Predicting Selectivity of Uranium vs. Vanadium from First Principles: Complete Molecular Design and Adsorption Modeling

Abstract

SUMMARY Background: Uranium is used as the basic fuel for nuclear power plants, which generate significant amounts of electricity and have life cycle carbon emissions that are as low as renewable energy sources. The extraction of this valuable energy commodity from the ground remains controversial, however, mainly because of environmental and health impacts. Alternatively, seawater offers an enormous uranium resource that may be tapped at minimal environmental cost. Currently, amidoxime polymers are the most widely considered adsorbent materials for large-scale extraction of uranium from seawater, but they are not perfectly selective for uranyl, UO22+. In particular, the competition between UO22+ and vanadium (VO2+/VO2+) cations poses a significant challenge to the efficient mining of UO22+. Thus, accelerating progress in the discovery and deployment of advanced materials for the recovery of uranium relies on the design of new ligands with high binding affinity and selectivity for uranium over competing metal ions. A cost-effective route to aid the discovery of new ligands is to apply computational methods to rapidly test attractive candidates and elucidate data-driven guidelines for rational design. Objectives: One of the key components in achieving rational design of highly selective ligands is the establishment of computational tools capable of assessing ligand selectivity trends. Therefore, the objectives of this study include: 1. Establish first-principles methods, based on computational chemistry techniques, to calculate stability constants for UO22+ and VO2+/VO2+ complexes. 2. Develop computational protocols to assess the binding strengths and selectivity of ligands that can be present in the actual poly(acrylamidoxime) adsorbents. 3. Develop adsorption models that can use information from first-principles computational methods to predict the adsorption behavior of uranium and vanadium by adsorbents synthesized at ORNL and compare results with experimental data. Results: In the first part of the study, we present an approach based on quantum chemical calculations that achieves high accuracy in reproducing experimental aqueous stability constants for UO22+ and VO2+/VO2+ complexes, providing the essential foundation for prospective screening of existing or even yet unsynthesized ligands with higher selectivity for uranium over vanadium. The developed computational protocols were used to assess the binding strengths and selectivity of aliphatic dicarboxylate ligands that can be present in the actual poly(acrylamidoxime) adsorbents. It was found that simple dicarboxylic functional groups possess low binding affinity and selectivity for uranyl, because their backbones present architectures that are poorly organized for the UO22+ complexation. In the second part of the study, adsorption models were developed and coupled with the results of the molecular studies in an effort to predict the adsorption of uranium and vanadium by the ORNL AF1 fiber adsorbent. These models can account for the effects of ligand surface density, specific surface area, surface charging, and other non-idealities of the adsorbent surface. It was found that by utilizing the reaction schemes and binding strengths proposed by the molecular studies, the adsorption model could accurately predict the uptake of uranium by both the acyclic amidoxime (AO) and cyclic imide dioximate (IDO) ligand in the ORNL laboratory studies, especially in the more neutral pH ranges (pH 5 to 9). The model, however, performed less adequately for predicting the uptake of vanadium for the same adsorbent. By exploring the causes behind the discrepancy between the adsorption model and the ORNL laboratory data, we found that the effect of surface charging was suppressing the total vanadium uptake predicted by the model. An investigation of literature revealed that the crystal structures of the 1:2 vanadium/IDO complex involved a sodium ion (Na+), which neutralized the charge of the adsorbed species. When this charge neutral species was included in the adsorption model, the predictions of the vanadium capacities were significantly improved across all pH ranges. This result suggests that, during adsorption, some surface species may closely associate with, or adsorb, counter-ions from solution to neutralize the charge of the adsorbent surface. Significance: This study is particularly significant when considering whether to produce an otherwise highly attractive ligand that may be synthetically challenging. If such a ligand is predicted by our calculations to achieve the desired uranium vs vanadium selectivity, this substantially reduces the risk of taking on such synthetic challenges. Moreover, the elimination of ligands that are unlikely to show a good uranyl binding affinity can release resources to focus on more promising UO22+- selective ligands. Furthermore, the results can successfully rationalize the experimentally observed loss in selectivity of amidoxime-based fibers as the number of adsorption/elution cycles increases, leading to the conversion of a significant amount of amidoxime to carboxylates. The obtained data enabled us to propose ligand design principles based on structural preorganization to achieve a dramatic enhancement in carboxylates binding affinity and selectivity toward UO22+ vs VO2+/VO2+. This concept was exemplified through the investigation of the complexes of the UO22+, VO2+, and VO2+ ions with the highly preorganized ligand PDA (1,10-phenanthroline-2,9-dicarboxylic acid) by a combination of quantum chemical calculations, along with fluorescence and absorbance techniques. The predictive power of the developed computational protocols was also demonstrated for amidoxime-type ligands, providing greater insights into new design strategies for the development of the next generation of adsorbents with high selectivity toward UO22+ over VO2+/VO2+ ions. In particular, our results suggest that the IDO functional group is the more preferable configuration for sequestration of uranium from seawater than the acyclic amidoxime (AO). However, at the same time IDO shows stronger binding affinity and higher selectivity for VO2+ over UO22+ and is likely responsible for the higher sorption of vanadium ions in marine tests, while AO does not appear to bind the VO2+ ions at all under seawater conditions. Furthermore, as indicated by the simulated speciation diagrams, vanadium forms complexes with cyclic imide dioximate over a large range of pH values, which may complicate efforts to strip vanadium from the fibers that are being developed to sequester uranium. Thus, selectivity of poly(acrylamidoxime) adsorbents toward UO22+ vs VO2+ could be improved by minimizing the formation of the cyclic imide dioximate moiety. Conclusions: Overall, the objectives of this milestone have been successfully met. We have established first-principles methods, based on computational chemistry techniques, to calculate stability constants for UO22+ and VO2+/VO2+ complexes. We have developed computational protocols to assess the binding strengths and selectivity of ligands that can be present in the actual poly(acrylamidoxime) adsorbents. Finally, we have developed adsorption models that can use information from first-principles computational methods to predict the adsorption behavior of uranium and vanadium by adsorbents synthesized at ORNL and compared results with experimental data. The culmination of molecular modeling studies and adsorption modeling techniques has shown significant promise with regards to ligand design and adsorbent performance prediction. We have demonstrated that the complexation reactions and binding strengths developed through molecular studies can yield the necessary parameters to predict the performance of a particular adsorbent when the appropriate structural parameters are also known (i.e., ligand surface density, specific surface area, mass, volume, etc.). Through these advancements, we now have a computational pathway to adsorbent fiber design, including the types of ligands to incorporate in the adsorbent and how much ligand should be on the surface versus the total available surface area of the fibers. Models developed here can also be used to predict the influence of environmental parameters such as temperature, pH, salinity, and concentration of competing ions on the uranium uptake performance of specific adsorbents. Thus, extensive deployment of the computational approaches presented in this report is expected to answer complex questions related to the extraction of uranium from seawater, with potentially superior adsorbent materials and performance, with respect to environmental parameters, as an end result.