Peter Knowles Flashcards
Isogyric
describes a chemical reaction in which the number of electron pairs is conserved
isodesmic
describes a chemical reaction in which the numbers of each type of chemical bond are
all equal for reactant and product. This means not only distinguishing C – C bonds from C – H,
but also counting separately single, double and triple bonds.
Why is there a large difference between the HF and MP2 atomisation energies
There is much more electron correlation energy in the molecules than in the atoms, because
in the atoms there are a number of unpaired electrons that do not have opposite-spin correlation.
Consequently, Hartree-Fock strongly underestimates the dissociation energy. MP2 improves on HF method by adding electron correlation effects
cc-pVTZ contains more functions than cc-pVDZ, therefore a larger fraction of the correlation energy
is recovered.
Explain why a DFT electronic structure method is more appropriate for this problem
than (i) Hartree-Fock; (ii) CCSD(T).
Hartree-Fock misses electron correlation effects; CCSD(T) includes them, but would be far too expensive.
Would you expect DFT to predict an activation energy for this reaction that is too high or
too low?
Like Hartree-Fock, DFT does not dissociate covalent bonds correctly, with an energy that
goes much too high. Transition state is part of the way towards broken bonds, and so,
compared to the equilibrium geometry, its DFT energy will be too high, i.e. activation energy
overestimated.
Why is the 6-31G(d) basis set appropriate for this study? You should consider the
desirability or otherwise of adopting specific alternatives, including minimal, 6-31G, and
cc-pVQZ sets.
Minimal vs 6-31G(d): No possibility for orbitals to grow or shrink; DZ set allows mixture of
diffuse and compact basis orbital.
6-31G vs 6-31G(d): No possibility for orbitals to bend; DZP allows polarisation of d by p.
cc-pVQZ: Would normally be used in CI-like calculations to support good representation of
2-electron wavefunctions, but for just the DFT orbitals it makes not much difference.
Why is there no mention of link atom, or other, termination techniques for the interface
between QM and MM regions?
Because there is a natural separation across the non-covalent interactions between solute
and solvent, so no need to cut any covalent bonds.
Variation method
Adjust the shape of the orbitals until the energy is as low as possible. These orbitals will be the best ones
Covalent bond
- As H+ approaches H the potential energy goes up
- At the same time, the Kinetic energy goes down as there is now a larger space for the electron to move. On reducing the kinetic pressure the molecule can then lower its energy further by shrinking
- Overall potential energy reduction
requirements for basis sets to lower energy
shrink, expand and bend
Error in Hartree Fock
Error largest when covalent bonds are stretched or pairs of electrons are created or destroyed. This is because correlation is absent in HF.
Method for treatment of electron correlation
DFT
What computational method is needed for:
- Electrostatic
- Induction
- Dispersion
- Electrostatic (HF or DFT)
- Induction (HF or DFT)
- Dispersion (correlated wavefunction)
HF and DFT do not represent dispersion interaction at all
Mode of Attraction for and distance dependence:
- Electrostatic
- Induction
- Dispersion
- Electrostatic (orientation dependent R-1/R-3)
- Induction (Attractive R-4/R-6)
- Dispersion (Attractive R-6)
Local correlation
- Even HF is quartic scaling because orbitals are delocalised.
- So all N^4 2e intergrals are non zero and must be calculated and manipulated.
- Instead mix the orbitals so they are as local as possible, then discard small interactions between orbitals on distant atoms.
- All methods can be reduced to NlogN
