Theoretical discovery of shape reactivity relationships in aluminum nanoclusters

Keggin-based aluminum nanoclusters have been noted to be efficient sorbents for the adsorption of arsenic, copper, lead, and zinc from water. Obtaining a molecular-level understanding of the adsorption processes associated with these molecules is of fundamental importance and could pave the way for rational design strategies for water treatment. Furthermore, due to their size and the availability of experimental crystal structures, Al nanoclusters are computationally tractable at the atomistic modeling level.

The adsorption of contaminants onto metal-oxide surfaces with nanoscale Keggin-type structural topologies has been established, but identification of the reactive sites and the exact binding mechanism are lacking. In more common surface studies the two main factors that affect reactivity have been found to be charge and functional group identity. Since Al nanoclusters each have a distinct shape we introduce the effects of shape as a third factor. In all the work presented in this dissertation, it is extremely apparent that the shape of the aluminum particle plays the most important role in nanoparticle reactivity studies.

We first focus on the reactivity of three aluminum polycations: [Al13O4(OH)24(H2O)12]7+ (Al13), [Al30O8(OH)56(H2O)26]18+ (Al30), and [Al32O8(OH)60(H2O)30]20+ (Al32). Using outer-sphere adsorption of sulfate and chloride as probe adsorbents, density functional theory (DFT) calculations determined that the reactivity can be represented as a function of particle topology, and not functional group identity or charge. Further exploring the shape-reactivity relationship of Al30 we reveal that cations and anions have opposing trends and ion reactivity can be generalized. It is determined that all cations favor the adsorption sites on the caps of Al30 and all anions favor adsorption in the beltway (middle) region. This result is supported by the visualization of the electrostatic potential of Al30 and three-dimensional induced charge density maps. The middle of the cluster is more positive than the caps, and this promotes anion adsorption in the beltway and cation adsorption on the caps.

Next we explore the reactivity of co-adsorption (outer-sphere anions and inner-sphere cations) onto Al30 through a collaborative approach. Al30 with two surface-bound Cu2+ cations (Cu2Al30-S) was experimentally crystallized in the presence of disulfonate anions; however, in the Cu2Al30-S structure the cations bind to the beltway region of the cluster. Using DFT we determined that the counter anions play a key (and governing) role in the crystallization of Cu2Al30-S. This result that outer-sphere adsorption dictates inner-sphere adsorption does not appear in surface calculations, it is unique to Keggin studies.

Seeing that all anions favor adsorption to the beltway region and all cations favor adsorption to the cap region we set out to determine if any reactivity patterns can be reversed. In order to do this inner-sphere As(V) and P(V) adsorption is modeled onto Al30 through another collaborative approach. The experimental crystal structure of (TBP)2[Al2(μ4-O8)(Al28(μ2-OH)56(H2O)26)]14+ (where TBP = t-butylphosphonate (CH3)3CPO3) has been synthesized, and using DFT calculations we can alter the R-group of P(V) or the DFT As(V) analogue to see if the inner-sphere anion ever prefers to bind to the cap region instead of the beltway. We observe that no matter the intrinsic properties of the R-group the anion always prefers to bind to the beltway region, which once again shows that the shape-reactivity relationship plays a major role in Keggin based structure reactivity.

Since As(V) is such a harmful ion we extend our As(V) adsorption studies to aluminum surfaces. As(V) has been experimentally shown to bind to aluminum surfaces in a bidentate binuclear configuration. By modeling a variety of configurations we can confirm and explain that the bidentate binuclear configuration is most stable due to the least amount of strain on the As(V) atom. Aluminum surfaces are common DFT models to study but are computationally expensive, due to this fact some people choose to model small Al octahedral cluster models instead. Comparing the reactivity of both systems we see a significant difference in energy magnitudes and ranges and can conclude that small Al octahedral cluster models cannot take place of aluminum surfaces.

All in all, the work presented in this dissertation provides an important contribution in our understanding of Keggin based Al compounds. Keggin based compounds are very sparsely studied computationally and this work helps to fill a knowledge gap. Hopefully the insights obtained from this work can help guide future Keggin based studies.