Bonding properties of thiolate-protected gold nanoclusters and structural analogs from X-ray absorption spectroscopy

1 Introduction to gold nanoclusters

The chemical and optical properties of gold have captivated humankind for hundreds of years. Modern day research has focused on nanosized gold (e.g., gold nanoparticles) for applications in areas such as medicine and catalysis [1, 2]. Gold has robust chemical properties in bulk form and in nanosize form, making it a suitable element to study the synthesis and properties of nanostructured materials [3]. Remarkably, immense progress has been made in the last 20–30 years to attain a more intimate knowledge of gold nanoparticle structure-property relationships. The versatility of nanostructured gold has been demonstrated through the synthesis and fabrication of various 1-D, 2-D, and 3-D nanomaterial morphologies (e.g., nanowires, nanoprisms, and nanocubes) leading to size- and shape-specific properties [4, 5]. Nevertheless, there is still a persistent motivation to understand the origin of gold nanomaterial properties on a fundamental level. A new and prominent vein of research includes the study of quantum-sized or ultra-small gold nanoparticles, known as gold nanoclusters [6]. On this size regime, the electronic and bonding properties are more molecule-like in nature, leading to unexpected chemical and physical properties [7].

Gold nanoclusters (Au NCs) are composed of only 10s or 100s of Au atoms, which typically have a particle diameter of <2 nm. Au NCs are often synthesized with a monolayer of ligands on the surface that protect the Au core [8, 9]. Although Au NCs can be formed with weaker protecting ligands (e.g., biomolecules, dendrimers, polymers), this review will focus on highly stable, small ligand-protected Au NCs. Recent synthetic strategies adapted from the widely used two-phase Brust-Shiffiren [10] method have been successful in obtaining single-size thiolate-protected Au NCs (the chemical formula [Au_n(SR)_m]^q is often used to describe the composition of single-sized Au NCs products; where n is number of Au atoms, m is the number of thiolate ligands (SR), and q is the total valence charge on the cluster, if any) [11]. As a result of their small size, discrete valence electronic states are found for Au_n(SR)_m instead of a quasi-continuous 5d band commonly seen for larger plasmonic Au NPs [6]. Therefore, electronic transitions exist between HOMO-LUMO levels, leading to the appearance of molecule-like properties (e.g., luminescence [12–14]). These unique properties are promising for widespread application in fields such as nanocatalysis [15] and bioimaging [16]. Interestingly, certain sizes of atomically precise Au NCs have been shown to follow a magic number series or superatom electronic shell-closing model providing a theoretical basis for their extraordinary stability [6, 8, 17, 18].

Despite the diverse range of protecting ligand types suitable for Au NC synthesis, thiolate ligands have demonstrated excellent stability and versatility in surface functionalization [17]. Since the elucidation of the Au₁₀₂(SR)₄₄ NC in 2007 [19], the catalog of possible Au core geometries and stabilizing surface structures from solved crystal structures and DFT predictions has grown significantly [20]. To date, icosahedral, face-centered-cubic (FCC)-ordered and smaller Au core (<13 Au atoms) structures have been reported crystallographically for various Au_n(SR)_m sizes [19, 21–30]. A couple of selenolate-protected Au NCs have also been recently elucidated with icosahedral or small Au₈ core structures [31, 32]. In the last few years, a number of surface structures in addition to the monomeric and dimeric staple-like motifs were identified such as a trimeric staple-like motif [28], tetrameric staple-like motif [29, 30], μ³ sulfide coordination [25], bridging thiolate motif [24], and ring-like motifs (long Au-SR oligomers) [27, 33]. Figure 1 presents the diversity of thiolate-protected Au NC structures. Currently, only the dimeric staple-like motif and ring-like motif have been confirmed to exist in structurally analogs selenolate-protected Au NCs [31, 32].

Figure 1:

Elucidated Au_n(SR)_m nanoclusters by X-ray crystallography (large yellow – core Au, small yellow – staple Au, red – S). Icosahedral-based: (A) Au₁₀₂(SR)₄₄, (B) Au₃₈(SR)₂₄, (C) [Au₂₅(SR)₁₈]^-,0, FCC based: (D) Au₃₆(SR)₂₄, (E) Au₃₀S(SR)₁₈, (F) Au₂₈(SR)₂₀, small or miscellaneous Au cores: (G) Au₂₄(SR)₂₀, (H) [Au₂₃(SR)₁₆]^-, (I) Au₂₀(SR)₁₆, and (J) Au₁₈(SR)₁₄. Carbon and hydrogen atoms are removed for clarity.

Uncovering the crystal structure of atomically precise Au NCs can provide important insights on the electronic structure, stability, and unique coordination environments of Au. Nevertheless, there is still plenty to uncover on the structural and electronic properties of Au NCs in response to chemical or physical processes such as ligand exchange, heteroatom metal doping, catalytic reactivity, and interactions/stability in complex biological systems. Intimate knowledge of these structure-property relationships could be vital in order for Au NC applications to come to fruition. Additionally, it is worthwhile to study Au NCs with well-defined core and surface structures in order to help predict the existence of such features in newly isolated Au NCs without an obtainable crystal structure. This significant obstacle originates from the inherently small nature of the Au NC size regime. The lack of a repeating gold lattice in Au NCs and the challenge of growing single crystals prevent or limit X-ray scattering and electron microscopy techniques from obtaining crystal packing information of the Au core. Additionally, the organoligand-Au interface is essentially undetectable with such techniques. A promising experimental approach in recent years has been to utilize X-ray absorption spectroscopy (XAS) to investigate the local structure of Au NCs from an element- and site-specific perspective.

