We present a method for quantifying the accuracy of extended X-ray absorption fine structure (EXAFS) fitting models. As a test system, we consider the structure of bare Au147 nanoparticles as well as particles bound with thiol ligands, which are used to systematically vary disorder in the atomic structure of the nanoparticles. The accuracy of the fitting model is determined by comparing two distributions of bond lengths: (1) a direct average over a molecular dynamics (MD) trajectory using forces and energies from density functional theory (DFT) and (2) a fit to the theoretical EXAFS spectra generated from that same trajectory. Both harmonic and quasi-harmonic EXAFS fitting models are used to characterize the first-shell Au-Au bond length distribution. The harmonic model is found to significantly underestimate the coordination number, disorder, and bond length. The quasi-harmonic model, which includes the third cumulant of the first-shell bond length distribution, yields accurate bond lengths, but incorrectly predicts a decrease in particle size and little change in the disorder with increasing thiol ligands. A direct analysis of the MD data shows that the particle surfaces become much more disordered with ligand binding, and the high disorder is incorrectly interpreted by the EXAFS fitting models. Our DFT calculations compare well with experimental EXAFS measurements of Au nanoparticles, synthesized using a dendrimer encapsulation technique, showing that systematic errors in EXAFS fitting models apply to nanoparticles 1-2 nm in size. Finally we show that a combination of experimental EXAFS analysis with candidate models from DFT is a promising strategy for a more accurate determination of nanoparticle structures.

The objective of the research described in this Account is the development of high-throughput computational-based screening methods for discovery of catalyst candidates and subsequent experimental validation using appropriate catalytic nanoparticles. Dendrimer-encapsulated nanoparticles (DENs), which are well-defined 1-2 nm diameter metal nanoparticles, fulfill the role of model electrocatalysts. Effective comparison of theory and experiment requires that the theoretical and experimental models map onto one another perfectly. We use novel synthetic methods, advanced characterization techniques, and density functional theory (DFT) calculations to approach this ideal. For example, well-defined core@shell DENs can be synthesized by electrochemical underpotential deposition (UPD), and the observed deposition potentials can be compared to those calculated by DFT. Theory is also used to learn more about structure than can be determined by analytical characterization alone. For example, density functional theory molecular dynamics (DFT-MD) was used to show that the core@shell configuration of Au@Pt DENs undergoes a surface reconstruction that dramatically affects its electrocatalytic properties. A separate Pd@Pt DENs study also revealed reorganization, in this case a core-shell inversion to a Pt@Pd structure. Understanding these types of structural changes is critical to building correlations between structure and catalytic function. Indeed, the second principal focus of the work described here is correlating structure and catalytic function through the combined use of theory and experiment. For example, the Au@Pt DENs system described earlier is used for the oxygen reduction reaction (ORR) as well as for the electro-oxidation of formic acid. The surface reorganization predicted by theory enhances our understanding of the catalytic measurements. In the case of formic acid oxidation, the deformed nanoparticle structure leads to reduced CO binding energy and therefore improved oxidation activity. The final catalytic study we present is an instance of theory correctly predicting (in advance of the experiments) the structure of an effective DEN electrocatalyst. Specifically, DFT was used to determine the optimal composition of the alloy-core in AuPd@Pt DENs for the ORR. This prediction was subsequently confirmed experimentally. This study highlights the major theme of our research: the progression of using theory to rationalize experimental results to the more advanced goal of using theory to predict catalyst function a priori. We still have a long way to go before theory will be the principal means of catalyst discovery, but this Account begins to shed some light on the path that may lead in that direction.

Bimetallic PdPt dendrimer-encapsulated nanoparticles (DENs) having sizes of about 2 nm were synthesized by a homogeneous route that involved (1) formation of a Pd core, (2) deposition of a Cu shell onto the Pd core in the presence of H2 gas, and (3) galvanic exchange of Pt for the Cu shell. Under these conditions, a Pd@Pt core@shell DEN is anticipated, but detailed characterization by in-situ extended X-ray absorption fine structure (EXAFS) spectroscopy and other analytical methods indicate that the metals invert to yield a Pt-rich core with primarily Pd in the shell. The experimental findings correlate well with density functional theoretical (DFT) calculations. Theory suggests that the increased disorder associated with <~2 nm diameter nanoparticles, along with the relatively large number of edge and corner sites, drives the structural rearrangement. This type of rearrangement is not observed on larger nanoparticles or in bulk metals.

