Thermodynamic Properties for Applications in Chemical Industry via Classical Force Fields - Multiscale Molecular Methods in Applied Chemistry - page 222

agreement with experimental data. However, the vapor pressure predictions are
typically slightly above experimental data at low temperatures and below experimen-
tal data at high temperatures [ 56 ], while the critical temperature is overestimated,
e.g., by 7 K for short-chain alkanes (ethane and pentane) [ 18 ].
4.1.5 GIBBS99 Force Field
The GIBBS99 exponential-6 force field [ 18 ] is a united atom representation that is
available for linear alkanes, cyclohexane, or benzene [ 134 ]. It differs from the
NERD and TraPPE force fields in the description of the Van der Waals interactions:
The Buckingham exponential-6 potential (7) was used instead of the LJ 12-6
potential (6). Similarly to the TraPPE force field, bond stretching was neglected.
However, the bond length between two methyl groups of the alkane chain was not
fixed as in the TraPPE force field, but depends on the molecular groups that form
the bond. Angle bending was represented by the harmonic potential (22) and the
torsional motion by a third order Fourier series (25). The force field parameters
were fitted to critical properties and saturated densities. The GIBBS99 force field
represents the vapor pressure and saturated densities for the alkanes from ethane
to n -dodecane with average deviations of around 2%. The experimental vapor
pressures of benzene and cyclohexane are reproduced with average errors of 2.6
and 1.7%, respectively [ 134 ].
4.1.6 Other Force Fields
Transferable force field families intended for biological applications are sometimes
applied in chemical engineering for the simulation of large molecules like poly-
mers. Also, some ionic liquids were parameterized in that framework [ 105 ]. Some
relevant force fields are chemistry at Harvard molecular mechanics (CHARMM)
[ 91 , 95 , 140 ], assisted model building with energy refinement (AMBER) [ 65 , 66 ,
146 , 147 ], Groningen molecular simulation (GROMOS) [ 148 - 150 ], condensed-
phase optimized molecular potentials for atomistic simulation studies (COMPASS)
[ 70 ], and consistent force field (CFF) [ 151 - 153 ], among many others. These force
fields best reproduce the data for which their parameters were optimized. AMBER,
CHARMM, and OPLS-AA overestimate the free energy of hydration of protein
functional groups [ 154 ]. Several works on the comparison of various of these force
field families for the simulation of proteins [ 155 - 159 ], deoxyribonucleic acids
[ 160 ], peptides [ 161 ], carbohydrates [ 162 ], or aqueous salt solutions [ 163 ] can be
found in the literature.
Some examples for transferable polarizable force fields are Drude [ 83 ], TraPPE-
pol, CHARMM-FQ [ 164 ], PIPF [ 165 - 167 ], and AMOEBA [ 168 ]. A review on
polarizable force fields can be found, e.g., in [ 45 ]. Many of the mentioned force
fields for biochemical applications as well as the polarizable force fields are being