On the automatic computation of global intermolecular potential energy surfaces for non-covalently bound systems

International audienceA detailed knowledge of the topography of a Potential Energy Surface (PES) is a highly desirable prerequisite for the simulation of any dynamical process. Topography on its own, however, does not fully determine the (quantum) behaviour of a system and dynamical calculations become mandatory [1, 2]. Characterisation of the topography of a PES is a tedious task and, for large dimensional systems, not an evident one. In recent years, a series of methodologies for the automatic determination of stationary points for covalently bound systems have been proposed. Among them, the so-called Transition State Search using Chemical Dynamics Simulations (TSSCDS) [3] relies on the efficient sampling of configuration space using a semiempirical level of theory combined with a graph-theory (adjacency matrix) based identification of guess structures to transition states (TS). The latter are subsequently optimized at an appropriate level of theory and Minimum Energy Paths are computed, thus generating the so-called reaction network (RXN). We have recently developed the vdW-TSSCDS method, a variant of TSSCDS, specifically designed for the automatic characterization of PESs of non-covalently bound systems [4]. The key idea has been the redefinition of the adjacency matrix in block-diagonal structure where diagonal blocks correspond to isolated fragments and off-diagonal blocks to intermolecular interactions. We have successfully applied this approach to semirigid-monomer complexes (e.g. benzene-water) as well as to reactive systems (e.g. naphthalene cation plus hydrogen). We present the vdW-TSSCDS method and its application to the construction of global PESs. More specifically, PESs in the form necessary for high-dimensional quantum dynamical calculations in the context of the Multi configuration Time-Dependent Hartree (MCTDH) method [5]. To this end, the information of the RXN is used to reparametrize a semiempirical Hamiltonian which is interfaced to the tensor decomposition algorithm Multigrid POTFIT [6]. This provides the PES in the correct mathematical form for MCTDH.[1] M. E. Tuckerman, D. Marx, M. L. Klein, M. Parrinello, Science 1997, 275, 817. [2] D. Peláez, H.-D. Meyer, Chem. Phys. 2017, 482, 100.[3] E. Martínez-Núñez, J. Comp. Chem. 2015, 36, 222.[4] S. Kopec, E. Martínez-Núñez, J. Soto, D. Peláez, (submitted).[5] M. H. Beck, A. Jäckle, G. A. Worth, H.-D. Meyer, Phys. Rep. 2000, 324, 1 [6] D. Peláez, H.-D. Meyer, J. Chem. Phys. 2013, 138, 014108