Atomically Precise Nanoclusters as Electrocatalysts

Abstract

This chapter summaries recent advances in electrocatalytic application of atomically precise metal nanoclusters (NCs). Metal nanoclusters with determined structures can serve as new model catalysts for electrochemical catalytic study at the atomic level and offer insights into the underlying mechanisms. In recent years, electrocatalysis by metal nanoclusters has been reported and shows promise in several important reactions, including oxygen reduction reaction, water splitting, and CO₂ reduction reaction. By tuning the structure/ligand of the metal nanoclusters, it is possible to achieve catalytic property modification at the atomic level. Overall, the new material of atomically precise metal nanoclusters holds great promise in precise control of catalytic properties and investigation of the fundamental catalytic mechanism at the atomic level.

2.1 Introduction

2.1.1 Atomically Precise Metal NCs

Atomically precise metal nanoclusters have attracted broad interest due to the crystal structure availability and unique properties in optical and catalysis applications [1,2,3]. Compared with the traditional plasmonic metal nanoparticles, the ultra-small NCs (<3 nm) show quantized electronic structures because of the quantum confinement effect [4]. As a result, a single atom change can significantly alter the properties of NCs. The UV-vis spectrum can be used as the “fingerprints” of NCs [5] because it shows certain distinct peaks for each size of NCs, rather than similar plasmonic peaks for regular nanoparticles. Similarly, the catalytic properties of various metal NCs can be totally different because of the major change in surface and electronic structure of NCs with subtle difference in atom numbers or size. Therefore, it is of great interest to correlate the structure and catalytic application using NCs as model catalysts [6]. Indeed, it is possible to build up a library of NCs structure–catalytic properties since the number of reported NCs is large enough. This library might offer great insights into the interpretation of catalytic process and reaction mechanism, and further offer some guidelines in the future design and synthesis of new NCs.

2.1.2 Electrochemical Catalysis with Atomically Precise Metal NCs

The global energy crisis and pollution issues have driven scientists to investigate the alternatives of fossil fuels. One of the strategies is using the secondary energy (such as solar energy and wind energy) derived electricity to split water for producing H₂ as a clean energy source [7]. In the water splitting system, hydrogen is produced at the cathode through hydrogen evolution reaction (HER), and oxygen is formed at the anode through oxygen evolution reaction (OER). Currently, Pt group metals are proved to be the most effective catalysts in HER, while Ir/Ru materials are successful in OER. However, the high cost of noble metals has motivated scientists to study alternative catalysts for these reactions.

On the other hand, the obtained hydrogen from HER and other fuels, such as methanol, can be utilized in the fuel cell system. The fuel cell is an electrochemical device to efficiently transform chemical energy of fuel without combustion [8]. Due to its high efficiency and environmentally friendly properties, fuel cells have been used in vehicles [9]. Currently, the disadvantage of fuel cell technique is the oxygen reduction reaction (ORR) in cathode electrode. It is believed that the ORR is the rate limiting reaction with very sluggish kinetics because of multi-electron transfer during the reaction [10]. Similar to the HER, ORR also favors Pt as the catalyst. Therefore, alternative catalysts are yet to be found to reduce the high cost of noble metals.

Another popular electrochemical catalytic reaction is the CO₂ electrochemical reduction reaction (CO₂RR). In the past decades, the global warming has been considered as a serious issue caused by massive CO₂ emission. To relieve the climate change pressure, one of the solutions is to utilize CO₂ as a resource to produce industrial chemicals and fuels [11, 12]. In the CO₂RR, catalytic materials are required to overcome the intrinsic inertness of CO₂ molecules [13]. Among the catalyst candidates, Au and Ag have been extensively studied because of their high selectivity toward CO [14]. Besides, Cu is also a good catalyst because of its versatility to form various carbon hydrates and low price.

To design better catalytic materials, it is essential to understand the mechanism behind these catalytic reactions. Thus, it is of great importance to find a system to correlate the structure and catalytic properties. Previously, several strategies, including size control and morphology control, have been used to investigate the relationships between structure and properties [15, 16]. For example, Zhu et al. [17] and Seoin et al. [18] reported the active site probing with Au catalysts of different morphologies from the view of experimental and computational modeling, respectively. Despite the well-designed experiment, the non-atomically monodispersed size of traditional nanomaterials significantly weakens the connection between the structure and properties.

In the past decades, the synthesis strategy of atomically precise metal NCs has been extensively investigated and a number of sizes of NCs between tens and hundreds of atoms (equivalent diameters ranging from sub-nanometer to ~2.2 nm) have been reported [1]. For applications as electrochemical catalytic materials, such NCs have several distinctive features such as high surface area and unique surface structure [19]. Besides, the atomic precision and crystal structure availability make metal NCs a perfect system to bridge the structure and properties.

In this chapter, several works about metal NCs as electrochemical catalysts are introduced with a focus on the atomic size effect, morphology effect, doping effect, and charge effect. The computational techniques used in the catalytic mechanism study are also summarized.

2.2 Synthesis and Structure Determination of Atomically Precise Metal NCs

2.2.1 Synthesis of Metal NCs

Here, we illustrate the size-focusing synthesis and structure determination using atomically precise Au₂₅(SR)₁₈ NCs as an example. Larger NCs such as Au₁₃₃(SR)₅₂ and Au₂₇₉(SR)₈₄ can also be synthesized by the size-focusing method [20, 21]. In the size-focusing method, a mixture of NCs with a controlled size distribution is first prepared by carefully controlling the ratio of gold precursor and reduction agent as well as other synthetic conditions. Then, the NCs mixture is subjected to size-focusing under harsh conditions, under which the unstable NCs decompose or convert to more stable ones. Eventually, only the most stable NCs can survive the size-focusing process [22].

