28.02.2013 Views

Handbook of Solvents - George Wypych - ChemTech - Ventech!

Handbook of Solvents - George Wypych - ChemTech - Ventech!

Handbook of Solvents - George Wypych - ChemTech - Ventech!

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.

472 Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli<br />

( rt)<br />

μψ��<br />

,<br />

∂V<br />

=− +∑Λ ψ ( rt , )<br />

*<br />

ik k<br />

[8.97]<br />

∂ψ<br />

i<br />

( rt , )<br />

k<br />

In eq. [8.97] the Λ are the Lagrange multipliers introduced to satisfy eq. [8.94]. It is worth<br />

noticing that, while the nuclear dynamics [8.95] can have a physical meaning, that is not<br />

true for the dynamics associated with the {αv} and {ψi}: this dynamics is fictitious, like the<br />

associated “masses” μ.<br />

Ifμ and the initial conditions {ψi} 0 and {dψi/dt} 0 are chosen such that the two classical<br />

sets <strong>of</strong> degrees <strong>of</strong> freedom (nuclear and electronic) are only weakly coupled, the transfer <strong>of</strong><br />

energy between them is small enough to allow the electrons to adiabatically follow the nuclear<br />

motion, then remaining close to their instantaneous BO surface. In such a metastable<br />

situation, meaningful temporal averages can be computed. The mentioned dynamics is<br />

meant to reproduce what actually occurs in real matter, that is, electrons adiabatically following<br />

the nuclear motion.<br />

QM potentials have been widely used in molecular dynamics simulation <strong>of</strong> liquid water<br />

using the CP DFT algorithm. See, for example, refs. [72,73].<br />

8.7.2.2 Semi-classical simulations<br />

Computer simulations are methods addressed to perform “computer experimentation”. The<br />

importance <strong>of</strong> computer simulations rests on the fact that they provide quasi-experimental<br />

data on well-defined models. As there is no uncertainty about the form <strong>of</strong> the interaction potential,<br />

theoretical results can be tested in a way that is usually impossible with results obtained<br />

by experiments on real liquids. In addition, it is possible to get information on<br />

quantities <strong>of</strong> no direct access to experimental measures.<br />

There are basically two ways <strong>of</strong> simulating a many-body system: through a stochastic<br />

process, such as the Monte Carlo (MC) simulation, 74 or through a deterministic process,<br />

such as a Molecular Dynamics (MD) simulation. 75,76 Numerical simulations are also performed<br />

in a hybridized form, like the Langevin dynamics 42 which is similar to MD except<br />

for the presence <strong>of</strong> a random dissipative force, or the Brownian dynamics, 42 which is based<br />

on the condition that the acceleration is balanced out by drifting and random dissipative<br />

forces.<br />

Both the MC and the MD methodologies are used to obtain information on the system<br />

via a classical statistical analysis but, whereas MC is limited to the treatment <strong>of</strong> static properties,<br />

MD is more general and can be used to take into account the time dependence <strong>of</strong> the<br />

system states, allowing one to calculate time fluctuations and dynamic properties.<br />

In the following, we shall briefly describe the main features <strong>of</strong> MD and MC methodologies,<br />

focusing the attention to their use in the treatment <strong>of</strong> liquid systems.<br />

Molecular dynamics<br />

Molecular Dynamics is the term used to refer to a technique based on the solution <strong>of</strong> the<br />

classical equation <strong>of</strong> motion for a classical many-body system described by a many-body<br />

Hamiltonian H.<br />

In a MD simulation, the system is placed within a cell <strong>of</strong> fixed volume, usually <strong>of</strong> cubic<br />

shape. A set <strong>of</strong> velocities is assigned, usually drawn from a Maxwell-Boltzmann distribution<br />

suitable for the temperature <strong>of</strong> interest and selected to make the linear momentum<br />

equal to zero. Then the trajectories <strong>of</strong> the particles are calculated by integration <strong>of</strong> the classical<br />

equation <strong>of</strong> motion. It is also assumed that the particles interact through some forces,

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

Saved successfully!

Ooh no, something went wrong!