Ammonia (NH3) is an important and basic material and plays a critical role in fertilizers for agricultural production. It is also an ideal carbon-free energy carrier with a high hydrogen energy density, a highly anticipated clean energy source for hydrogen conversion in the future and a key route to achieving carbon neutrality1,2,3,4,5,6,7,8,9. The Haber–Bosch process is currently the primary method used for industrial production of NH3. However, this process requires high temperatures (>400 °C) and pressures (>150 bar) to activate the chemically inert nitrogen (N2; N ≡ N) and force the reaction of N2 with H2 to produce NH3; this results in a serious waste of resources, energy inefficiency, and pollution. Many researchers have attempted to develop an environmentally friendly, energy efficient, and pollution-free method for NH3 production at ambient temperatures and pressures. Recently, NH3 production using an electrocatalytic nitrogen reduction reaction (ECNRR) under ambient conditions has been considered an attractive alternative to the Haber–Bosch process3,4,5,6,7,8,9,10,11. Practical application of the ECNRR for mass NH3 production is still a great challenge because of limited N2 solubility in water, the high splitting energy of the N ≡ N bond and the competing hydrogen evolution reaction (HER). These scientific issues are limitations and result in low selectivity and NH3 yield rate (YNH3). There is no doubt that finding more reactive nitrogen sources with higher water solubility to replace N2 would be an effective way to solve these issues and achieve large-scale NH3 production.

Among nitrogen-containing molecules, nitrate (NO3−) is an important alternative to N2 as a nitrogen source due to its unlimited solubility in water, much lower activation energy than N2, and easy adsorption on the surfaces of catalysts, which are beneficial in alleviating mass transfer barriers in the electrolytic process and improving the overall performance for NH3 production. In addition, nitrate is one of the most serious water pollutants and is harmful to human health and the environment11. Significantly, the electrocatalytic nitrate reduction reaction (ECNtRR) would provide a promising and economical approach to ammonia production with advantages such as high efficiency, low energy consumption, and pollution-free operation at ambient temperatures and pressures, and it would simultaneously consume the nitrates in industrial wastewater, polluted groundwater and surface water and maintain the balance of the nitrogen cycle12,13,14. However, the ECNtRR is a complex eight-electron and nine-proton process during which it is easy to produce byproducts, and this results in lower reaction selectivity and efficiency. It is imperative to develop new electrocatalysts with high selectivity and high NH3 yields (YNH3) under ambient conditions.

Recently, a number of electrocatalysts have been reported for ECNtRRs run under ambient conditions12,13,14,15,16,17. However, there is still a large gap between the performance of these catalysts and practical commercialization. Heterojunction catalysts are advanced catalytic systems that have undergone rapid development in recent years, and they provide a direction for catalyst development in the future18,19,20,21,22,23,24,25,26,27,28. Heterojunction electrocatalysts have high selectivity, high activity and stability, which is the key to developing high-performance ECNtRR electrocatalysts. The d-orbital electrons enable electron transfer into the lowest vacant molecular orbital of nitrate29. The catalysts, including metal elements with highly occupied d-orbitals and open d-orbital shells (e.g., Fe, Co, Ni, etc.) have exhibited superior catalytic performance. Prussian blue analogs (PBAs) have attracted considerable interest for application in catalysis and energy conversion because of their easy syntheses, adjustable structures, and promising electrochemical properties30. However, their low conductivities and limited structural stabilities limit application in the field of catalysis.

Combining traditional materials with carbon materials to form heterogeneous interface structures is a very promising way to obtain new catalysts with targeted reaction selectivity, activity and stability31. Among all reported carbon materials, graphdiyne (GDY) is the first sp- and sp2-cohybridized two-dimensional carbon material, and it exhibits many unique and distinguishing properties, such as a large highly π-conjugated network, abundant carbon chemical bonds, extremely uneven surface charge, natural pores and band gap, high intrinsic charge carrier mobility, high electrical conductivity, and excellent stability14,19,27,28,32,33,34,35,36,37,38,39,40,41,42,43,44,45. Compared with traditional carbon materials, GDY has an unparalleled advantage in that it can be grown controllably on the surface of any substrate under low temperatures and mild conditions. Benefiting from these unique properties, a number of GDY-based materials have been fabricated and have shown potential in various fields, including catalysis, energy conversion and storage, batteries, solar cells, intelligence, and life sciences5,9,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61. These advantages of GDY make it an ideal material for constructing of interface structures with a significant numbers of active sites, facilitated charge transfer, enhanced catalytic activity and long-term stability.