In the case of Au₂₅ [23, 24], the Au(III) salt is initially reduced to Au(I) by thiols at 0 °C in the first step. The as-obtained Au(I)-SR complex is then reduced by adding a NaBH₄ aqueous solution. Polydisperse NCs protected by thiolate are obtained after the reduction process. During the following size-focusing process, it can be observed from the evolution of the optical absorption spectra that the monodispersed Au₂₅ NCs gradually become dominant, as shown in Fig. 2.1a. The mass spectrum also illustrates the molecular purity of Au₂₅ (Fig. 2.1b).

Fig. 2.1: a

Adapted from Ref. [24]. Copyright 2009 Royal Society of Chemistry

Evolution of the UV-vis spectra of the reaction product in the Au₂₅ synthesis, b mass spectrometry analysis of Au₂₅.

2.2.2 Structure Determination of Metal NCs

The crystal structures of NCs can be determined by X-ray crystallography. In the case of Au₂₅, the structure comprises a Au₁₃ icosahedral core and a Au₁₂(SR)₁₈ shell [23]. The Au₁₂ shell can be dissected into six dimeric –S–Au–S–Au–S– staple motifs (Fig. 2.2). Due to the atomic precision and the determined crystal structure of metal NCs, it is of great interest to use NCs as catalysts for catalytic mechanism study. Especially for electrochemical catalysis, it is challenging to capture the intermediates during the catalytic process and thus very little is known about the mechanism. However, the well-defined structure of nanocluster catalysts can now facilitate the computational modeling, thus providing opportunities to reveal the mechanism behind electrochemical catalysis.

Fig. 2.2

Adapted from Ref. [5]. Copyright 2012 American Chemical Society

Dissection of the Au₂₅ cluster into Au₁₃ kernel and six surface motifs.

2.3 Hydrogen Evolution Reaction with Metal NC Catalysts

The hydrogen evolution reaction (HER) occurs at the cathode when an external voltage is applied. The reaction can be described in three steps:

Volmer reaction:

$${\text{H}}^{ + } + {\text{M}} + {\text{e}}^{ - } \to {\text{MH}}_{\text{ads}} \quad \left( {\text{acidic}} \right)$$

(2.1)

$${\text{H}}_{2} {\text{O}} + {\text{M}} + {\text{e}}^{ - } \to {\text{MH}}_{\text{ads}} + {\text{OH}}^{ - } \quad \left( {\text{alkaline}} \right)$$

(2.2)

In the Volmer reaction, hydrogen absorbs on the catalytic material to form a MH_ads intermediate, followed by a Heyrovsky reaction or Tafel reaction. In the Heyrovsky reaction, the dihydrogen is formed through an electrochemical desorption:

$${\text{MH}}_{{{\text{ad}}s}} + {\text{H}}^{ + } + {\text{e}}^{ - } \to {\text{M}} + {\text{H}}_{2} \quad \left( {\text{acidic}} \right)$$

(2.3)

$${\text{MH}}_{\text{ads}} + {\text{H}}_{ 2} {\text{O}} + {\text{e}}^{ - } \to {\text{M}} + {\text{OH}}^{ - } + {\text{H}}_{2} \quad \left( {\text{alkaline}} \right)$$

(2.4)

The hydrogen can also undergo a chemical desorption process through the Tafel reaction:

$$2\;{\text{MH}}_{\text{ads}} \to 2\;{\text{M}} + {\text{H}}_{2}$$

(2.5)

The binding energy of hydrogen to the catalyst is the key factor of HER activity. Among the HER catalysts, noble metals such as Pt and Pd have moderate binding energy with hydrogen, thus showing excellent HER activities [25]. However, the high cost and stability issue of Pt catalysts motivate the scientists to find alternative materials for HER.

Here, we summarize the doping effects and synergetic effects of Au NCs in HER. We also introduce how the computational technique is used in these cases to explain differences in catalytic activity and verify proposed mechanisms. These works offer some insights into the catalytic reactions, which is expected to further pave the way for future design of catalyst materials.

2.3.1 Pt or Pd Doped Au₂₅ NCs in HER

The study of Au NCs as HER catalysts was reported by Kyuju et al. in 2017 [26]. A molecular-like Pt₁Au₂₄ nanocluster was prepared and used as catalysts in homogeneous HER. Following this work, the same group further studied the Pd₁Au₂₄, Pd₂Au₃₆ and Pt₂Au₃₆ nanoclusters [27]. In the case of Au₂₅ nanocluster, the doping atom (Pd or Pt) exclusively replaces the central gold atom in the nanocluster. The overall structure of nanoclusters remains unchanged after doping, while the mass spectrum and optical absorption spectrum obviously changed (Fig. 2.3). In the mechanistic study of nanoclusters as HER catalysts, Voltammetry was used to study the electron transfer properties (Fig. 2.4). The redox potentials are drastically changed after Pt doping. Also, compared with Au₂₅, Pt₁Au₂₄ has more positive onset potential and higher current. Combining the voltammetry and linear weep voltammograms (LSVs) with different concentration of trifluoroacetic acid (TFA), it can be seen that the [PtAu₂₄]^1−/2− peak at 1.10 V drastically rises with increasing TFA concentration, indicating that the [PtAu₂₄]²⁻ is the major contributor for enhanced HER activity. Also, the production rate with the Pt₁Au₂₄ catalyst is significantly higher than that of the commercial Pt/C catalyst. Additionally, the charge-state-dependent catalytic activity results show that the catalytic currents at potentials negative to the [PtAu₂₄]^1−/2− exhibit a linear correlation with [PtAu₂₄] and [TFA]^1/2, corresponding to the Volmer–Heyrovsky mechanism [28, 29]. Meanwhile, the currents exhibit a linear correlation with [TFA] and [PtAu]^3/2 at −1.0 V where [PtAu₂₄]¹⁻ is dominant, corresponding to the Volmer–Tafel mechanism [28]. These results are reasonable considering the charge state of Pt₁Au₂₄ at different potential. To be specific, the dominant [PtAu₂₄]²⁻ at negative potential adsorbs a proton to form [H–PtAu₂₄]¹⁻ with negative charge, which is prone to react with another proton to evolve H₂. On the other hand, the dominant [PtAu₂₄]¹⁻ at −1.0 V will form [H–PtAu₂₄]⁰ after adsorbing a proton. Therefore, the Tafel pathway is preferred at this potential.

