Thermodynamic Properties for Applications in Chemical Industry via Classical Force Fields - Multiscale Molecular Methods in Applied Chemistry - page 226

5 Molecular Simulation Methods
Given an adequate force field, molecular simulation is in principle capable of
yielding predictions of thermodynamic properties for a broad range of thermody-
namic conditions. To this end, different simulation techniques can be employed,
which can be divided in MD and MC. Here, some simulations tools for predicting
thermodynamic properties that are important for chemical engineering, i.e., vapor-
liquid equilibrium and transport properties, will be addressed briefly.
5.1 Molecular Dynamics
MD is a technique in which the time evolution of the molecular motions is simulated
following the laws of classical mechanics. Therefore, the physical variable time
must be considered explicitly. In this way, the dynamic evolution of coordinates and
moments, i.e., the trajectory of the system, is calculated by numerically solving
Newton's equations of motion. This trajectory, together with the associated energies
and forces, leads to the static and dynamic thermodynamic properties of the studied
system via statistical analysis methods. MD is also a powerful tool to understand
dynamic processes at the atomistic level that involve fluids or materials [ 9 ].
In MD, a set of second order differential equations is solved by finite difference
techniques. This can be done with a variety of integration algorithms, such
as Verlet, velocity Verlet, Leap-Frog, or Gear predictor-corrector. Although
the microcanonical ( NVE ) ensemble is the most natural one for MD simulations,
generally the canonic ( NVT ) or the isobaric-isothermal ( NpT ) ensembles are
applied. Particularly in chemical engineering, physical properties are needed for
specified thermodynamic conditions like temperature or pressure. Several methods
exist to control temperature and pressure during simulation, e.g., velocity scaling,
Anderson thermostat, Berendsen thermostat, Nos´-Hoover thermostat, Nos´-
Hoover chains thermostat, or Berendsen barostat. A description of these algorithms
can be found, e.g., in [ 11 , 180 ].
An MD simulation yields a significant amount of useful information for chemi-
cal engineering applications [ 11 ]. E.g., it is employed to study dynamic processes,
like diffusion, adsorption, or glass transition. A review of MD applications can be
found, e.g., in [ 9 ].
5.2 Monte Carlo
MC is a stochastic method that samples the configuration space of a system with a
specified Hamiltonian [ 181 ]. In MC simulations, the transition between states or
configurations is achieved by a random generation of a new state, evaluating a
probabilistic acceptance criterion, and accepting or rejecting the perturbation.