07.04.2013 Views

Essentials of Computational Chemistry

Essentials of Computational Chemistry

Essentials of Computational Chemistry

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

9.1 PROPERTIES RELATED TO CHARGE DISTRIBUTION 327<br />

is, one can compute the energy in the absence <strong>of</strong> the perturbation, then modify the Hamiltonian<br />

to include the perturbation (e.g., introduce an electric-field term), then compute the<br />

property as<br />

∂<br />

〈|H|〉−〈|H<br />

〈|H|〉 = lim<br />

∂X X→0<br />

(0) |〉<br />

X<br />

(9.34)<br />

where H is again the complete Hamiltonian and H (0) is the perturbation-free Hamiltonian.<br />

This procedure is called the ‘finite-field’ approach. In practice, one must take some care to<br />

ensure that computed values are numerically converged (a balance must be struck between<br />

using a small enough value <strong>of</strong> the perturbation that the limit holds but a large enough value<br />

that the numerator does not suffer from numerical noise).<br />

Note that Eq. (9.34) can be generalized for higher derivatives, but numerical stability now<br />

becomes harder to achieve. Moreover, the procedure can be rather tedious, since in practice<br />

one must carry out a separate computation for each component associated with properties that<br />

are typically tensors. It is computationally much more convenient when analytic expressions<br />

can be found that permit direct calculation <strong>of</strong> these higher-order derivatives in a fashion that<br />

generalizes the procedure by which Eq. (9.33) is derived (not shown here).<br />

As for the utility <strong>of</strong> different levels <strong>of</strong> theory for computing the polarizability and hyperpolarizability,<br />

the lack <strong>of</strong> high-quality gas-phase experimental data available for all but the<br />

smallest <strong>of</strong> molecules makes comparison between theory and experiment rather limited. As<br />

a rough rule <strong>of</strong> thumb, ab initio HF theory seems to do better for these properties than<br />

for dipole moments – at least there does not appear to be any particular systematic error.<br />

Semiempirical levels <strong>of</strong> theory are less reliable. DFT and correlated levels <strong>of</strong> MO theory do<br />

well, but it is not obvious for the latter that the improvement over HF necessarily justifies<br />

the cost, at least for routine purposes.<br />

9.1.6 ESR Hyperfine Coupling Constants<br />

When a molecule carries a net electronic spin, that spin interacts with the (non-zero) spins <strong>of</strong><br />

the individual nuclei. The energy difference between the two possibilities <strong>of</strong> the electronic and<br />

nuclear spins being either aligned or opposed in the z direction can be measured by electron<br />

spin resonance (ESR) spectroscopy and defines the isotropic hyperfine splitting (h.f.s.) or<br />

hyperfine coupling constant. If we were to pursue computation <strong>of</strong> this quantity using the<br />

approach outlined in the last section, we would modify the Hamiltonian to introduce a spin<br />

magnetic dipole at a particular nuclear position. The integral that results when Eq. (9.33) is<br />

used to evaluate the necessary perturbation is known as a Fermi contact integral. Isotropic<br />

h.f.s. values are determined as<br />

aX = (4π/3)〈Sz〉 −1 ggXββXρ(X) (9.35)<br />

where 〈Sz〉 is the expectation value <strong>of</strong> the operator Sz (1/2 for a doublet, 1 for a triplet,<br />

etc.), g is the electronic g factor (typically taken to be 2.0, the approximate value for a free<br />

electron), β is the Bohr magneton, gX and βX are the corresponding values for nucleus X,

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!