Abstract
Reactive nitrogen is becoming a prevalent and threatening pollutant in the biosphere, with the rate of pollution continuing to increase. Thus, an efficient method for denitrification is essential. One promising method is the use of electrocatalysis to remove nitrate, a common form of reactive nitrogen, from water. CuPd nanoparticles (clusters) have shown great promise as a catalyst for selective nitrate electroreduction to dinitrogen (N2), a benign form of nitrogen. In order to enable the use of CuPd catalysts as a denitrification method, their catalytic ability must be improved. For this to happen, we must understand more about two key factors: the mechanisms of nitrate reduction toward different products, and the structure of the CuPd cluster catalysts. These are the two broad aims of the research presented in this thesis.
First, using density functional theory, we determine the mechanisms by which NO electroreduction products are formed, on planar and stepped transition metal surfaces, as the reduction of NO is key to determining the product selectivity of the overall nitrate reduction reaction. There are a number of possible reduction products in addition to N2, with the common products on most transition metals being ammonia (NH4+), hydroxylamine (H3NOH+), and nitrous oxide (N2O), which are the products considered here. In addition, the theoretical onset potentials for each product on the different metals are calculated, and scaling relations and a limiting potential volcano are constructed to investigate the trends across the range of metals.
On the planar metal surfaces, it was found that the mechanism toward ammonia on the majority of metals goes via an NOH intermediate, while for hydroxylamine both NOH and HNO are key intermediates. In addition, it was found that the onset potential for NO reduction is not strongly influenced by the energy of N binding, and thus is relatively invariant across the range of metals. For nitrous oxide formation, two mechanisms are considered: an Eley-Rideal mechanism is found to be possible on all of the metals studied, while a Langmuir-Hinshelwood mechanism was only found to be possible on Ag, Au and Cu. The influence of the choice of solvation correction, surface NO coverage, and exchange-correlation functional were also investigated, and were found to have minimal influence on the overall results and trends.
Following on from the planar surfaces, stepped metal surfaces were investigated in order to determine the impact of surface morphology on the reaction mechanisms and onset potentials. The steps represent a more complex surface geometry, and have been shown to have higher catalytic activity than planar surfaces, and are potentially present on clusters. The largest impact of the surface morphology was on the mechanisms by which the products are formed; for ammonia and hydroxylamine, reaction pathways via both HNO and NOH intermediates are competitive, whereas for planar surfaces, NOH was more favoured. For nitrous oxide formation, it was found that both the Langmuir-Hinshelwood and Eley-Rideal mechanisms are plausible for most metals, which is a distinct difference from the planar surfaces, where the Eley-Rideal mechanism was the only plausible mechanism on most metals. The onset potentials and scaling relations were found to be relatively insensitive to the surface morphology, and were found to be similar to the planar surfaces.
On transition metals, formation of N2 in significant amounts is only observed on Pd. Here we investigate the mechanism of N2 formation on Pd, and the adsorption and reaction of relevant species, in comparison with Pt, a metal that does not form any significant amount of N2, despite its similarity to Pd for many other catalytic applications. It was found that the potential dependent adsorption of hydrogen is a key difference on the metals; on Pt, hydrogen may inhibit the adsorption and formation of key precursors to N2 formation, while on Pd, this inhibition is unlikely to occur.
Understanding the structure of CuPd clusters is essential to understanding their nitrate reduction ability, due to the intrinsic link between a cluster's structure and catalytic ability. Here we use a genetic algorithm, with energies calculated using an empirical potential, to investigate the structure of CuPd clusters at multiple sizes and Cu:Pd ratios. First, we consider 38 atom clusters, and explore the influence of the chosen parameters in the empirical potential. It is found that using multiple parameterisations across multiple genetic algorithm runs improves the search for low energy cluster structures, and increases the chance of success in finding the global minimum energy structure. At the 38 atom size, it is found that for compositions close to either Cu38 or Pd38, a structure similar to the monometallic cluster structure, a truncated octahedron, is favoured. For intermediate compositions (Cu5Pd33 to Cu25Pd13), a "pancake" icosahedron (a type of poly-icosahedron) is most stable, which is a structural motif that is particularly stable for bimetallic clusters, and is potentially interesting for catalytic purposes due to its unique structure, especially compared to the monometallic clusters.
We then consider larger clusters, extending the work from 38 atom clusters to include clusters up to 309 atoms. At these larger sizes, it was found that the overall trends are similar to those seen for 38 atoms; the most stable clusters in the Pd- and Cu-majority compositions are found to have a structure similar to the monometallic cluster structure and in the intermediate composition range are found to adopt poly-icosahedral structures. It was also found that in Pd-majority regions, Cu atoms occupied the core or vertex positions, while in intermediate regions, the segregation was mixed, however in the Cu-majority regions, the Pd atoms occupied the faces of the clusters. Given that similar structures and segregations were seen for CuPd clusters across multiple sizes, the structures found here may be representative of experimental CuPd clusters in this size range, and therefore provide an excellent platform for future catalytic studies.