Fig. 2.3

Adapted with permission from Ref. [26]. Copyright 2017 Springer Nature

a Structures of Au₂₅ and PtAu₂₄ NCs (golden, Au atoms of the kernel; olive, Au atoms of the shell; gray, sulfur), b Mass spectrometry analysis of Au₂₅ (red) and Pt₁Au₂₄ (blue) NCs, c UV-vis-NIR absorption spectra of Au₂₅ (red) and Pt₁Au₂₄ (blue) NCs.

Fig. 2.4

Adapted with permission from Ref. [26]. Copyright 2017 Springer Nature

a Square-wave voltammetry (SWV) of Au₂₅ (red) and PtAu₂₄ (blue) NCs, b LSV of Au₂₅ (red) and PtAu₂₄ (blue) NCs, c LSVs of PtAu₂₄ in THF in the presence of 0. 4. 8, 12, 17, 21, 27, 34, 45, 55, and 60 mM of trifluoroacetic acid (TFA), d H₂ production rates per mass of metals in catalyst at various overpotentials on PtAu₂₄ (blue) and Pt/C electrodes (red), e Dependence of the catalytic current Ic on the concentration of TFA in the presence of PtAu₂₄ (I mM), f Dependence of the catalytic current Ic on the concentration of PtAu₂₄ in TFA (1 M) solution at −1.3 V (blue), −1.8 V (green) and −2.2 V (purple), insets show dependence of the Ic on the concentration of e TFA and f PtAu₂₄ at −1.0 V, g calculated reaction energies for HER on PtAu₂₄.

DFT calculations are also utilized to explain the enhanced catalytic activity of Pt₁Au₂₄ (Fig. 2.4g). The results show that the adsorption of the first proton on Pt₁Au₂₄ is thermodynamically neutral, while the second proton adsorption is endothermic. This result is consistent with the charge-state-dependent catalytic activity. On the other hand, the geometry optimization shows that the proton is prone to bind with the Pt atom in the icosahedral center. Therefore, the stronger H–Pt interaction with respect to H–Au is a key factor for the enhanced HER activity. Using the same strategies, the group also studied the HER activities of Pd₁Au₂₄, Pt₂Au₃₆ and Pd₂Au₃₆ [27]. These works proved the versatility of metal NCs in catalytic mechanism study. Especially, the introduction of SWV and DFT calculation makes the metal NCs a perfect tool to correlate the structure and properties.

2.3.2 Boosting HER Activity with Au NCs/MoS₂ Composite

Besides the doping effects, metal NCs are also reported to show strong synergetic effects when loaded on other materials. In 2017, the synergetic effect was reported by Zhao et al. [30]. In this work, Au₂₅(SR)₁₈ and Au₂₅(SePh)₁₈ NCs are loaded on the MoS₂ ultra-thin nanosheets. These two NCs have a similar kernel structure but different protecting ligands. The TEM images and XPS spectra indicate the successful loading of Au NCs on the surface of MoS₂ (Fig. 2.5). The Mo 3d XPS of composites exhibits negative shifting compared with MoS₂, while the Au 4f spectra of composites show obvious positive shifting compared with Au₂₅ NCs. The XPS results clearly indicate that the electron density transfers from Au₂₅ to MoS₂. In the HER activity test, the thiolate-protected Au₂₅ NCs exhibit a more positive onset potential and higher current density compared with MoS₂ nanosheets. On the other hand, the benzeneselenolate-protected Au₂₅ NCs loaded MoS₂ nanosheets show similar synergetic effects. However, the enhancement is less obvious compared with the thiolate-protected Au₂₅ NCs. To explain the different enhancement in HER catalytic activity, Au 4f XPS spectra of Au₂₅(SR)₁₈/MoS₂, Au₂₅(SePh)₁₈/MoS₂ and pure MoS₂ are obtained. The Au₂₅(SR)₁₈/MoS₂ composites exhibit more positive shifting compared with the Au₂₅(SePh)₁₈/MoS₂ composites, indicating stronger electron density transfer effects of thiol protected Au₂₅. Therefore, the electron interaction between MoS₂ nanosheets and Au NCs is a key factor for the HER activity. Based on these results, the authors proposed a dual interfacial effect, where the core/ligand interface of Au NCs and the MoS₂/Au NCs interface are both important in the HER catalytic activity.

Fig. 2.5

Adapted from Ref. [30]. Copyright 2017 Wiley-VCH

a TEM image of the MoS₂ nanosheets, b HAADF-STEM image of the Au₂₅/MoS₂ composite, c high-resolution Mo 3d XPS spectra of MoS₂ nanosheet and Au₂₅/MoS₂ composite, d high-resolution Au 4f XPS spectra of Au₂₅ NCs and Au₂₅/MoS₂ composite, e HER LSV curves, f high-resolution Au 4f XPS spectra of MoS₂, Au₂₅(SePh)₁₈/MoS₂ and Au₂₅(SR)₁₈/MoS₂ composite.

Du et al. also studied the synergetic effects between Au₂Pd₆ NCs and MoS₂ [31]. In this work, DFT calculation is used to investigate the origin of the enhanced HER activity. The DFT calculation shows the _ΔG for a proton adsorbed at Au₂Pd₆/MoS₂ composite is more negative than that of MoS₂, indicating better HER activity of the composites. Besides, it is found that in the Au₂Pd₆ composites, both Au atoms and S atoms have appropriate _ΔG for proton adsorption. In contrast, only the Au–Pd bridge site has proper _ΔG in Au₂Pd₆ NCs, and no proper active site for proton adsorption can be found in defect-free MoS₂. Therefore, the significant increase in active sites in the composites is a key factor for boosted HER activity. Meanwhile, the DOS analysis also explained the enhanced activity of composites. The Au₂Pd₆ composites have a defect state near the Fermi level. This unique defect state narrows the band gap, leading to a better electronic conductivity.

