Green synthesis of a magnesium single-atom catalyst from Spinacia oleracea chlorophyll extracts for sustainable electrocatalytic nitrate reduction to ammonia

Abstract

Single atom-based catalysts (SACs), due to their exceptional electrocatalytic behavior, have been explored for numerous applications such as the oxygen reduction reaction, thermo-catalytic CO₂ reduction, and other catalytic activities. Ammonia is one of the most used chemical compounds, and its electrochemical synthesis is a promising alternative as it is simple and cost-effective and shows selective tunability towards the synthesis. Utilizing single-atom electrocatalysts with comparatively low metal mass loading yet exceptional activity could be a better approach to maximize ammonia production. Herein, we report a practically viable magnesium SAC (MgN_xC) for the promising reduction of nitrate to ammonia. The catalyst was prepared using green leaf extracts of Spinacia oleracea via simple one-step pyrolysis. We optimized their synthesis temperature to scrutinize the effect of SAC formation and their variation on the catalysis efficacy. The MgN_xC650 catalyst, anchored on a defective graphitic matrix, exhibits the best-optimized potential of −0.58 V vs. RHE, a faradaic efficiency of 81.5 ± 2.9% and a yield rate of 392.5 ± 41.2 μmol h⁻¹ mg⁻¹_cat. with excellent repeatability. A comprehensive study of the nature of the heterojunction formed at the reactive interface of MgN_xC catalysts was carried out by Mott–Schottky analysis to probe the band structure of the intrinsically induced metal–semiconductor junction in the MgN_xC catalysts, followed by the analysis of parameters like flat band potential and carrier density correlation. DFT is employed to optimize the most stable reactive site and various reaction pathways for favorable nitrate reduction reaction with probable reaction intermediates were explored. Collectively, our work demonstrates a simple, cost-effective, and convenient way to synthesize SACs. Moreover, it provides clear evidence that chlorophyll moieties can be used as a template to prepare metal catalysts singly anchored on the graphitic carbon matrix.

---

1. Introduction

Due to the excessive use of nitrogen-based fertilizers and various chemicals in agriculture, the discharge of industrial wastewater effluents has significantly increased nitrate ion concentrations in our environment. This poses significant environmental challenges, particularly impacting water bodies and soil quality, ultimately causing ecological damage and harm to human health. In particular, nitrate ions cause serious health issues in living organisms, including non-Hodgkin lymphoma, methemoglobinemia, and blue baby syndrome, and are even carcinogenic.^1–3 While the World Health Organization (WHO) has established a maximum permissible nitrate (NO₃⁻) level in drinking water at 50 mg L⁻¹ (approximately 11 mg L⁻¹ as N-NO₃⁻), it is a matter of concern to note that over 45 million individuals may be at risk of exposure to water containing nitrate concentrations exceeding the permissible limit.³ Therefore, proper management practices for waste are the need of the hour for a clean and sustainable future. Various methods have been developed and implemented to manage and convert biomass wastes containing vegetable residues and other biodegradable waste into value-added products such as composite-based functional materials, carbon-based materials for electrodes (like graphitic carbon, activated carbon), and novel adsorbents for heavy metals like Cd²⁺ and Pb²⁺.^4–7 One of the advantages associated with nitrate is that it can be converted into ammonia electrochemically, offering numerous other benefits such as feasibility in storage, transportation, and large-scale ammonia production under eco-friendly conditions needed for the Haber–Bosch process.¹ It may be noted that ammonia (NH₃) has gained considerable interest over the years because 70% of it is used in various fertilizer industries as a chemical and a potential carbon-free hydrogen carrier. However, there is an emission of 1.8 tons of CO₂ for every 1 ton of NH₃ produced by the Haber–Bosch process.¹

Electrochemical catalysts for ammonia production from nitrate are at the forefront of research, expanding the horizons with various electrocatalysts and many experimental parameters. Over the past decade, a wealth of research reports have focused intently on the efficient production of ammonia by reducing nitrogen-rich sources, primarily in the form of dinitrogen (N₂) or nitrate (NO₃⁻) ions. In the context of N₂ and NO₃⁻ sources, the combination of the low solubility of N₂ in water with the strong triple bond strength of N≡ (941 kJ mol⁻¹) imposes significant limitations on the faradaic efficiencies, reaction rate, and NH₃ selectivity, resulting in an extremely meagre ammonia generation rate of less than 10 mmol h⁻¹ g_cat.⁻¹.^1,3,8 Fortunately, NO₃⁻ is an alternative source of nitrogen with a low dissociation energy of N=O (204 kJ mol⁻¹) and high solubility in aqueous solvents.^1,9

Single-atom catalysts (SACs) have recently emerged as prominent catalysts for a variety of applications, including electrocatalytic applications for ammonia production.^10,11 The demand for SACs is increasing day by day. Typically, the catalytic performance of SACs is contingent on factors such as size, distribution, interaction with surface atoms of the support, and structural modifications.^10–12 However, one of the challenges associated with SACs is their synthesis.^13–16 Methodologies such as traditional mixing-and-activation (wet/dry), metal–organic framework (MOF) conversion, thermal transformation, ultrahigh vacuum deposition, electrodeposition, and galvanic replacement have been demonstrated for the synthesis of SACs.^13,14 However, traditional mixing-and-activation and MOF conversion use harmful and flammable organic solvents or MOFs as precursors which also require additional carefully supervised processing steps to obtain SACs.¹⁴ Thermal transformation requires higher temperatures, whereas ultrahigh vacuum deposition requires a high vacuum with indispensable expensive apparatus. Also, electrodeposition and galvanic replacement methods have a very low production rate.^13,14 The detailed comparison can be found in Table S1(a) of the ESI.†

In the context of non-precious metals, alkaline earth metals have gained considerable attention due to their ability to stabilize single atoms by anchoring with nitrogen or carbon atoms and their high electro-catalytic activity. Magnesium is non-toxic as well as naturally abundant. Various applications have been reported in terms of its electrocatalytic activity, such as the oxygen reduction reaction (ORR), thermo-catalytic CO₂ reduction, and other catalytic activities.^10,11,17 However, the synthesis methodologies mainly involve either the sublimation-migration-anchoring strategy, traditional mixing-and-activation, or Mg-based MOF pyrolysis at a very high temperature of 900 °C. The detailed comparison can be found in Table S1(b) of the ESI.†^14,17,18

To mitigate the huge demand, it is of utmost importance to synthesize SACs in a single step at low temperature and with less toxic chemicals. Here, we have demonstrated one-step synthesis of a magnesium SAC from green leaves at 650 °C. It should be noted that green leaves of Spinacia oleracea (spinach) have been used as the precursor for Mg. Furthermore, the magnesium-based SAC has been explored to produce one of the highly value-added products, i.e. ammonia, by reducing the considerably higher concentration of nitrate ions. In this work, we have achieved an outcome by following the “Hit two targets with one arrow” strategy by using spinach residue to prepare electrocatalysts for converting nitrate (NO₃⁻), one of the major water pollutants, to the valuable product ammonia. Comparison of our catalyst with those reported in the literature indicates that our catalyst not only has excellent stability but also has quite impressive faradaic efficiency and yield rate.

