2 +c3 f3 +c4 f4 + … where is the molecular orbital wavefunction and fn represent atomic orbital wavefunctions. Variational Principle The energy calculated from any approximation of the wavefunction will be higher than the true energy. The better the wavefunction, the lower the energy (the more closely it approximates reality). Changes are made systematically to minimize the calculated energy. At the energy minimum (which approximates the true energy of the system), dEcalc-real ~ 0. Basis sets A basis set is a set of mathematical equations used to represent the shapes of spaces (orbitals) occupied by the electrons and their energies. Basis sets in common use have a simple mathematical form for representing the radial distribution of electron density. Most commonly used are Gaussian type basis sets, which approximate the better, but more complicated Slater-Type orbitals (STO). Slater-type orbitals (STO) Slater-type orbitals describe the electron distribution quite well, but they are not simple enough to manipulate mathematically. Several Gaussian-type orbitals can be added together to approximate the STO. Here 4 GTO’s mimic 1 STO fairly well. Basis Sets STO-3G (Slater-type orbitals approximated by 3 Gaussian functions)… a minimal basis set, commonly used in Semi-Empirical MO calculations. (L-click here) Hartree-Fock Self-Consistent Field (SCF) Method... Computational methodology: • guess the wavefunction (LCAO orbital coefficients) • • • of all occupied orbitals compute the potential (repulsion) each electron would experience from all other electrons (taken as a group in the H-F approximation) solve for Fock operators to generate a new, improved wavefunction (orbital coefficients) repeat above two steps until the new wavefunction is not much improved; at this point the field is called self-consistent. (SCF theory) Semi-empirical MO Calculations: Further Simplifications Neglect core (1s) electrons; replace integral for Hcore by an empirical or calculated parameter Neglect various other interactions between electrons on adjacent atoms: CNDO, INDO, MINDO/3, MNDO, etc. Add parameters so as to make the simplified calculation give results in agreement with observables (spectra or molecular properties). Steps in Performing a Semiempirical MO Calculation Construct a model or input structure from MM calculation, X-ray file, or other source (database) optimize structure using MM method to obtain a good starting geometry select MO method (usually AM1 or PM3) specify charge and spin multiplicity (s = n + 1), where n = # unpaired electrons, usually 0, so s usually is 1. select single point or geometry optimization set termination condition (time, cycles, gradient) select keywords (from list of >100) if desired. Comparison of Results Mean errors relative to experimental measurements MINDO/3 MNDO AM1 PM3 Hf, kcal/mol 11.7 6.6 5.9 -- IP, eV -- 0.69 0.52 0.58 , Debyes -- 0.33 0.24 0.28 r, Angstroms -- 0.054 0.050 0.036 adegrees -- 4.3 3.3 3.9 More results... Enthalpy of Formation, kcal/mol ethane propane cyclopropane cyclopentane cyclohexane MM3 -19.66 -25.32 12.95 -18.87 -29.95 PM3 -18.14 -23.62 16.27 -23.89 -31.03 Exp’t --24.8 12.7 -18.4 -29.5 Some Applications... Calculation of reaction pathways (mechanisms) Determination of reaction intermediates and transition structures Visualization of orbital interactions (formation of new bonds, breaking bonds as a reaction proceeds) Shapes of molecules including their charge distribution (electron density) …more Applications QSAR (Quantitative Structure-Activity Relationships) CoMFA (Comparative Molecular Field Analysis) Remote interactions (those beyond normal covalent bonding distance) Docking (interaction of molecules, such as pharmaceuticals with biomolecules) NMR chemical shift prediction. As we will see, ab initio and DFT calculations generally give better results than Semi-empirical MO calculations