2.4 Oxygen Evolution Reaction with Metal NCs

Oxygen evolution reaction (OER) is the other half reaction in water splitting and is indeed critical. Oxygen can be formed through several proton/electron-coupled steps in OER. The reaction can be described as follows:

$$2\;{\text{H}}_{ 2} {\text{O}} \to 4\;{\text{H}}^{ + } + {\text{O}}_{2} + 4\;{\text{e}}^{ - } \quad \left( {\text{acidic}} \right)$$

(2.6)

$$4\;{\text{OH}}^{ - } \to 2\;{\text{H}}_{2} {\text{O}} + {\text{O}}_{2} + 4\;{\text{e}}^{ - } \quad \left( {\text{alkaline}} \right)$$

(2.7)

In OER, the formation of oxygen requires a four-electron transfer, and the reaction kinetically favors single electron transfer at each step [32]. Therefore, catalysts are required to overcome the energy barrier and lower the high overpotential in the sluggish OER [33, 34]. To reduce the cost of catalyst materials, cheaper and efficient alternative catalytic materials are extensively studied to replace the current Ir-based materials. In previous reports, anchoring a small amount of gold onto cobalt-based materials can enhance the OER activity [35, 36]. However, the mechanism for the improvement was not well understood due to the variability and complicacy of the gold-loaded composites. In this section, the synergetic effects between Au NCs and CoSe₂ nanosheets are introduced. This unique composite may provide valuable insights into the mechanistic study by taking advantage of the precise atomic structures of Au NCs [1].

2.4.1 Au_n NCs Promote OER at the Nanocluster/CoSe₂ Interface

The OER performance of metal NCs was first reported by Zhao et al. in 2017 [37]. In this work, composites of Au₂₅ and ultra-thin CoSe₂ nanosheets were synthesized and tested as OER catalysts. TEM images clearly show the ultra-thin nanosheet structure of CoSe₂ (Fig. 2.6a–c). In the high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM), it can be observed that Au nanoclusters are homogeneously dispersed on the surface of CoSe₂ nanosheets.

Fig. 2.6

Adapted from Ref. [37]. Copyright 2017 American Chemical Society

a TEM image of CoSe₂ nanosheets, b, c HAADF-STEM images of Au₂₅/CoSe₂ composite, d OER polarization curves of Au₂₅/CoSe₂, CoSe₂, Pt/C and Au₂₅/C, e overpotential at the current density of 10 mA cm⁻², and the current density at the overpotential of 0.45 V for Au₂₅/CoSe₂, CoSe₂, Pt/C and Au₂₅/C catalysts, f stability test of Au₂₅/CoSe₂,HER LSV curves, g UV-vis spectra of Au₂₅ NCs before and after the stability test.

In the electrochemical test (Fig. 2.6d–f), the Au₂₅/CoSe₂ composites show much smaller onset potential (1.406 V vs. RHE) and higher current density than CoSe₂ nanosheets and Au₂₅-loaded carbon. At 1.68 V, Au₂₅/CoSe₂ composites achieve a current density of 11.78 mA cm⁻², which is 2.4 times that of CoSe₂ nanosheets (4.92 mA cm⁻²) and 20.7 times that of Au₂₅-loaded carbon (0.57 mA cm⁻²). Also, the composites exhibit higher current density and smaller overpotential than commercial Pt/C catalysts. In the stability test, the polarization curve and UV-vis spectra of the Au₂₅/CoSe₂ composites exhibit the same features before and after 1000 cycles, indicating excellent stability of the composites as OER catalysts (Fig. 2.6g).

The XPS and Raman analysis of CoSe₂ and composites were conducted to explain the enhanced OER activity of Au₂₅/CoSe₂ composites (Fig. 2.7). The binding energy of Co 2p in the composites shows a ~1 eV decrease compared with CoSe₂, indicating electronic interaction between the Au₂₅ and CoSe₂ nanosheet. Also, the Raman peak at ca. 657 cm⁻¹ exhibits a shift toward higher wavenumber, suggesting the electronic interaction. It is believed that such an electronic interaction is a key factor that stabilizes the hydroperoxyl intermediates and optimizes interaction between CoSe₂ and oxygen.

Fig. 2.7

Adapted from Ref. [37]. Copyright 2017 American Chemical Society

a High-resolution Co 2p XPS spectra of CoSe₂ and Au₂₅/CoSe₂ composites, b Raman spectra of CoSe₂ and Au₂₅/CoSe₂ composites.

2.4.2 Au_n NC Size Effect in OER

The size of gold nanoclusters is also important for the catalytic activity. To study the potential size dependence of Au_n NCs for OER, Zhao et al. compared gold nanoclusters of Au₁₀(SPh-^tBu)₁₀, Au₂₅(SR)₁₈, Au₁₄₄(SR)₆₀ and Au₃₃₃(SR)₇₉, with the latter three being protected by the same phenylethanethiolate ligand. These NCs were loaded onto CoSe₂ (all at 2.0 wt%, denoted as Au_n/CoSe₂). The OER polarization curves show a moderate increase of OER activity with an increase in cluster size. The Au₃₃₃/CoSe₂ catalyst possesses the smallest overpotential (~0.41 V for 10 mA cm⁻²) and the largest current density (15.44 mA cm⁻² at the overpotential of 0.45 V).

