- Email: [email protected]
It’s an Interesting MOF, but Is It Stable?
Preview
It’s an Interesting MOF, but Is It Stable? Randall Q. Snurr1,* The mechanical stability of metal-organic framework (MOF) structures is key to their functionality. In this issue of Matter, Moghadam et al. use high-throughput simulations and machine learning to derive structure-mechanical properties and predict bulk moduli of MOFs from a few physical parameters.
Interest in metal-organic frameworks (MOFs) has grown tremendously since the term was coined almost 25 years ago.1 MOFs are interesting for their diverse chemistry, the beauty of their crystal structures, and their potential technological applications, which range from gas storage and chemical separations to catalysis and photochemistry to drug delivery and medical applications.2 In many of these applications, the mechanical stability of the material is an important factor. For example, mechanical strength is important in pelletizing MOF powders into larger, shaped particles that are required in industrial gas separation. Despite its importance, there are relatively few studies on MOF stability in the open literature, and a general understanding of how MOF topology, the length of the organic linkers, and the coordination geometry of the building blocks affect the mechanical stability of the resulting MOF is not available. As MOFs move toward commercialization, this lack of understanding becomes an increasingly important knowledge gap. In this issue of Matter, Moghadam et al.3 present an interactive ‘‘map’’ of the structure-mechanical landscape of MOFs. To derive general relationships that should apply to a diverse set of MOFs, they turned to high-throughput molecular simulations and analyzed
26
3,385 MOFs containing 41 network topologies. For each MOF, they calculated the bulk, shear, and Young’s moduli using classical molecular mechanics. Then, to understand how the mechanical properties are related to the structural properties (linker length, gravimetric surface area, pore-limiting diameter, etc.), they developed an interactive visualization tool. The webbased tool is publicly available and allows the user to analyze the results interactively with 15 structural and mechanical features in 5D plots. Using this tool, Moghadam et al. could answer questions such as whether linker length or the coordination number and topology is a more important factor in determining the mechanical stability. The results indicate that extension of the organic linkers induces bigger reductions in stability for network topologies with high coordination numbers such as fcu and ftw. Moghadam et al.3 also used their large dataset to train and validate an artificial neural network (ANN) to predict the bulk modulus from four MOF structural descriptors: the density, gravimetric surface area, largest cavity diameter, and void fraction. Adding the topology as a fifth descriptor dramatically improved the accuracy of the ANN, indicating that topological features are essential in accurately predicting the bulk modulus. The ANN can be used to rapidly estimate the bulk
Matter 1, 17–38, July 10, 2019 ª 2019 Elsevier Inc.
modulus for existing or new MOFs from five readily obtained descriptors, making it a valuable tool for the MOF community. To complement the molecular mechanics calculations (zero Kelvin results), Moghadam et al. examined a small number of materials at finite temperatures and pressures using molecular dynamics simulations with an ab initio-derived force field. The results indicate that the bulk moduli from the high-throughput calculations (zero Kelvin and simple force field) are representative of the critical pressure these materials can withstand at more typical operating conditions. It should be noted that the study focuses on mechanical stability and does not address whether a structure can withstand moisture, which can lead to degradation of some MOFs. Developing our understanding of the hydrothermal stability of MOFs remains an open challenge. Beyond its contribution to the development of mechanically stable MOFs, this paper illustrates some encouraging trends that have the potential to vastly accelerate research in the coming years. First, the 11 authors come from six different institutions and three countries. Rather than viewing each other as ‘‘competitors,’’ the authors have come together to solve an important problem, each research group bringing different expertise to the project. Second, the project both uses and creates open-source computational tools, including a database of MOFs,4 molecular modeling software,5 the ANN, and the web-based
1Department
of Chemical & Biological Engineering, Northwestern University, Evanston, IL 60208, USA *Correspondence: [email protected] https://doi.org/10.1016/j.matt.2019.06.014
interactive data visualizer. Such sharing of codes, databases, and other tools can drastically speed up research and has other benefits as well, such as the potential for improved reproducibility of results.6 Finally, machine learning and other methods from data science—while perhaps overhyped at the moment— truly can lead to new ways of doing research. The results and insights developed in the work of Moghadam et al. were only possible because of the large amount of data generated by molecular modeling. It is possible that we are at the beginning of a new era of scientific research due to this shift toward team research, open-source tools, and data science.