Herein, we demonstrate facile fabrication of GDY-based multi-interface heterostructures via in situ growth of GDY films on the surfaces of hollowed FeCoNi PBA nanocubes (h-FeCoNi PBA@GDY). The experimental results demonstrate that incomplete charge transfer between metal atoms and GDY significantly improved the charge transfer ability of h-FeCoNi PBA@GDY, enhanced the conductivity, and increased the number of active sites, which resulted in excellent ECNtRR performance with a YNH3 of 1015.5 μmol h−1 cm−2 and a FE of 95.1% with low overpotentials. These values compare favorably to previously reported values. Furthermore, h-FeCoNi PBA@GDY exhibits long-term stability over 14 continuous cycles.

Results and discussion

As shown in Fig. 1, h-FeCoNi PBA@GDY was prepared via a facile three-step method. Typically, the FeCoNi PBA nanocubes were first synthesized via coprecipitation by using carbon cloth (CC) as the growth substrate (step I). Then, the freshly prepared FeCoNi PBA nanocubes were added to sodium sulfide solution and converted into hollowed FeCoNi PBA nanoboxes (h-FeCoNi PBA) through an anion-exchange-etching process (step II). Finally, h-FeCoNi PBA@GDY (mass loading: 0.12 mg cm−2) was obtained by in situ growth of a GDY layer on the h-FeCoNi PBA surface.

Fig. 1: Schematic illustration of the synthetic routes.
figure 1

The multistep reactions to synthesize h-FeCoNi PBA@GDY involved the coprecipitation of FeCoNi PBA, anion exchange synthesis of h-FeCoNi PBA, and in-situ growth of GDY to form h-FeCoNi PBA@GDY.

Full size image

Field-emission scanning electron microscopy (SEM) images showed that the CC substrate had a 3D porous architecture woven from the carbon fiber with a smooth (Fig. 2a, b) and hydrophobic surface (Fig. 2c). After the coprecipitation reaction, FeCoNi PBA nanocubes with smooth surfaces were uniformly and densely grown on the surface of the CC (Fig. 2d−f). The cubic morphology of the FeCoNi PBA was well maintained after treatment with the sodium sulfide solution (Fig. 2g−i). The as-synthesized samples were next used as substrates for in situ growth of the GDY. Remarkably, Fig. 2j−l shows that h-FeCoNi PBA@GDY had a cubic morphology but a rougher surface, which could be attributed to in situ growth of the GDY. These morphological characteristics effectively enlarged the active surface area and therefore increased the number of active sites, which enhance the electrocatalytic activity. Energy-dispersive spectroscopy (EDS) mapping results confirmed the presence and uniform dispersion of Fe, Co, Ni, C, N and S elements in h-FeCoNi PBA@GDY (Fig. 2m). Contact angle measurements revealed that h-FeCoNi PBA@GDY has a superhydrophilic surface (Fig. 2n), which improved the mass/ion transport capability and reduced contact resistance, thereby improving the overall electrocatalytic performance.

Fig. 2: Scanning electron microscopy and contact angle analyses.
figure 2

a Low- and b high-magnification scanning electron microscopy (SEM) images of CC. c Contact angle measurements of CC. Low- and high-magnification SEM images of d–f FeCoNi PBA, g–i h-FeCoNi PBA and j-l h-FeCoNi PBA@GDY. m EDX maps of Fe, Co, Ni, C, N, and S in h-FeCoNi PBA@GDY. n Contact angle measurements of h-FeCoNi PBA@GDY.

