A ground state potential energy surface for HONO based on a neural network with exponential fitting functions

• DOI: 10.1039/c7cp04010e
• Publication Date: 2017-08-04T13:33:18Z
• Authors: Ekadashi Pradhan; Alex Brown
• Keywords: Potential energy surface

The minimum energy structures, i.e., trans-HONO, cis-HONO, HNO2, and OH + NO, as well the corresponding transition states, TStrans↔cis, TS1,2H-shift, TS1,3H-shift, on ground state potential surface (PES) of HONO have been characterized at CCSD(T)-F12/cc-pVTZ-F12 level theory. Using same theory, a six-dimensional (6D) PES, encompassing trans- cis-isomers associated state, is fit in sum-of-products form using neural network exponential fitting functions. A second PES developed based ab initio data from CCSD(T) computations extrapolated to complete basis set (CBS) limit. fits, 90 neurons, are accurate (RMSEs ≈ 10 cm-1) up 000 cm-1 above minimum. PESs validated by computing vibrational energies block improved relaxation with multi configuration time dependent Hartree (MCTDH) approach. frequencies obtained compared available experimental measurements, previous theoretical CCSD(T)/cc-pVQZ(-g functions) anharmonic MP2/aug-cc-pVTZ CCSD(T)/aug-cc-pVTZ levels theory second-order perturbation results suggest that these best for HONO, thus, should be suitable variety dynamics studies, including quantum MCTDH where can exploited computational efficiency.