Ru and RuO x decorated carbon nitride for e ﬃ cient ammonia photosynthesis †

Photocatalytic ammonia synthesis is a promising strategy for sustainable development compared to the energy-intensive industrial Haber – Bosch approach. Herein, a ternary heterostructure that consists of ruthenium species and carbon nitride (C 3 N 4 ) was rationally explored for ammonia photosynthesis. Compared to the small ammonia yield from the g-C 3 N 4 and Ru/g-C 3 N 4 system, the Ru/RuO 2 /g-C 3 N 4 system represents 6 times higher activity with excellent stability under full-spectrum irradiation. Such an enhancement is not only due to e ﬃ cient transfer of electrons and holes to Ru and RuO 2 , respectively, facilitating both the reduction and oxidation reaction, but also taking advantage of Ru for N u N activation.


Introduction
Ammonia is one of the most fundamental and essential feedstocks in the chemical industry due to its application in fertilizer production, medicaments and biological molecules. 1,2The main industrial process of ammonia production is the Haber-Bosch process, which requires extremely high temperatures and pressures, leading to an energy-intensive and highly carbon-emitting process. 3Ammonia photosynthesis is believed to be a low cost and environmentally friendly approach which could be regarded as a promising next-generation ammonia synthesis technique. 4,5However, the efficiency obtained to date has been very moderate, mainly due to the high recombination of photo-induced charges and the lack of efficient activation sites for cleavage of the NuN triple bond.
Graphitic carbon nitride (g-C 3 N 4 ), a thermally and chemically stable nitride semiconductor photocatalyst, is one of the most promising polymer photocatalysts for ammonia photosynthesis. 6,7It not only has good visible light absorption ability but also has enough driving force for the reduction reaction of N 2 .Many g-C 3 N 4 based photocatalysts, including g-C 3 N 4 /Cs x WO 3 , 8 g-C 3 N 4 /ZrO 2 , 9 g-C 3 N 4 /ZnMoCdS 10 and so on, have been applied in ammonia photosynthesis.However, they mainly suffered from low efficiency of charge separation, thus resulting in a low yield in photocatalytic N 2 reduction.Engineering heterojunctions has been proved to be one of the most effective ways for the spatial separation of photo-induced carriers. 11,12Besides, the precise design of suitable active sites and matching of functionalities during heterojunction construction certainly aid in the specific photocatalytic reaction. 13xperimental and theoretical studies indicated that N 2 could be easily adsorbed and activated on the Ru catalyst surface to form N 2 H x species to facilitate the cleavage of the NuN triple bonds in thermal catalysis. 14,15Recently, Ru has been used as an efficient cocatalyst for photo-assisted ammonia synthesis in aqueous solutions such as Ru/GaN and Ru/TiO 2 due to the electron trapping sites of Ru. 16,17 It is more efficient if a dual cocatalyst could be loaded for efficient trapping of both electrons and holes and also facilitate both reduction and oxidation reactions.
Ruthenium oxide (RuO 2 ) has been widely reported to be efficient as a hole acceptor, specifically it could extract photogenerated holes from excited photocatalysts efficiently and therefore suppresses charge recombination.Hence, it is of particular interest to rationally design the Ru/RuO 2 /g-C 3 N 4 ternary heterojunction by coating RuO 2 and Ru clusters on g-C 3 N 4 nanosheets as dual co-catalysts for efficient ammonia photosynthesis.The 2D g-C 3 N 4 nanosheets, with short bulk-tosurface diffusion length, accelerate the transfer of photoinduced carriers.Moreover, it also provides large surface area and abundant unsaturated sites (e.g.-NH 2 ) for coordination of active components.Based on this excellent platform, the suitable choice of Ru as an electron acceptor not only takes advantage of its unique interaction and activation with NuN, but also makes good use of the formed oxide species (RuO 2 ) as a well-known hole acceptor.This ternary heterostructure Ru/ RuO 2 /g-C 3 N 4 exhibits a much higher NH 3 generation rate (13.3 µmol g −1 h −1 ) than Ru/g-C 3 N 4 (2.5 µmol g −1 h −1 ) and g-C 3 N 4 (almost zero).The optimised catalyst can also remain stable for at least four cycles.