ACKNOWLEDGMENTS The author is grateful to the US Department of Energy, Office of Science, Office of Basic Energy Sciences for support under award DE-FG02-08ER15967.
DECLARATION OF INTERESTS R.Q.S. has a commercial interest in the company NuMat Technologies, which is seeking to commercialize metalorganic frameworks. 1. Yaghi, O.M., Li, G., and Li, H. (1995). Selective binding and removal of guests in a microporous metal-organic framework. Nature 378, 703–706. 2. L.R. MacGillivray, ed. (2010). Metal-Organic Frameworks: Design and Application (John Wiley & Sons).
3. Moghadam, P.Z., Rogge, S.M.J., Li, A., Chow, C.-M., Wieme, J., Moharrami, N., Aragones-Anglada, M., Conduit, G., Gomez-Gualdron, D.A., Van Speybroeck, V., et al. (2019). Structural-mechanical stability relations of metal-organic frameworks via machine learning. Matter 1, this issue, 219–234. 4. Colo´n, Y.J., Go´mez-Gualdro´n, D., and Snurr, R.Q. (2017). Topologicallyguided, automated construction of MOFs and their evaluation for energy-related applications. Cryst. Growth Des. 17, 5801–5810. 5. Verstraelen, T., Vanduyfhuys, L., Vandenbrande, S., and Rogge, S.M.J.. (n.d.). Yaff, Yet Another Force Field. https://molmod.ugent.be/ software/. 6. Cummings, P.T., and Gilmer, J.B. (2019). Open-source molecular modeling software in chemical engineering. Curr. Opin. Chem. Eng. 23, 99–105.
Preview
See-Through Teeth, Clearly
the shell and under the animal’s soft body, where nobody can see it during the animal’s life. In dragonfish teeth, instead, the function of transparent teeth is clear. Pun intended.
Pupa U.P.A. Gilbert1,2,* and Cayla A. Stifler1 The teeth of the deep-sea dragonfish are sharp, hard, stiff, and transparent. Here we compare them to other teeth and their structure, which may determine both light scattering and mechanical behavior of teeth in diverse animals. If you want to catch fish in deep sea, where no sunlight penetrates you need to develop tools highly specialized for the task. The dragonfish is a master of such tricks. It evolved to be dark like the abyss surroundings; it grew a barbel protruding from its chin and holding a bioluminescent lantern to attract its prey, it adapted its jaw so that it can open much more than comparable size fish so it can prey on fish up to 50% its body mass; and it evolved a set of teeth that are the latest and greatest in biomaterial science. As cleverly discovered by Velasco-Hogan et al.1, they are not only razor-sharp, hard, and stiff, but they are completely transparent when wet, that is, their index of refraction must exactly match that of seawater, and they do not
scatter light at all. Why bother? Because if they scattered light, when illuminated by bioluminescent light either from the dragonfish itself or by prey bioluminescence, the prey would discover the ferocious teeth and quickly swim away to save themselves. Thus, besides their structure and mechanical properties, the dragonfish teeth must have sophisticated optical behavior, as cleverly discovered and shown by Velasco-Hogan et al.1 Transparent teeth are not per se unique; they exist in other animals too, e.g., the radula teeth of the red abalone (Figure 1A), but in abalone teeth transparency is accidental, not functional, since the mouth radula is located under
So, dragonfish teeth are transparent, and usefully so. But how is transparency achieved? The first-discovery paper by Velasco-Hogan et al.1 addresses this point by presenting transmittance and reflectance data, and scattering calculations based on simplified assumptions, which hint at the possible role of nanocrystal size. Extremely informative density and refractive index measurements, however, remain to be done. So does the full characterization of the teeth surface: the enamel-like layer. One of the open questions is: How does a mineralized material match the density and refractive index of
1University
of Wisconsin–Madison, Department of Physics, Madison, WI 53706, USA
2University
of Wisconsin–Madison, Departments of Chemistry, Materials Science, and Geoscience, Madison, WI 53706, USA *Correspondence: [email protected] https://doi.org/10.1016/j.matt.2019.06.015
Matter 1, 17–38, July 10, 2019 ª 2019 Elsevier Inc.
