Use of molecular dynamics simulations to study the impact of weak to strong non-covalent chemical interactions

Molecular dynamics,Ionic liquids

Mehta, Neil A.

关键词: Molecular dynamics,Ionic liquids;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/107838/MEHTA-DISSERTATION-2020.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

In this thesis, we have used Molecular Dynamics (MD) to study the role of van der Waals and Coulomb interactions on two fundamental processes, namely, gas-surface interactions and electrospray propulsion, respectively. For the gas-surface interaction work, we proved that conventionally used gas surface interaction models were not adequate in reproducing the lobular scattering behavior obtained from the beam molecular experiments. Trajectory MD simulations were used to predict angular distributions and average translational energies. The translational energy and angular distributions of the scattered N$_2$ were obtained for incidence velocities of 1,453 and 2,220 $\mathrm{m\, s^{-1}}$, and incidence angles of 30$\mathrm{^o}$, 45$\mathrm{^o}$, and 70$\mathrm{^o}$ and a surface temperature of 677 K. The trajectories of scattered nitrogen molecules were found to fall into three main categories, i.e., single collision, multiple collisions with escape, and multiple collisions without escape. While the conventional GSI models did not match the translational energy and angular distributions obtained from the experiments, the results obtained from MD simulations were found to be in good agreement. The MD simulations also showed that the number of surface layers used to model the HOPG surface and the carbon-nitrogen Lennard-Jones potential are important in improving the agreement between the simulations and the experiments.Trajectory MD simulations were also performed on quartz surface to study the difference between atomistically smooth (HOPG) and rough surfaces (quartz). The type of surface affected collision statistics for the three collision categories, providing a strong correlation between incidence speed, angle, surface topology and the probability distribution of the energy accommodation coefficients. Based on the scattering characterisics, we developed a direct velocity sampling (DVSM) GSI model, which was then coupled to a DSMC code. A hypersonic flow over a flat plate was simulated using DSMC coupled with DVSM. The coupled simulation predicted flow, such as velocity contours, and surface properties, such as, heat flux in agreement to those observed experimentally for the HOPG surface. When the parameters of quartz were used for the DVSM model, the flow predicted was comparable to the Maxwell gas-surface interaction (GSI) model, highlighting the accuracy and versatility of the DVSM model in modeling flows over smooth and rough surfaces.As a natural progression of our trajectory MD simulation of diatomic N$_2$ gas, we used trajectory simulations to model collision of ice-like argon and amorphous silica aggregates on the HOPG and quartz surface. It was found that at all incidence velocities, the quartz surface was stickier than the HOPG surface. The sticking probabilities and elastic moduli obtained from MD were then used to model surface evolution at a micron length scale using kinetic Monte Carlo (kMC) simulations. Rules were derived to control the number of sites available for the process execution in kMC to accurately model erosion of HOPG by atomic oxygen (AO) attack and ice-nucleation on surfaces. It was observed that the effect of defects was to increase the material erosion rate while that of aggregate nucleation was to lower it. Similarly, simulations were performed to study the effects of AO attack and N$_2$ adsorption-desorption on surface evolution and it was found that N$_2$ adsorption-desorption limits the surface available for erosion by AO attack.With respect to the electrospray work, we performed MD electrospray simulations of 1-ethyl-3-methylimidazolium Tetrafluoroborate (EMIM-BF$_4$) ionic liquid with the goal of evaluating the influence of long-range Coulomb models on ion emission characteristics. The direct Coulomb (DC), shifted force Coulomb sum (SFCS), and particle-particle particle-mesh (PPPM) long-range Coulomb models were evaluated in terms of emission products predicted by these three models. The DC method with a sufficiently large large cut-off radius was found to be the most accurate approach for modeling electrosprays, but, it is computationally expensive. The Coulomb potential energy modeled by the DC method in combination with the radial electric fields were found to be necessary to generate the Taylor cone. The differences observed between the SFCS and the DC in terms of predicting the total ion emission suggested that the former should not be used in MD electrospray simulations. Furthermore, the common assumption of domain periodicity was observed to be detrimental to the accuracy of the capillary based electrospray simulations and therefore, the PPPM model could not be used for the electrospray. simulations.As an efficient alternative, we developed a new octree-based Coulomb interaction model. For the octree-based method, Coulomb interactions were categorized as intra- and inter-leaf Coulomb interactions based on a criterion related to the Bjerrum length of the IL. The octree-based method was found capable of reproducing Coulomb energy in agreement with established and computationally more expensive models, such as the DC and the Damped Shifted Force (DSF) method in the absence of an external electric field. In the presence of an external electric field, the octree-based method produced distinctly different results compared to that obtained by the DC method. The time required to form Taylor's cone was shorter for the octree method compared to the DC approach. While no emission larger than monomers was observed from the DC simulation, emission of larger species such as dimers and trimers was observed when the octree-based Coulomb interaction model was used. Furthermore, the octree-based model formed a smaller ion emission cone compared to that from the direct Coulomb method, showcasing the advantage of using the octree-based model for electrospray simulations.We also used MD simulations as a predictive tool to study the disintegration of large droplet emitted during the electrospray process. We considered an isolated droplet of EMIM-BF$_4$ as a test case to understand the effects of normal and radial electric fields on the final mass, volume, and charge of the droplet. The disintegration process was not found to be continuous but occurred as discrete events during the initial time steps for a range of normal and radial electric fields. Increasing the field strengths led to decrease in the emission of secondary droplets and larger aggregates but increased the emission of smaller species. The volume decay rate of the primary droplet was found to depend linearly on the normal and radial electric field strengths. The radii of the ion aggregates emitted from the primary droplet and the emission break-up times agreed well with those from the Rayleigh instability theory and the Coulomb fission model.The MD and kMC algorithm developed for this work were specific for the HOPG and quartz surfaces, these methods are general and can be applied to study any gas or aggregate interaction with other surface types. Similarly, the octree-based Coulomb interaction model was used to perform electrospray simulations of IL, it can be used to model simulations of large domains with non-homogeneous charge distributions.