Experimental section
g-C 3 N 4 was synthesized using urea as the precursor.In detail, 10 g urea powder was calcined in a muffle furnace (Carbolite, CWF 1300) at 550 °C for 4 h with a rate of 5 °C min −1 .The obtained yellow product was washed with water, 0.1 M HCl, 0.1 M NaOH and water three times, respectively, then dried in an oven at 70 °C overnight.RuO 2 and/or Ru loaded carbon nitride was synthesized via wet impregnation.Typically, 200 mg bulk C 3 N 4 and 45 wt% RuCl 3 •H 2 O were suspended in 40 ml water followed by ultra-sonication (Fisherbrand) for 30 min.After that, the suspension was dried at 100 °C and calcined at 300 °C for 2 h in air and under a 10% H 2 /Ar atmosphere, and were labelled as Ru/RuO 2 /g-C 3 N 4 and Ru/g-C 3 N 4 , respectively.Ru loaded on g-C 3 N 4 by photodeposition was labelled as Ru/ g-C 3 N 4 -p.The obtained materials were washed several times with water to remove RuCl 3 residues.A series of Ru/RuO 2 / g-C 3 N 4 samples were labelled as Ru/RuO 2 /g-C 3 N 4 -5% x% (x = 0.5-7.5),where x is the Ru loading weight percentage (nominal ratio).
The powder X-ray diffraction (XRD) was carried out using a Bruker D4 diffractometer equipped with a Cu-Kα source (Kα1 = 1.540562Å and Kα2 = 1.544398Å).Transmission electron microscopy (TEM) (JEOL2010) was used to analyse the morphology and composition of the sample.The surface elemental composition analyses was characterized by X-ray Photoelectron Spectroscopy (XPS) (Thermoscientific XPS K-alpha).UV-vis-NIR diffuse reflectance spectra (DRS) were recorded using an Agilent Carry 3500 UV-Vis-NIR spectrophotometer equipped with a diffuse reflectance unit.Steady-state photoluminescence (PL) and Raman spectra were examined using a Renishaw inVia Raman microscope.Photocurrent testing was carried out using a closable three-electrode electrolytic cell, using an electrochemical workstation (IVIUM) control voltage with 0.1 mol L −1 Na 2 SO 4 solution as the electrolyte, where the FTO electrode coated with the catalyst was used as the working electrode, the platinum plate was used as the counter electrode, and the Ag/AgCl electrode was used as the reference electrode.
In a typical photocatalytic N 2 reduction reaction, a certain amount of photocatalyst was well dispersed in 30 ml of aqueous solution containing a sacrificial electron donor and then transferred to a 100 ml reactor.An in-house-built 100 mL two-necked flat-bottomed quartz container was used as the reactor, 50 mL of deionized water was used as the solvent and proton source for the reaction, and 0.03 g of the obtained photocatalyst was added into the solution.The reactor was sealed, purged with N 2 gas for 20 min and then irradiated.The N 2 and Ar control experiment was irradiated under LED irradiation (365 nm).Other control and cycling experiments were irradiated under full arc using a 300 W xenon lamp.The solution was taken out at regular intervals to detect ammonia concentration using ion chromatography.
Photocurrent testing was carried out using a three-electrode electrolytic cell, using an electrochemical workstation (IVIUM) control voltage.0.1 M Na 2 SO 4 solution was used as the electrolyte, the FTO electrode coated with the catalyst was used as the working electrode, the platinum plate served as the counter electrode, and the Ag/AgCl electrode acted as the reference electrode.A 150 W xenon lamp was used as the full arc light source.The photocurrent test was carried out using a potentiostatic technique with a switching period of 5 s, which mainly measured the photocurrent density of the photocatalysts.The alternating current impedance spectrum was measured using a three-electrode system with the following parameters: at −0.4 V vs. Ag/AgCl, high frequency 10 kHz, low-frequency 1 Hz, amplitude 10 mV.The test was carried out under irradiation with 100 mL of 0.1 M Na 2 SO 4 solution as the electrolyte.