2.5 Oxygen Reduction Reaction with Au NCs

The ORR is the rate-determining step of the fuel cell system because of its sluggish kinetics [38, 39]. In both acidic and alkaline electrolyte, different mechanisms have been documented for ORR. These processes can be described as follows:

$${\text{Acidic:}}\quad {\text{O}}_{2} + 4\;{\text{H}}^{ + } + 4\;{\text{e}}^{ - } \to 2\;{\text{H}}_{2} {\text{O}}$$

(2.8)

$${\text{O}}_{2} + 2\;{\text{H}}^{ + } + 2\;{\text{e}}^{ - } \to {\text{H}}_{2} {\text{O}}_{2}$$

(2.9)

$${\text{H}}_{2} {\text{O}}_{2} + 2\;{\text{H}} + 2\;{\text{e}}^{ - } \to 2\;{\text{H}}_{2} {\text{O}}$$

(2.10)

$${\text{Alkaline: }}\quad {\text{O}}_{2} + {\text{H}}_{2} {\text{O}} + 4\;{\text{e}}^{ - } \to 4\;{\text{OH}}^{ - }$$

(2.11)

$${\text{O}}_{2} + {\text{H}}_{2} {\text{O}} + 2\;{\text{e}}^{ - } \to {\text{HO}}_{2} {^{ - }} + {\text{OH}}^{ - }$$

(2.12)

$${\text{HO}}_{2^{ - }} + {\text{H}}_{2} {\text{O}} + 2\;{\text{e}}^{ - } \to 3\;{\text{OH}}^{ - }$$

(2.13)

From the equations, it can be seen that two possible pathways can be observed in ORR in both electrolytes. One of them is the direct 4e⁻ pathway where oxygen is reduced to H₂O in acidic electrolytes or OH⁻ in alkaline electrolytes. The other pathway is the 2e⁻ mechanism where H₂O₂ or HO₂⁻ is first formed before the sequent reduction to H₂O or OH⁻ with another 2e⁻ transfer. It is believed that the commercial Pt/C electrode favors the direct 4e⁻ pathways. However, the complicated surface structure of Pt/C catalysts makes it challenging to figure out the reaction occurring in the catalytic process [40,41,42]. To understand the ORR mechanism, several noble metal nanoparticle-based catalytic materials with either direct 4e⁻ or 2e⁻ pathways have been extensively studied [43,44,45,46,47,48,49]. Among these catalysts, Au has shown several unique properties in ORR. In 2007, Zhang et al. found that Pt catalysts can be stabilized against dissolution by modification with Au NCs [50]. In their electrochemical study, the Au NCs-modified Pt catalysts exhibit ultra-high stability where the polarization curve remains unchanged after 30,000 cycles. Later, Yin et al. reported in 2012 the Au NCs/graphene hybrids for high-performance ORR [51]. The hybrid catalytic materials exhibit high current density and excellent stability comparable to that of the commercial Pt/C catalysts. Yet, no study about combining the atomically precise Au NCs with DFT calculations is reported. The precise structure and ultra-small size make the Au NCs a perfect system to study the size effect in ORR. In this section, the reports on the size effects of atomically precise Au NCs are introduced.

2.5.1 Nanocluster-Derived Ultra-Small Nanoparticles for ORR

The size effect has been extensively studied for Au nanoparticles in the past decades [52, 53]. However, reports are rare for ultra-small Au nanoparticles (i.e., core diameter <2 nm) for ORR. In 2016, Wang et al. reported porous carbon-supported ultra-small nanoparticles as ORR catalysts using thiolate-capped Au₂₅, Au₃₈ and Au₁₄₄ NCs as precursors [54]. The average diameters of Au nanoparticles were estimated to be 3.7 ± 0.9 nm for Au₂₅-derived catalyst (AuPC-1), 4.9 ± 1.1 nm for Au₃₈-derived one (AuPC-2), and 5.8 ± 1.25 nm for Au₁₄₄-derived one (AuPC-3), as shown in Fig. 2.8. All the Au nanoparticles are larger than the sizes of the original nanoclusters because of aggregation of clusters during the calcination.

Fig. 2.8

Adapted with permission from Ref. [54]. Copyright 2016 American Chemical Society

TEM images of a AuPC-1, c AuPC-2, and e AuPC-3, with the corresponding size distribution in panels b, d, and e.

In the electrochemical test (Fig. 2.9), it is found that the AuPC-1 sample exhibits a peak current density similar to that of commercial Pt/C (0.57 mA cm⁻²). The rotation ring and disk electrode (RRDE) measurements show that the onset potential is 0.95, 0.91, and 0.89 V for AuPC-1, AuPC-2, and AuPC-3, respectively. Also, the diffusion-limited current density of AuPC-1 (3.61 mA cm⁻²) is obviously higher than that of AuPC-2 (3.21 mA cm⁻¹) and AuPC-3 (3.16 mA cm⁻²). It is noted that the ORR activity of AuPC-1 is similar to that of the commercial Pt/C catalysts (0.95 V for onset potential and 4.98 mA cm⁻² for limiting current density). Taking all the results together, one can find that the ORR activity increases with the decreasing Au nanoparticle size. Especially, the smallest AuPC-1 sample shows a comparable activity with the commercial Pt/C electrode.

Fig. 2.9

Adapted with permission from Ref. [54]. Copyright 2016 American Chemical Society

a CV curves, b the ORR polarization curves, c Tafel plots, d Chronoamperometric profiles of AuPC-1, AuPC-2, AuPC-3 and Pt/C catalysts.

In the Tafel plots (Fig. 2.9c), it can be seen that the specific activity increases with the decrease of AuPC nanoparticle size. At 0.8 V, the current density increases in the order of AuPC-3 (0.16 mA cm⁻²) < AuPC-2 (0.195 mA cm⁻²) < AuPC-1 (0.612 mA cm⁻²) < commercial Pt/C (0.615 mA cm⁻²). Additionally, similar features of the Tafel plots can be observed for all the catalysts. Two clear linear regions are displayed at low and high overpotentials. In the low overpotential region, the slopes of these catalysts are all close to 60 mV dec⁻¹, indicating that a pseudo-two-electron reaction might be the rate-determining step. However, the similar slopes of approximately 120 mV dec⁻¹ for the four catalysts suggest that the rate-determining step is probably the first electron transfer to oxygen molecules. Also, the stability test of AuPC and commercial Pt/C catalysts indicate superior stability of Au nanoparticles in ORR. The relative current of AuPC-1, AuPC-2, and AuPC-3 shows a loss of 19.2%, 15.6%, and 22.7%, respectively, after 8 h. While for commercial Pt/C catalysts 35% loss of current is observed. The authors ascribe the superior performance of nanoclusters-derived ultra-small nanoparticles to the low-coordination surface Au atoms of small-sized nanoparticles and the synergetic effects between carbon and Au.

