Hydrothermal Self‐Assembly of Gold Nanoparticles Embed on Carbon Felt for Effective Nitrogen Reduction

Gold‐based electrocatalysts are potential candidates for electrocatalytic nitrogen reduction reaction (e‐NRR) application due to their high conductivity and low hydrogen evolution tendency. However, the role of nonmetallic Au surfaces is not widely studied. Herein, the self‐assembly of gold nanoparticles confined in carbon felts (Au NPs@CFs) containing two different states of Au is reported, showing effective e‐NRR performance. The effects of surfactant concentration on the morphology and electrochemical performance of the Au NPs are analyzed. X‐ray photoelectron spectroscopy indicates two valence states of Au (Au0 and Au3+) in Au NPs@CFs. Interestingly, it is found that the e‐NRR properties of Au NPs@CFs enhance with the increase of Au3+ content. A typical sample Au NPs@CF‐1.0 with the highest Au3+ content exhibits the best e‐NRR performance (NH3 yield of 66.1 μg h−1 mg−1cat. and Faradaic efficiency of 24.9% at −0.3 V vs reversible hydrogen electrode), which, in part due to the presence of Au3+, is conducive to nitrogen adsorption. Theoretical calculation results find that a stable Au3+ surface can adsorb N2 strongly, which in turn can lead to energetically favorable formation of ammonia. This study provides new insights into the relation between oxidation states and performance of gold‐based electrocatalyst with high e‐NRR performance.


Introduction
Ammonia is one of the most crucial nitrogen-containing raw materials.[3] The electrochemical nitrogen reduction reaction (e-NRR) resulting in ammonia (as proposed in the 1960s) constantly remained in focus due to the mild reaction conditions, easy operation, and environmental friendliness.[6] However, the energy of the N≡N triple bond of a molecular N 2 is too high (%940 kJ mol À1 ), reducing the likelihood of N 2 activation on an electrocatalytic surface. [7]Besides, the reaction potentials of e-NRR and hydrogen evolution reaction (HER) are very close, leading to a competition in electron consumption and low ammonia yields. [8]herefore, promising e-NRR catalysts should possess high N 2 adsorption and activation capacities and good electrical conductivities.11][12] The number of active sites of an electrocatalyst is one of the key factors affecting its performance.Au nanostructures' electrochemical performance is known to be superior to bulk Au due to the stronger N 2 adsorption and activation capacity and the higher amount of exposed active sites.This has been reported by preparing a series of monodisperse gold nanoparticles (Au NPs) as the electrocatalysts for e-NRR.Chen's group [13] reported that Gold-based electrocatalysts are potential candidates for electrocatalytic nitrogen reduction reaction (e-NRR) application due to their high conductivity and low hydrogen evolution tendency.However, the role of nonmetallic Au surfaces is not widely studied.Herein, the self-assembly of gold nanoparticles confined in carbon felts (Au NPs@CFs) containing