2. Results and discussion

2.1. Structural characterization

2.1.1. X-ray diffraction. Four samples have been prepared by varying the pyrolysis temperature and are named MgN_xC550, MgN_xC650, MgN_xC750, and MgN_xC850. First of all, the samples were characterized with X-ray diffraction (XRD) studies. The diffraction pattern, as depicted in Fig. 1(a), is found with no prominent peak for MgN_xC650, while, as the pyrolysis temperature increases, the peaks corresponding to Mg nanoparticles start appearing. MgN_xC650 has no prominent peak, but a broad peak near 25° indicates the presence of a defective graphene matrix. On the other hand, MgN_xC850 shows the presence of various peaks. The peaks at 36.53° and 64.96° correspond to the (101) and (220) planes of Mg (PDF #35-0821), whereas the peak at 49.21° is attributed to the (102) plane of cubic Mg(OH)₂ and 74.24° for the (311) cubic plane of MgO.^19,20 An increase in the pyrolysis temperature increases Mg nanoparticles with increased MgO moieties and a decreased Mg(OH)₂. This might be due to the dissociation of Mg(OH)₂ into byproducts of surface-adsorbed OH/H₂O moieties.^21–23

Fig. 1 (a) XRD patterns and (b) Raman spectra of all MgN_xC catalysts. (c) HRTEM image of MgN_xC650, (d) HAADF-STEM image of MgN_xC650 where typical Mg single atoms are marked in red circles, and (e–j) HAADF-STEM EDS elemental mapping of MgN_xC650 with Mg, O, C, N, and combined distribution of elements, respectively.

2.1.2. Raman spectroscopy. Further understanding of the structure, crystallization, and defects in the MgN_xC samples as well as validation of the nitrogen doping in our catalyst was achieved through Raman studies. In all cases, the D and G peaks were observed near 1343.1 and 1592.3 cm⁻¹, respectively, as shown in Fig. 1(b). The intensity ratios between the D peak and G peak (I_D/I_G) for MgN_xC550, MgN_xC650, MgN_xC750, and MgN_xC850 were 1.003, 1.015, 0.996, and 0.994, respectively. The lower the ratio, the higher the graphitization, indicating fewer defective states and impurities in the graphene layers.²³ Normally, the D peak remains inactive in perfect graphite or graphene but becomes active in the presence of defects.^24,25 The observed peaks provide strong evidence for the formation of defective graphite structures, which is further supported by the appearance of a weak and broad peak centered at around 2800 cm⁻¹, corresponding to the overtone of the D band.²⁴ In summary, it reflects the dominance of defective states in our samples, which significantly improves surface activity and accounts for electrochemical reactivity.^26,27

2.2. Morphological characterization

2.2.1. Scanning electron microscopy (SEM). SEM images of MgN_xC650 and MgN_xC850, as shown in Fig. S1(a–d)†, reveal different morphologies, that is, a flower petal-like sheet structure and a ruptured sheet structure, respectively.^24,28 The pyrolysis temperature has significantly impacted the morphology of the MgN_xC samples. The SEM-EDS elemental mapping of a broad region of interest, depicted in Fig. S1(e–i)†, shows the uniform presence of different elements and confirms the uniform catalytic activity throughout the catalyst.

2.2.2. Transmission electron microscopy (TEM). The samples were also intensively subjected to TEM to probe the internal structure and the effect of pyrolysis temperature, as represented in Fig. 1(c) and S2.† MgN_xC650 is found to have a sheet-like structure without nanoparticles, whereas in the case of MgN_xC850, nanoparticles are clearly seen (Fig. S2†). We performed aberration-corrected TEM (Ab-TEM) for critical investigation. Fig. 1(c) shows the high-resolution TEM (HRTEM) image of MgN_xC650, confirming the porous graphitic carbon. High-angle annular dark-field atomic-resolution scanning TEM (HAADF-STEM) images in Fig. 1(d) and S4† reveal that there is a large amount of atomically dispersed metal atoms (the bright dots) anchored in MgN_xC650, whereas MgN_xC850 consist of nanoparticles, as shown in Fig. S3.†

Elemental mapping of MgN_xC650 and MgN_xC850, as manifested in Fig. 1(d–j), S4, and S5,† confirms the presence of atomically dispersed magnesium atoms and agglomerated nanoparticles, respectively. The average size of nanoparticles is ∼7 nm for MgN_xC850 (Fig. S6†).

2.3. Compositional characterization

2.3.1. X-ray photoelectron spectroscopy (XPS). The compositional analysis was performed through XPS studies. From the XPS spectra of MgN_xC650, as shown in Fig. 2, the C 1s core level peak is resolved into graphitic carbon (–C–C/–C=C), –C=N, –C–O/–C–N, and O–C=O groups at 284.6, 285.3, 286.5, and 288.8 eV, respectively,²⁵ whereas the N 1s peak exhibits pyridinic N, pyrrolic N, and graphitic N at 398.02 eV, 398.99 eV, and 400.77 eV, respectively.²⁴ The O 1s spectrum is resolved into carbonyl (–C=O), carboxyl (–COO) and hydroxyl (–OH) groups in the graphitic matrix found at 530.88, 532.68, and 533.58 eV, respectively, whereas above 532 eV, all the peaks correspond to physically adsorbed surface water molecules and hydroxyl groups.²³ For MgN_xC850 (Fig. S7†), C 1s shows peaks corresponding to –C=, –C=N, –C–N, and –COO (carboxyl) groups at 284.5, 285.5, 286.9, and 288.5 eV, respectively.²⁹ MgN_xC650 show Mg 1s with a peak at 1304.3 eV, which corresponds to the Mg–N bond, whereas MgN_xC850 show a peak at 1303.3 eV for MgO.¹² Moreover, the relative atomic concentrations of different elements were analysed, as shown in Fig. S8.†

Fig. 2 XPS spectra of MgN_xC650 including (a) Mg 1s, (b) C 1s, (c) N 1s, and (d) O 1s. BET analysis showing the (e) adsorption–desorption isotherm of MgN_xC650 and (f) correlations among the multipoint surface area, average pore size and total pore volume.

Also, the FTIR spectrum shown in Fig. S9† of MgN_xC850 has a peak near 550 cm⁻¹ for stretching frequency corresponding to the MgO bond, signifying the presence of MgO nanoparticles, whereas no such peak is observed in the case of MgN_xC650.

2.3.2. Thermogravimetric analysis (TGA). TGA is executed on MgN_xC650 and the data are presented in Fig. S10.† It clearly shows, in two different controlled environmental media, viz., air and N₂ gas, a temperature range from room temperature to 900 °C with a ramping rate of 10 °C min⁻¹. Under air, as the temperature is increased to 200 °C, the physically adsorbed water moieties and maximum moisture contents get eliminated and cellulose starts decomposing, leading to about 16% weight loss. Between 200 and 400 °C, around 39% weight loss was noticed, which could be attributed to the decomposition of complex aromatic structures, and over 400 °C to 900 °C, ∼28% devolatilization of other biomass components, such as cellulose and lignin, occurred, with incombustible residues of around 17%.^30,31 Under a N₂ atmosphere, about 15.4% weight loss was observed up to 200 °C due to the loss of moisture and entrapped water molecules. About 17% weight loss was observed between 200 and 400 °C which may be due to the high-temperature volatiles. Finally, about 52% of non-combusted residues remained between 400 and 900 °C.³¹ It signifies the presence of complex aromatic and carbonaceous components which help in the formation of a highly conductive graphitic bed.