There are several advantages for using XAS including the measurement of subnanometer particle sizes (or seemingly amorphous materials), variable experimental conditions (solution-phase, temperature, in situ redox), multiple absorption edges for different elements (Au, S, Se, etc.) can be probed, and data analysis can reveal both metal-metal and metal-ligand structural environments. This review will summarize the capabilities of using XAS to study the structural and electronic properties of atomically precise, thiolate-protected Au NCs. Recently synthesized Au_n(SR)_m NC structural analogs containing selenolate and tellurolate ligands will also be covered. A comprehensive review on some of our group’s thiolate-protected Au NC research was recently published [34], where XAS and/or XPS studies on Au₁₄₄(SR)₆₀, Au₃₈(SR)₂₄, Au₃₆(SR)₂₄, Au₂₅(SR)₁₈, Au₂₄Pt(SR)₁₈, and Au₁₉(SR)₁₃ NCs have been discussed. Before presenting some of the current XAS work in the community of thiolate-protected Au NC research, a brief description of the X-ray absorption process and data analysis is given first.

2 Introduction to X-ray absorption spectroscopy

XAS is inherently a local structure experimental technique commonly used to probe the coordination number and bond distance of the nearest atomic neighbors to the absorbing element. The X-ray absorption measurement is conducted in a relatively straightforward manner where modulation in the absorbing atom’s X-ray absorption coefficient is measured and plotted as a function of incident X-ray energy. The first region of the XAS spectrum, known as X-ray absorption near-edge structure (XANES), is where the incident X-ray energy is increased from below the desired absorbing atom’s core level (e.g., Au L₃-edge (2p_3/2) at 11,919 eV, known as the absorption edge energy (E₀)) to about 40–50 eV above. A few eV after the absorption edge energy, the first commonly observable feature is historically known as the white-line intensity. This transition occurs due to the promotion of excited core electrons to unoccupied valence levels via a dipole-allowed transition [35]. The white-line intensity can, therefore, be used quantitatively or qualitatively to determine the electronic structure of the valence level for the absorbing element. The intensity of the white-line feature is an average of all the absorbing elements in the measured sample, so it does not offer a site-specific perspective of the valence electronic structure. Near-edge features following the white-line can also be indicative of the absorbing atom coordination environment or geometrical orientation of nearest neighboring atoms and other available electronic transitions due to neighboring atoms. As the energy of the incident X-rays increases past the near-edge region of the spectrum, the excited core electrons will leave the absorbing atom in the form of a photoelectron wave. Emitted photoelectron waves will backscatter off electron shells of neighboring atoms with some photoelectron waves returning to the absorbing atom, quenching the core-hole excited state. Again, XAS is inherently a local structure technique as the emitted photoelectron wave will decay over time and distance due to inelastic losses. Therefore, interpretation of the nearest neighbors is largely limited to the first few scattering shells around the absorbing atom. As the energy is modulated, the photoelectron waves will experience constructive and deconstructive interference due to various backscattering paths. This will, in turn, influence the X-ray absorption coefficient of the element over the energy range measured creating the oscillatory part of the XAS spectrum known as the extended X-ray absorption fine structure (EXAFS) region. A rigorous mathematical treatment of the X-ray absorption process can be found in other works [35, 36].

Post-edge oscillations (often isolated as k-space spectra) can be further analyzed to determine the contribution of different scatterers to the overall EXAFS signal. The local scattering is typically represented in radial space (R-space) by applying a Fourier transform (FT) to a desired range of the k-space. This provides a qualitative view of the local structure environment around the absorbing atom without fitting the EXAFS data. Although it is possible to obtain structural details of the sample by fitting longer-range single scattering or multiple scattering features in the EXAFS spectrum [37], this can be difficult for Au NC samples due to the small particle size and short-range order. In this review, single scattering shells are mainly interpreted for various metal-ligand and metal-metal scattering paths. Modern EXAFS fitting methods use theoretical scattering paths generated from ab initio calculations to obtain backscattering amplitude functions and phase-shift values for different scattering environments in order to represent the major scattering shells in the EXAFS data [38, 39]. Using a theoretical scattering path to fit the EXAFS data, quantitative structural parameters can then be extracted from the EXAFS equation to provide the coordination number (CN), bond distance (R), and the Debye-Waller factor, a measure of static and thermal disorder (σ²=σ²_static+σ²_thermal). One caveat of EXAFS fitting is the limited number of independent parameters available to work with. Meaning, not every unique scattering path can be incorporated into the fit as each scattering path will contain up to four free-running parameters (aforementioned CN, R, and σ², along with ΔE₀). In order to maximize the number of independent parameters, a long k-range can be selected for the FT to R-space given that the late k-space oscillations are distinct and not overruled by experimental noise. As an aside, if a multi-shell EXAFS analysis is to be conducted on a Au NC sample, where weaker and long distance single scattering paths are used to fit the data (e.g., aurophilic interactions), the XAS experiment should be conducted at low temperature (∼77 K) and with a large beam spot on the sample to enhance the late k-space oscillations and avoid beam damage, ensuring the EXAFS data is reproducible. A wide R-space fitting window can also be used to increase the number of independent parameters. However, it is recommended that the number of free-running parameters should be less than the total number of independent parameters, if not around half of the number of independent parameters.