two different states of Au is reported, showing effective e-NRR performance.The effects of surfactant concentration on the morphology and electrochemical performance of the Au NPs are analyzed.Xray photoelectron spectroscopy indicates two valence states of Au (Au 0 and Au 3þ ) in Au NPs@CFs.Interestingly, it is found that the e-NRR properties of Au NPs@CFs enhance with the increase of Au 3þ content.A typical sample Au NPs@CF-1.0with the highest Au 3þ content exhibits the best e-NRR performance (NH 3 yield of 66.1 μg h À1 mg À1 cat.and Faradaic efficiency of 24.9% at À0.3 V vs reversible hydrogen electrode), which, in part due to the presence of Au 3þ , is conducive to nitrogen adsorption.Theoretical calculation results find that a stable Au 3þ surface can adsorb N 2 strongly, which in turn can lead to energetically favorable formation of ammonia.This study provides new insights into the relation between oxidation states and performance of gold-based electrocatalyst with high e-NRR performance.
a typical Au NP electrode exhibited a high NH 3 yield of 17.49 mg h À1 mg À1 Au and an Faradaic efficiency (FE) of 5.79% due to the optimal surface-to-step site ratio of Au NPs.Although electrodes modified with Au NPs can provide high e-NRR performance, they are often unstable: Au NPs can aggregate during the preparation and electrocatalytic processes, which usually require diverse stabilization and dispersion methods.Carbon felt (CF) is a textile material composed of randomly oriented short carbon fibers.It is often used as an electrode material due to its high electrical conductivity and chemical stability. [14]he high permeability of CF allows the N 2 and electrolytes to flow freely in their internal space, increasing the contact area between the electrode and reactants.Therefore, depositing Au NPs on the surface of CF can combine their unique characteristics, which can be beneficial to e-NRR.To the best of our knowledge, research concerning the self-assembled growth of Au NPs on CF electrode for electrochemical NRR is rarely reported.
Herein, we developed a coordination-assisted self-assembly growth and carbonization method to synthesize monodisperse Au NPs inside the surface of CFs (Au NPs@CFs).Cetyl trimethyl ammonium bromide (CTAB), hexamethylenetetramine (HMTA), HAuCl 4 , and CF acted as the structure-directing, coordination, and stabilizing agent, gold precursor, and carbon-based carrier, respectively.The particle size and valence distribution of Au NPs can be adjusted in a particular range via the CTAB concentration.The sample electrode with the smallest particle size (%114 nm) and highest Au 3þ /Au 0 ratio exhibited an excellent e-NRR performance, with an NH 3 yield of 66.1 μg h À1 mg À1 cat.
and an FE of 24.9% at À0.3 V versus revisable hydrogen electrode (RHE).Density functional theory (DFT) calculations show that high oxidation states Au 3þ can also be an active phase with strong N 2 adsorption and activation capacities, which in turn leads to energetically favorable ammonia formation.