2.3.3. BET surface area. As the surface area of a catalyst significantly affects the performance and catalytic activities, we carried out BET surface area measurements of all the prepared catalysts, viz., MgN_xC550, MgN_xC650, MgN_xC750, and MgN_xC850. Fig. 2(e) presents the nitrogen adsorption–desorption isotherm of MgN_xC650, which corresponds to a type IV isotherm with a type H4 hysteresis loop.³² The presence of mesoporous structures has been previously reported and these structures are known to enhance the diffusion and transport of reactants to the active sites, ultimately promoting greater efficiency by enhancing the kinetics of reactions.²⁴ The details of BET analysis with various parameters are provided in Table S3.† The surface areas of MgN_xC550, MgN_xC650, MgN_xC750, and MgN_xC850 are 611.31, 369.44, 301.01, and 83.75 m² g⁻¹, respectively.

2.4. Electrocatalytic NO₃⁻ reduction

To explore the NO₃⁻reduction reaction (NO₃⁻RR) performance of the MgNx_xC, we first conducted linear sweep voltammetry (LSV) using 1 M KOH as the basic medium supporting electrolyte with and without 1 M KNO₃ as the nitrate source, as shown in Fig. S11.† The strikingly increased current density in the presence of NO₃⁻ relative to that without NO₃⁻ signified the remarkable occurrence of NO₃⁻RR. Fig. 3(a–c) compares the onset potential and overpotential vs. RHE, which reveals that MgN_xC650 is the best catalyst among the synthesized electrocatalysts. As the potential increases to more negative, the hydrogen evolution reaction (HER) competes with the nitrate reduction reaction (NO₃⁻RR). The HER is highly competitive as it needs fewer electrons than the NO₃⁻RR. Further characterization of the reaction products revealed that the production rate of ammonia is up to 393.14 μmol h⁻¹ mg⁻¹_cat. with a faradaic efficiency (FE) reaching the maximum value of 78.82 ± 1.58% by MgN_xC650, whereas MgN_xC550, MgN_xC750, and MgN_xC850 exhibit ammonia production rates of 87.78 ± 6.48, 334.71 ± 34.49, and 224.82 ± 28.71 μmol h⁻¹ mg⁻¹_cat., with faradaic efficiencies of 22.58 ± 1.67, 51.54 ± 5.31, and 26.00 ± 3.32%, respectively (Fig. 3(d–f)).

Fig. 3 (a) Linear sweep voltammetry (LSV) measurements, (b) current density (j/mA cm⁻²) comparison (@0.58 V), (c) onset potentials and overpotential of MgN_xC electrocatalysts @0.58 V, (d) NH₃ and NO₂ faradaic efficiency, (e) yield rate, (f) comparison of faradaic efficiencies and yield rates of different MgN_xC catalysts, (g) MgN_xC650 potential optimization for faradaic efficiencies of NH₃ and NO₂⁻, (h) yield rate of NH₃ and NO₂⁻ of MgN_xC650 and (i) comparison of the faradaic efficiency and yield rate of the MgN_xC650 catalyst.

Considering MgN_xC650 as our best-performing electrocatalyst, we proceeded further with detailed kinetic studies. We began by looking into the behaviour of the electrocatalyst at different concentrations of 0.1, 0.5, and 1 M KNO₃ with 1 M KOH serving as the supporting electrolyte and found that 1 M KNO₃ has the best nitrate reduction activity as compared to 0.1 M KNO₃ and 0.5 M KNO₃, as depicted in Fig. S12.† Therefore, we kept 1 M KNO₃ as the optimum concentration to perform all further experiments.

All electrolysis studies were performed using chronoamperometric measurements in an H-type cell. To evaluate the optimum potential for NO₃⁻RR, all the chronoamperometry measurements were conducted with different potentials for 1 hour. As the potential became more negative, the ammonia production also enhanced, indicated by the increasing current density shown in Fig. 3. We found that the optimum potential for MgN_xC650 for nitrate reduction is −0.58 V vs. RHE, with a faradaic efficiency of 81.46 ± 2.92% and a yield rate of 392.51 ± 41.16 μmol h⁻¹ mg⁻¹_cat.

The best catalyst, MgN_xC650, was subjected to a repeatability test at the optimized potential for 30 minutes for each cycle. It showed superior performance for ten continuous cycles using the same drop-cast catalytic ink without any degradation (Fig. S13†). Moreover, we conducted time-dependent experiments at different intervals for 5 h. It was observed that as the time increased, the yield of nitrite reduced and the concentration of ammonia increased in a sequence (Fig. S14†).

Overall, MgN_xC650 performs outstandingly in the electrochemical reduction of nitrate to ammonia.

The reaction products at all potentials were analysed using three different techniques, viz., UV-visible spectroscopy, ion-exchange chromatography, and (¹H) NMR spectroscopy. All the quantifications were processed by rooting the calibration curve as a reference, as shown in the ESI.†

2.5. Quantification of NO₃⁻RR products

2.5.1. UV-visible spectroscopy. Here, we employed Nessler's method to detect the presence of ammonia, which utilizes mercury(II) iodide or potassium iodide in an alkaline solution, with sodium or potassium hydroxide serving as the alkaline agent. When ammonia (NH₃) reacts with this reagent, it forms a yellow compound (ESI†). This compound can be quantified by using a spectrophotometer in the wavelength range between 410 and 425 nm.^33,34 Detailed procedural information is provided in the ESI.†

As the concentration of ammonia increases, the yellow colour becomes more intense, transitioning from a pale yellow to a deeper yellow-orange or even reddish-brown hue, as can be seen in Fig. S14.† To achieve stable sensitivity and allow concurrent measurements for ammonia quantification, the reaction products were subjected to reagent exposure for 2 hours. Fig. S15† shows the calibration curve for the detection of ammonia. Reports have shown the formation of intermediate nitrite (NO₂⁻) during the ammonia conversion, so it is important to identify the presence of nitrite in our reaction product.^1,35 Here, we utilised the modified Griess reagent for our study. Griess's method, i.e., Griess's reagent, has been reported to quantify nitrite anions accurately.³⁶ The absorbance of the resulting colour is measured spectrophotometrically after the reagent exposure time of 20 minutes at a specific wavelength of around 540–546 nm, as seen in Fig. S16.†

2.5.2. Nuclear magnetic resonance (NMR) spectroscopy. Here, we have employed ¹H NMR spectroscopy to quantify the presence of ammonia in the reaction product. To confirm the origin of the produced ammonia, chronoamperometric electrolysis experiments were conducted using both nitrogen isotopes, namely, ¹⁴N and ¹⁵N, which were sourced from K¹⁴NO₃ and K¹⁵NO₃, respectively, all under identical conditions with 1 M KOH, as shown in Fig. 4(a). All the NMR measurements were conducted using DMSO-d₆ as a solvent and maleic acid as an internal standard in a highly acidic medium with a pH range between 2 and 3 to maximize the formation of ammonium cations (NH₄⁺) with an ideal time duration of 20 min for all respective experiments.