Full size image

Transmission electron microscopy (TEM), high-resolution TEM (HRTEM) and scanning TEM (STEM) were used to study the morphologies of the samples. The STEM (Fig. 3a) image revealed a solid cubic morphology and smooth surface for FeCoNi PBA. The HRTEM image (Fig. 3b) showed the monocrystalline nature of FeCoNi PBA with a lattice spacing of 0.517 nm. Elemental mapping images showed the uniform distribution of Fe, Co, Ni, C and N throughout the FeCoNi PBA nanocubes (Fig. 3c). In contrast to FeCoNi PBA, h-FeCoNi PBA contains distinct hollow interiors, which also confirmed the success of the anion-exchange-etching process (Fig. 3d). Interestingly, h-FeCoNi PBA showed two lattice spacings of 0.560 nm and 0.242/0.289 nm (Fig. 3e), which correspond to the structures of PBA and metal sulfides. In addition, elemental mapping images for a single h-FeCoNi PBA nanobox show the inner distributions of Co and S and the outermost distributions of Fe, Co, Ni, C and N, along with the presence of less S (Fig. 3f). Subsequently, a uniform GDY layer was successfully grown on the surface of h-FeCoNi PBA (Fig. 3g, i). The intimate interactions between the h-FeCoNi PBA and GDY were clearly observed (Fig. 3h). The HRTEM images (Fig. 3j–m) of h-FeCoNi PBA@GDY showed lattice fringes with distances of 0.363 nm, 0.279 and 0.242 nm, and 0.521 nm, which were ascribed to GDY, metal sulfides, and PBA, respectively.

Fig. 3: Transmission electron microscopy analyses.
figure 3

a STEM, b HRTEM and c elemental mapping images for FeCoNi PBA. d STEM, e HRTEM, and f elemental mapping images for h-FeCoNi PBA. g STEM, h, j, k, l, m HRTEM and i elemental mapping images for h-FeCoNi PBA@GDY.

Full size image

Raman spectroscopy and X-ray photoelectron spectroscopy (XPS) were used to further investigate the elemental compositions and chemical states of the samples. Figure 4a shows the Raman spectra of GDY and h-FeCoNi PBA@GDY. GDY exhibited four typical peaks for the D band (1380 cm−1), G band (1599 cm−1) and vibrations of the conjugated diynes (1950 and 2152 cm−1). Compared with GDY, an additional peak at 1023 cm−1 corresponding to the h-FeCoNi PBA species was observed, which demonstrated the successful synthesis of the h-FeCoNi PBA@GDY heterostructure. The relative intensity ratio for the D and G bands (ID/IG) reflects the relative number of defects in the sample62. h-FeCoNi PBA@GDY had a higher ID/IG value (0.86) than pure GDY (0.67), indicating the presence of more defects in h-FeCoNi PBA@GDY, which would promote electron delocalization and accelerate charge transfer63. The XPS survey spectrum (Fig. S1) confirmed the presence of Fe, Co, Ni, C, N, and S in the samples. The C 1 s XPS spectrum of GDY (Fig. 4b) exhibited four peaks at 284.0, 284.6, 286.1, and 288.1 eV, which corresponded to −C = C − (sp2-C), − C≡C − (sp-C), − C − O, and −C = O species, respectively. For h-FeCoNi PBA@GDY, the C 1s XPS spectrum was deconvoluted into six peaks for −C = C − (284.3 eV), C≡N (284.9 eV), −C≡C − (285.6 eV), −C − O (287.0 eV), −C = O (288.6 eV) and π-π* (289.9 eV). The presence of a peak for the π-π* transition reflected the interaction between GDY and h-FeCoNi PBA. The area ratio for the sp2-C/sp-C peaks of h-FeCoNi PBA@GDY was 0.5, which is consistent with that of GDY, indicating that the structure of GDY remained intact in the heterostructure and demonstrating the robustness of the GDY structure. The sp-C peak shifted by 0.25 eV to a higher binding energy than the pristine GDY, which indicated charge transfer from GDY to the h-FeCoNi PBA species. The Fe 2p XPS spectrum (Fig. 4c) indicated the coexistence of the Fe2+/Fe3+ redox couple in h-FeCoNi PBA and the presence of only Fe2+ in h-FeCoNi PBA@GDY. This result confirmed charge transfer from the GDY to the Fe species in h-FeCoNi PBA@GDY. As shown in Fig. 4d, the Co 2p XPS spectrum of h-FeCoNi PBA@GDY displayed two spin–orbit triplets at 783.0, 798.1, 781.3, 796.3, 778.4, and 793.9 eV as well as two satellite peaks at 786.6 and 802.4 eV. The triplets are attributable to Co3+, Co2+, and Co0 species. Compared with those for pure h-FeCoNi PBA, the Co 2p1/2 and 2p3/2 peaks for h-FeCoNi PBA@GDY shifted to lower binding energies, which confirmed the charge transfer from GDY to Co species in h-FeCoNi PBA@GDY. The high-resolution Ni 2p XPS spectrum for h-FeCoNi PBA@GDY (Fig. 4e) exhibited two 3d3/2/3d1/2 doublet peaks at 856.2/858.5 and 873.7/875.6 eV, which were ascribed to Ni2+and Ni3+ species and confirmed the coexistence of Ni2+ and Ni3+ species. The Ni 2p binding energy also exhibited a negative shift relative to that of pure h-FeCoNi PBA. Figure 4f shows the S 2p XPS spectra of h-FeCoNi PBA@GDY and h-FeCoNi PBA. In the S 2p spectrum, the peaks at 163.55, 164.18, and 165.22 eV were assigned to terminal S22−, bridging S22− and apical S2− of h-FeCoNi@GDY. These results demonstrated the enhanced charge transfer from GDY to the metal species in h-FeCoNi PBA@GDY, which enhanced the intrinsic activity.

