Global Analytical Potential Energy Surface for Large Amplitude Nuclear Motions in Ammonia

Abstract

An analytical, full-dimensional, and global representation of the potential energy surface of NH(3) in the lowest adiabatic electronic state is developed, and parameters are determined by adjustment to ab initio data and thermochemical data for several low-lying dissociation channels. The electronic structure is calculated at the CASPT2 level within an [8,7] active space. The representation is compared to other recently published potential energy surfaces for this molecule. The present representation is distinguished by giving a qualitatively correct description of the potential energy for very large amplitude displacements of the nuclei from equilibrium. Other characteristic features of the present surface are the equilibrium geometries r(eq)(NH(3)) approximately 101.24 pm, r(eq)(NH(2)) approximately 102.60 pm, alpha(eq)(NH(3)) approximately 106.67 degrees, and the inversion barrier at r(inv)(NH(3)) approximately 99.80 pm and 1781 cm(-1) above the NH(3) minimum. The barrier to linearity in NH(2) is 11,914 cm(-1) above the NH(2)((2)B(1)) minimum. While the quartic force field for NH(3) from the present representation is significantly different from that of the other potential energy surfaces, the vibrational structures obtained from perturbation theory are quite similar for all representations studied here.

Global Analytical Potential Energy Surfaces for High‐resolution Molecular Spectroscopy and Reaction Dynamics

“…Those representations were used to calculate vibrational term values and transition dipole moments that agree very well with experimental data from high-resolution spectroscopy (deviations are a few cm −1 to up to the near-infrared). Drawbacks of these representations are, first, that they are not global, because the CCSD(T) data have a local definition range: it was shown (Marquardt et al 2005) that indeed the data correlate asymptotically with a reaction channel that corresponds to a higher electronic state. Secondly, these representations are not sufficiently compact, because of the rather long polynomial expansion used.…”

“…In this section, we review some theoretical developments on new, compact analytical forms that were devised to fill this gap (Marquardt and Quack 1998, Cuvelier et al 2004, Marquardt et al 2005, Marquardt and Sagui 2007, Marquardt et al 2010) and discuss applications regarding global analytical representations of molecules of the type XY n , with emphasis on methane Quack 1998, 2004) and ammonia (Marquardt et al 2003a(Marquardt et al ,b, 2005.…”

“…Two major differences exist, however, which make the global analytical representation of the adiabatic lowest potential energy surface considerably more difficult than in the case of methane, despite the lower, total dimensionality: these are the very shallow barrier for inversion, and a conical intersection involving the reaction channel into ground-state azanyl ( 2 B 1 ) NH 2 : Figure 4 shows the lowest ammonia dissociation channels, as obtained from a CASPT2 calculation (Marquardt et al 2005) for planar ammonia and 120 • valence angles; only one NH bond is elongated (abscissa r NH = r), while the remaining NH bonds are kept fixed at roughly 100 pm.…”

“…Following a more generalized, kinetic point of view, a global, smooth representation of the global analytical representation of a PES might be conceivable in cases when the nuclei will move on the lowest adiabatic PES, while approaching and departing from the regions defining a conical intersection, Lowest adiabatic singlet (♦) and triplet ( * ) states of ammonia at different values of the NH bond length; HNH angles and the two remaining NH bond lengths are kept fixed at 120 • (planar ammonia) and 100 pm, roughly. Points are energies calculated at the CASPT2 level of theory on an (8,7) complete active space of a triple zeta-type basis set (Marquardt et al 2005), the connecting lines are guide-of-the-eye cubic spline interpolations of the data.…”

“…Very little is known, however, about the spectroscopic accuracy of these representations. In contrast to this, the representation AMMPOT2 was developed by Marquardt et al (2005), with the aim of describing both reaction channels and bound vibrational motion. A variational calculation yields a vibrational spectrum that differs 50 cm −1 , on the average, from experimental term values to up to 6000 cm −1 , where deviations for individual transitions may be as large as 30 cm −1 , which is not very good for a fitted surface.…”