Results and discussion
The crystal structure of the obtained samples along with pure carbon nitride was firstly examined by powder X-ray diffraction (Fig. 1a).The characteristic diffraction peak at 27.4°assigned to the interlayer stacking of g-C 3 N 4 in all three samples remains unchanged, excluding the insertion of Ru species at the interlayer. 18The Ru/g-C 3 N 4 -5% sample shows a series of peaks at 38.0°, 41.9°, 43.6°, 57.8°, 68.7°and 77.7°which are ascribed to the (1 0 −1 0), (0 0 0 2), (1 0 −1 1), (1 0 −1 2), (2 -1 -1 0) and (1 0 −1 3) lattice planes of the hexagonal Ru crystal structure, which corresponds to the commercial Ru.In contrast, no peaks attributed to RuO x species can be observed over the Ru/RuO 2 /g-C 3 N 4 -5% sample which may be due to the low amount or high dispersion of Ru species.FTIR and Raman spectroscopy were used to characterize the stable chemical structure of the obtained samples.For FTIR spectra (Fig. 1b), the typical signal at 810 cm −1 is assigned to the out-of-plane bending vibration characteristic of heptazine.The multiple peaks at 1200-1600 cm −1 represent the stretching vibration of tri-s-triazine heterocyclic stretches. 19Moreover, the absorption band in the range of 3000-3300 cm −1 is ascribed to the -NH 2 stretching. 18Raman spectroscopy was carried out to examine the fine structure of g-C 3 N 4 and Ru/RuO 2 /g-C 3 N 4 -5% in detail (Fig. 1c).In the range of 1200-1700 cm −1 , a series of peaks are attributed to C-N stretching vibrations, specifically "G" and "D" band profiles of structurally disordered graphitic carbon and other carbon/nitrogen layered compounds. 20,21The peak at 980 cm −1 could be assigned to the symmetric N-breathing mode of heptazine, whilst the peak at 690 cm −1 corresponds to the in-plane bending vibrations of the tri/heptazine C-N-C linkages. 22All the above characterizations confirm the preservation of carbon nitride's structure after loading Ru species, while the specific peaks assigned to Ru species are not observed, again due to the highly dispersed and low amount of Ru loaded. 23,24EM was applied to analyse the microstructures of photocatalysts after loading Ru and RuO 2 (Fig. 2).The thin g-C 3 N 4 nanosheets are firmly assembled with small nanoparticles, constructing stable 0D/2D heterojunctions.The size of Ru species is around 2 nm, as shown in Fig. S1.† Fig. 2b shows the high-resolution image along with the fast Fourier transform (FFT) image.The transformed lattice fringe with a spacing of 0.207 nm matches well with the (101) plane the metallic Ru, while the spacing of 0.18 nm corresponds to the (211) lattice plane of RuO 2 . 25,26Ru and RuO 2 formation was believed to be due to calcination of the Ru precursor on carbon nitride at 300 °C in air.Similarly, the formation of Ru and RuO 2 by thermal decomposition in air was reported, where a mixture of RuO 2 as the dominant phase and Ru was found in the sample to be calcined between 200 °C and 400 °C. 27The reason is likely due to the organic substrate working as a reductant. 28The EDS mapping (Fig. 2c) shows the homogeneous distribution of the corresponding elements (C, N, O and Ru), indicating the high dispersion of RuO x on the g-C 3 N 4 .The high dispersion of Ru species can be attributed to two reasons: (i) long duration of sonication can help disperse the Ru precursor during uniform precipitation on the carbon nitride; (ii) carbon nitride possesses rich NH 2groups, which are widely reported to coordinate and anchor transition metals, avoiding the aggregation.Furthermore, Ru and O are ubiquitously dispersed in the selected region of Ru/RuO 2 / g-C 3 N 4 -5%, indicating successful formation of RuO x .
