Supramolecular isomerism and structural flexibility in coordination networks sustained by cadmium rod building blocks

Bifunctional N-donor carboxylate linkers generally afford dia and sql topology coordination networks of general formula ML2 that are based upon the MN2(CO2)2 molecular building block (MBB). Herein, we report on a new N-donor carboxylate linker, β-(3,4-pyridinedicarboximido)propionate (PyImPr), which afforded Cd(PyImPr)2via reaction of PyImPrH with Cd(acetate)2·2H2O. We observed that, depending upon whether Cd(PyImPr)2 was prepared by layering or solvothermal methods, 2D or 3D supramolecular isomers, respectively, of Cd(PyImPr)2 were isolated. Single crystal X-ray diffraction studies revealed that both supramolecular isomers are comprised of the same carboxylate bridged rod building block, RBB. We were interested to determine if the ethylene moiety of PyImPr could enable structural flexibility. Indeed, open-to-closed structural transformations occurred upon solvent removal for both phases, but they were found to be irreversible. A survey of the Cambridge Structural Database (CSD) was conducted to analyse the relative frequency of RBB topologies in related ML2 coordination networks in order to provide insight from a crystal engineering perspective.


Crystal structures
Cd(PyImPr) 2 -2D-α Figure S1: Cd(PyImPr)2-2D-α as viewed along the c-axis. Hydrogen atoms have been omitted for clarity. Figure S2: Topological representation of Cd(PyImPr)2-2D-α with the points of extension -the C10 carboxylate carbon and the pyridine centroid in black and purple, respectively. These link to form edge-sharing octahedra shown in red that form 2D sheets.     Figure S8: Topological representation of Cd(PyImPr)2-hlz-α along the a-axis with the points of extension -the C10 carboxylate carbon and the pyridine centroid in black and purple, respectively. These link to form edge-sharing octahedra shown in red that form a 3D net resulting in the hlz topology. The angle formed between RBB units as measured using planes formed from the pyridine centroid points of extension is also shown. Figure S9: Topological representation of Cd(PyImPr)2-hlz-α at an offset from the a-axis with the points of extension -the C10 carboxylate carbon and the pyridine centroid in black and purple, respectively. These link to form edge-sharing octahedra shown in red that form a 3D net resulting in the hlz topology.

24.0°
Figure S10: Select close interactions between pore DMF molecules and the 3D framework in Cd(PyImPr)2-hlz-α. Figure S12: Topological representation of Cd(PyImPr)2-hlz-β along the a-axis with the points of extension -the C10 carboxylate carbon and the pyridine centroid in black and purple, respectively. These link to form edge-sharing octahedra shown in red that form a 3D net resulting in the hlz topology. The angle formed between RBB units as measured using planes formed from the pyridine centroid points of extension is also shown.

6.7°
Figure S13: Topological representation of Cd(PyImPr)2-hlz-β at an offset from the a-axis with the points of extension -the C10 carboxylate carbon and the pyridine centroid in black and purple, respectively. These link to form edge-sharing octahedra shown in red and that form a 3D net resulting in the hlz topology.

PXRD Data
Cd(PyImPr) 2 -2D Figure S15: Overlay of experimental and calculated PXRD patterns of Cd(PyImPr)2-2D-α and Cd(PyImPr)2-2D-β. Calculated PXRD patterns were generated from crystals collected at 100 K while experimental patterns were collected at RT. A discrepancy is notable for Cd(PyImPr)2-2D-α whereby the peak at 7.89° (corresponding to the 001 plane) has very low intensity in the calculated PXRD (red) pattern. This arises from the fact that the single crystal structure has the disordered solvent removed from the model by the 'SQUEEZE' function. Adding electron density to the pore results in a calculated PXRD (blue) with a larger 001 peak intensity which aligns better with the peak seen in the experimental PXRD (black).

Database mining
Example of shortcoming in ConQuest search function for periodic structures Figure S25: Example of shortcoming for ConQuest search whereby for a periodic structure, a search query must match a section of the diagrammatic representation of the structure which results in too broad a search if only a small section is used (as in search criteria 1), or the desired structure is not found at all if the search criteria goes beyond the diagrammatic representation (as in search criteria 2 and 3).