Characterizations of Au NPs@CFs
The Au NPs@CFs composites were obtained as shown in Scheme 1 and detailed in the previous section.The CF was made from the support materials (i.e., carbon fibers).The gold precursor molecules were adsorbed on the CF aided by a surfactant and stabilizer agent.During the heat treatment process, the gold precursor was reduced to monodispersed Au NPs by an in situ carbon thermal reduction.Therefore, the uniformly dispersed Au NPs encapsulated in a CF surface was obtained.
The morphology and structure of the as-prepared Au NPs@CFs were first characterized by field emission scanning electron microscopy (FESEM).The Au NPs were distributed on the CF surface (Figure 1).The average nanoparticle sizes were 122.3 AE 30.8, 126.8 AE 21.9, 188.8 AE 78.1, and 114.8 AE 33.3 nm for Au NPs@CF-0.3,Au NPs@CF-0.5,Au NPs@CF-0.7,and Au NPs@CF-1.0,respectively.Besides, the distribution of Au NPs in the sample Au NPs@CF-1.0 is more uniform compared to the other three samples, and most of the Au NPs were embedded in the CF framework.The results demonstrated that a relatively high CTAB concentration is favorable to uniformly form the Au NPs on the felt surface.
This embedded structure can prevent the NPs from aggregation during electrocatalysis, resulting in more active sites and higher electrocatalytic performance.The high-resolution transmission electron microscopy (HRTEM) micrograph of sample Au NPs@CF-1.0showed a characteristic lattice spacing of 0.235 nm (Figure S2, Supporting Information), corresponding to the (111) crystallographic plane of Au.This was confirmed later by the X-ray diffraction (XRD) patterns.As shown in Figure 2a, the diffraction peaks at 25.8°in all samples can be assigned to the amorphous carbon in the CF.The other diffraction peaks at 38.2°, 44.4°, 64.6º, and 77.6º can be indexed to the (111), ( 200), (220), and (311) crystallographic planes of the Au cubic phase (JCPDS no.04-0784), respectively. [15]These characterizations revealed that Au NPs@CFs were successfully synthesized.
To further analyze the surface chemistry of the samples, X-ray photoelectron spectrometry (XPS) measurements were carried out.The Au 4f spectra of these samples can be fitted into four peaks at 84.01, 85.81, 87.66, and 89.90 eV, respectively (Figure 2b).The ones at 84.01 and 87.66 eV belong to Au 0 , whereas the others belong to Au 3þ . [16]It should be noted that the latter concentration in Au NPs@CF-1.0(17.55%) is much higher than those of Au NPs@CF-0.3(9.23%), Au NPs@CF-0.5 (11.86%), and Au NPs@CF-0.7 (10.38%) (Table S1, Supporting Information).Sample Au NPs@CF-1.0with more Au 3þ could provide more adsorption sites for N 2 than Au during the electrocatalytic process, resulting in an enhanced electrochemical performance. [17]Figure 2c shows the O 1s XPS spectra.The peaks at 530.5, 532.1, and 533.6 eV can be attributed to the Au-O, adsorbed oxygen (O 2 ), and C-O-C, respectively.The composition via this element is summarized in Table S2, Supporting Information.The sample Au NPs@CF-1.0showed the most significant Au-O content among the samples.This result was reinforced by the presence of Au 3þ content based on the Au 4f spectra.The high-resolution C 1s spectra of all samples were Scheme 1. Schematic illustration of the synthesis of Au NPs@CF.(j) (k) (l) Figure 1.SEM images of a,e) Au@CF-0.3, b,f ) Au@CF-0.5, c,g) Au@CF-0.7,d,h) Au@CF-1.0, and i-l) the corresponding particle size distributions of these samples.acquired as well (Figure 2d).The following three species were detected: 284.7 eV (C═C), 285.4 eV (C-O), and 288.7 eV (COOH), [18] respectively, which indicate the existence of carbon in the Au NPs@CF samples.