27
It’s an Interesting MOF, but Is It Stable? Randall Q. Snurr1,* The mechanical stability of metal-organic framework (MOF) structures is key to their functionality. In this issue of Matter, Moghadam et al. use high-throughput simulations and machine learning to derive structure-mechanical properties and predict bulk moduli of MOFs from a few physical parameters.
Interest in metal-organic frameworks (MOFs) has grown tremendously since the term was coined almost 25 years ago.1 MOFs are interesting for their diverse chemistry, the beauty of their crystal structures, and their potential technological applications, which range from gas storage and chemical separations to catalysis and photochemistry to drug delivery and medical applications.2 In many of these applications, the mechanical stability of the material is an important factor. For example, mechanical strength is important in pelletizing MOF powders into larger, shaped particles that are required in industrial gas separation. Despite its importance, there are relatively few studies on MOF stability in the open literature, and a general understanding of how MOF topology, the length of the organic linkers, and the coordination geometry of the building blocks affect the mechanical stability of the resulting MOF is not available. As MOFs move toward commercialization, this lack of understanding becomes an increasingly important knowledge gap. In this issue of Matter, Moghadam et al.3 present an interactive ‘‘map’’ of the structure-mechanical landscape of MOFs. To derive general relationships that should apply to a diverse set of MOFs, they turned to high-throughput molecular simulations and analyzed
26
3,385 MOFs containing 41 network topologies. For each MOF, they calculated the bulk, shear, and Young’s moduli using classical molecular mechanics. Then, to understand how the mechanical properties are related to the structural properties (linker length, gravimetric surface area, pore-limiting diameter, etc.), they developed an interactive visualization tool. The webbased tool is publicly available and allows the user to analyze the results interactively with 15 structural and mechanical features in 5D plots. Using this tool, Moghadam et al. could answer questions such as whether linker length or the coordination number and topology is a more important factor in determining the mechanical stability. The results indicate that extension of the organic linkers induces bigger reductions in stability for network topologies with high coordination numbers such as fcu and ftw. Moghadam et al.3 also used their large dataset to train and validate an artificial neural network (ANN) to predict the bulk modulus from four MOF structural descriptors: the density, gravimetric surface area, largest cavity diameter, and void fraction. Adding the topology as a fifth descriptor dramatically improved the accuracy of the ANN, indicating that topological features are essential in accurately predicting the bulk modulus. The ANN can be used to rapidly estimate the bulk
Matter 1, 17–38, July 10, 2019 ª 2019 Elsevier Inc.
modulus for existing or new MOFs from five readily obtained descriptors, making it a valuable tool for the MOF community. To complement the molecular mechanics calculations (zero Kelvin results), Moghadam et al. examined a small number of materials at finite temperatures and pressures using molecular dynamics simulations with an ab initio-derived force field. The results indicate that the bulk moduli from the high-throughput calculations (zero Kelvin and simple force field) are representative of the critical pressure these materials can withstand at more typical operating conditions. It should be noted that the study focuses on mechanical stability and does not address whether a structure can withstand moisture, which can lead to degradation of some MOFs. Developing our understanding of the hydrothermal stability of MOFs remains an open challenge. Beyond its contribution to the development of mechanically stable MOFs, this paper illustrates some encouraging trends that have the potential to vastly accelerate research in the coming years. First, the 11 authors come from six different institutions and three countries. Rather than viewing each other as ‘‘competitors,’’ the authors have come together to solve an important problem, each research group bringing different expertise to the project. Second, the project both uses and creates open-source computational tools, including a database of MOFs,4 molecular modeling software,5 the ANN, and the web-based
1Department
of Chemical & Biological Engineering, Northwestern University, Evanston, IL 60208, USA *Correspondence: [email protected] https://doi.org/10.1016/j.matt.2019.06.014
interactive data visualizer. Such sharing of codes, databases, and other tools can drastically speed up research and has other benefits as well, such as the potential for improved reproducibility of results.6 Finally, machine learning and other methods from data science—while perhaps overhyped at the moment— truly can lead to new ways of doing research. The results and insights developed in the work of Moghadam et al. were only possible because of the large amount of data generated by molecular modeling. It is possible that we are at the beginning of a new era of scientific research due to this shift toward team research, open-source tools, and data science.