There are two general approaches that can be applied to fit the EXAFS of Au_n(SR)_m NCs and other metal NC systems. Au L₃-edge EXAFS will be focused on, which is more commonly probed. If the total structure of the Au NC has not been solved crystallographically, the number of scattering shells used to fit the data will be limited. It is difficult to reliably determine what range of bonding types the scattering shell will encompass (center to surface and/or surface to surface) without knowledge of the core or surface structure. Thus, two scattering shells, Au-ligand and Au-Au core interactions, can be incorporated confidently without contribution from longer Au-ligand, longer Au-Au, or multiple scattering paths. The EXAFS-determined CN values for each scattering shell provide structural information on the size and relative thiolate:Au ratio. If the Au_n(SR)_m NC composition is known from mass spectrometry (MS) measurements, CN and R values for each shell can be used to help infer more local structural details. The second EXAFS fitting method expands on the interpretation of Au_n(SR)_m NC local structure by using an available total structure model to identify distinct scattering paths that account for core, surface, and possibly aurophilic interactions (two to three Au-Au scattering shells total). Scattering shells can be assigned confidently if there is a clear separation in bond distance for each scattering environment. For example, Au₂₅(SR)₁₈ NCs have Au-Au interactions on the surface around 2.95 Å and between surface and staple sites around 3.16 Å. The bond length distribution for each shell is completely separated to ensure little overlap of different scattering environments. The EXAFS-determined R values for each shell should then be compared with the crystal structure model to ensure the correct site-specific representation of each distinct scattering path. This second fitting method allows interpretation of Au_n(SR)_m NC EXAFS data from a site-specific perspective. With a highly pure sample and known structural framework, CN values can be calculated and fixed in the EXAFS fitting analysis to lower the number of free-running parameters. ΔE₀ parameters for similar scattering paths can also be correlated to minimize the number of free-running parameters, if needed. This multi-shell fitting methodology was proven to be effective for detecting the structural response of Au₂₅(SR)₁₈ NCs to solvent and temperature by monitoring changes in bond length and disorder [40]. See Figure 2 for a simulated Au L₃-edge EXAFS spectrum, indicating each distinct scattering shell that can be interpreted in the EXAFS spectrum of Au₂₅(SR)₁₈ NCs. Au-Au CN values are also extremely sensitive to a small change in Au NC composition given their subnanometer size [41]. As the EXAFS signal is averaged over all absorbing atoms, larger Au NPs will not see such a dramatic change in Au-Au coordination with a small change in composition due to the higher amount of atoms in the core versus on the surface. A comprehensive review on the geometrical and compositional characterization of NPs from EXAFS can be found by Frenkel et al. [42].

Figure 2:

Distinct Au-Au bonding environments of the Au₂₅(SR)₁₈ NC (top) and simulated EXAFS spectrum for Au₂₅(SR)₁₈ (phase-corrected) with scattering features identified for each site-specific bonding type (bottom). Reproduced with permission from Ref. [40].

3 Thiolate-protected gold nanoclusters

Atomically precise phosphine- and halide-protected Au NCs first sparked an interest in many researchers for studying the composition-dependent properties of Au NPs. The famous Schmid cluster, Au₅₅(PPh₃)₁₂Cl₆ [43] was one of the first size-specific Au NCs studied with XAS in the 1990s [44–46]. Au L₃-edge EXAFS fitting results from a few studies suggested the Au core is likely cuboctahedral instead of icosahedral. Further analysis of the Au₅₅(PPh₃)₁₂Cl₆ with EXAFS by Cluskey et al. suggested a possible arrangement of phosphine and chloride ligands on the surface [44]. Marcus et al. probed the thermal vibrational properties using temperature-dependent XAS measurements [46]. This early use of XAS to study Au₅₅ proved to be a beneficial approach for investigating the structure and properties of ultra-small Au NCs. Nevertheless, the synthesis of thiolate-protected gold nanoparticles had made exceptional progress through the late 1990s and early 2000s with the Brust-Schiffrin synthetic procedure [10]. This breakthrough led to the successful synthesis and crystallization of two highly stable, thiolate-protected Au NCs, Au₁₀₂(SR)₄₄ and Au₂₅(SR)₁₈, reported in 2007 and 2008, respectively [19, 22, 23]. Attention had then shifted to the exciting revelation of icosahedral core structures and staple-like Au-SR oligomer motifs on the surface (shown in Figure 1).

