Applications of the Methods of
Computational Chemistry And Material Science
In this lecture, we will applying the methodologies to obtain some
important properties and discuss some example systems. We will
present properties such as:
- Optimized molecular geometries
- Electron population analysis
- Electronic multipole moments
- Spectroscopy, including IR and vibrational spectra
- Harmonic force constants
- Energy barriers of reactions
- Mapping sections of the Born-Oppenheimer potential surface
- Chemical reactivity and Reaction Energetics
- Ionization potentials
- Solvation
Structures
Molecular structures are obtained from an energy calculation, either
classical or quantum or a combination of the two methods. The process
is generally considered a refinement from an experimental structure or
a 'guess' structure. This initial structure is necessary as
you recall from previous lectures.
A refinement process, based on increasing accuracy and complexity of the
calculatio, could be outlined as follows:
- Molecular Mechanics
- Molecular Dynamics
- Semi-Empirical
- Quantum Mechanics
So, one question leaps to mind, How do you know when you are
done? Several answers exist, for most cases some measure of
Gradient Accuracy is required. This may be just looking
at differences of a scalar value (such as energy).
In any refinement process, both the cost and accuracy of the energy
calculations and some geometry optimization must be considered. For
the energy calculations consider the following:
- Select different starting configurations to ensure that you
are at (or near enough) to the global minimum.
- Only energy differences have real value
- Energy functions across the different methodologies have vastly
different (arbitrary) forms.
Classical modeling can:
- be a good for guess structures, on the average, structures are
good to about 5-10%.
- be very dependent on parameters for the energy function.
- very often be the only mechanism for large molecules.
Semi-Empirical approaches can:
- be a good initial guess for ab-initio and Density Functional
calculations.
- have less variability in the parameters.
- yield better structures due to the inclusion of quantum
mechanics.
In doing complex calculations, observe the behavior of families of
molecules. The behavior can indicate parameterization effects.
That is if the parameterization of a classical force field or the
semi-empirical parameter sets have some problems. In addition,
energy differences can indicate movement along the energy surface
with changes in structures and other properties.
- Local Minima Problem present exists all methodologies
- Require both energy and gradient testing
- For local minima problems, look at dynamics, simulated
annealing or a conformational search.
So, how do you know when you are finished?
- Comparison to experimental values,
- Tolerance of experimental values.
- Tolerance of computed values.
- Run out of computer time
Overview of the scheme
- Get initial structure
- Refine
- Molecular Mechanics
- Semi-Empirical
- Quantum Mechanics
- Check for behavior of other molecules
- Continue until tolerances are met
Spectroscopy
Some points to consider
- How it is useful
- How is it calculated
- Ionization Potentials
- Koopman's Theorem
- DFT and Transition State Theory
- Electronic Spectra
- Infrared
- Visible
- Ultra Violet
Reaction Energetics
Reaction energetics combine computing the energy and following
a particular path (the reaction path) along the Born-Oppenheimer
energy surface.
Computational Programs for
Computational Chemistry and Material Science
- ab initio - Third party codes
- GAUSSIAN
- GAMESS
- HONDO7
- SPARTAN
- CADPAC
- DFT codes
- DMOL - BIOSYM Technologies Inc.
- DGAUSS - MSI Research
- Semi-Empirical codes
- Classical Modelling - codes
- DISCOVER - BIOSYM Technologies Inc.
- SYBYL - Tripos Associates
- CLAMPS
Author: Ken Flurchick