To create truly effective electrocatalysts for the cathodic reaction governing proton exchange membrane fuel cells (PEMFC), namely the oxygen reduction reaction (ORR), necessitates an accurate and detailed structural understanding of these electrocatalysts, especially at the nanoscale, and to precisely correlate that structure with demonstrable performance enhancement. To address this key issue, we have combined and interwoven theoretical calculations with experimental, spectroscopic observations in order to acquire useful structural insights into the active site geometry with implications for designing optimized nanoscale electrocatalysts with rationally predicted properties. Specifically, we have probed ultrathin (∼2 nm) core-shell Pt∼Pd9Au nanowires, which have been previously shown to be excellent candidates for ORR in terms of both activity and long-term stability, from the complementary perspectives of both DFT calculations and X-ray absorption spectroscopy (XAS). The combination and correlation of data from both experimental and theoretical studies has revealed for the first time that the catalytically active structure of our ternary nanowires can actually be ascribed to a PtAu∼Pd configuration, comprising a PtAu binary shell and a pure inner Pd core. Moreover, we have plausibly attributed the resulting structure to a specific synthesis step, namely the Cu underpotential deposition (UPD) followed by galvanic replacement with Pt. Hence, the fundamental insights gained into the performance of our ultrathin nanowires from our demonstrated approach will likely guide future directed efforts aimed at broadly improving upon the durability and stability of nanoscale electrocatalysts in general.

Here we outline a new method for synthesizing fully reduced Pt dendrimer-encapsulated nanoparticles (DENs). This is achieved by first synthesizing Cu DENs of the appropriate size through sequential dendrimer loading and reduction steps, and then galvanically exchanging the zerovalent Cu DENs for Pt. The properties of Pt DENs having an average of 55, 140, and 225 atoms prepared by direct chemical reduction and by galvanic exchange are compared. Data obtained by UV–vis spectroscopy, X-ray absorption spectroscopy, X-ray photoelectron spectroscopy, and high-resolution electron microscopy confirm only the presence of fully reduced Pt DENs when synthesized by galvanic exchange, while chemical reduction leads to a mixture of reduced DENs and unreduced precursor. These results are significant because Pt DENs are good models for developing a better understanding of the effects of finite size on catalytic reactions. Until now, however, the results of such studies have been complicated by a heterogeneous mixture of Pt catalysts.

We demonstrated the capability of combined EXAFS and DFT calculations for characterizing the structural and thermal properties of Au147 clusters.

The oxygen reduction reaction (ORR) is a major factor that drives galvanic corrosion. To better understand how to tune materials to better inhibit catalytic ORR, we have identified an in silico procedure for predicting elemental dopants that would cause common, natively formed titanium oxides to better suppress this reaction. In this work, we created an amorphous TiO2 surface model that is in good agreement with experimental radial distribution function data and contains reaction sites capable of replicating experimental ORR overpotentials. Dopant performance trends predicted with our quantum chemistry model mirrored experimental results, and our top three predicted dopants (Mn, Al, and V, each present at doping concentrations of 1%) were experimentally verified to lower ORR currents under alkaline conditions by up to 77% vs the undoped material. These results show the robustness of calculated thermodynamic descriptors for identifying poor, TiO2-based ORR catalysts. This also opens the possibility of using quantum chemistry to guide the design of coating materials that would better resist the ORR and presumably galvanic corrosion.

a b s t r a c t 6 a r t i c l e i n f o 7 8Available online xxxx This paper summarizes several studies correlating the structure and function of nanoparticle catalysts. Three 17 types of alloy nanoparticles are considered, random alloy, core@shell and alloy-core@shell structures. In the 18 first two cases, the focus is to build theoretical models to understand previous experimental results. In the latter 19 case, calculations play a greater role in leading the development of nanoparticle catalysts. We demonstrate that 20 iteration between theory and experiment can facilitate an understanding of nanoparticle catalysts and reduce the 21 time and effort involved in the design of new catalysts. 22