Au_n(SR)_m NCs are being recognized as promising catalysts for a variety of oxidation/reduction reactions [15, 47, 48]. For larger metal NP catalysts, XAS characterization has been utilized to detect changes in surface local structure after catalytic treatments [49, 50]. Such structural information is crucial for identifying and further understanding the catalytically active locations on the surface of the metal NP or NC. A study by Shivhare et al. demonstrated that partially or completely removing the protecting ligands from alkanethiolate-protected Au₂₅ NCs (supported on carbon powder) by calcination increased the catalytic reduction of 4-nitrophenol [51]. Au L₃-edge EXAFS spectra for Au₂₅(SC₈H₉)₁₈ and Au₂₅(SC₆H₁₃)₁₈ were collected before thermal treatment and fitted using four shells to confirm the core and surface local structure of the NC samples. Results were consistent with a previous EXAFS study on Au₂₅(SC₈H₉)₁₈ [40]. TEM was used in conjunction with Au L₃-edge EXAFS measurements (temperature-dependent EXAFS spectra are depicted in Figure 3) to monitor particle growth with thermal treatment. It was found from EXAFS fitting results that Au₂₅ NCs undergo partial cluster sintering at a mild calcination temperature of 150°C where the Au-S CN decreased, and the core Au-Au bond length increased to 2.85 Å from 2.76–2.80 Å at room temperature. At higher temperatures of 200°C the Au-Au CN increased to nine and the Au-S CN decreased to 0.6–0.8, for both thiolate-protected Au₂₅ NCs. The small increase in Au NC particle size and EXAFS fitting results suggest there was partial removal of surface thiolate ligands and small Au core growth around 200°C. It is around this temperature that Au₂₅ NCs showed the highest rate constant for the reduction of 4-nitrophenol. This finding supports the hypothesis that the Au core surface is an important location for catalytic activity. It was also evident by qualitative comparison of the two thiolate-Au₂₅ systems in Figure 3 that phenylethanethiolate ligands are thermally removed before hexanethiolate ligands, indicating the type of thiolate ligand could impact the catalytic activity of thermally treated Au_n(SR)_m NCs. A similar study by Liu et al. compared the activity of [Au₂₅(PPh₃)₁₀(SC₁₂H₂₅)₅Cl₂]²⁺ and [Au₂₅(SC₈H₉)₁₈]^- (Au₂₅(SR)₁₈ before and after calcination on a SiO₂ support [52]. Au L₃-edge EXAFS results for Au₂₅(SR)₁₈ on SiO₂ indicated the thiolate ligands were mostly removed by 200°C in air and by 300°C under He gas. From their analysis of the Au-Au EXAFS fitting results, they suggested that the Au core grew after the removal of thiolate ligands with the Au-Au bond distance remaining around 2.82–2.84 Å, while remaining much shorter than bulk Au (2.88 Å). After performing the styrene oxidation reaction with Au₂₅(SR)₁₈, the Au-Au CN was found to decrease only slightly from 9.5 to 8.5. Monitoring the structure of Au_n(SR)_m NCs in situ catalysis reaction or post-catalytic treatment with EXAFS should be considered in future work in this area as the local structure and element-specific perspective from XAS can be used to help detect small changes on the Au NC surface or in the core.

An alkanethiol ligand-exchange-induced formation of Au₃₈(SR)₂₄ NCs from polydisperse Au_n(SG)_m NCs (SG – glutathione) was reported by Stellwagen et al. [53]. The authors used FT-IR to confirm that all tested alkanethiol molecules over various lengths were bound to the Au NC; moreover, in an all-trans configuration. Au L₃- and S K-edge XANES results (shown in Figure 4) were used to further demonstrate that all lengths of tested alkanethiol ligands were completely ligand exchanged. From the Au L₃-edge XANES, mainly the white-line feature can be interpreted to deduce the occupation of 5d electronic states, but for S K-edge, the near-edge region is sensitive to more changes in the structural and electronic environment, which can be observed through the E₀ shift, white-line intensity, and post- and pre-edge features. Formation of S-metal bonds produces a noticeable change in the near-edge region due to the appearance of vacant electronic states from the metal atom. For example, quadrupole transitions from orbital mixing of S 1s and the vacancies in the nd level of the bonded metal atom (e.g., 5d from Au) can be observed as pre-edge features in S K-edge XANES [54]. All thiolate-protected Au₃₈ NCs showed a broadening of the white-line feature from the pure thiol reference due to the formation of S-Au interactions but were all still narrower than the white-line feature for Au₂S reference, indicating the more molecule-like Au(I)-SR interactions in staple-like motif structures. Therefore, the first S K-edge XANES feature can serve as a quick indication of a thiolate or sulfide environment. Anomalously, all alkanethiolate-protected Au₃₈ NCs except for hexanethiolate-protected Au₃₈ exhibited a pre-edge shoulder. This anomaly could be related to impurities, which the authors pointed out with discussion of weak UV-Vis absorption features. This pre-edge feature for S K-edge has been shown to be size dependent in other work and likely related to the size of the Au core [55], but was not further inspected in this work by Stellwagen et al. Overall, S K-edge near-edge features and the relative white-line intensity from Au L₃-edge XANES supported the presence of staple-like Au-SR motifs and charge transfer from Au to S for varied lengths of alkanethiolate-protected Au₃₈ NCs.

Figure 3:

Temperature-dependent Au L₃-edge EXAFS of phenylethanethiolate-Au₂₅ (A) and hexanethiolate-Au₂₅ (B) NCs supported by carbon powder. Reprinted with permission from Ref. [51].