To further investigate the chemical states of Ru species in the as-prepared sample, X-ray photoelectron spectroscopy (XPS) measurement was carried out (Fig. 3).The C 1s spectra overlap with Ru 3d spectra (Fig. 3a), illustrating three main peaks at 288.0 eV, 286.2 eV and 284.6 eV corresponding to the sp 2 -bonded carbon (N-CvN), C-O and adventitious C-C bond, respectively. 29The signals at 280.2 eV and 281.7 eV are ascribed to Ru 0 and RuO 2 , respectively. 30The peaks in N 1s spectra (Fig. 3b) at 398.5 eV and 400.0 eV are assigned to heptazine C-N-C and N-(C) 3 , while the signal centred at 401.2 eV corresponds to the C-N-H bond.In order to unravel the coexistence of Ru and RuO 2 , we further analysed the Ru species by XPS (Fig. 3c).The peaks at 461.9 eV and 484.1 eV are assigned to metallic Ru, while those at 463.7 eV and 485.9 eV are correlated to RuO 2 . 31,32The average ratio of Ru 0 /RuO 2 in Ru/RuO 2 / g-C 3 N 4 is calculated to be 0.18 by the curve fit of Ru 3p spectra.The peaks at 531.2 eV and 532.3 eV in O 1s spectra are related to Ru-O-Ru and -C-O, respectively.The binding energy at 534 eV is assigned to the adsorbed H 2 O (Fig. 3d). 33n addition, the XPS Ru 3p spectra of Ru/RuO 2 /g-C 3 N 4 -5% and Ru/g-C 3 N 4 -5% are shown in Fig. S2.† Compared with the Ru 3p spectra of Ru/RuO 2 /g-C 3 N 4 -5%, Ru/g-C 3 N 4 -5% shows peaks at 461.9 eV and 484.2 eV, corresponding to Ru 3p 3/2 and Ru 3p 1/2 of metallic Ru.The absence of RuO 2 in Ru/g-C 3 N 4 -5% is due to  the reduction condition of H 2 .Combining the TEM and XPS analysis of the Ru/RuO 2 /g-C 3 N 4 -5% sample, Ru species exist as Ru and RuO 2 clusters on carbon nitride nanosheets.
The photocatalytic activity for ammonia synthesis is shown in Fig. 4. Control experiments were conducted and are shown in Fig. 4a.No ammonia is detected from the pure g-C 3 N 4 and in the absence of illumination, indicating that the Ru species and light irradiation are both indispensable to the photocatalytic ammonia photosynthesis.A small concentration of 15 µmol g −1 NH 4 + was initially observed in the Ru/RuO 2 / g-C 3 N 4 -5% sample under an Ar atmosphere while no further increase of NH 4 + can be observed in the following 5 hours.The production of NH 4 + in the first hour is likely due to the remaining NH x species adsorbed on g-C 3 N 4 during sample preparation rather than photosynthesis of N 2 .In the absence of the Ru cocatalyst, this adsorbed NH x cannot be easily desorbed, while with the loading of the Ru cocatalyst, this small amount of adsorbed NH x is readily removed within one hour.This is an interesting note for ammonia synthesis if there were NH x groups produced during photocatalyst synthesis.When changing to a N 2 atmosphere, the ammonia yield on the Ru/ RuO 2 /g-C 3 N 4 -5% sample linearly increases, on average 13.3 µmol g −1 h −1 , indicating that N 2 gas is the main source of ammonia photosynthesis.Then, the Ru cocatalyst amount was investigated between 0.5 wt% and 7.5 wt% to optimise its loading.All samples present enhanced activity compared with pure C 3 N 4 .The optimum amount of the loaded co-catalyst has been determined to be 5%, showing the highest N 2 fixation activity with an NH 4 + concentration of 13.3 µmol g −1 h −1 under full arc irradiation.The ammonia generation rate is comparable to that of the reported Ru system photocatalysts which have been list in Table S1.† 34,35 A further increase in the amount of loaded co-catalysts results in decreased activity, because excess co-catalyst may block the light absorption of g-C 3 N 4 . 