Nanoporous Materials Explorer

From Wiki
Revision as of 22:05, 28 November 2014 by Emmhald (talk | contribs) (Hypothetical zeolites)

Jump to: navigation, search


Nanoporous materials are defined as materials having a porosity at the scale of less than 100 nm and often have pores comparable to the size of individual molecules. This gives rise to a series of unique properties, making nanoporous materials useful for industrially important applications such as gas storage, separations, catalysis, et cetera. A vast number of unique nanoporous materials can be created, varying in chemical composition and pore topology. Thousands such materials have already been synthesized and hundreds of thousand hypothetical materials have been computationally predicted. In addition, a considerable number of computational screening studies have appeared in the literature that examine the potential of nanoporous materials for a series of applications. This has generated a substantial amount of data that cannot be presented efficiently by traditional publications. To address this, the Nanoporous Explorer App provides a platform for the aggregation and presentation of data related to nanoporous materials and their properties in an interactive way. The app aims to ease the access to the available information in a way that was not previously possible and enable the identification of promising materials based on their performance and properties. The data for the Nanoporous Explorer App are predicted, measured, and/or maintained by the Nanoporous Materials Genome Center (NMGC), This manual covers the description of the classes of materials and type of properties currently accessible.

Using the Nanoporous Explorer App

The search interface of the Nanoporous Explorer App is built around an interactive 2-dimensional scatter plot. In order to initiate a search and generate a scatter plot, you need to select one or more of the available material and adsorbate classes from the drop down menu at the top of the screen and hit the “Explore Nanoporous Materials” button. The default selection includes all the available material classes and no adsorbates.

Interfacing with the data

Once a scatter plot has been generated based on a set of selected material and adsorbate classes, you can click and drag the mouse cursor on the plot to zoom in on a specific area. When zoomed in, a “reest zoom” button will appear to go back to the default plot size. The sliders on the right of the screen can adjust the pressure and temperature range of the data. You can click on a point in the scatter plot to access a list of materials in the vicinity of the chosen point. The materials close to chosen point will appear as a table below the scatter plot, and you can obtain detailed information about the material by clicking on each material in the table The details page for each material provides all the properties available for this specific material. Such details include pore characteristics and material properties for all the materials as well as adsorption properties (Henry’s constants, adsorption isotherms and the heats of adsorption) for a subset of the materials.

Material Classes


Computation-Ready, Experimental Metal-Organic Framework Database (CoRE MOF Database)

Computation-Ready, Experimental MOF database was constructed by a collaboration within the Nanoporous Materials Genome Center. The database is derived from the Cambridge Structural Database (CSD), which contains a large number of small organic and inorganic molecules as well as periodic structures. After the construction of the database, a large-scale GCMC simulation was carried out on all the structures in the database for methane storage and working capacity [1]

Database Construction Procedure

Four steps were taken to effectively make the crystal structures from the CSD computation-ready:

  1. Removal of free solvents and coordinated solvents.
  2. Judicious retention of charge-balancing ions in the framework atoms to make the crystal charge neutral.
  3. Text Mining to recover structures with no framework disorder.
  4. Manual editing of substantially disordered structures.

An example of the solvent removal and charge retention procedures is provided with accompanying python scripts.

Graph-based representation of Molecular Structures

Solvent removal and ion retention procedures rely on the construction of molecular graph representation of a MOF crystal. To construct the periodic adjacency matrix for each structure the Atomic Simulation Environment NeighborList module is used. Two atoms are considered bonded if the distance between them is less than the sum of their CSD covalent radii plus a skin distance of 0.3 Å. The skin distance is chosen to be slightly smaller than the CSD definition (0.4 Å) such that the terminal atom connected to the metal atom does not form another bond with other nearby atoms. The adjacency matrix is passed to the SciPy connected components module to identify the bonded components in each structure.

Removal of Free Solvents and Coordinated Solvents (Text excerpt from Ref. [1])

All bonded components in the molecular graph of each structure other than the MOF framework and charge-balancing ions were removed. The MOF framework was defined as the highest molecular weight bonded component of the graph. Interpenetrated MOF frameworks were retained by identifying the number of atoms, N, in the largest bonded component in the structure and retaining all additional components having at least 0.5N atoms. The bonded component corresponding to the MOF framework often includes undesirable solvent bound to unsaturated metal centers. To remove these coordinated solvent molecules, we performed a trial “cut” on all bonds between metal centers oxygen atoms. If the number of bonded clusters detected by the connected component algorithm remained constant, the bond was restored. If the number of bonded components increased, the entire new component was considered a solvent molecule and removed. An exception was built into the algorithm to retain hydroxyl groups bonded to metal centers.

Retention of Charge-Balancing Ions (Text excerpt from Ref. [1])

Many MOF structures with associated charge-balancing ions also contain undesirable neutral solvent molecules. To discriminate between ionic species and neutral solvent molecules, the elemental compositions of the bonded components in a molecular graph of each structure were compared to the chemical formulas reported by the CSD using an in-house python script. The bonded components are the independent “molecules” within each structure; these include the MOF framework, the ionic species, and any neutral solvent molecules. The bonded components with elemental compositions matching the composition of the ions reported by the CSD were exempted from deletion in the solvent removal step.