Figure 4:

S K-edge XANES spectra of n-alkanethiolate-Au₃₈ NCs after thiol-etching of polydisperse Au_n(SG)_m NCs. Reproduced with permission from Ref. [53].

On the topic of Au-SR surface interactions, Au L₃-edge EXAFS was employed by Tsukuda et al. to investigate a unique binding motif for sterically demanding thiolates for Au₄₁(S-Eind)₁₂ (S-Eind-1,1,3,3,5,5,7,7-octaethyl-s-hydrindacene-4-thiol) [56]. The total structure of Au₄₁(S-Eind)₁₂ was not elucidated in this work or in previous work. Thus, only a two-shell EXAFS fit was conducted to probe the relative Au core size and the amount of thiolate-Au bonding. A high Au-Au CN was found from EXAFS fitting of Au₄₁(S-Eind)₁₂ (4.1±0.4), which is much higher than Au₃₈(SC₁₈H₃₇)₂₄ (1.7±0.2) and, in fact, more comparable to Au_∼43 NCs (4.4±0.4) stabilized by weakly coordinating polyvinylpyrrolidone (PVP). A comparison of each R-space spectrum is shown in Figure 5. This EXAFS fitting result of the main Au-Au scattering shell suggested most of the Au atoms are consolidated in the Au core with few, if any, Au atoms held away from the Au core in Au-SR oligomer-protecting structures (i.e., staple-like motifs). A μ³-like bonding motif was proposed, considering the average CN for the Au-S scattering shell (0.9±0.4) and composition of Au₄₁(S-Eind)₁₂, where each sterically demanding S-Eind ligand is bonded to three Au surface atoms on average. It should be noted that a μ³-like bonding motif was recently elucidated on the Au₃₀S(SR)₁₈ NC, where only one sulfur atom interacts with the Au NC surface, but as a sulfide instead of a thiolate [25]. Finally, although each Au NC measured in this work has a similar number of Au atoms (Au₃₈, Au₄₁, and Au_∼43), it is clear that the amount of ligand-Au interactions dramatically affects the scattering peak intensities in the EXAFS spectrum, as demonstrated in Figure 5.

Figure 5:

Sterically demanding thiol ligand (A) used to protect Au₄₁(SR)₁₂ NCs. Au L₃-edge EXAFS comparison (B) of similar size Au NCs but with varying ligand-Au interfaces. Reproduced with permission from Ref. [56].

A recent study on the identification of a highly luminescent Au₂₂(SG)₁₈ NC (SG – glutathione) was reported by Yu et al. [57]. The strong luminescence property was thought to originate from longer Au-SR oligomeric motifs on the NC surface based on previous studies of aggregation-induced emission [58]. Au L₃-edge XAS analysis was used to help predict the structural environment of Au₂₂(SG)₁₈ in combination with a proposed DFT-optimized model. Of the generated DFT-optimized structures, a Au₂₂(SR)₁₈ NC with one RS-[Au-SR]₃ and one RS-[Au-SR]₄ motif were predicted to interlock with each other on each end of a prolate Au₈ core (four Au-SR oligomeric structures total). EXAFS-determined CN values from the Au-S scattering shell and the shortest Au-Au scattering shell were consistent with the proposed model within the reported fitting uncertainties. The average Au-Au bond length in the predicted Au₈ core was found to be 2.67(1) Å, which is significantly shorter than Au-Au bonding in most known Au core structures. Together, MS and DFT/EXAFS characterization provided convincing evidence for a Au₈ core with interlocking Au-SR oligomers as the strongly luminescent Au(SR) NC product.

A more recent study on the bonding properties of Au₂₈(SR)₂₀ NCs from XAS was reported by our research group. Temperature-dependent bonding properties of Au₂₈(SR)₂₀ were examined by fitting Au L₃-edge EXAFS data measured at 90 and 300 K. At both measured temperatures, short Au-Au interactions related to Au₄ core structures remain around 2.74 Å in distance. A second, longer Au-Au scattering shell representing Au-Au interactions between tightly bonded Au4 units and on the surface of the Au core fitted in the EXAFS spectra appeared to contract at the 300 K (from 2.99 (3) to 2.93 (3) Å). A similar thermal contraction property was observed for the larger FCC-ordered relative, Au₃₆(SR)₂₄. Au₂₈(SR)₂₀ NCs were found to have a highly occupied 5d electron valence state compared with similar-sized Au₂₅(SR)₁₈, consistent with the previous XAS study on Au₃₆(SR)₂₄ when compared to similar-sized icosahedral-based Au₃₈(SR)₂₄ [59]. This XAS work on FCC-ordered Au_n(SR)_m NCs communicates the apparent influence of smaller Au₄ core structures on the electronic and temperature-dependent bonding properties of FCC-ordered Au_n(SR)_m NCs. It is anticipated that XAS elucidation of these unique structural and electronic characteristics could be used to distinguish FCC-ordered Au NCs from icosahedral-based Au NCs when the total structure is unknown. Importantly, this should be tested with additional FCC-ordered Au NCs to confirm the potentially identifiable characteristics of Au₄ units.

