Metal oxo clusters, from theory to innovation; Synthesis, mechanism & novel application in recyclable polymers

The world is currently facing major challenges such as climate change, which can be tackled by developing new materials that improve on current processes. One area showing great potential is the nanoparticle field which has been developing rapidly over the past 30 years. A nanoparticle can be thought of as a hybrid inorganic-organic object. The inorganic core dictates the physical properties such as e.g. luminescence, while the organic ligand shell provides colloidal stability and solubility. These materials are promising candidates for catalysis due to their high surface to volume ratio. However, size control remains one of the major challenges in this field. In the best case, these nanoparticles have a polydispersity of 5 %, which means that particles with an average size of 5 nm will also have particles with sizes of 4.75 and 5.25 nm. At the same time as the nanoparticle field, also metal oxo clusters were being reported in the literature. These materials are very similar to the previously mentioned nanoparticles, as they are also hybrid objects consisting of a core and a ligand shell. However, they are usually smaller and have the added advantage of being atomically precise. The latter means that their polydispersity is zero, making them excellent building blocks.
However, these metal oxo clusters were mainly characterized using single crystal XRD to obtain structural data. This limited the possible synthesis to short and rigid ligands and introduced tedious and long crystallization processes. These limitations are the reason why this field has been dormant for the last 1-2 decades. Due to the clear advantage of having atomically precise building blocks, we sought to revive this field, by developing a new characterization toolbox that eliminates the need for crystallization. We first optimized the synthesis, after which the formation mechanism was studied. Using the knowledge gained from these projects, the clusters have been developed and used as tunable inorganic monomers in both free radical polymerizations and covalent adaptable networks.
Firstly, the cluster synthesis was standardized as the reaction conditions in literature were quite divergent. We found that when a metal (Zr or Hf) alkoxide is reacted with 8 equivalents acetic acid, the M12-acetate cluster is consistently formed. After optimizing this for a short carboxylic acid, we performed the same reaction with longer carboxylic acids, similar to the nanoparticle field. After purification, we elucidated their structure via PDF measurements, thus eliminating the need for crystalline material. Fitting the data proved that clusters are formed regardless of the carboxylic acid used during synthesis. The dimerization of the clusters is controlled only by the sterical hindrance on the alpha-position of the carboxylic acid, not by it's length. If there is something different from a -CH2 on this position the monomeric M6 clusters will form, else the dimeric M12 clusters are formed. Through ligand exchange under the appropriate conditions (vacuum at 70 °C) it was possible to convert monomers into dimers and vice versa. The organic ligand shell was further characterized using NMR, FTIR and TGA. We found that, on top of the coordinated ligand shell, which display different binding modes (bridging & chelating), additional H-bonded ligands are present. Applying our toolbox for hafnium clusters we confirmed that the same conclusions are valid. Finally, we tested Zr12-oleate as catalysts for the esterification reaction between oleic acid and ethanol, since they can be seen as the smallest possible nanoparticle. 5 nm ZrO2 nanoparticles have been used successfully as esterification catalysts in the past. Interestingly, our clusters showed a 5-fold increase in reaction conversion due to their increased surface to volume ratio, creating a better, cheaper and more sustainable catalyst.
Secondly, the formation mechanism was studied. Using NMR and FTIR we learned that the first 2 equivalents of carboxylic acid exchange with Zr(OPr)4. It appears that this exchange does not go towards completion but is an equilibrium. Only when the third equivalent is added a signal for free acid appears together with an ester signal. In situ EXAFS taught us, despite the large error, that the Zr complex after exchange with 1 equivalent of acid is most likely a dimer, while the 2 equivalent sample seems to fit a trimeric structure. By following the ester formation over time by NMR, while varying multiple reaction conditions, we found that the Zr concentration should be high in order to have sufficient ester formation. Increasing the length of the carboxylic acid and/or alkoxide or adding sterical hindrance has a strong negative effect on the ester formation. The Zr-Zr degeneracy, which is 4 in the final cluster, increases simultaneously with the ester formation. Finally, by combining our data a preliminary reaction mechanism was proposed where the initial ligand exchange is followed by a fast ester formation, after which a slow ester formation occurs and finally the Zr6 cluster is formed.
Zr12-oleate and -linoleate clusters were used unsuccessfully as tunable inorganic monomers for radical polymer synthesis. Under our current conditions, reacting the clusters with AIBN, dicumyl peroxide or without initiator, no polymer networks were formed. It is possible that some low molecular weight polymers were formed but since our objective was to create a polymer network this was not investigated further. Instead, we used 10-undecenoic acid as a ligand, which has a terminal alkene. By reacting these clusters with 10 w% dicumyl peroxide solid polymer networks with excellent insoluble fractions were obtained. This result shows us that the alkene functionalities in oleic and linoleic acid are shielded by the remaining ligand tail, inhibiting successful polymerization. We then switched from alkene ligands to (meth-)acrylate ligands which are more reactive towards free radical polymerization. Using our previous knowledge, samples were synthesized with different amounts of reactive ligands on the surface. It was found that samples where the clusters contained 6 reactive ligands or more on the surface resulted in good insoluble fractions, indicative of a polymer network. When fewer reactive ligands were present on the surface, the insoluble fractions were too high. Whether the high insoluble fractions are due to low cluster functionalization or low cluster loading, is still unclear. Remarkably, the Tg did not change significantly despite the large variation in sample composition, not for the alkene-cluster networks nor the (meth-)acrylate-cluster networks. For the (meth-)acrylate capped clusters, polymer samples synthesized from mono-2-(acryloyloxy)ethyl succinate containing clusters show the most promising features. However, further research should be done to mechanically characterize these materials.
Finally, the clusters were used as tunable monomers in covalent adaptable networks. The cluster surface was functionalized with different amounts of custom-made epoxy ligands, after which the clusters were reacted into a polymer network. It was found that the Tg ranges from approximately -5 °C to 40 °C and the insoluble fractions were good for almost all samples. The samples were able to relax stress very rapidly, but in a dissociative manner contrary to the intended associative transesterification. We postulate that a ligand exchange on the cluster core is responsible for the fast relaxation of the polymer networks. So far, no clear trend could be observed upon changing the polymer composition by adding co-monomer, different amounts of catalyst or different amount of epoxide ligands. However, DMA measurements showed that the addition of the clusters, even in small amounts, had a positive effect on the mechanical properties. So not only did we switch the reversible chemistry from a transesterification towards a ligand exchange mechanism, but we also improved the mechanical properties of the materials.