Fig. 4: Structural characterization.
figure 4

a Raman spectra of h-FeCoNi PBA@GDY and GDY. b C 1s XPS spectra of h-FeCoNi PBA@GDY and GDY. c Fe 2p, d Co 2p, e Ni 2p and f S 2p XPS spectra of h-FeCoNi PBA@GDY and h-FeCoNi PBA.

Full size image

Electrocatalytic nitrate reduction performance

The ECNtRR catalytic capabilities of the samples were studied in a customized H-type cell under alkaline conditions at a constant potential for 1 h. Figure 5a shows the linear sweep voltammetry (LSV) curves for h-FeCoNi PBA@GDY, which were recorded in 0.5 M KNO3 + 1.0 M KOH and 1.0 M KOH electrolyte solutions, respectively. h-FeCoNi PBA@GDY exhibited larger current densities in 0.5 M KNO3 + 1.0 M KOH than in 1.0 M KOH at the same applied potentials, which indicated that the ECNtRR was facilitated in the presence of nitrate. Ultraviolet‒visible (UV‒Vis) spectrophotometry and nuclear magnetic resonance spectroscopy (NMR) were employed to detect the synthesized NH3. The UV‒Vis adsorption spectrum for pure NH4Cl in KOH solution (Fig. 5b) and the associated calibration curves (Fig. 5c) were first determined. As shown in Fig. 5d, h-FeCoNi PBA@GDY displayed high FEs in excess of 90% over a wide potential window ranging from −0.232 V to −1.032 V versus RHE; the highest FE of 95.1% was seen at −0.432 V vs. RHE and exceeded those for most of the reported electrocatalysts (Fig. 5e; Table S1). Figure 5f shows that the YNH3 for h-FeCoNi PBA@GDY increased with increasing applied potential and reached a maximum YNH3 of 1015.5 μmol h−1 cm−2 at −1.032 V vs. RHE. A series of control experiments were next conducted to determine the N source of the NH3 product. As shown in Fig. 5g, large amounts of NH3 were produced in the presence of nitrates, while negligible NH3 was detected in the absence of nitrates. This confirmed that the NH3 product resulted from electroreduction of nitrate. Previous reports demonstrated that the rate-limiting step for ECNtRR is the sluggish NO3−-to-*NO2 conversion process64,65. Almost no NO2− (Figs. 5h, S2) or N2H4 (Figs. 5i, S3 and S4) were detected during the ECNtRR process, which indicated the high selectivity of h-FeCoNi PBA@GDY for the ECNtRR. The electrolytes remaining after the ECNtRR were also tested with proton nuclear magnetic resonance spectroscopy (1H NMR, Figs. 5j–l, S5). As shown in Fig. 5k, three peaks corresponding to 14NH4+ clearly indicated its presence in the electrolyte after 1 h of electrolysis. The experimental results showed that the FE values determined by 1H NMR (Fig. 5l) were consistent with those determined by UV‒Vis spectroscopy, which confirmed the precision and accuracy of quantitative concentration determinations for the generated NH3.