In summary, the small-sized Au particles are beneficial for the activation of oxygen, thus increasing the catalytic activity of ORR.

2.5.2 Size Effect of Au NCs in ORR

Au NCs were first reported for ORR by Chen et al. in 2009 [55]. Four Au NCs with different sizes: Au₁₁, Au_~25, Au_~55, and Au_~140 are synthesized. (On a note, we found that the UV-vis spectrum of “Au₁₁” [55] instead resembles that of the Au₂₅ rod cluster [1]). In the electrochemical test, it was found that the “Au₁₁” exhibits the highest limiting current density and smallest overpotential in ORR. Overall, the catalytic activity decreases as the cluster size increases. However, the crystal structures of Au NCs were not obtained at that time. Therefore, the assignment of precise atom numbers was preliminary. Later, Jones et al. reported in 2018 a series of t-butylthiolate protected Au NCs with increasing sizes for ORR [56]. These four nanoclusters, namely Au₂₃, Au₃₀, Au₄₆, and Au₆₅, show distinct UV-vis spectra as shown in Fig. 2.10, and different colors were observed for these four nanoclusters. The same t-butylthiolate ligand and different core sizes make it ideal to compare the atomically precise size effect with this series of Au NCs.

Fig. 2.10

Adapted with permission from Ref. [56]. Copyright 2018 American Chemical Society

a UV-vis spectra, b photon energy plot (eV), and c photograph of Au₂₃, Au₃₀, Au₄₆ and Au₆₅.

In the electrochemical test (Table 2.1), the Au₆₅ exhibits a transfer of 3.2 electrons, which is higher than that of other NCs (approximately 2 electrons). Also, the potential at −1 mA cm⁻² shows a trend of Au₆₅ < Au₄₆ < Au₃₀ < Au₂₃, indicating that the ORR catalytic activity increases as the nanocluster size grows. Therefore, it can be concluded that larger NCs can facilitate the ORR with smaller overpotential, higher diffusion-limiting current and higher selectivity toward OH⁻ production.

Table 2.1 Experimental parameters for Au NC-catalyzed ORR (n: the number of transferred electrons)

2.5.3 Charge-State-Dependent ORR Activity of Au₂₅ NCs

In 2007, Negishi et al. reported the charge state of Au₂₅ NCs can be tuned between −1, 0 and +1 [57]. This unique property provides an ideal model to study the charge-state effect of Au NCs in electrochemical catalysis [58,59,60]. Later in 2014, Lu et al. synthesized these atomically precise Au₂₅ NCs protected by dodecanethiolate with different charge states (−1, 0 and +1) for ORR [61]. The UV-vis spectra clearly show the different features of the as-prepared NCs. In addition, the Au 4f_7/2 binding energy shows a positive shift when the charge state becomes more positive, further indicating the different charge state of Au₂₅ NCs (Fig. 2.11).

Fig. 2.11

Adapted with permission from Ref. [61]. Copyright 2014 Royal Society of Chemistry

a UV-vis spectra and b XPS spectra of Au₂₅ NCs with different charge states.

The electrochemical test results show that the Au₂₅⁻ shows a more positive onset potential and higher diffusion-limiting current density compared with Au ⁰₂₅ and Au₂₅⁺ (Fig. 2.12). Also, the H₂O₂ production percentages show a trend of Au₂₅⁻ (86%) > Au ⁰₂₅ (82) > Au₂₅⁺ (72%), indicating that the two-electron pathway is dominant with the Au₂₅ nanoclusters. Thus, the Au₂₅⁻ can be used as a promising catalyst for H₂O₂ production in ORR. The combined experimental results and previous DFT calculations suggest that charging the cluster can increase the chemical activity with respect to O₂ [62]; the authors proposed that the strong charge-state effects on H₂O₂ production can be attributed to electron transfer from the anionic Au₂₅ core into the LUMO (π^*) of O₂, activating the O₂ molecule and generating peroxo-like species.

Fig. 2.12

Adapted with permission from Ref. [61]. Copyright 2014 Royal Society of Chemistry

a CV of the ORR on Au nanoparticles and Au₂₅ NCs with different charge states, b RRDE voltammograms recorded on glass carbon electrode and the Au₂₅ NCs, c selectivity of the H₂O₂, d the electron transfer number as a function of the applied potentials.

2.6 CO₂ Reduction Reaction with Metal NCs

CO₂ reduction reaction (CO₂RR) has been extensively investigated in order to remediate the global climate change issues during the past decades. As a multiproton and multi-electron process, the CO₂ RR is a complicated process with several products produced at various voltages as shown in Table 2.2 [63]. Especially, the formation of CO₂⁻ key intermediate consumes a large amount of energy. On the other hand, the competing HER also hinders the efficient CO₂RR in aqueous solutions. Therefore, highly efficient catalysts are critically required to lower the energy barrier in CO₂RR [64,65,66].

Table 2.2 Reduction potentials (vs. RHE) of various products in CO₂ reduction reactions

Among the catalytic materials, Au has been extensively studied due to its high selectivity toward CO formation [14]. On the other hand, Cu is also attractive because of its versatility in forming various hydrocarbon products [67]. In this section, we summarize the Au and Cu NCs as catalysts for CO₂RR. The application of atomically precise NCs offers an opportunity for correlating the structure and catalytic properties, hence providing insights into the mechanism and also fundamental rules for future design of advanced catalytic materials for CO₂RR.

2.6.1 Au₂₅ for CO₂RR

In 2012, Kauffman et al. first reported atomically precise Au₂₅ NCs as catalysts for CO₂RR [68]. The electrochemical results show that Au₂₅ have much higher activity in CO₂RR than Au nanoparticles and bulk Au as shown in Fig. 2.13. To be detailed, the Au₂₅ exhibits higher current density in LSV and higher CO formation rate than Au nanoparticles and bulk Au.

