Chemical Physics Letters 625 (2015) 186–192 Contents lists available at ScienceDirect Chemical Physics Letters jou rn al h om epa ge: www.elsevier.com/locate/cplett Organic solvent simulations under non-periodic boundary conditions: A library of effective potentials for the GLOB model Giordano Mancini a,b,∗ , Giuseppe Brancato a,b,∗ , Balasubramanian Chandramouli a , Vincenzo Barone a,b a Scuola Normale Superiore, Piazza dei Cavalieri 7, 56126 Pisa, Italy b Istituto Nazionale di Fisica Nucleare (INFN) sezione di Pisa, Largo Bruno Pontecorvo 3, 56127 Pisa, Italy a r t i c l e i n f o Article history: Received 25 September 2014 In final form 2 March 2015 Available online 6 March 2015 a b s t r a c t We extend the library of solvents that can be treated using the GLOB (general liquid optimized boundary) method, that allows to perform MD simulations under non-periodic boundary conditions (NPBC) opti- mizing effective potentials between explicit molecules and the boundary for four organic solvents: CHCl 3 , CCl 4 , CH 3 OH and CH 3 CN. We show that GLOB allows reducing the number of explicit solvent shells to be included, while yielding results comparable with PBC and significant advantages over simulations with- out explicit boundaries. Finally, we provide polynomial fittings for all available GLOB effective potentials (including SPC water) to simplify their implementation in NPBC MD simulations. © 2015 Elsevier B.V. All rights reserved. 1. Introduction Periodic boundary conditions (PBC) are the de facto standard for performing molecular dynamics (MD) simulations of bulk systems minimizing artifacts related to the finite dimensions of the simula- tion box by letting particles which leave the box through one face re-enter it through the opposite face [1]. In some instances there are, however, good reasons to employ non-periodic boundary con- ditions (NPBC) [2–8]. First, PBC can be unnecessarily costly since they require a simulation box whose shape is space-filling when periodically replicated, for example, a cube. Whenever the simu- lated process involves a globular macromolecule such as a protein in solution or corresponds to a localized chemical event, a spheri- cal simulation box may demand a lower number of explicit solvent molecules as compared to a cubic box. In addition, the necessity to correct for long range interactions, non-vanishing across box boundaries (e.g. the Ewald [9] or PME [10] methods) most often accounts for the greatest share of the computational cost needed to perform a simulation. Thus, NPBC were originally designed for these cases [11–13]. Second, PBC may introduce spurious features in structural or dynamical properties in certain systems involving, for instance, charged species; examples of these artifacts include ∗ Corresponding authors at: Scuola Normale Superiore, Piazza dei Cavalieri 7, 56126 Pisa, Italy. E-mail addresses: giordano.mancini@sns.it (G. Mancini), giuseppe.brancato@sns.it (G. Brancato). effects on the counter ion distribution, conformational equilibria and energetic bias in the simulation of electrolytic solutions and/or biological systems [14–22]. This behavior follows from a periodicity that is artificially imposed onto intrinsically non-periodic systems such as fluids and amorphous polymers. We have an additional motivation for employing NPBC, which is related to our interest in multi-scale methods in which the inner- most layer is treated by quantum mechanical (QM) methods based on localized basis functions; thus, contrary to approaches based on plane waves, periodicity is not intrinsic to the model and its intro- duction would require additional methodological complications and computational costs. On the other hand, the latest implemen- tations of the polarizable continuum model (PCM) [23] offer an inexpensive yet reliable treatment of NPBC for the electrostatic part of the intermolecular potential. The PCM cavity can be also employed to force elastic reflection at the boundary, without any additional cost. Thus, it is quite natural to perform NVT simulations under such conditions provided that an additional phenomeno- logical potential (hereafter referred to as W vdW ) is included to enforce constant density (and possibly other constraints) near the boundary. This philosophy is at the heart of the GLOB (general liq- uid optimized boundary) [7,8] model we have been developing in recent years for MM, QM/MM and full QM extended Lagrangian simulations. Several works [24–31] have shown that the GLOB model delivers remarkable results for different systems and pro- cesses in aqueous solutions and its range of applications can be further extended to constant pressure processes [32] and to polar- izable force fields [33,34]. On these grounds, the main aim of the http://dx.doi.org/10.1016/j.cplett.2015.03.001 0009-2614/© 2015 Elsevier B.V. All rights reserved.