Fig. 4 (a) NMR of K¹⁵NO₃ and K¹⁴NO₃ as a source of nitrogen (controlled experiments), (b) double layer capacitance (C_dl) analysis, (c) ECSA measurements, (d) band structure of the Mott–Schottky heterojunction formed at the interface of MgN_xC650 with the KOH electrolyte after attaining equilibrium, (e) band structure when we perturb the catalyst with an applied potential equal to the flat band potential (V_fb), (f) inversion of band at the junction on applying a potential exceeding V_fb, (g) M–S measurement for MgN_xC650, (h) comparison of the flat band potential and carrier density of the MgN_xC catalyst in 1 M KOH, and (i) M–S measurements before and after introducing 1 M KNO₃.

The isotopes ¹⁴N and ¹⁵N yield distinctive triplet and doublet signals in ¹H NMR spectra.³³ Consequently, detecting a doublet signal resulting from the formation of ¹⁵NH₃ serves as conclusive evidence, verifying that the ammonia originated from K¹⁵NO₃ and not from other impurities or sources. We have quantified ammonia concentration and conducted a comparative analysis with other quantification techniques. Detailed results and information can be found in the ESI,† along with the standard calibration curve in Fig. S17.†

2.5.3. Ion exchange chromatography. We analysed the ions present in the reaction products collected from the cathodic compartment following electrolysis using chronoamperometry. Our investigation focused on identifying all possible ions, with particular emphasis on NH₃/NH₄⁺ and NO₂⁻ ions, using appropriately concentrated solutions of the reaction products.

The examination started with the preparation of a standard calibration curve (Fig. S18 and 19†) and then carefully diluting the concentrated reaction product (pH 10–12) from the cathodic compartment after electrolysis. A detailed and comprehensive protocol outlining our methodology can be found in the ESI.†

2.6. Cyclic voltammetry, C_dl, and ECSA measurements

The electrochemical surface area (ECSA) for all the catalysts was estimated using cyclic voltammetry (CV) measurements in the non-faradaic region of the potential window, depicting the closed symmetric loop cyclic voltametric curves for all the MgN_xC catalysts (Fig. S20†). To calculate the double-layer capacitance (C_dl), the relationship between the current density (Δj/mA cm⁻²) and the scan rate at different intervals was considered, as shown in Fig. 4(b and c). From Fig. 4(b), the slope (C_dl) is found to be 1.1, 42.84, 22.69, and 5.59 mF cm⁻² for MgN_xC550, MgN_xC650, MgN_xC750, and MgN_xC850, respectively. C_dl values are proportional to the ECSA of a material. MgN_xC650 accounted for the maximum C_dl, which further helps increase activity and promotes faster reaction kinetics. The variation of ECSA is shown in Fig. 4(c). The ECSA is maximum for MgN_xC550.

2.7. Mott–Schottky (M–S) analysis

The structural and chemical characterization of the best-performing MgN_xC650 catalysts revealed that the Mg single atoms are isolated and evenly distributed over the nitrogen-doped carbon matrix. At the interface of the electrode–electrolyte medium, the formed heterojunction is expected to behave like a Mott–Schottky (M–S) junction. The Mg SAC with excessive nitrogen in the C–N matrix behaves like an n-type semiconductor, which creates an electrolyte–semiconductor interface. The electron flow from the MgN_xC centre, having a comparatively higher Fermi level, is expected towards the lower lying level of the electrolyte medium, resulting in band bending at the interface, as shown in Fig. 4(d). The system is perturbed with a small, applied potential with a constant frequency of 1 kHz and an amplitude of 20 mV in the non-faradaic window. The potential thus needed to make the band flat from the equilibrium state is denoted as the flat band potential (V_fb), as depicted in Fig. 4(e and f). This parameter can be extracted from the X-intercept of our M–S plot. The lower flat band potential favours easy charge transfer between the catalyst and the electrolyte at the interface and hence favours better reaction kinetics.^26,27,37,38 The slope of the inverse capacitance square value (1/C²) vs. potential plot is inversely proportional to the charge carrier density of the MgN_xC catalysts. Fig. 4(g) shows the M–S measurement for MgN_xC650 with clearly visible zones of accumulation, depletion, and inversion in 1 M KOH electrolyte at a constant frequency of 1 kHz. This confirms the M–S behaviour at the catalyst–electrolyte interface. The comparison of the M–S plots of MgN_xC550, MgN_xC650, MgN_xC750, and MgN_xC850 is shown in Fig. 4(h). We can extract two parameters, the flat band potential and the charge carrier density for comparison. It is found that there is a gradual shift in the X-intercept of the MgN_xC650 catalyst towards the positive side of the applied potential, indicating the lower flat-band potential and minimal band bending at the semiconductor-electrolyte interface. Also, the effective charge carrier density of all the MgN_xC catalysts in the presence of 1M KOH lies almost in the same range, indicating their almost equal dopant concentration.

To understand the effect of the addition of our reactant KNO₃ species on the band structure, we have performed a similar M–S analysis with all parameters the same in a 1 M KOH + 1M KNO₃ electrolyte medium. As shown in Fig. 4(i) and S21,† we could observe a clear deviation in the slope of the M–S plots in the linear region for all the MgN_xC samples. This observation essentially points to the fact that on introducing KNO₃, the electrolyte's redox energy level is altered, changing the degree of band bending and the effective charge carriers available for the redox process at the interface. When we compare this effect of electrolyte change in all three samples, we could observe that MgN_xC650 adjusted itself to a state of less band bending with more effective concertation of charge carriers at the interface compared to MgN_xC550, MgN_xC750, and MgN_xC850. The change in the slope of the M–S plots in the bare KOH electrolyte and the mixture of KOH and KNO₃ also indicates that MgN_xC650 has better selectivity towards nitrate reduction over the hydrogen evolution reaction. Thus, overall, M–S gives us a clear picture of the band structure at the interface, verifying the observed activity trend of the MgN_xC samples towards nitrate reduction.

2.8. Theoretical studies of NO₃⁻ reduction

To determine the most favorable type of N-doped structure, formation energies of g-N, pyrrolic-N (pyl), and three different pyridinic structures with two (pyn2), three (pyn3) and four (pyn4) N atoms were calculated, and the results show that pyridinic-N is more stable than pyrrolic-N (Fig. S22†), in line with the experimental observations showing a large presence of g-N followed by pyridinic-N. Qinghong Yuan et al. also observed a similar stability trend for N-doped graphene structures.³⁹ Furthermore, Mg doping favors the formation of a tetracoordinated Mg–N₄ structure (Fig. S23†) with significant charge transfer between Mg (+1.61) and N (−1.43). For pyn-3, the Mg atom moves out of plane, resulting in less effective coordination and charge transfer between the Mg (+1.42) and N (−1.36) atoms, lowering its stability (Fig. S24†). The final substrate structure was prepared by adding some graphitic nitrogen atoms to resemble the experimental N ratio in the substrate (Fig. S24(g)†).