Fig. 2.13

Adapted from Ref. [68]. Copyright 2018 American Chemical Society

a LSV of Au₂₅/CB, b potential-dependent H₂ and CO formation rates for Au₂₅/CB, c LSV of various Au catalysts in quiescent CO₂ saturated 0.1 M KHCO₃, d potential-dependent CO formation rates for the various Au catalysts.

To explain the superior activity of the Au₂₅ nanocluster, the same group used DFT calculations to obtain the free energy diagram of the CO₂RR process (Fig. 2.14) [69]. They proposed that partial ligand removal would occur in order to expose the active sites for CO₂ adsorption. The free energy diagrams of both fully ligand-protected Au₂₅ and singly dethiolated Au₂₅ cluster were obtained. In the energy diagrams, it can be seen that the most endergonic step for both cases is the ^*COOH formation. The U_onset for fully ligand-protected Au₂₅ is −2.04 V, much larger than that of singly dethiolated Au₂₅ cluster (−0.34 V) and experimentally value (−0.193 V), indicating that their proposal of ligand removal is correct. Therefore, they concluded that the cluster can facilitate the reduction of CO₂ by partial removal of the thiolate ligand. The exposed Au site can reduce the free energy of the COOH intermediate formation, thus lowering the overpotential for CO formation. This group also reported the long-term stability of Au₂₅ NCs in CO₂RR. The results show that Au₂₅ can catalyze the CO₂RR for 6 days with steady production rate of 745 ± 59 L/(g_Au h) and CO selectivity of 86 ± 5%, indicating the exceptional stability of Au₂₅ NCs in CO₂RR [70].

Fig. 2.14

Adapted from Ref. [69]. Copyright 2016 the American Institute of Physics

Optimized structure of the model: a fully ligand-protected Au₂₅(SCH₃) ⁻₁₈ NC and b singly dethiolated Au₂₅(SCH₃) ⁻₁₇ NC. Free energy diagram for electrochemical reduction of CO₂ to CO: c over the fully ligand-protected Au₂₅(SCH₃) ⁻₁₈ NC, and d over the singly dethiolated Au₂₅(SCH₃) ⁻₁₇ NC. White, gray, blue and golden balls represent H, C, S, and Au atom, respectively.

2.6.2 Atomic-Level Morphology Effects in CO₂RR

Previously, Au nanomaterials with different morphology have been extensively studied in testing the catalytic activity of facet, edge, and corner. Despite some interesting results, the nonavailable atomic-level structure of these nanomaterials made it difficult to connect the structure and the catalytic properties. To find more solid evidence of morphology effects of Au catalysts in CO₂RR, Zhao et al. prepared atomically precise Au₂₅ nanosphere and nanorod and tested their electrochemical performance as CO₂RR catalysts [71]. These two NCs exhibit distinct features in UV-vis spectra, corresponding to their spectroscopic fingerprints (Fig. 2.15). The Au₂₅ nanosphere comprises an icosahedral Au₁₃ core protected by six dimeric surface staples (–SR–Au–SR–Au–SR–), showing a spherical morphology; while the Au₂₅ nanorod comprises two Au₁₃ icosahedra fused together by sharing one vertex gold atom, and the rod is protected by five bridging thiolates (–SR–) at the rod’s waist, 5 phosphine ligands and one chloride on each end of the nanorod.

Fig. 2.15

Adapted from Ref. [71]. Copyright 2018 American Chemical Society

UV-vis spectra of: a Au₂₅ nanosphere and b Au₂₅ nanorod; c Atom packing structures of Au₂₅ nanosphere and nanorod.

The electrochemical results in Fig. 2.16 show that Au₂₅ nanosphere has higher CO Faradaic efficiency around 70% than the Au₂₅ nanorod (30–60%). The Au₂₅ nanosphere also exhibits a much higher CO formation rate. Especially, at high potential of −1.17 V, the CO formation rate of Au₂₅ sphere (33.3 μL min⁻¹) is 2.8 times that of Au₂₅ nanorod (11.7 μL min⁻¹). The larger CO FE and higher CO formation rate of the Au₂₅ nanosphere indicate its high catalytic performance compared with the Au₂₅ nanorod.

Fig. 2.16

Adapted from Ref. [71]. Copyright 2018 American Chemical Society

a Total current density of CO₂ reduction and b Faradaic efficiency (FE) for CO production over the Au₂₅ nanosphere and nanorod; c FE for CO and H₂ at the potential of −1.07 and −1.17 V over Au₂₅ nanosphere and nanorod, d CO formation rates over Au₂₅ nanosphere and nanorod.

DFT calculations are used to evaluate the free energy of reaction steps to understand the mechanism of the better performance of the Au₂₅ nanosphere (Fig. 2.17). First, the _ΔG values for ligand removal from NCs are calculated. For the Au₂₅ nanosphere, removal of a single –SCH₃ is considered, while for the nanorod, removal of –SCH₃, –Cl and PH₃ is calculated. The results show that the desorption of –PH₃ and the removal of –Cl from the Au₂₅ nanorod have the same _ΔG: 0.54 eV. For the –SCH₃ removal energy of the nanosphere and the nanorod, _ΔG is calculated to be 0.49 eV and 0.95 eV, respectively. These results indicate that the removal of –PH₃ and –Cl is more favored for the nanorod. However, the ligand removal from the nanosphere is less endergonic than that from the nanorod. The free energy diagram after ligand removal shows that ^*COOH, an important intermediate in CO₂ reduction to CO on Au, is more stabilized on the Au₂₅(SCH₃)₁₇ nanosphere with one –SCH₃ ligand removed compared with any of the other ligand-removed systems of the nanorod. Therefore, it is concluded that the energetically favorable removal of –SCH₃ from the Au₂₅ nanosphere to expose active sites and the stabilization of ^*COOH intermediates on the obtained Au₂₅(SCH₃)₁₇ nanosphere contribute to the superior catalytic performance of the Au₂₅ nanosphere. This work has successfully correlated the atomic-level morphology with catalytic performance, explaining the factors that determine the CO₂RR activities with the aid of DFT calculations. It has shed light on the mechanism for the CO₂RR in the future.