Python Scripts & Example of Cleaning Procedure

To Be Updated

Hypothetical MOFs

Wilmer and co-workers at Northwestern University have created a database of 137,953 hypothetical Metal-organic frameworks . Current analysis of this database includes Grand Canonical Monte Carlo (GCMC) simulations to estimate the methane storage and working capacity of MOFs using Universal Force Field parameters (UFF). Among the top 300 hypothetical MOFs, Wilmer and co-workers identified target materials for synthesis and measured gas adsorption characteristics and found that excellent agreement between simulation and experiment. The hypothetical MOF database has also been used to define the structure-property relationship for CO2 and N2 separation , Xe/Kr separation , and hydrogen storage applications .

Building Blocks

A novel, bottom-up algorithm was developed to speed up structure enumeration. One hundred and two building blocks with varying degree of geometry and number of acids sites (e.g., COOH- sites) were used. The building blocks are divided into three main groups: metal nodes, organic linkers, and functional groups. Table 1 and Figure 1 summarizes the building blocks used in structure generation algorithm for Reference 1.

Types of Building Blocks Number of Connection Points Number of Building Blocks
Metal nodes 12 1
6 2
6 2
4 3
Organic Linkers 2 30
3 4
1 8
Functional Groups 1 13

Table 1: A summary of building blocks used in hypothetical MOF generation algorithm.

Figure 1: List of building blocks used in the hypothetical MOF database. Excerpt from Ref. [1].

To make the hypothetical MOF database relevant for hydrogen storage, only magnesium atoms were incorporated into the linker as a functional group, and new types of linkers were used as building blocks. The two hydrogen atoms on the phenyl ring group were substituted with two oxygen atoms and one magnesium atom. For each MOF structure, 0%, 50% and 100% magnesium functionalization was used. The list of building blocks and functionalization scheme is shown in Figure 2.

Figure 2: Building blocks used in Ref. 2. All organic linkers are the unfunctionalized versions. The linkers may also be terminated with nitrogen atoms instead of carboxylic acid groups where appropriate. The degree of functionalization of an organic linker is shown in the bottom. 50% functionalization signifies 50% of all possible functionalization sites are functionalized and 100% functionalization means 100% of all possible sites are functionalized. Reproduced from Figure S2 and S3 from Reference 2.

Generation Algorithms

The generation procedure creates hypothetical MOFs by recombining building blocks derived from crystallographic data of already synthesized MOFs. Atoms are grouped into building blocks based on reagents used in reported synthesis procedures as shown in Table 1. A building block can be combined with any other building block provided that the local geometry and chemical composition are the same as in crystallographically determined structure. Building blocks are combined in a piece-wise manner. When atomic overlap occurs at a particular step, a different building block or connection site is chosen until all possibilities are exhausted. Note that there is no force field (or quantum mechanical) energy minimizations involved; the pieces are connected according to geometric rules that govern how the building blocks are connected in already synthesized MOFs. Figure 2 summarizes the generation step discussed above.

Figure 2: A flowchart for the hypothetical MOF generation procedure. The upper and lower limits of i, j, k, and m refer to the numbered building blocks in Figure S2. In the “Select nth composition/arrangement” step, the total number and arrangement of building blocks is encoded in an enumerable string. In the particular library of building blocks used, functional groups could be connected in any location where a hydrogen atom is otherwise bonded to a carbon atom provided no atomic collisions occur. In the following “collisions between atoms” step, structures were considered colliding if any two atoms were closer than one angstrom. This distance was used so as not to discard potentially interesting MOFs due slight structural errors introduced in the generation process. Text and figure excerpt form the Supporting Information Figure S6 in Ref 1.

Hypothetical Zeolites

Zeolites are crystalline nanoporous material made from tetrahedrally coordinated silicon or alumnimum atoms connected by oxygen atoms. Zeolites are naturally occuring, but are usually produced synthetically for industrial applications in adsorption and catalysis. The International Zeolite Association database lists 218 silaceous zeolite structures that have been synthesized in the laboratory 6. Synthesis of new zeolite structures is an active area of research. Deem et al. generated a large database of hypothetical silica zeolite structures that could serve as targets for experimental synthesis 7,8. First, graphs of possible framework were enumerated by placing tetrahedral nodes (“T-atoms”) in all 230 symmetry groups over a wide range of lattice constants. These candidate structure were then annealed with the Sander-Leslie-Catlow interatomic potential to yield over 300,000 structures within 30 kJ mol-1 of quartz. The large discrepancy between the number of experimentally observed and hypothetical zeolites is an active area of research 9.

Other Hypothetical Materials

In addition to MOFs and Zeolites currently a set of 10,000 computational predicted porous polymeric networks (PPNs) are available.

Material Classes

Henry's Constants

Adsorption Isotherms

Heats of Adsorption

DDEC Point Charges

Pore Descriptors

P-XRD Patterns


  1. 1.0 1.1 1.2 1.3 Y.G. Chung, J. Camp, M. Haranczyk, B.J. Sikora, W. Bury, V. Krungleviciute, T. Yildirim, O. K. Farha, D. S. Sholl, and R. Q. Snurr, “Computation-Ready, Experimental Metal-Organic Frameworks: A Tool to Enable High-Throughput Screening of Nanoporous Materials,” Chemistry of Materials, 2014, 26 (21), 6185 – 6192