Nitrate reduction to ammonia (NRA) is a complex pathway involving a large number of reaction intermediates. We investigated different reaction pathways leading to the formation of NH₃, NO₂, NO, and NH₂OH, Fig. 5, Fig. S25, and S26.† For the reduction to initiate, suitable adsorption of the initial nitrate species on the catalyst surface is the primary requirement. The nitrate ion adsorbs on the substrate in a bidentate manner through two oxygen atoms and is thermodynamically favorable (ΔG_NO3* = −0.69 eV). Bader charge analysis of *NO₃ shows efficient charge transfer from the substrate to the intermediate, reducing the charge on the N atoms in the substrate from −1.42 to −1.34 e- and increasing the N–O bond length from 1.24 Å in NO₃⁻ to 1.30 Å in *NO₃ (Fig. S27(a)† before and (b) after adsorption). The elongation of N–O bonds in *NO₃ suggests the activation of the adsorbate at the active site. To determine the minimum energy pathway for the reaction, three probable pathways are explored in depth as schemes NRA1 (Fig. 5), NRA2 (Fig. S25†), and NRA3 (Fig. S26†), where scheme NRA1 is found to be more favorable, i.e., NRA1: *NO₃ → *HNO₃ → *NO₂ → *HNO₂ → *NO → *NHO → *NHOH → *NH → *NH₂ → *NH₃, as shown in Fig. 5.

Fig. 5 (a) Adsorbed configurations of the different intermediates leading to NH₃ as the main product on the Mg-SAC in the minimum energy pathway and (b) free energy diagram for nitrate reduction to ammonia on the Mg-SAC. The blue dashed line represents the minimum energy pathway with a limiting potential of U_L = −0.76 V vs. RHE. The red dashed lines show desorption of NO₂(g) and NO(g).

After activation of *NO₃ on the substrate, its further reduction requires the transfer of a proton–electron pair to the activated species. Further hydrogenations of *NO₃ to form *HNO₃ and *NO₂ to form *HNO₂ are endothermic, with potential barriers of 0.76 eV and 0.64 eV, respectively. Subsequent protonation and dehydration of *HNO₃ and *HNO₂ to form *NO₂ and *NO are exothermic with ΔG of −2.59 and −0.20 eV, respectively. These findings agree with a previous work on the Fe–N₄ SAC by Wu et al.³⁵ and other SACs.⁴⁰ The desorption of *NO₂ to NO₂(g) is more energy demanding with a barrier of 0.90 eV as compared to the hydrogenation of *NO₂ to form *HNO₂ (+0.64 eV) (Fig. 5). Protonation and subsequent loss of a water molecule from the *HNO₂ molecule give *NO, which has the possibility to bind to Mg through either the O or N atom. *NO binding to Mg gives rise to the NRA1 and NRA2 pathways, while *ON binding leads to pathway NRA3. Considering NRA1, the desorption barrier for NO(g) from *NO is endothermic, which needs a large potential of 2.24 eV. We would like to emphasize that in our experimental findings as well, NO₂(g) was detected as an intermediate product, while NO(g) was not observed. This agrees with the DFT calculations, showing a large potential barrier of NO desorption (2.24 eV) compared to the desorption of NO₂ (0.90 eV). The following proton–electron transfers to *NO leading to *NHO (NRA1, ΔG = −0.67 eV) and *NOH (NRA2, ΔG = +0.56 eV) are reported to be the potential determining steps (PDS) in many previous reports.^41,42 The reaction is likely to proceed through NRA1 to *NHO as it is thermodynamically more favorable. Furthermore, proton–electron transfer to the oxygen atom of *NHO leads to *NHOH, which upon subsequent protonation and dehydration leads to *NH. From *NH, continuous hydrogenations, which are all exothermic, lead to *NH₃ as the final product (Fig. 5). The adsorbed configurations of intermediates for the minimum energy path (NRA1) are shown in Fig. 5(a).

As per NRA3, the intermediate *ON species is adsorbed on the Mg-SAC through the O atom. The adsorption of *ON to Mg (G = −2.26 eV) is stronger and more favorable than *NO (G = −2.08 eV), owing to the stronger affinity of Mg towards the oxygen atom.¹⁸ Consequently, the Mg–O distance in *ON is shorter (2.02 Å) relative to the Mg–N distance (2.08 Å) in *NO. However, the subsequent proton–electron transfer to *ON to form either *HON or *ONH has a large potential barrier of 1.17 eV or 1.94 eV, respectively (Fig. S26†). As a result, the progress of the reaction is hindered through this pathway. The various reactions and their respective referencing parameters can be found in Tables S4 and S5.†

In conclusion, our DFT calculations reveal the different reaction pathways for nitrate reduction to ammonia on the Mg-SAC. The protonation of *NO₃ to form *HNO₃ is the potential limiting step with a limiting potential of −0.76 V (vs. RHE). Excellent selectivity towards ammonia results from the strong adsorption of the reaction intermediates onto the catalyst. The detection of NO₂(g) may be attributed to its low desorption barrier as compared to NO(g).

3. Experimental section

3.1. Essential materials and synthesis of electrocatalysts

3.1.1. Essential experimental components and supplies. A comprehensive inventory of various chemicals, reagents, and solutions used in our study is as follows: chlorophyll extracts from Spinacia oleracea (fresh spinach leaves), coffee powder (70% coffee + 30% chicory) from Nestle, India, ammonium chloride (NH₄Cl, mol. wt. 53.491 g mol⁻¹) (CAS 12125-02-9), ammonium molybdate ((NH₄)₆Mo₇O₂₄·4H₂O, mol. wt. 1235.86 g mol⁻¹) (CAS 12054-85-2), from Sigma-Aldrich Co., USA, potassium hydroxide (KOH, mol. wt. 56.11) (CAS 1310-58-3), potassium nitrate AR (KNO₃ or K¹⁴NO₃, mol. wt. 101.10 g mol⁻¹) (CAS 7757-79-1), maleic acid AR (C₄H₄O₄, mol. wt. 116.072 g mol⁻¹) (CAS 110-16-7), ethanol absolute (C₂H₅OH, mol. wt. 46.069 g mol⁻¹, 99.9%) (UN:1170), DMSO-d₆ (C₂D₆OS, mol. wt. 84.17 g mol⁻¹, 99.8 atom% D) (CAS 2206-27-1), hydrochloric acid (HCl, mol. wt. 36.46 g mol⁻¹, 32%**) (CAS 7647-01-0), sulphuric acid (H₂SO₄, mol. wt. 98 g mol⁻¹, 98%**) (CAS 7664-93-9), hydrofluoric acid (HF, mol. wt. 20.006 g mol⁻¹, 40%) (CAS 7664-39-3), sodium nitrite (NaNO₂, 68.9953 g mol⁻¹, >98.5%(T)) (CAS 7632-00-0)], Nessler's reagent (volume: 125 ml) (SAP code: 11007LM125), Griess's reagent (modified) (EC Number 215-981-2, perfluorinated resin solution containing Nafion-1100 W (CAS 31175-20-9), potassium nitrate nitrogen isotope labelled (K¹⁵NO₃) (mol. wt. 102.10 g mol⁻¹, 98 atom% ¹⁵N) (CAS 57654-83-8).