3.1 Metal dopants in thiolate-protected Au NCs

Correlating the structural and electronic properties of Au NCs with their catalytic activity and optoelectronic characteristics is undoubtedly an important step forward in preparing atomically precise Au NCs for application purposes. In addition to modifying the ligand, surface structure or size, an interesting area in Au(SR) NC research is tuning the electronic properties by introducing a metal dopant, usually noble or late transition metals. Although a number of heteroatoms can be added to the Au NC composition, identifying the most thermodynamically favorable site in Au NCs for a single heteroatom is an intriguing and important question from the nanochemists’ perspective. Few heteroatom-doped Au NCs also represent some of the smallest nanoalloys synthetically possible [60, 61]. With only a few crystal structures of heteroatom-doped Au NCs presently known [62], the preferred location of the metal dopant and its effect on the Au NC structure can be investigated through XAS measurements. Conveniently, XAS experiments can probe both the Au L₃-edge and the L₃- or K-edge of a 3d, 4d, or 5d heteroatom to provide complimentary data, pinpointing the location of the dopant site (center, surface, or staple). Here, a few studies on heteroatom-doped thiolate-protected Au₂₅ are summarized.

The dopant location and electronic properties of monopalladium-doped Au₂₅ NCs (Au₂₄Pd(SR)₁₈) were investigated with Pd K-edge XAS, using [197] Au Mossbauer spectroscopy for complementary data of the Au local structure [63]. By comparing the Pd K-edge EXAFS oscillation pattern of the sample with the theoretical pattern of each dopant location scenario (Pd in the center, surface, or staple, as seen in Figure 6) and fitting the EXAFS data, it was clear that Pd was located in the center position with a high Pd-Au CN of 10.7. Consistent with the Au₂₄Pt(SR)₁₈ sample studied by Christensen et al. [65], both d⁹ metals caused a small contraction of the icosahedral core where the center M-Au bond length averages ∼2.75 Å. XANES spectra at both absorption edges were used to deduce intracluster charge transfer from the central Pd site to surrounding Au by observing a decrease in Au L₃ white-line intensity and a positive shift in Pd K-edge E₀ position relative to bulk Pd. In the latter experimental observation, a small shift of the core electron-binding energy (shown by plotting the first derivative of Pd foil, Au₂₄Pd, and Na₂PdCl₄) to higher-energy supports valence electrons are transferred to neighboring Au atoms.

Figure 6:

Possible metal dopant locations in Au₂₅(SR)₁₈ NCs (top). Cu K-edge k-space (A), FT-EXAFS (B) and XANES (C). Experimental data for Au_23.6Cu_1.4(SR)₁₈ at (a) 300 K and (b) 8 K along with simulated spectra for (c) center, (d) surface, and (e) staple dopant sites. (XANES spectrum (a) is from Cu foil.) Reproduced with permission from Ref. [64].

Au₂₅ NCs doped with Cu or Ag atoms were most recently studied using XAS in conjunction with DFT calculations [64]. Au_23.8Ag_1.2(SR)₁₈ and Au_23.6Cu_1.4(SR)₁₈ NCs were synthesized (protected by phenylethanethiol), having compositions very close to single atom-doped Au₂₅ NCs. First, DFT-optimized structures were determined for each system (Au₂₄M(SR)₁₈, where M is Ag or Cu) and for each possible dopant site location (center, surface, and staple). From these models, Ag K-edge and Cu K-edge EXAFS spectra were simulated to compare with experimental data using a R² factor method to determine the agreement between simulated and experimental EXAFS. For the monosilver system, EXAFS fitting results and comparison to simulated EXAFS of a Ag surface site dopant site for Au₂₅ suggested Ag prefers the surface of Au₂₅ from agreeable Ag-Au and Ag-S CN values. The EXAFS-determined Ag-S bond length increases to 2.443 Å, consistent with an increase from the DFT model. For the monocopper system, both EXAFS and XANES results (shown in Figure 6) from the Cu K-edge perspective show the Cu atom occupies the staple site in Au₂₅ NCs instead of the surface or core. Convincingly, only Cu-S bonding around 2.2 Å was evident from the fit of the FT-EXAFS spectrum. As can be seen from these summarized Au₂₄M(SR)₁₈ studies, XAS is an invaluable characterization technique for determining the favorable location of various metals and could be extended to other metal-based NC systems besides Au₂₅. Moreover, this latter study exemplifies the useful collaboration of DFT structural optimization, XAS simulation, and experimental XAS studies to elucidate the prominent metal dopant site in Au(SR) NCs.

