An enhanced sampling QM/AMOEBA approach: The case of the excited state intramolecular proton transfer in solvated 3-hydroxyflavone

Abstract

We present an extension of the polarizable quantum mechanical (QM)/AMOEBA approach to enhanced sampling techniques. This is achieved by connecting the enhanced sampling PLUMED library to the machinery based on the interface of Gaussian and Tinker to perform QM/AMOEBA molecular dynamics. As an application, we study the excited state intramolecular proton transfer of 3-hydroxyflavone in two solvents: methanol and methylcyclohexane. By using a combination of molecular dynamics and umbrella sampling, we find an ultrafast component of the transfer, which is common to the two solvents, and a much slower component, which is active in the protic solvent only. The mechanisms of the two components are explained in terms of intramolecular vibrational redistribution and intermolecular hydrogenbonding, respectively. Ground and excited state free energies along an effective reaction coordinate are finally obtained allowing for a detailed analysis of the solvent mediated mechanism.

Grassmann extrapolation of density matrices for Born-Oppenheimer molecular dynamics

“…The tests performed with a 10 −5 convergence threshold are representative of a standard, dft ground state bomd simulation. When performing a more sophisticated quantum mechanical calculation, such as a bomd on an excited state pes, 16 such a convergence threshold may not be sufficient to guarantee the stability of the simulation, as the scf solution is used to set up the linear response equations and the numerical error can be amplified, resulting in poorly accurate forces.…”

“…The first system is 3-hydroxyflavone (3hf) in acetonitrile. 16 Two systems (ocp and appa) are chromophores embedded in a biological matrix -namely, a carotenoid in the orange carotenoid protein (ocp) and flavine in acid phosphatase (appa), a blue light-using flavine photoreceptor. [33][34][35] The fourth system is dimethylaminobenzonitrile (dmabn) in methanol.…”

“…Unfortunately, the performances of the xlbo-based bomd scheme are not so good when a tighter scf convergence is required, which can be the case when one wants to perform md simulations using post Hartree-Fock (hf) methods or for excited states described in a timedependent dft framework. 14,16 In fact, such methods require the solution to a second set of qm equations which are typically non-variational, making them more susceptible to numerical errors and instabilities. Computing the forces for non-scf energies requires therefore a more accurate scf solution.…”

Ultrafast Transient Infrared Spectroscopy of Photoreceptors with Polarizable QM/MM Dynamics

“…28,29 Our implementation relies on a highly optimized linear-scaling machinery 30 and has been successfully extended to Born− Oppenheimer MD (BOMD) simulations. 31,32 In our implementation, QM/AMOEBA molecular dynamics is achieved through the interplay of the Tinker 33,34 and Gaussian 35 software packages. The Tinker MD package is used to compute the bonded and dispersion-repulsion contributions to the energy and forces and to propagate the MD trajectories.…”

“…In particular, our group has been focused on the development of a versatile and efficient polarizable QM/MM strategy − based on density functional theory (DFT) and the highly accurate AMOEBA polarizable force field. , Our implementation relies on a highly optimized linear-scaling machinery and has been successfully extended to Born–Oppenheimer MD (BOMD) simulations. , In our implementation, QM/AMOEBA molecular dynamics is achieved through the interplay of the Tinker , and Gaussian software packages. The Tinker MD package is used to compute the bonded and dispersion-repulsion contributions to the energy and forces and to propagate the MD trajectories.…”