3.1.2. Synthesis of electrocatalysts. We have utilized the green leaves of Spinacia oleracea (spinach leaves) and fine-filtered coffee powder as major precursors in addition to ammonium chloride and ammonium molybdate. The role of each precursor has been outlined in the Experimental section in the ESI.† In summary, the materials and steps involved in our procedure can be briefly outlined as follows (Fig. S28–36†):

1. Chlorophyll extract preparation:

Adequate volume (ml) of chlorophyll extracts was obtained from fresh spinach leaves using the ethanol-mediated solvent extraction method. The chlorophyll extracts were concentrated by heating them in a hot oven at 80 °C until they formed a thick slurry paste.

2. Coffee filtration and preparation:

A certain amount of coffee powder was taken and sieved to filter out finer particles. The coffee powder was ground for 30 minutes. To prevent solidification upon exposure to moisture, approximately 2 ml of ethanol was added to make it more viscous and easier to grind.

3. Mg-intermediate preparation:

Slurry solutions of chlorophyll extract (25 ml) and the coffee filtrate (20 g) were mixed. A small amount of ammonium chloride (0.15 g) and ammonium molybdate (0.002 g) was added. The mixture was placed in a muffle furnace at 250 °C for 5 hours and then allowed to cool at room temperature naturally. The resulting intermediate material, referred to as the “Mg-intermediate”, was collected.

4. Mg-intermediate processing:

The Mg-intermediate (approximately 10 g) was used for X-ray analysis. It was washed with deionized water and acetone to remove impurities until consistent X-ray patterns were achieved.

5. Pyrolysis and final products:

The Mg-intermediate was subjected to pyrolysis at different temperatures (550 °C, 650 °C, 750 °C, and 850 °C) in a N₂ environment. The ramping rate was set at 10 °C per minute with a dwelling time of 3 hours. After pyrolysis, the resulting products were analyzed by X-ray diffraction. They were leached with a dilute acidic solution of hydrofluoric acid until the X-ray diffraction pattern matches with the previous diffraction pattern, which indicates the removal of impurities. Furthermore, the samples were thoroughly washed with deionized water unless clear work-up water with bottom-settled black powder was obtained and finally vacuum-dried. These products were named MgN_xC550, MgN_xC650, MgN_xC750, and MgN_xC850, corresponding to the respective pyrolysis temperatures of 550 to 850 °C, respectively. The corresponding X-ray diffraction patterns with the complete procedure, including the preparation of the working electrode, can be found in the ESI.†

3.2. Characterization of electrocatalysts

The analytical approach involved the use of an X-ray diffractometer, specifically the PANalytical instrument, which employed monochromatized Cu Kα radiation with a wavelength of 1.54 Å.

We employed advanced microscopy techniques to gain direct insights into the sample's physical morphology. With a scanning electron microscope (SEM), the FESEM FEI inspected 50 models for surface visualization. Additionally, a transmission electron microscope (TEM), the Tecnai T20 model, operating at an accelerating voltage of 200 kV, provided detailed internal structural information. Furthermore, to achieve atomic-level visualization of our catalysts, we employed HRTEM, STEM, and aberration-corrected transmission electron microscopy (Ab-TEM) techniques using 60–300 kV low base Titan® Themis™ with a monochromator, a CEOS probe corrector for Cs aberration-corrected STEM and a Quantum 965 Gatan Imaging filter (post column) for EELS analysis, allowing for even finer structural and atomic level insights.

For a comprehensive understanding of the catalytic material's composition, including the elemental composition, oxidation states, and the nature of chemical bonds present, we utilised the capabilities of the X-ray photoelectron spectrometer AXIS ULTRA DLD Kratos. This state-of-the-art instrument featured a monochromatic Al Kα radiation source emitting at 1486.6 eV, enabling high-resolution X-ray photoelectron spectroscopy (HR-XPS) analysis at the elemental level.

All the UV-Vis spectrophotometric measurements were performed using a Hitachi U-2900 double-beam spectrophotometer. Ion-exchange chromatographic measurements were performed using Eco IC Metrohm.

NMR spectra were recorded using the AV400, Bruker 400 MHz highresolution multinuclear FT-NMR spectrometer, and analysis was completed using the software Mestrenova.

3.3. Electrochemical measurements

The electrochemical investigations were carried out utilizing a CHI. Model 750E Inc. electrochemical workstation based in Austin, USA. This sophisticated instrument enabled a three-electrode configuration. A commercial glassy carbon electrode (GCE) with a geometric surface area of 0.196 cm² was employed as the working electrode for nitrate reduction, while an Ag/AgCl electrode served as the reference electrode. The counter electrode was a platinum wire. Additionally, an H-type electrochemical cell was implemented, featuring two compartments connected by a proton conductive membrane, which was modified to create a three-electrode working system by incorporating Nafion as a semi-permeable membrane (Fig. S37†).

To ensure an impeccably smooth electrode surface, the rotating ring-disk electrode (RRDE) underwent meticulous polishing using fine white alumina powder with an average particle size of 0.005 μm for a duration of 10 minutes. Subsequently, approximately 8 mg of catalyst material was introduced into a volume of 1000 μL of the ethanol–water mixture with a 1 : 1 ratio. To enhance the dispersion of the electrocatalyst, 200 μL of a 20 wt% Nafion-1100W solution, a perfluorinated resin solution, was added to 1000 μL of the previous catalyst mixture. This solution was then subjected to 20 minutes of sonication and named electrocatalytic ink. For each electrochemical measurement, a 5 μL aliquot of the prepared electrocatalyst ink was utilized by drop casting on the RRDE, with careful attempt to maintain a geometrical surface area of 0.196 cm², and dried well for 10 minutes. To minimize uncompensated solution resistance and potential interferences, iR compensation was applied during each measurement. All linear sweep voltammetry (LSV) measurements were conducted at a scan rate of 5 mV s⁻¹, employing the rotating disk electrode (RRDE), unless specified otherwise.

3.4. Mott–Schottky measurements

The measurements were done in an 1 M KOH electrolyte solution purged with Ar gas to remove the unwanted gaseous impurities. A constant current frequency of 1 kHz and amplitude of 20 mV in the potential range of −0.5 to 1.5 V were used for the M–S measurements. The system's impedance was measured by perturbing the system with a certain voltage and then converted into the capacitance of the space charge region. The inverse square of capacitance has a linear relationship with the applied potential, as shown in the famous M–S equation,

Here, ‘C’ represents the space charge capacitance, typically determined via impedance measurements, ‘A’ stands for the cross-sectional area of the depletion region, ‘ε’ denotes the dielectric constant of the semiconductor material, ‘ε₀’ signifies the permittivity of free space, ‘e’ symbolizes the elementary charge of an electron, ‘N_d’ represents the carrier density, ‘V’ indicates the applied potential, ‘V_fb’ corresponds to the flat band potential, ‘k_β’ is the Boltzmann constant, and ‘T’ stands for the absolute temperature. These parameters play crucial roles in understanding and describing various aspects of the semiconductor behavior of our heterojunction. According to this equation, the M–S plot will be linear, with the slope inversely proportional to the carrier density, and the X-intercept will provide the flat band potential (V_fb).

