Rice U team uses new computational methodology to identify high capacity MOFs for on-board natural gas storage

18 December 2014

Researchers from Rice University, Lawrence Berkeley National Laboratory and UC Berkeley have developed a computational methodology to support the experimental exploration of potential high-capacity metal organic frameworks (MOFs) for use in on-board storage of natural gas. The advantages to using MOFs as a storage medium are many and start with increased capacity over the heavy, high-pressure cylinders in current use.

In a paper in the ACS Journal of Physical Chemistry C, they report identifying 48 materials with higher predicted deliverable capacity (at 65 bar storage, 5.8 bar depletion, and 298 K) than MOF-5—the currently best available for the natural gas storage application. The best material identified by the researchers has a predicted deliverable capacity 8% higher than that of MOF-5.

<!——>

MOFs are nanoscale compounds of metal ions or clusters known as secondary building units (SBUs) and organic binding ligands, or linkers. These linkers hold the SBUs together in a spongy network that can capture and store methane molecules in a tank under pressure. As the pressure is relieved, the network releases the methane for use.

Due to their high porosities, high surface area, and tunable chemistry, MOFs are regarded as a promising class of nano-porous materials. Potential applications of MOFs include drug delivery, sensing, purification, catalysis, and gas storage. In the gas storage application, in particular, MOFs appeal as a competitive alternative to other materials, such as zeolites, because of their potentially higher performance and adjustability. Computation predictions have extended the number of potential MOFs to 100, 000.

From a practical point of view, one can only synthesize and test a small fraction of all possible MOF materials, and computation predictions are useful for suggesting promising sets of MOFs to synthesize. Existing prediction methods focus on simulation of the self-assembly process, placing known SBUs into candidate periodic networks. A limitation of these methods is that only pre-existing linkers are considered, with little or no exploration of the space of possible organic linkers.
We here develop a de novo evolutionary algorithm to explore the composition and configuration space of linker molecules to optimize methane deliverable capacity in predicted MOFs.

Since the linker, SBU, and topology can all vary, the chemical search space is nearly infinite. This poses a fundamental problem for the current methods of library generation, which are all based on brute force enumeration, as the number of compounds grows exponentially with length and branching structure of the linkers. The vast majority of these potential have poor methane delivery performance. Thus, it is desirable to efficiently sample the part of the MOF composition space with favorable materials properties. We tackle this issue by using an evolutionary algorithm to rapidly explore MOF linker composition space, among MOFs with high predicted methane deliverable capacity.

One of 48 metal organic frameworks discovered through an algorithm developed at Rice to explore compounds that excel at storing methane. Here, molecules known as secondary building units (top left) and organic binding ligands, or linkers (top right) can be used in a chemical process to produce the metal organic framework seen at bottom. Courtesy of the Deem Research Group. Click to enlarge.

The team led by Rice bioengineer Michael Deem used a custom algorithm not only quickly to design new MOF configurations able to store compressed natural gas with a high “deliverable capacity,” but also to design ones that can be reliably synthesized from commercial precursor molecules. The algorithm also keeps track of the routes to synthesis.

The program adhered to standard DOE conditions that an ideal MOF would store methane at 65 bar (atmospheric pressure at sea level is one bar) and release it at 5.8 bar, all at 298 kelvins (about 77 degrees Fahrenheit). That pressure is significantly less than standard CNG tanks, and the temperature is far higher than liquid natural gas tanks that must be cooled to minus 260 degrees F.

Lower pressures mean tanks can be lighter and made to fit cars better, Deem said. They may also offer the possibility that customers can tank up from household gas supply lines.

The Deem group’s algorithm was adapted from an earlier project to identify zeolites. The researchers ran Monte Carlo calculations on nearly 57,000 precursor molecules, modifying them with synthetic chemistry reactions via the computer to find which would make MOFs with the best deliverable capacity.

Our work differs from previous efforts because we’re searching the space of possible MOF linkers specifically for this deliverable capacity. We’re very keen to work with experimental groups, and happy to collaborate. We have joint projects underway, so we hope some of these predicted materials will be synthesized very soon.

The researchers hope to begin real-world testing of their best MOF models.

Yi Bao, a graduate student in Deem’s lab at Rice’s BioScience Research Collaborative, is lead author of the paper. Co-authors are Richard Martin and Maciej Haranczyk of the Lawrence Berkeley National Laboratory and Cory Simon and Berend Smit of the University of California-Berkeley. Deem is chair of Rice’s Department of Bioengineering and the John W. Cox Professor of Biochemical and Genetic Engineering.

The DOE Office of Basic Energy Sciences, Division of Chemical Sciences, Geosciences and Biosciences, supported the research. The researchers utilized the National Science Foundation-funded DAVinCi supercomputer administered by Rice’s Ken Kennedy Institute for Information Technology.

Resources

• Yi Bao, Richard Luis Martin, Cory M Simon, Maciej Haranczyk, Berend Smit, and Michael W Deem, “In Silico Discovery of High Deliverable Capacity Metal-Organic Frameworks,” J. Phys. Chem. C, Just Accepted Manuscript
  doi: 10.1021/jp5123486