Here, we describe the data sources for the crystal structures in the QMOF Database.
Most materials in the QMOF Database are taken from the Cambridge Structural Database (CSD) with their free solvents removed. To determine which materials in the CSD are sufficiently MOF-like, we used the list of materials in the CSD MOF Subset and the 2019 CoRE MOF Database. Unless otherwise stated, these structures were taken directly from the CSD so that we could use the valuable CSD meta-data to choose which materials to discard in our pre-screening process. Note that given the loose definition of a MOF in the CSD MOF Subset, some of these materials may be better classified as coordination polymers. You may wish to filter the materials by void fraction or pore size if you are specifically interested in porous materials.
All materials taken from the CSD have _FSR at the end of their name. The source flag is set to 'CSD'.
Since v8 of the QMOF Database, some MOFs were taken directly from the 2019 CoRE MOF Database uploaded on Zenodo. This was done in one of two ways. In the first approach uploaded with v8, a curated subset of CoRE MOFs discussed in Kancharlapalli et al. was adopted to reduce the likelihood of obtaining erroneous crystal structures. In the second approach uploaded with v9, a curated subset of CoRE MOFs presented in Chen and Manz was adopted, after removing entries with CSD-identified charge-balancing ions, to reduce the likelihood of obtaining erroneous crystal structures. In both cases, only the FSR (free solvent removed) version of the CoRE MOF Database was used.
All materials taken directly from the CoRE MOF Database have core at the start of their name. The source flag is set to 'CoRE'. Note that there are some materials in the CoRE MOF Database that were instead downloaded from the CSD (as described above) rather than taken from the CoRE MOF Database itself. These structures will not have core in their filenames. Cross-check the refcodes in the QMOF Database with the list of structures in the parent CoRE MOF Database if you are specifically looking for all CoRE MOFs.
A few pyrene-containing MOFs were taken from prior work using the data uploaded to the Materials Cloud.
All materials taken from this dataset have pyrene in their name. The source flag is set to 'Pyrene'.
Since v6 of the QMOF Database, we have now included hypothetical MOFs from various sources.
Several of the MOFs in the database were constructed using the ToBaCCo code described in a series of prior studies here and here. These are described below.
A database of ToBaCCo-constructed Zr-MOFs was adopted from prior work by Andereson and Gómez-Gualdrón. See here for the dataset. We also exchanged the Zr species for Hf to include hypothetical Hf-MOFs as well.
All materials taken from this dataset have tobacco and _SR_ in their names. The source flag is set to 'Anderson'.
A database of Cu triangle MOFs was adopted from prior work by Colón, Gómez-Gualdrón, and Snurr. These structures had H atoms in the center of the Cu triangles, which were removed before relaxing their structures with DFT. See here for the dataset.
All materials taken from this dataset have tobacco in their names. The source flag is set to 'ToBaCCo'.
Several of the MOFs in the database were constructed using the TOBASCCO code as described in prior work by Boyd and Woo. These were obtained from prior work by Boyd et al. using the Materials Cloud dataset here. We made modifications to several of these MOFs prior to structure relaxation to diversify our dataset. For instance, we occasionally exchanged the metals in the inorganic node, and we constructed Al rod MOFs by exchanging the metals in the pre-existing V rod MOFs and protonating the bridging oxo ligands.
All materials taken from this dataset have boydwoo in their names. The source flag is set to 'BoydWoo'.
Hypothetical MOFs were also taken from the genomic MOF (GMOF) database here. All materials taken from this dataset have gmof in their names. The source flag is set to 'GMOF'.
Several of the MOFs in the database were obtained from prior work by Maciej Haranczyk and coworkers.
Hypothetical MOF-5 analogues were obtained from prior work by Haranczyk and colleagues. See here for the dataset.
All materials taken from this dataset have MOF-5 in their names. The source flag is set to 'Haranczyk_MOF5'.
Hypothetical MOF-74 analogues were obtained from prior work by Haranczyk and colleagues.
All materials taken from this dataset have mof74 in their names. The source flag is set to 'Haranczyk_MOF74'.