3.5. Faradaic efficiency and yield rate calculations

Faraday efficiency plays a crucial role in analysing the performance of any electrocatalyst, especially in our study focused on nitrate reduction. The faradaic efficiency (FE) is calculated using the following formula for all measurements:³

Additionally, the yield rate (production) of ammonia is determined as follows:

Similarly, for the Faraday efficiency concerning nitrite (NO₂⁻), the following formula is applied:

Similarly, the yield rate of the nitrite anion is calculated as follows:

In both nitrate and nitrite reduction analyses, 8 and 2, respectively, account for the number of electrons involved in the process. F represents Faraday's constant (96485 C mol⁻¹), C denotes the concentration determined from calibration plots, V_H-Cell signifies the cathodic compartment volume during electrolysis, Q represents the charge obtained from chronoamperometric plots, t stands for the electrolysis time, and m_cat. is the mass loading of the electrocatalyst used in each respective measurement.

3.6. Theoretical studies

All electronic structure calculations were performed using density functional theory, as implemented in the Vienna Ab initio Simulation Package (VASP) by considering spin polarization.⁴³ All electron projector-augmented wave (PAW) pseudopotentials for Mg, C, N and H atoms were used to describe the electron–ion interactions.⁴⁴ Electronic exchange and correlation were approximated by Perdew–Burke–Ernzerhof generalized gradient approximation (GGA).⁴⁵

The plane wave kinetic energy cutoff of 520 eV was used to optimize all the substrate structures and the reaction intermediates. The Brillouin zone was sampled using a 3 × 3 × 1 Monkhorst–Pack k-point grid for the monolayer two-dimensional graphene with a 7 × 7 supercell, while a 2 × 2 × 1 Monkhorst–Pack grid was used for the intermediates. The graphene monolayer, Mg-doped substrate structures, and the reaction intermediates were relaxed using a conjugate gradient scheme until the energies and the forces converged to 10^–8 eV and 0.001 eV Å⁻¹, 10^–5 eV and 0.01 eV Å⁻¹, and 10^–5 eV and 0.01 eV Å⁻¹, respectively.

The images were separated by a minimum vacuum of 20 Å along the z-direction to prevent spurious interactions between the periodic images. The DFT-D3 method of Grimme was used to calculate the van der Waals correction.⁴⁶ A Γ-centered mesh was used for the optimization of gaseous molecules HNO₃ and H₂, for referencing the intermediates, in a cubic box of dimensions 15 Å × 15 Å × 15 Å. The details of computational and adopted formulations can be found in the ESI.†

4. Conclusions

We successfully derived a green synthesis of the magnesium single-atom catalyst from chlorophyll extracts for sustainable electrocatalytic nitrate reduction to ammonia. MgN_xC650 has shown the best electrochemical activity towards the conversion of nitrate to ammonia. It has shown excellent activity due to magnesium serving as a single atom electrocatalyst at the optimum potential of −0.58 V vs. RHE with a faradaic efficiency of 81.46 ± 2.92% and a yield rate of 392.51 ± 41.16 μmol h⁻¹ mg⁻¹_cat. The effect on the morphology and performance by tuning the pyrolysis temperature is thoroughly studied. The catalyst's band structure and charge transfer efficiency were analysed through Mott–Schottky analysis. The novelty of this work lies in the cost-effective, green synthesis of electroactive catalysts from green resources. It avoids the use of any heavy metals and utilizes the environment-friendly magnesium metal using the green leaves extracts of Spinacia oleracea. This green route can be used as a template to synthesize other metal single atoms for respective catalyst preparation.

Author contributions

K.K.: primarily responsible for conducting the majority of experiments, encompassing materials synthesis, methodology development, data curation, analysis, image processing, and manuscript writing and editing. P.T.: theoretical nitrate reduction studies and mechanistic analysis, and manuscript writing and editing. G.R.: XPS data acquisition and analysis, SEM imaging, Mott–Schottky analysis, and contributed to manuscript writing and editing. D.K.: conducted TEM imaging and its analysis. D.B.G.: conducted electrochemical measurements and manuscript writing and editing. N.G.M: ion-exchange chromatography characterization and analysis. K.R.: provided ion-exchange chromatography characterization facilities. S.K.M.: TEM imaging and related funding acquisition. A.K.S.: theoretical studies supervision and funding acquisition. K.K.N.: conceptualization, funding acquisition, supervision, scientific discussions, project administration, and manuscript review and editing. All authors collectively approved the final version of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

K. K. extends sincere gratitude to the Ministry of Education (MoE) for providing the valuable research scholarship. The authors also wish to express their appreciation to the Department of Science and Technology (DST), India, for generously funding the BET adsorption–desorption facility through DST-SERB (EMR/2016/005843) and DST-FIST (SR/FST/PS11-009/2010). Furthermore, the authors acknowledge the Materials Research Centre (MRC) for providing the XRD facility, the Centre for Nanoscience and Engineering (CeNSE), Advanced Facility for Microscopy and Microanalysis (AFMM), and the Inorganic and Physical Chemistry (IPC) department at IISc for facilitating access to XPS, SEM, TEM and NMR facilities. The authors thank the Material Research Centre (MRC) and the Supercomputer Educational and Research Centre (SERC), Indian Institute of Science (IISc), Bangalore for the computational facilities. P. T. and A. K. S. acknowledge the support from the Science and Engineering Research Board (SERB), Department of Science and Technology (DST), Government of India (File Number: CRG/2021/000633). Special thanks to Dr Ravi Nandan, Dr Omeshwari Bisen and Dr Hemam Rachana Devi at the Materials Research Centre, IISc Bangalore, for their invaluable support with insightful scientific discussions. A portion of this research was conducted at the Indian Institute of Technology, Madras (IITM) and remaining at the Indian Institute of Science (IISc), Bengaluru, India.