Mining Methodology
The datamining strategy used in this work to identify single-linker ML2 MOFs (M = metal ion, L = bifunctional N-donor carboxylate ligand) and identify their topologies is comprised of 6 steps schematically illustrated in Figure S26. Each unique refcode was treated as an individual entry.
Step 1. A list of MOFs containing 116981 refcodes was obtained from the Cambridge Structural Database, CSD 'MOF subset' (version: Sept 2022). 1 Step 2. Structures with ML2 stoichiometry were identified using an analysis of the compound names in the list from step 1. Compound names were obtained from the CSD using the Application Programming Interface (CSD Python API). 2 Analysis of the compound names was performed using custom-written Python script which implements the workflow briefly described below. Each compound name in the CSD can be broken down to parts corresponding to the linker, metal, and solvent. A few representative examples are provided in Figure S27 to illustrate this. Stoichiometric coefficients for the linker and metal were extracted from the name text and so stoichiometric linker to metal (L/M) ratio was determined. The list from step 1 was narrowed down to single-linker structures having L/M = 2 (6943 refcodes).
We note that structures wherein the compound names in the database do not follow this naming convention were not included in these results. For example, the hlz structure, Mn(3-(pyridin-4-yl)acrylic acid)2 (refcode: OTIQAL), has the compound name "catena-((µ-3-(pyridin-4yl)acrylato)-manganese dimethylformamide solvate)" in the CSD giving an L/M ratio of 1, despite in-fact having 2 ligands per metal. These structures have not been included to ensure consistency.
Step 3. Structures having M(N)2(O)2 coordination were identified using the ConQuest 3 query shown in figure S28. This narrows the list from step 2 to 2980 structures.
Step 4. Structures having linkers with 'N-donor' & 'carboxylate' groups both coordinated to different metal atoms were identified to find single-linker structures with bifunctional N-donor carboxylate linkers. This analysis was performed using custom-written Python script which implements the algorithm reported by us previously. 4 Additionally, exceptions were handled in the following way: Structures involving M 3+ ions involved additional counter anions attached to the metal and were excluded: ETALUJ, LIRCUN, LOTJUB, ZECGIB, ZECHEY, ZECHIC, ZECHOI, ZECHUO.
Structures based on racemate linkers were treated as single-linker ML2 structures and so were included in the results: AVOGID, SUCMEL, WUWVOA.
Due to single-crystal structure disorder in the linker, 29 structures were automatically identified by the code as being mixed-linker structures. Manual inspection confirmed these to be single-linker structures and so were included.
This resulted in 1138 structures being selected for further analysis.
Step 5. The list of refcodes from step 4 was matched with topologies reported in the TOPOS TTO database (version: Dec 2021) 5 using a custom-written Python script. Valence-bonded MOF topology determinations in either the standard or cluster representation were used. In the rare examples where the standard representation and the cluster representation resulted in differing topologies (as in AHIFAB, ARUWUH, BEPTUM, BEQJEP, DAFZAM, ECIHOQ, HUDQUT,  IWORET, MEDQOC, NUNVOI01, TUVYAL, WOFVET, XEWGEN) the standard representation was used as the "automated topology determination".
Step 6. The list of refcodes from step 4 was manually inspected to detect RBBs and assign updated topologies.
For structures with finite molecular building blocks (MBBs), this would typically equate to the metal centres and ligands or, where clusters were involved, points that link metal clusters together, such as the carboxylate carbons surrounding the copper paddlewheel MBB in HKUST-1. 6 In the case of RBBs, metal centres are linked through common points of extension that form the nodes for the resultant network topology. 7 For the ML2 bifunctional N-donor carboxylates discussed here, this would typically include the carboxylate carbons and N-donors of the linker. ( Figure S29) Net simplification and topology determination is then performed on Topospro, 5 giving the final topology.
Lead coordination compounds required additional consideration. A common coordination mode of Pb 2+ involves so-called hemi-directional 8 bonding wherein a Pb ion forms 6 shorter interactions and 2 marginally longer interactions. There is inconsistency in the literature and also in mercury as to whether the longer interactions should be considered as formal bonds (see NICOPB/NICOPB02/NICOPB04 and NICOPB01/NICOPB03). For the purposes of topology determination, all such interactions, where they are evident, are considered as formal bonds. Where this is present the row is marked with †.     ML 2 structures based on N-donor carboxylate linkers with edge-sharing octahedra RBBs Step 6 of database mining identified 165 ML2 refcodes based on N-donor carboxylate linkers and edge-sharing octahedra RBBs. Table S2 lists the structures with RBBs. The structures presented in this manuscript have been included at the end of the table.  **At the time of writing, the publication associated with NAFTIA mis-assigns atom types on the linker and so linker name was determined from the reaction procedure used.