3.2 Selenolate- and tellurolate-protected analogs

Some studies have suggested that a heavier chalcogen head group (e.g., Se) could improve the stability of ligand-protected Au NCs and NPs [66, 67]. The isolation and structural characterization of dodecaneselenolate-protected Au₃₈ clusters (Au₃₈(SeC₁₂H₂₅)₂₄) was reported first by Kurashige et al. [68]. In this work, the local structure of Au₃₈(SeR)₂₄ was mainly characterized by Au L₃-edge EXAFS, as shown in Figure 7. Au₃₈(SC₁₂H₂₅)₂₄ and Au foil were used to compare with selenolate-protected Au₃₈ NCs. The scattering shell for Au-Se was shifted by about 0.1 Å (fitted Au-Se bond distance of 2.43 Å) and much broader compared to a Au-S scattering shell from Au₃₈(SC₁₂H₂₅)₂₄, likely due to the larger atomic radius of Se. Despite the overlap in Au-Se and Au-Au scattering, fitting the Au-Au scattering shell indicated a Au-Au core distance of 2.77 Å and average CN value similar to thiolate-protected Au₃₈, indicating a similar structural framework to well-defined Au₃₈(SR)₂₄ NCs. Au₂₅(SeC₁₂H₂₅)₁₈ NCs were also measured with XAS in this work showing a shifted, broader Au-Se scattering shell with comparable EXAFS fitting results to Au₂₅(SR)₁₈. EXAFS results from both NC sizes infer the Au core structure in each selenolate-protected Au NC does not significantly expand, contract, or change composition. Nevertheless, additional Au-Au scattering shells were not used to inspect the surface and aurophilic interactions of selenolate-protected Au₂₅ and Au₃₈. Interestingly, the white-line intensities of selenolate-protected Au₃₈ and Au₂₅ were, in fact, very near the intensity of bulk Au (as shown in Figure 7) indicating selenolate ligands dramatically affect the 5d valence electronic structure of Au, which may help account for enhanced stability of the Au NC.

Figure 7:

Au L₃-edge EXAFS and XANES of Au₃₈(SeR)₂₄ with Au₃₈(SR)₂₄ and Au foil for reference. Reproduced with permission from Ref. [68].

Intriguing bonding behavior of benzeneselenolate-protected Au₂₅ NCs (Au₂₅(SeR)₁₈) was reported by our research group [69]. Au L₃-edge and Se K-edge XAS results were used to investigate the local structure and temperature-dependent bonding properties of Au₂₅(SeR)₁₈ NCs with the XAS data collected presented in Figure 8. Similar to the dodecanceselenolate-Au₂₅ and -Au₃₈ studied by Kurashige et al. [68], Au-Se bonding produced a broad scattering feature that partially overlapped with Au-Au scattering. Comparing EXAFS oscillations in the k-space at 50 K (Figure 8, top left), late k-space (8–13 Å^-1) oscillations were significantly more intense for the selenolate-protected Au₂₅, potentially from more tightly ordered Au-Au bonding. An understanding of the Au local structure was extended in this work by employing a four-shell EXAFS fit following a previously established fitting approach for Au₂₅(SR)₁₈ NCs [40]. At 50 K, Au-Se and core Au-Au bond distances from EXAFS fitting were consistent with Kurashige et al. [68]. The second Au-Au scattering shell indicated that the Au-Au interactions on the surface of the Au₁₃ core were comparable to the thiolate-protected Au₂₅. However, fitting the third Au-Au scattering shell (from staple Au to surface Au) gave a long Au-Au distance suggesting substantially weaker aurophilic interactions compared with Au₂₅(SR)₁₈. When Au₂₅(SeR)₁₈ NCs were measured at 300 K, all three site-specific Au-Au bonding environments show a thermal contraction, while Se-C bonding appeared to expand with increasing temperature (Se K-edge EXAFS) (Figure 8 bottom right). Unexpected temperature-dependent trends of the Debye-Waller factors for each Au-Au shell were also noted. Molecular dynamics simulation of the Au₂₅(SePh)₁₈ cluster at 300 K showed staple Au is drawn closer to the core by a conformational change in the dimeric staple structure, providing a potential mechanism of thermal contraction. For the electronic properties, the white-line intensity for Au₂₅(SePh)₁₈ at the Au L₃-edge was surprisingly even lower than bulk Au. From the Se K-edge perspective, an increase in the white-line intensity and small positive shift of the E₀ energy demonstrates the electron transfer from 4p when bonding with Au. As few studies have investigated the Se K-edge XANES for Au NCs, ab initio simulations of the near-edge region and density of states for Au, Se, and C sites were conducted to identify the origin of near-edge transitions. Important near-edge features include contributions from Se p-DOS, Se d-DOS, and C p-DOS from the organoselenolate ligand.

Figure 8:

Au L₃-edge EXAFS in (A) k-space and (B) R-space for Au₂₅(SeR)₁₈ and Au₂₅(SR)₁₈. Temperature-dependent (C) Au L₃-edge and (D) Se K-edge EXAFS for Au₂₅(SeR)₁₈. Reproduced with permission from Ref. [69].

Thus far, only one XAS study has been conducted on tellurolate-protected Au NCs. Kurashige et al. demonstrated that benzenetellurol (HTePh) could be partially exchanged onto Au₂₅(SR)₁₈ forming various mixed ligand products of [Au₂₅(TePh)_m(SR)_18-m]^- where m=∼1, ∼3, and ∼7 [70]. Unique to tellurolate-protected Au NCs, no sign of oxidized telluride was found from Te K-edge XAS results, whereas oxides have been detected on larger tellurolate-protected Au NPs. From various EXAFS fitting approaches (two- and three-shell fitting), Au-Te bonds were found to exist around 2.58–2.59 Å in distance. By including one Au-Au scattering shell to account for shorter Au-Au interactions in the Au₁₃ core, EXAFS fitting results indicated the Au-Au distance and coordination to be similar to pure thiolate-protected Au₂₅ crystal structure and only slightly lengthened compared to the measured Au₂₅(SR)₁₈ reference. Selenolate- and tellurolate-protected Au NCs are excellent materials for XAS characterization as high-quality EXAFS oscillations can be collected at Se K- and Te K-edges, gaining or confirming the local structure of the Au-ligand interface.

