For the first time, to our knowledge, a lyotropic liquid crystal (LLC) system is studied atomistically using molecular dynamics simulations.<br/>LLC's are a class of nanostructured materials that can be modified, assembled and crosslinked into stable membranes with uniform straight<br/>pores that can perform solute specific separations and offer the ability to control pore architecture at the atomic lengthscale. Pore diameters on the order of 1 nm make LLC membranes well-suited for aqueous separations such as desalination and biorefinement. A molecular model which is consistent with experimentally measured structural features and material properties can provide a link between monomer structure and macroscopic separation performance. We have used simulations with experimental data to create such a model. In our model, we have developed a method to simulate the crosslinking mechanism, a key step during synthesis. We have also adopted and compared multiple methods for calculating ionic conductivity from atomistic simulations. Simulated X-ray diffraction patterns based on atomic coordinates generate 1D and 2D diffraction patterns containing all major features present experimentally. Using these methods to characterize the molecular model, we have discovered an additional metastable state which may form under certain experimental conditions. Each configuration gives different structural properties but persists for 100's of simulated nanoseconds. We have also studied the transport of a range of solutes with various charges and hydrodynamic radii. With a clear understanding of mechanisms of transport in these complex self-assembled systems, one can choose monomers to achieve specific separation goals. We can use this information to draw correlations between pore structure and selective preferences. These studies will help guide monomer choice for separation-specific objectives. To generate robust and reproducible statistics for each type of calculation, long simulations, up to 1000 ns, must be run. Simulating the system of 66K atoms with GROMACS is facilitated by the use of NVIDIA Tesla K80 GPU accelerators. Parallelization, with linear scaling up to 12 GPUs, yields 125 ns/day of simulation time.

Understanding the microscopic structure of lyotropic liquid crystal membranes using molecular dynamics simulations

254th National Meeting (2017)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.