Jared Dickson and Dr. Branton Campbell, Physics and Astronomy

Open-framework compounds, most notably the aluminosilicate zeolites, play a critical role in modern technology. Their robust crystalline structures contain large cavities and channels of molecular dimensions which make them useful as molecular sieves (desiccants, membrane filters, gas-separators), ion-exchangers (soaps, detergents, water softeners, radioactive waste sequestering agents), chemical catalysts (e.g. petroleum refinement), and nanomaterial growth templates. The many uses of open-framework materials arise from having size and shape-selective pore systems that determine which molecules can enter and what happens to them while inside. Once their structure-property relationships are well characterized, researchers commonly tune the useful properties of open-framework materials by modifying their atomic structures in a controlled way.

Single-crystal x-ray diffuse scattering has recently been used to characterize complex nanoscale defect structures in zeolite mordenite (B.J. Campbell et al., J. Appl. Cryst. 37, 187-192, 2004), which may prove to be a useful means of modifying their adsorptive and catalytic properties. However, most industrial and academic researchers interested in exploring these novel defect structures do not have ready access to sophisticated diffuse scattering tools.

My research efforts were aimed at simulating the effects of these newly-discovered defect structures in common x-ray powder diffraction (PXD) data so that standard laboratory techniques can be used to identify and quantify the framework defects in their materials. The ability to generate powder patterns from defect-structure models is a new feature of the DISCUS software package (Th. Proffen and R.B. Neder, J. Appl. Cryst. 32, 838, 1999). After receiving a beta version of this feature from Professor Reinhard Neder at the Institut fuer Mineralogie at the University of Wurzberg in Germany and verified that it works on small-scale models, we developed an algorithm that employs both the Discus and Mathematica software packages to create large-scale three-dimensional defect models models in mordenite. Because this effort required the resources of the Marylou supercomputers provided by the BYU College of Engineering, we compiled and configured DISCUS to work on maryloux.

To achieve sufficient reciprocal-space resolution to observe defect-induced broadening in mordenite, spherical models containing about several million atoms were needed. Using the standard Fourier Transform option provided by DISCUS, we found that this would take about four months on a single maryloux node. Prof. Neder then pointed out that the columnar shape of the mordenite defects should permit us to greatly simplify the calculations by taking advantage of the translational periodicity parallel to the defect columns. This reduces the model size to about 250,000 atoms, so that simulations can be completed in under an hour. However, this clever “trick” employs an alternative routine that has not been fully coded yet to allow for sufficient flexibility in the output — an issue that requires modifications to the freely-available DISCUS source code. Further work should be aimed at completing these modifications and completing a family of simulations with varying defect concentrations.

The opportunities provided by this research were very rich and rewarding. I especially enjoyed the ability to solve problems that were not in a textbook and as such had been solved countless times before. Our research has not yet produced our desired results, we have eliminated one pathway to solving our problem and our other method looks extremely promising with more development. With continued work I am excited to continue the research to completion, as the results will have an actual use and will not just be a report on a shelf.