Metal-Organic Frameworks (MOFs) are well known for their gas adsorption qualities, gained from their exceptional porosities and high internal surface areas. along with the ability to tune MOFs with reticular chemistry. To understand how gases bind within the MOFs a variety of techniques can be used. Gas adsorption isotherms and infra-red spectrometry can provide information on bulk behaviour and binding affinities but single-crystal X-ray diffraction (scXRD) can provide detailed information on the exact mechanisms and geometries of gas binding. However, owing to their high porosities MOFs are typically poorly diffracting and as the gas loading is done in situ a gas cell is required which interferes with the x-ray beam. Therefore, obtaining data of high enough resolution and intensity to model gas binding requires synchrotron radiation. In our case we used the excellent resources available at Diamond Light Source, U.K., on beamline I-19.

In this study we looked at two MOFs, Ni-CPO-27 and the closely related Co-4,6-dihydroxyterephthalic acid. (Co-4,6-dhip). They both have 1.D. hexagonal channels and a high density of open metal sites that make them excellent candidates for the storage of gases that bind to metals. However, these open metal sites don’t like being coordinatively unsaturated and rapidly adsorb water whenever possible, making activation of the MOF to remove residual water an important first step. Activation requires high temperatures, low pressures and patience, with several hours needed to activate the samples. To do this, the crystal is mounted within a gas cell and put under vacuum while an N₂ cryostream provides temperature control. Activation has to be performed on the beam line as removing the sample from vacuum for even a short period causes water to be readsorbed. Even then, during the experiment we could not fully dehydrate these MOFs, with 6% water occupancy the best we could achieve in the time available. We also found that morphology can play a part in the ease of activation, with thin needle like crystals proving harder to dehydrate than broader plates, indicating that the hexagonal channels of the two MOFs run down the length of the needles.

Once the MOFs are activated they can be loaded with gases. Our gases of choice were nitric oxide (NO) and carbon monoxide (CO). These are both toxic gases in high concentrations but in small amounts they have useful medical properties, being important signalling molecules in mammalian physiology. By studying the MOFs under atmospheres of NO and CO, at two different temperatures, we could learn a lot about their binding. Both MOFs bind the gases to their open metal sites, Figure 1, with the different metals and linkers seeming to have very little effect on the gas binding at the 2.5 bar pressure we used. NO binds via the N atom, retained a bond length similar to that of free NO and bound in a bent fashion. CO binds in a linear fashion through the C atom and retains a bond length similar to that of free CO. We also found that CO binds more weakly as it is less able to compete with water adsorption. This indicates neither gas undergoes any rearrangement of electrons, with both simply binding through the lone pairs one would predict from a molecular orbital model of the two molecules. When it came to modelling the disorder of the bound gases there were some differences. NO could be modelled with the oxygen split into 5 positions at 450 K but only 4 positions at 300 K. The disorder in CO could only be sensibly modelled with one large oxygen position, suggesting it is highly disordered at the temperatures used in this study. 

These results not only provide us with valuable information on the MOF gas interaction, but they form part of a dataset from which it should be possible to build new computer models to better predict how gases will interact with MOFs. This in turn should allow us to compute what MOFs will be the best at adsorbing these gases, and therefore improve the efficiency of screening the large number of possible MOF structures.

For more detail on the experiment and results, please read our paper published in Communications Chemistry.