ACKNOWLEDGMENTS The author is grateful to the US Department of Energy, Office of Science, Office of Basic Energy Sciences for support under award DE-FG02-08ER15967.
DECLARATION OF INTERESTS R.Q.S. has a commercial interest in the company NuMat Technologies, which is seeking to commercialize metalorganic frameworks. 1. Yaghi, O.M., Li, G., and Li, H. (1995). Selective binding and removal of guests in a microporous metal-organic framework. Nature 378, 703–706. 2. L.R. MacGillivray, ed. (2010). Metal-Organic Frameworks: Design and Application (John Wiley & Sons).
3. Moghadam, P.Z., Rogge, S.M.J., Li, A., Chow, C.-M., Wieme, J., Moharrami, N., Aragones-Anglada, M., Conduit, G., Gomez-Gualdron, D.A., Van Speybroeck, V., et al. (2019). Structural-mechanical stability relations of metal-organic frameworks via machine learning. Matter 1, this issue, 219–234. 4. Colo´n, Y.J., Go´mez-Gualdro´n, D., and Snurr, R.Q. (2017). Topologicallyguided, automated construction of MOFs and their evaluation for energy-related applications. Cryst. Growth Des. 17, 5801–5810. 5. Verstraelen, T., Vanduyfhuys, L., Vandenbrande, S., and Rogge, S.M.J.. (n.d.). Yaff, Yet Another Force Field. https://molmod.ugent.be/ software/. 6. Cummings, P.T., and Gilmer, J.B. (2019). Open-source molecular modeling software in chemical engineering. Curr. Opin. Chem. Eng. 23, 99–105.
Preview
See-Through Teeth, Clearly
the shell and under the animal’s soft body, where nobody can see it during the animal’s life. In dragonfish teeth, instead, the function of transparent teeth is clear. Pun intended.
Pupa U.P.A. Gilbert1,2,* and Cayla A. Stifler1 The teeth of the deep-sea dragonfish are sharp, hard, stiff, and transparent. Here we compare them to other teeth and their structure, which may determine both light scattering and mechanical behavior of teeth in diverse animals. If you want to catch fish in deep sea, where no sunlight penetrates you need to develop tools highly specialized for the task. The dragonfish is a master of such tricks. It evolved to be dark like the abyss surroundings; it grew a barbel protruding from its chin and holding a bioluminescent lantern to attract its prey, it adapted its jaw so that it can open much more than comparable size fish so it can prey on fish up to 50% its body mass; and it evolved a set of teeth that are the latest and greatest in biomaterial science. As cleverly discovered by Velasco-Hogan et al.1, they are not only razor-sharp, hard, and stiff, but they are completely transparent when wet, that is, their index of refraction must exactly match that of seawater, and they do not
scatter light at all. Why bother? Because if they scattered light, when illuminated by bioluminescent light either from the dragonfish itself or by prey bioluminescence, the prey would discover the ferocious teeth and quickly swim away to save themselves. Thus, besides their structure and mechanical properties, the dragonfish teeth must have sophisticated optical behavior, as cleverly discovered and shown by Velasco-Hogan et al.1 Transparent teeth are not per se unique; they exist in other animals too, e.g., the radula teeth of the red abalone (Figure 1A), but in abalone teeth transparency is accidental, not functional, since the mouth radula is located under
So, dragonfish teeth are transparent, and usefully so. But how is transparency achieved? The first-discovery paper by Velasco-Hogan et al.1 addresses this point by presenting transmittance and reflectance data, and scattering calculations based on simplified assumptions, which hint at the possible role of nanocrystal size. Extremely informative density and refractive index measurements, however, remain to be done. So does the full characterization of the teeth surface: the enamel-like layer. One of the open questions is: How does a mineralized material match the density and refractive index of
1University
of Wisconsin–Madison, Department of Physics, Madison, WI 53706, USA
2University
of Wisconsin–Madison, Departments of Chemistry, Materials Science, and Geoscience, Madison, WI 53706, USA *Correspondence: [email protected] https://doi.org/10.1016/j.matt.2019.06.015
Matter 1, 17–38, July 10, 2019 ª 2019 Elsevier Inc.
27