4 Outlook and summary

It is anticipated that the use of XAS for the study of Au NC structural and electronic properties will continue to increase in popularity. It is challenging to predict the core and surface structure of Au NCs and to interpret the electronic properties with XAS measurements alone. As seen from the reviewed studies, having the support of MS, DFT ab initio simulations and/or single crystal X-ray structure, can provide the researcher with a basis to test the EXAFS fitting with various structural models. Mass spectrometry results can help account for the fitted CN values, based on the determined composition. With DFT-predicted structural models and ab initio XAS simulations, distinct scattering paths can be identified and utilized for EXAFS fitting purposes. The combination of MS, DFT, and EXAFS can be powerful when predicting the structure of the studied Au_n(SR)_m NCs, but should be conducted with caution if there is no clear favorable structural isomer from DFT calculation. Unfixed Au-S or Au-ligand CN values from the EXAFS fit are a good first indication of the similarities between the data and the predicted model. In any case, the crystallographic total structure of a Au_n(SR)_m NC is the most useful piece of information that can be used to explore the capabilities of EXAFS fitting. Importantly, if two scattering shells with the same absorbing and backscattering atoms are used to fit the experimental EXAFS data, the researcher should provide a means to clearly explain the scattering environment for each path used to fit the EXAFS. With confident assignment of each scattering path and enough representative scattering paths to account for the core and surface structure, the study of Au_n(SR)_m NC bonding properties can be significantly enriched, allowing further exploration of the Au_n(SR)_m NC structural response to a number of experimental conditions available at synchrotron beamline facilities.

EXAFS fitting methodologies and XAS experiments developed from studies on Au_n(SR)_m NCs can easily be extended to other metal-based NC systems. Therefore, there will be a strong need for unifying or creating a standard fitting procedure for the study of NC systems so that results can easily be compared across various reports. Careful and well-designed XAS experimentation will hopefully lead to an improved understanding of important mechanisms concerning Au NC formation, solvent-cluster interactions, ligand exchange, and interactions with biological systems. These fundamental questions that have surrounded Au_n(SR)_m NCs for many years could be answered in full, or in part, with XAS experimentation or characterization. Select studies on Au NCs that utilize the versatility of the experimental conditions or the in situ capabilities of an XAS experimental setup are briefly discussed below.

The role of the solvent type on the structure of Au NCs and smaller Au NPs has been studied recently with XAS [71, 72]. Through EXAFS analyses of Au NCs or Au NPs in various solvent conditions, such works have been able to identify solvent-induced transformations of the Au core structure or the thiolate surface coverage. On the formation of Au NCs and NPs, in situ or quick-XAS experiments are promising for observing the nucleation and subsequent directed growth of Au particles. Monitoring the reduction of Au³⁺ in the presence of thiol molecules via Au L₃-edge XANES and EXAFS by Ohyama et al. has shown evidence of Au₄ clusters in the early stages of nucleation [73]. Using the weaker protecting ligand PVP, Yao et al. observed the formation of Au_nCl_n+x complexes in the initial nucleation stage of Au NCs or Au NPs [74]. Further interpretation of their in situ EXAFS data leads to kinetic insights on three different stages of Au NC formation. Another step forward in EXAFS characterization of Au NC surface structure or the thiolate-Au interface is possible by combining DFT-molecular dynamics simulations and ab initio calculation of EXAFS spectra. Yancey et al. took this experimental-theoretical approach to study the surface structural disorder of Au_∼147 clusters protected by dendrimer molecules with increasing amounts of terminal thiols (12, 24, 50, and 72 thiols per Au_∼147 cluster) [75]. Remarkable similarities were reported for each Au_∼147 cluster with varied thiolate-Au interactions. This approach opens up new opportunities to more carefully interpret experimental EXAFS data for Au NCs with unknown surface and core structures. Finally, advanced XAS-related techniques such as X-ray excited optical luminescence [76] and high-energy resolved fluorescence detection [77] may offer new insights into the element-specific luminescence and time-resolved high-resolution valence electronic structure of Au(SR) NCs, respectively. Both techniques are employed when the absorbing atom’s near-edge region is probed and can be compared with the conventional XANES.

In summary, XAS can serve as an excellent characterization technique to study the local structure and electronic properties of Au NCs from a site- and element-specific perspective. EXAFS is particularly useful for smaller Au NC systems (# of Au atoms<∼50) as more distinct scattering shells can be resolved in the data without a strong contribution from the bulk of a large metal core. This review has demonstrated the utility of using XAS for investigating Au_n(SR)_m NCs in catalysis, thiolate-Au surface bonding motifs, unknown Au_n(SR)_m structures, temperature-dependent bonding properties, metal dopant location, and local structural/electronic properties of selenolate- and tellurolate-protected Au NCs.