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.

13.3 BOUNDARIES THROUGH BONDS 473<br />

Of course, Eq. (13.6) admits to further generalization. Rather than dividing a system into<br />

large and small models, there may be instances where a division into large, medium, and<br />

small models may be advantageous, with increasingly smaller regions treated with increasingly<br />

higher levels <strong>of</strong> theory. Svensson et al. (1996) generalized their geometry optimization<br />

scheme to this more general case, demonstrating the method for MO/MO/MM combinations,<br />

and refer to it as ONIOM, where the acronym, representing ‘our own n-layered integrated<br />

molecular orbital molecular mechanics’ scheme, is meant to emphasize the typical inwardto-outward,<br />

near-spherical layering <strong>of</strong> models that is typically chosen and is reminiscent<br />

<strong>of</strong> the almost eponymous lachrymatory bulb. To provide yet another layer <strong>of</strong> modeling<br />

when condensed-phase effects are <strong>of</strong> interest, the combination <strong>of</strong> ONIOM with the PCM<br />

continuum solvation model has been described (Vreven et al. 2001) as has a model for<br />

permitting explicit solvent molecules to morph from MM to QM while passing through a<br />

buffer region surrounding the QM subsystem (Kerdcharoen and Morokuma 2002).<br />

13.3.2 Link Atoms<br />

Situations arise where the influence <strong>of</strong> the MM region on the QM region to which it is<br />

bonded cannot be regarded simply as steric. In a large protein, for instance, polar and<br />

possibly charged residues in an MM region inevitably will polarize a QM region in the same<br />

protein. The only way to eliminate such QM/MM coupling is to include the entire protein<br />

in the QM region, and such an approach is extremely impractical for anything other than a<br />

possible single-point calculation at a fairly low level <strong>of</strong> electronic structure theory.<br />

Of course, the strong coupling invoked here between the two regions is in no manner<br />

different than that dealt with in Section 13.2.2. What is different is that now there are<br />

interaction energy terms between the QM and MM regions that are not non-bonded terms,<br />

these new terms being associated with the bonds cut by the QM/MM boundary. In practice,<br />

coupled QM/MM calculations involving link atoms tend to adopt the following protocols<br />

for computation <strong>of</strong> the various terms.<br />

1. HQM is computed for the QM region capped with hydrogen atoms at every bond cut<br />

by the QM/MM boundary. The Fock operator may be like that defined in Eq. (13.5).<br />

However, since the capping hydrogen atom is not really a part <strong>of</strong> the system, the third<br />

term on the r.h.s. is not evaluated when µ or ν is a basis function on a capping hydrogen;<br />

similarly, no nuclear repulsion between the capping hydrogen nucleus and the MM atoms<br />

is computed.<br />

2. The energies <strong>of</strong> bonds cut by the QM/MM boundary are evaluated using the<br />

standard MM bond-stretching term (i.e., as though the QM atom were an MM<br />

atom). In addition, a very large force constant is applied to the fictitious bond angle<br />

MM–atom–QM–atom–capping–H so that it remains essentially zero (note that this<br />

connectivity choice avoids the difficulty <strong>of</strong> working with bond angles near π radians).<br />

3. Angle bending energies involving two MM atoms and one QM atom are computed using<br />

the standard force-field formulation. Angle bending terms involving one MM atom and

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

Saved successfully!

Ooh no, something went wrong!