Electrochemical Performance of Au NPs@CF Electrodes
To evaluate the e-NRR performance of Au NPs@CF electrodes, linear sweep voltammetry (LSV) tests were carried out in Ar-or N 2 -purged 0.1 M Na 2 SO 4 solutions.Under the same potential, all samples showed a current density difference between Ar and N 2 saturated electrolytes (Figure 3), indicating that the investigated electrodes may have a specific e-NRR efficiency during the electrochemical process.As the applied potential was more negative, the difference in current density was more significant.Sample Au NPs@CF-1.0showed the highest current density difference at the same negative potential, demonstrating its best e-NRR activity.
To further evaluate the e-NRR performance of the electrodes, chronoamperometric measurements were carried out to analyze the NH 3 yield and FE.After 2 h at À0.3 V versus RHE in N 2 saturated 0.1 M Na 2 SO 4 , a specific amount of indophenol indicator was added to the electrolyte.The samples were collected, and their UV-vis absorption spectra were recorded (Figure 4a).The highest absorbance value (0.18) at 655 nm was achieved by the Au NPs@CF-1.0electrode.In contrast, the other samples showed a lower value (0.069, 0.117, and 0.072), suggesting a lower NH 3 generation performance than the previously mentioned sample.The excellent activity of the highlighted sample may have multiple origins.Its confined structure (based on the SEM micrographs) can inhibit the aggregation of Au NPs and provide stable active sites for e-NRR.The XPS analysis also showed that the Au NPs@CF-1.0had the highest Au 3þ content compared with the other three samples, which could promote the adsorption of N 2 .Therefore, the Au NPs@CF-1.0electrode was further analyzed.
The TG curves of Au NPs@CF-1.0were used to determine the Au mass loaded on CF (Figure 4b), and several weight-loss steps were found.Water evaporation was the dominant phenomenon in the initial stage (below %200 °C in the air).As the temperature rose, a noticeable weight loss was observed at 200-300 °C due to the reduction of Au 3þ to Au NPs.The final step was observed between 500 and 700 °C, and it was associated with the reaction of CF with oxygen.When the temperature passed above 700 °C, the mass of the Au NPs@CF-1.0electrode did not change, suggesting that the final product was achieved (Au).Therefore, the Au NPs content was 0.64 wt%, corresponding to 0.056 mg Au in 8.7 mg of electrode mass.
Furthermore, the effect of the applied potential on the e-NRR performance of Au NPs@CF-1.0electrode was examined.The electrolyte's UV-vis absorption (Figure 5a) was registered after staining the electrolyte with an indoxyl indicator after 2 h of electrolysis (under different potentiostatic conditions).The results showed that Au NPs@CF-1.0electrode had a high e-NRR activity under the potential range from À0.1 to À0.5 V versus RHE (Figure 5b).The most significant NH 3 yield of 66.1 μg h À1 mg À1 cat and FE of 24.9% were obtained at À0.3 V versus RHE.However, an increase in HER activity can be achieved at negative potential values, suppressing the e-NRR process.Therefore, the applied potential was optimized toward the e-NRR process.No N 2 H 4 was detected after 2 h of electrolysis (the Watt and Chrisp method) (Figure 5c), revealing that the Au NPs@CF-1.0electrode had excellent selectivity toward the production of NH 3 .Besides, the optimized e-NRR performance of Au NPs@CF-1.0was higher than the other reported metalbased and nonmetallic catalysts for e-NRR (Table 1).
To analyze whether the detected NH 3 was generated via the electrocatalytic transformation of   Au NPs@CF-1.0;potential: À0.3 V versus RHE; electrolyte: 0.1 M N 2 -saturated Na 2 SO 4 .
The first control experiment indicated that the applied potential was necessary for e-NRR.The second suggested that the active centers for the e-NRR process were located on Au NPs.Finally, the other two control experiments demonstrated that N 2 was the origin of the obtained NH 3 .Also, it should be mentioned that no NH 4 þ was detected in the first three control experiments.In addition, long time e-NRR tests were performed under continuous N 2 and Ar bubbling (alternating).After 12 h of electrolysis, no ammonia was detected in the presence of Ar bubbling, while ammonia was detected in the presence of N 2 bubbling, indicating that the N-source of ammonia produced was not from the surrounding environment (Figure S3, Supporting Information).These results revealed that the detected NH 3 was generated by an electrocatalytic reaction of N 2 on the Au NPs@CF-1.0electrode surface.Durability under long-term electrolysis is a criterion to evaluate the e-NRR performance.Figure 6b shows the chronoamperometry curves of the Au NPs@CF-1.0electrode, which was tested at different potentials.It was found that all the timedependent current density curves remained almost constant after 2 h electrolysis.Based on the above results, seven consecutive experiments were conducted under the optimal potential of À0.3 V versus RHE. Figure 6c shows no noticeable decrease in the NH 3 yields and FEs after 14 h.Besides, compared with the SEM images of Figure S4, Supporting Information (postelectrocatalysis for 6 h) and Figure 1 (before electrolysis), we can find that the particle size distribution of AuNPs remained almost unchanged.These results demonstrated that the as-prepared Au NPs@CF-1.0electrode have a good stability.Furthermore, the XPS results of Au@CF-1.0 electrode after electrolysis (Figure 6d) found that there was a slight decrease of Au 3þ after NRR (from 17.55% to 15.76), which is attributed to the relatively low reaction potential that led to the reduction of Au.

