Protonentransport und Wasserstoffbrücken-Netzwerke in komplexen Umgebungen mit Hilfe von beschleunigten Quanten-Simulationen verstehen
Zusammenfassung der Projektergebnisse
The main goal of this project was to develop efficient atomistic simulation methods to accurately investigate hydrogen bond networks and proton transfer reactions in complex, condensed-phase environments. A better understanding of these environments on the molecular level could for instance lead to improved proton exchange membranes, optimized electrochemical reactions, and the developments of new drugs that target enzyme active sites. In all these systems disordered hydrogen bond environments are encountered, whose accurate description requires to explicitly account for electronic and nuclear quantum effects. While in principle this can be done employing ab initio molecular dynamics coupled with a path-integral treatment of the nuclei, those methods are computationally very demanding and thus limited to small systems and short simulation times. In the course of the project I have developed reactive interatomic potentials based on machine learning techniques trained to the results of accurate ab initio calculations. The efficiency afforded by the developed potentials allowed me to reach length- and timescales previously not accessible, which made it possible to investigate the vibrational properties of hydrogen bond networks across a wide range of temperatures and to shed light onto the impact of nuclear quantum effects on proton defect dynamics in solution. To investigate the dynamic properties of hydrogen bond networks I have led a collaborative research project in which simulations where combined with experimental Raman measurements of water in the ambient and supercooled regimes. In a first study, we have shown that the machine learning accelerated simulations accurately capture the subtle temperature-dependent Raman spectra of liquid water across the entire frequency range. We then employed a 2D correlation analysis of the simulated spectra to assign Raman band and could show that several spectral regions exhibit a strong dependence on the local tetrahedral order. In a second study, we combined measurements of liquid and supercooled water with extended simulations to identify a previously unrecognized peak in the OH stretching region of the spectrum. By a decomposition analysis of the simulated spectra we were able to show that this feature corresponds to overcoordinated hydrogen bond environments, structures that play a crucial role in fundamental processes such as the immobilization of water around hydrophobic solutes and the anomalous diffusion in supercooled water. To study proton transport in solution I have developed a reactive machine learning potential that is able to simultaneously describe the transport of hydroxide and hydronium defects and the dissociation of water molecules in solution. Performing extensive simulations over hundreds of nanoseconds allowed me to investigate the influence of quantum effects, system size, and exact exchange on the diffusion behaviour and to predict the self-ionization constant of water in excellent agreement with experiment.
Projektbezogene Publikationen (Auswahl)
- The Interplay of Structure and Dynamics in the Raman Spectrum of Liquid Water over the Full Frequency and Temperature Range. J. Phys. Chem. Lett. 9, 851 (2018)
T. Morawietz, O. Marsalek, S. R. Pattenaude, L. M. Streacker, D. Ben-Amotz, T. E. Markland
(Siehe online unter https://doi.org/10.1021/acs.jpclett.8b00133) - Hiding in the crowd: Spectral signatures of overcoordinated hydrogen bond environments
T. Morawietz, A. S. Urbina, P. K. Wise, X. Wu, W. Lu, D. Ben-Amotz, T. E. Markland
(Siehe online unter https://doi.org/10.1021/acs.jpclett.9b01781)