Fig. 5: Electrocatalytic performance.
figure 5

a LSV curves for h-FeCoNi PBA@GDY obtained in 1.0 M KOH solution, with and without 0.5 M KNO3, respectively. b UV–vis spectra determined with various NH3 concentrations after incubation for 1 h under ambient conditions. c Corresponding calibration curves. d FEs of the samples obtained at different potentials. e Comparison of the ECNtRR performance of h-FeCoNi PBA@GDY with those of reported catalysts. f YNH3 for h-FeCoNi PBA@GDY at different potentials. g YNH3 for h-FeCoNi PBA@GDY in the electrolyte with or without NO3−. h FE and YNO2- at different potentials. i Absorption spectra for N2H4 determined at different potentials. j Calibration curves for determining NH3 concentrations using the 1H NMR method. k 1H NMR spectra for 14NH4+ obtained at different potentials from h-FeCoNi PBA@GDY. l FEs for NH3 production with h-FeCoNi PBA@GDY, determined at different potentials.

Full size image

The long-term stability of h-FeCoNi PBA@GDY was determined with continuous cycling tests. As shown in Fig. 6a, the YNH3 and FE for the h-FeCoNi PBA@GDY sample showed slight fluctuations but remained stable over 14 cycles under ambient conditions, indicating the excellent stability of h-FeCoNi PBA@GDY during operation of the ECNtRR. In addition, the SEM image (Fig. 6b) and XPS data (Figs. 6c–f and S6) revealed that the morphology and structure of h-FeCoNi PBA@GDY were well maintained after the stability tests, which confirmed the robust structural stability of the catalyst. Compared with freshly prepared h-FeCoNi PBA@GDY, the Fe 2p, Co 2p, and Ni 2p binding energies of the used samples exhibited obvious shifts, which indicated that significant charge transfer occurred during electrocatalysis. In addition, the Co3+/Co2+ and Ni3+/Ni2+ intensity ratios decreased from 1.49 to 0.76 and from 0.58 to 0.49, respectively. To gain deep insight into the origins of the excellent catalytic performance, electrochemically active surface area (ECSA) and electrochemical impedance spectra (EIS) measurements were carried out. The ESCAs proportional to the double-layer capacitance (Cdl) were estimated by using a CV method (Fig. S7). h-FeCoNi PBA@GDY showed a capacitance of 13.35 mF cm−2, larger than those of FeCoNi PBA (5.26 mF cm−2) and h-FeCoNi PBA (5.47 mF cm−2) (Fig. 6g). This implied that h-FeCoNi PBA@GDY has the largest number of active sites, which increases electrocatalytic activity. The Nyquist plots (Figs. 6h, S8 and Table S2) were fitted to a R(QR)(QR) equivalent-circuit model consisting of resolution resistance (Rs) and charge transfer resistance (Rct). h-FeCoNi PBA@GDY showed the smallest Rs and Rct (Rs = 4.23 Ω, Rct = 0.11 Ω), which indicated improved charge transfer kinetics and the highest conductivity. These factors determined the intrinsic electrochemical properties of h-FeCoNi PBA@GDY.

Fig. 6: Stability tests.
figure 6

a Stability test of the h-FeCoNi PBA@GDY catalyst at −0.432 V versus RHE. b SEM image of h-FeCoNi PBA@GDY after the stability tests. c Fe 2p, d Co 2p, e Ni 2p, f C 1s XPS spectra of h-FeCoNi PBA@GDY after the ECNtRR. g Dependence of capacitive current on scan rate. h Nyquist plots of samples.

Full size image

Conclusion

In summary, we demonstrated a rational and facile strategy for controlled synthesis of h-FeCoNi PBA@GDY heterostructures for selective and efficient catalysis of the ECNtRR in alkaline solution at ambient temperature and pressure. Our results demonstrated that incomplete charge transfer between metal atoms and GDY at the interface greatly improved the charge transfer ability and conductivity of the electrocatalyst, leading to formation of more active sites and therefore greatly increasing the NH3 production efficiency. When used for the ECNtRR, the electrocatalyst exhibited high catalytic performance with an FE of 95.1% and YNH3 of 1015.5 μmol h−1 cm−2 at low overpotentials. This work provides a promising approach for design and fabrication of efficient and inexpensive electrocatalysts supporting ammonia synthesis under ambient conditions and future applications in the renewable energy field.