07.04.2013 Views

Essentials of Computational Chemistry

Essentials of Computational Chemistry

Essentials of Computational Chemistry

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

13.4 EMPIRICAL VALENCE BOND METHODS 479<br />

For the charged VB wave function, we take<br />

−αOH(r<br />

H1 = DOH 1 − e ObH∗ −rOH,eq) 2 <br />

+<br />

+ <br />

H∈b<br />

+ <br />

1<br />

2 kHOH(θ HO b H − θHOH,eq) 2<br />

qiqj<br />

εrij<br />

i∈a j∈b<br />

+ <br />

i∈a j∈b<br />

4εij<br />

H=H ∗<br />

σij 12 rij<br />

1<br />

2 kOH(rOH − rOH,eq) 2<br />

<br />

6<br />

σij<br />

− + Ɣ2<br />

rij<br />

(13.11)<br />

where, with the exception <strong>of</strong> the final term, differences between Eqs. (13.10) and (13.11)<br />

simply reflect the differences in O–H ∗ connectivity between Eqs. (13.8) and (13.9). The final<br />

term, Ɣ2, is a parameter that is adjusted to make the relative energies <strong>of</strong> the two VB wave<br />

functions correct at their respective minima. In the current example Ɣ2 may be determined<br />

as the energy <strong>of</strong> proton transfer from one water molecule to another in the gas phase (we are<br />

still considering only a two-molecule system), which energy is available from mass spectral<br />

measurements.<br />

If we were to partially optimize the geometries <strong>of</strong> the systems having the two possible<br />

connectivities, holding fixed only the difference in the two bond lengths from O a and O b<br />

each to H∗ , we would obtain the set <strong>of</strong> energy curves for the two VB Hamiltonians shown<br />

in Figure 13.7. A crude definition <strong>of</strong> the energy for the reaction coordinate would then be<br />

simply to take the minimum <strong>of</strong> H1 or H2 as one proceeds from left to right in the proton<br />

transfer process. Mathematically, that process is equivalent to taking as our energy the lowest<br />

eigenvalue <strong>of</strong> the 2 × 2matrix <br />

H11 H12<br />

H =<br />

H21 H22<br />

(13.12)<br />

where H11 is taken to be H1, H22 is taken to be H2, andH12 and H21 are taken to be zero.<br />

The matrix then being diagonal, the lowest eigenvalue is simply the lower <strong>of</strong> H1 or H2.<br />

However, Eq. (13.12) is not simply an odd exercise in matrix algebra. Instead, it suggests a<br />

more chemical approach to obtaining the energy <strong>of</strong> the reacting system. Along the way from<br />

the VB structure <strong>of</strong> Eq. (13.8) to that <strong>of</strong> Eq. (13.9), the system obviously passes through<br />

a region where it is best described as a mixture <strong>of</strong> these two extreme resonances. The<br />

mathematical way in which this mixing can be accomplished is to allow the <strong>of</strong>f-diagonal<br />

matrix elements in Eq. (13.12) to be non-zero. The most widely used approximation for<br />

these <strong>of</strong>f-diagonal elements in a case like the one discussed thus far is<br />

H12 = H21 = Ae −B(rOO−rOO,‘eq’)<br />

(13.13)<br />

where A, B, andrOO,‘eq’ are parameters to be optimized against experimental data. Thus,<br />

the coupling is designed to decrease exponentially as the two heavy atoms separate from<br />

some maximally coupled distance (for more complicated approaches to model H12 see, for<br />

example, Chang, Minichino, and Miller 1992 and Kim et al. 2000). With a non-zero coupling,<br />

diagonalization <strong>of</strong> H in Eq. (13.12) will give a curve for the lowest eigenvalue, H0 shown<br />

in Figure 13.7, that smoothly connects the two VB extrema. In addition, the eigenvector

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

Saved successfully!

Ooh no, something went wrong!