References

1. J. Wang, C. Cai, Y. Wang, X. Yang, D. Wu, Y. Zhu, M. Li, M. Gu and M. Shao, ACS Catal., 2021, 11, 15135–15140.
2. J. Gao, B. Jiang, C. Ni, Y. Qi, Y. Zhang, N. Oturan and M. A. Oturan, Appl. Catal., B, 2019, 254, 391–402.
3. S. Garcia-Segura, M. Lanzarini-Lopes, K. Hristovski and P. Westerhoff, Appl. Catal., B, 2018, 236, 546–568.
4. C. Zhou and Y. Wang, Sci. Technol. Adv. Mater., 2020, 21, 787–804.
5. M. Feizi and M. Jalali, J. Taiwan Inst. Chem. Eng., 2015, 54, 125–136.
6. N. Tripathi, C. D. Hills, R. S. Singh and C. J. Atkinson, npj Clim. Atmos. Sci., 2019, 2, 35.
7. K. Wang and J. W. Tester, Green Energy Resour., 2023, 1, 100005.
8. L. Mi, Q. Huo, J. Cao, X. Chen, H. Yang, Q. Hu and C. He, Adv. Energy Mater., 2022, 12, 1–10.
9. G. F. Chen, Y. Yuan, H. Jiang, S. Y. Ren, L. X. Ding, L. Ma, T. Wu, J. Lu and H. Wang, Nat. Energy, 2020, 5, 605–613.
10. B. Mohanty, B. K. Jena and S. Basu, ACS Omega, 2020, 5, 1287–1295.
11. B. Zhang, T. Fan, N. Xie, G. Nie and H. Zhang, Adv. Sci., 2019, 6, 1901787.
12. X. B. Shao, Y. Nian, S. S. Peng, G. S. Zhang, M. X. Gu, Y. Han, X. Q. Liu and L. B. Sun, Sci. China: Chem., 2023, 66, 1737–1743.
13. Y. Chen, S. Ji, C. Chen, Q. Peng, D. Wang and Y. Li, Joule, 2018, 2, 1242–1264.
14. F. D. Speck, J. H. Kim, G. Bae, S. H. Joo, K. J. J. Mayrhofer, C. H. Choi and S. Cherevko, JACS Au, 2021, 1, 1086–1100.
15. J. Shin, Y. J. Lee, A. Jan, S. M. Choi, M. Y. Park, S. Choi, J. Y. Hwang, S. Hong, S. G. Park, H. J. Chang, M. K. Cho, J. P. Singh, K. H. Chae, S. Yang, H. Il Ji, H. Kim, J. W. Son, J. H. Lee, B. K. Kim, H. W. Lee, J. Hong, Y. J. Lee and K. J. Yoon, Energy Environ. Sci., 2020, 13, 4903–4920.
16. J. Jones, H. Xiong, A. T. DeLaRiva, E. J. Peterson, H. Pham, S. R. Challa, G. Qi, S. Oh, M. H. Wiebenga, X. I. P. Hernández, Y. Wang and A. K. Datye, Science, 2016, 353, 150–154.
17. O. Y. Bisen and K. K. Nanda, Appl. Mater. Today, 2020, 21, 100846.
18. S. Liu, Z. Li, C. Wang, W. Tao, M. Huang, M. Zuo, Y. Yang, K. Yang, L. Zhang, S. Chen, P. Xu and Q. Chen, Nat. Commun., 2020, 11, 938.
19. L. Kumari, W. Z. Li, C. H. Vannoy, R. M. Leblanc and D. Z. Wang, Ceram. Int., 2009, 35, 3355–3364.
20. Z. Yang, X. Gong, B. Wang, D. Yang, T. Fu and Y. Liu, RSC Adv., 2018, 8, 35179–35186.
21. M. Shaw, D. Samanta, S. Bera, M. K. Mahto, M. A. S. Shaik, S. Konar, I. Mondal, D. Dhara and A. Pathak, Inorg. Chem., 2022, 61, 14568–14581.
22. H. Idriss, Surf. Sci., 2021, 712, 2–7.
23. D. V. Sivkov, O. V. Petrova, S. V. Nekipelov, A. S. Vinogradov, R. N. Skandakov, K. A. Bakina, S. I. Isaenko, A. M. Ob'edkov, B. S. Kaverin, I. V. Vilkov and V. N. Sivkov, Appl. Sci., 2022, 12, 7744.
24. X. Liu, C. Culhane, W. Li and S. Zou, ACS Omega, 2020, 5, 24367–24378.
25. S. K. Tiwari, D. Chen, Y. Chen, K. Thummavichai, O. Ola, Z. Ma, G. Liu, N. Wang and Y. Zhu, ACS Omega, 2022, 7, 3230–3239.
26. G. Raj, R. Nandan, K. Kumar, D. B. Gorle, A. B. Mallya, S. M. Osman, J. Na, Y. Yamauchi and K. K. Nanda, Mater. Horiz., 2023, 10, 5032–5044.
27. G. Raj, D. Das, B. Sarkar, S. Biswas and K. K. Nanda, Sustainable Mater. Technol., 2022, 33, e00451.
28. I. Shariff, S. Dwiratna and B. M. Yamin, IOP Conf. Ser. Earth Environ. Sci., 2020, 443, 012021.
29. Y. C. G. Kwan, G. M. Ng and C. H. A. Huan, Thin Solid Films, 2015, 590, 40–48.
30. M. Protić, A. Miltojević, B. Zoraja, M. Raos and I. Krstić, Teh. Vjesn., 2021, 28, 1762–1768.
31. G. S. Mary, P. Sugumaran, S. Niveditha, B. Ramalakshmi, P. Ravichandran and S. Seshadri, Int. J. Recycl. Org. Waste Agric., 2016, 5, 43–53.
32. M. Thommes, K. Kaneko, A. V. Neimark, J. P. Olivier, F. Rodriguez-Reinoso, J. Rouquerol and K. S. W. Sing, Pure Appl. Chem., 2015, 87, 1051–1069.
33. W. P. Utomo, H. Wu and Y. H. Ng, Energies, 2023, 16, 27.
34. Z. Zhang, K. Yao, L. Cong, Z. Yu, L. Qu and W. Huang, Catal. Sci. Technol., 2020, 10, 1336–1342.
35. Z. Y. Wu, M. Karamad, X. Yong, Q. Huang, D. A. Cullen, P. Zhu, C. Xia, Q. Xiao, M. Shakouri, F. Y. Chen, J. Y. (Timothy) Kim, Y. Xia, K. Heck, Y. Hu, M. S. Wong, Q. Li, I. Gates, S. Siahrostami and H. Wang, Nat. Commun., 2021, 12, 1–10.
36. G. Wen, J. Liang, Q. Liu, T. Li, X. An, F. Zhang, A. A. Alshehri, K. A. Alzahrani, Y. Luo, Q. Kong and X. Sun, Nano Res., 2022, 15, 972–977.
37. O. Y. Bisen, R. Nandan, G. Raj, A. K. Yadav and K. K. Nanda, ACS Appl. Energy Mater., 2022, 5, 14019–14034.
38. R. Nandan, G. Raj and K. K. Nanda, ACS Appl. Mater. Interfaces, 2022, 14, 16108–16116.
39. S. Bu, N. Yao, M. A. Hunter, D. J. Searles and Q. Yuan, npj Comput. Mater., 2020, 6, 128.
40. D. Y. Shin and D. H. Lim, Chem. Eng. J., 2023, 468, 143466.
41. T. Hu, C. Wang, M. Wang, C. M. Li and C. Guo, ACS Catal., 2021, 11, 14417–14427.
42. Q. Fang, H. Yin, X. Mao, Y. Han, C. Yan, A. P. O'Mullane and A. Du, J. Phys. Chem. Lett., 2023, 14, 2410–2415.
43. G. Kresse, J. Non-Cryst. Solids, 1995, 192–193, 222–229.
44. D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775.
45. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
46. S. Grimme, J. Antony, S. Ehrlich and H. Krieg, J. Chem. Phys., 2010, 132, 154104.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4gc01422g

---