Theoretical Analysis
[21] Its high e-NRR activity and efficiency were attributed to the hindered HER and the highly active low-coordinated Au site.In contrast, the effects of nonmetallic Au phase were not well understood.Herein, we conducted DFT calculations of e-NRR on a stable Au 2 O 3 (100) surface (Figure 7), with the computational methods described in the Supporting Information.Three typical pathways were analyzed, including the alternating pathway, the distal pathway, and hybrid pathway.Interestingly, this surface possesses strong adsorption and activation capacities of N 2 , leading to an energetically favorable e-NRR process.[24] Furthermore, our surface Pourbaix diagram calculations [25] (Figure S1, Supporting Information) found that Au 2 O 3 (100) tends to remain pristine under the potential of interest (i.e., a potential from À1.5 to À0.5 V vs RHE), which will neither form oxygen vacancy nor be poisoned by the oxygen generated from water dissociation.All these results show that Au 2 O 3 may possess relatively good e-NRR activity, and Au 3þ is not easy to be reduced under e-NRR potentials, in good qualitative agreement with experimental observations in Figure 5 and 6.We consider that the highly effective e-NRR found on Au@CF was due to the presence of highly active Au and Au 3þ after synthesis.

Conclusion
A surfactant-assisted hydrothermal self-assembly method was successfully applied to obtain Au NPs@CFs, which contained Au NPs uniformly embedded into the CF surface.The effects of the CTAB concentration on the morphology of the Au NPs size and the electrocatalytic properties of Au NPs@CFs were systematically investigated.The typical catalyst, Au NPs@CF-1.0,exhibited the best e-NRR activity with a high NH 3 yield of 66.1 μg h À1 mg À1 cat and FE of 24.9% at À0.3 V versus RHE.The high performance was attributed to the uniformly embedded structure of Au NPs and the presence of highly active Au 3þ , which facilitated the adsorption and activation of N 2 .The highly active e-NRR performance of Au 3þ was further confirmed by DFT calculations.This research reveals that the mixture of high-valent Au ions and zero-valent Au could be more beneficial than zero-valent Au for electrochemical ammonia synthesis, providing new insights into the design of efficient Au-based NRR catalysts.Preparation of Catalysts: Pretreatment of CFs: first, the CFs were immersed in a mixed solution (H 2 SO 4 , 98%: H 2 O 2 , 30% = 1:3) and ultrasonicated in ethanol for 30 min.Then they were kept in 80 °C water bath.Finally, the CFs were washed alternately with ethanol and water 3 times and dried.

Experimental Section
In a typical procedure, 1.0 g CTAB was dissolved in 60 mL distilled water under constant ultrasonication, and then 2 mL HAuCl 4 (2 wt%) was added under stirring, forming a yellow solution.After 30 min, 0.14 g HMTA was added to the mixture.After stirring vigorously for another 30 min, the mixture was transferred to a 100 mL Teflon-lined autoclave.Simultaneously, the pretreated carbon felt was immersed into the solution vertically and kept at 180 °C for 12 h in an oven.After cooling to room temperature, it was washed with deionized water and absolute ethanol until pH = 7.The product was dried in an oven at 70 °C overnight.Finally, the sample was calcined at 800 °C for 2 h under an Ar flow with a heating rate of 2 °C min À1 .The samples were named as follows: Au@CF-x, where x (x = 0.3, 0.5, 0.7, and 1.0 g) denotes the mass of CTAB.
Characterizations: FESEM (ZEISS-GeminiSEM300) and TEM (JEOL-JEM2100) were used to characterize the morphology and microstructure of the samples.LabX XRD-6000 X-ray diffractometer recorded XRD patterns with Cu Kα radiation (λ¼ = 1.5406Å).XPS (Thermo Scientific K-Alphaþ) was used to analyze the chemical component and elemental valences of the sample.The thermal stability, phase transition temperature, and content of certain components of materials were analyzed by thermogravimetric analysis (TGA).UV-vis spectra were recorded on a Metash UV 6100 s spectrophotometer to estimate the NH 3 yield.All electrochemical measurements were carried out at the CHI660D Electrochemical Workstation using an H-type electrochemical device, in which working electrode, reference electrode, and counter electrode are the catalyst@CF, Hg/HgCl 2 , and blank carbon paper, respectively.0.1 mol L À1 (0.1 M) Na 2 SO 4 was used as the electrolyte.The preparation of electrolyte and establishment of standard curves (NH 3 and N 2 H 4 ) were the same as our recent work. [5]

Figure 5 .
Figure 5. a) UV-vis absorption spectra of the electrolytes stained with indophenol indicator after 2 h electrolysis on Au@CF-1.0 in N 2 -saturated solution at various potentials, the b) corresponding NH 3 yields and FEs, and c) UV-vis absorption spectra of N 2 H 4 .All the tests were performed in 0.1 mol L À1 Na 2 SO 4 under ambient conditions.

Figure 6 .
Figure 6.a) NH 3 yields of Au@CF-1.0 in N 2 -and Ar-saturated solutions, N 2 -saturated solution at open circuit, and N 2 -saturated solution at a blank carbon paper.b) Chronoamperometry test of Au@CF-1.0 at various potentials.c) Cycling test of Au@CF-1.0 at À0.3 V versus RHE.All the tests were conducted in 0.1 mol L À1 Na 2 SO 4 under ambient conditions.d) XPS spectrum of Au@CF-1.0 electrode after electrolysis.

Table 1 .
Comparison of the e-NRR performance from reported electrocatalysts.