Fig. 2.17

Adapted from Ref. [71]. Copyright 2018 American Chemical Society

a ΔG for ligand removal (eV) from the NCs at 0 V versus RHE, b free energy diagrams for CO₂ reduction to CO on the ligand-removed NCs at 0 V versus RHE.

2.6.3 Cu NC-Catalyzed CO₂RR

Cu catalysts are attractive for their ability to reduce CO₂ to hydrocarbon products. However, the mechanisms of CO₂ reduction on nanostructured Cu catalysts are not well understood yet. In 2017, Tang et al. reported a copper-hydride nanocluster for CO₂RR to study the mechanism of Cu catalysts [72]. This Cu₃₂H₂₀L₁₂ NC (L = S₂P(OiPr)₂) comprises a distorted hexacapped rhombohedral core of 14 Cu atoms sandwiched by two Cu₉ triangular cupola fragments of Cu atoms, while the hydrides and L ligands are homogeneously distributed on the surface of the nanocluster (Fig. 2.18). The study on the CO₂ reduction mechanism shows that the key initial step of CO₂ reduction is where the first hydrogen is added: C or O of CO₂. The H addition on C would facilitate the formation of HCOOH, otherwise, CO would occur. Based on the structure of the Cu nanocluster, the authors proposed two possible channels to form the critical HCOO^* intermediate: (1) the non-electrochemical absorption of CO₂ on lattice hydrides (lattice-hydride channel); (2) the electrochemical CO₂ reaction with proton and electron (proton-reduction channel). The free energy diagram of both channels for HCOOH and CO production shows that in both cases the lattice-hydride mechanism exhibits more energy downhill compared with the proton-reduction mechanism, suggesting the CO₂ reduction favors the lattice-hydride channel over this Cu NC. After the confirmation of reaction channel, the free energy diagram of CO and HCOOH formation is calculated following the lattice-hydride mechanism (Fig. 2.19a). The results indicate the HCOOH pathway is more favorable than the CO pathway over the Cu cluster.

Fig. 2.18

Adapted with permission from Ref. [72]. Copyright 2017 American Chemical Society

Atomic structure of the Cu₃₂H₂₀L₁₂ NC (L = S₂PH₂). Orange, Cu; green, hydride; yellow, S; purple, P; white, H.

Fig. 2.19

Adapted with permission from Ref. [72]. Copyright 2017 American Chemical Society

a Free energy diagrams for HCOOH and CO formation on the Cu₃₂H₂₀L₁₂ NCs via the lattice-hydride mechanism, b overall mechanism of HCOOH formation from CO₂ reduction on Cu₃₂H₂₀L₁₂ NCs via the lattice-hydride channel, orange, Cu; green, hydride; red, oxygen; gray, carbon.

Based on all the DFT calculation results, the authors proposed a complete catalytic mechanism as shown in Fig. 2.19d. It can be seen that the HCOOH is formed through the non-electrochemical lattice-hydride pathway. Additionally, the Cu NC can be recovered by the electrochemical reaction with two protons and electrons. Electrochemical tests are conducted to verify the theoretical prediction that HCOOH is favored over the Cu NC (Fig. 2.20). It can be seen that HCOOH is the dominant product at low potential with selectivity higher than 80%, while H₂ becomes dominant at high potential due to the competing HER. Only a small amount of CO is formed throughout the potential window. Therefore, these electrochemical results have successfully verified the accuracy of theoretical prediction. This work has demonstrated the methods of mechanism study using atomically precise metal NCs by combining DFT calculation and electrochemical experiment. Compared with traditional nanoparticles, the precise structure of nanoclusters makes the computational modeling much more facile and convincing. It is anticipated that the application of metal NCs in CO₂RR can offer insights in the mechanism study in the future.

Fig. 2.20

Adapted with permission from Ref. [72]. Copyright 2017 American Chemical Society

a Average current densities (black) and cumulative FE for H₂, HCOOH and CO, b product selectivity for H₂, HCOOH and CO.

2.7 Summary and Future Perspective

In this chapter, we have summarized the literature work about metal NCs as electrochemical catalysts. Compared with the metallic nanoparticles, the metal NCs exhibit discrete electronic energy levels due to the quantum size effect. This unique electronic property, plus atomically precise structures, as well as their various atom-packing structures, render the nanoclusters great potential in catalytic applications.

Metal nanoclusters (homogold and doped ones) have been demonstrated in several important electrochemical reactions including HER, OER, ORR, and CO₂RR. The results show that doping effects, synergetic effects, size effects, thermostability effects, charge effects, and morphology effects all play important roles in the catalytic effects. One of the most important advantages of nanoclusters in catalysis is the feasibility of computational modeling due to the available structure. For example, the catalytic mechanism of CO₂RR over Cu₃₂H₂₀L₁₂ NCs are predicted by DFT calculations and successfully verified by the electrochemical experiment. Metal nanoclusters are expected to be a promising class of model catalysts for correlating the structure and properties, providing exciting opportunities for the understanding of catalysis mechanism at the atomic level.

Future work in NCs electrocatalysis should investigate the following aspects:

• (i) The doping effects. Several bimetallic NCs have been reported. However, the catalytic properties of bimetallic NCs are still rarely studied. The investigation of bimetallic NCs may be helpful to understand the impact of doping atoms, further revealing the fundamental catalytic mechanisms;

• (ii) The synergetic effects. Au NCs-loaded composite materials are reported to show enhanced catalytic activity. It is essential to understand the interaction between supporting materials and metal NCs with more precise interfaces;

• (iii) Metal NCs in other catalytic reactions. Owing to the characterized structure and tunable properties, metal NCs should be broadened to other catalytic reactions to study the fundamental catalytic mechanisms, such as nitrogen reduction reactions (NRR). The new electrocatalytic reactions remain to be explored.