36In order to elucidate the important role of RuO 2 in ternary heterostructures, other methods including photodeposition and H 2 reduction have been used to load Ru species on g-C 3 N 4 (denoted Ru/g-C 3 N 4 -5%-p and Ru/g-C 3 N 4 -5%) for comparison (Fig. 4c).All samples exhibited a linear increase in ammonia production with increasing time.Ru/g-C 3 N 4 -5% and Ru/g-C 3 N 4 -5%-p present an ammonia yield of less than 10 µmol g −1 after 3 h full arc irradiation, only one-fourth of that achieved by Ru/RuO 2 /g-C 3 N 4 -5%.On the former samples, only Ru metallic species were observed as mentioned above, which can act as an electron acceptor and activation sites for N 2 reduction. 12The extraordinary high activity of Ru/RuO 2 / g-C 3 N 4 -5% can be ascribed to the synergistic effect between metallic Ru and RuO 2 on g-C 3 N 4 for both electron and hole extraction as well as for N 2 activation as discussed below.In addition, a little change is observed from cycling experiments in Fig. 3d, which is believed to be due to the loss of a photocatalyst amount as we filtered the sample from the solution for cycling.Furthermore, the XRD and XPS of Ru/RuO 2 /g-C 3 N 4 -5% before and after the reaction are presented in Fig. S3 and Fig. S4.† No obvious differences in the XRD peaks and Ru 3p XPS spectra of the used photocatalyst are observed compared with that of the fresh photocatalyst.The ratio of the Ru metal and RuO 2 before and after the reaction has been compared to further confirm the stability of Ru/RuO 2 /g-C 3 N 4 -5%.The Ru/ RuO 2 ratio after the reaction is estimated to be 0.19, which is similar to that before the reaction (0.20).All these results prove that the two cocatalysts remain very stable.UV-Vis DRS analysis was implemented to investigate the light absorption of all samples (Fig. 1d).g-C 3 N 4 exhibits light absorption from the UV to visible range, and the sharp edge suggests that the visible light absorption is due to the band gap transition.After loading Ru species, there is no obvious difference in the absorption edge compared with the pristine g-C 3 N 4 , indicating that loading Ru and RuO 2 does not have a great influence on the band structure of g-C 3 N 4 .In the visible light region, the light absorption intensities of Ru/g-C 3 N 4 -5% and Ru/RuO 2 /g-C 3 N 4 -5% samples are strengthened gradually as Ru species increase, which may be due to the light scattering of the Ru species.Photoluminescence (PL) spectra of the samples were recorded to study the behaviour of photogenerated charge carriers, shown in Fig. 5a.The g-C 3 N 4 exhibits broad and robust doublet peaks at 440 nm and 500 nm.The emission centre at around 440 nm is ascribed to π-π* transitions, which are usually observed in conjugated ring systems including heterocyclic aromatic compounds.The peak at around 500 nm is explained as the n-π* transition, 37 which involves lone pairs on the N atom on the edge of the triazine/ heptazine ring.After introducing Ru species, Ru/g-C 3 N 4 -5% exhibits a similar emission peak profile to that of pure g-C 3 N 4 with decreased intensity.Ru/RuO 2 /g-C 3 N 4 -5% also shows the same trend but the lowest intensity compared with Ru/g-C 3 N 4  To investigate charge transfer kinetics among different samples, in situ photoelectrochemical and electrochemical impedance spectroscopy measurements were performed on all three samples in N 2 purged solution.From Fig. S5, † Ru/RuO 2 / g-C 3 N 4 -5% exhibits higher reduction photocurrent density (e.g. 2 µA cm −2 ) compared to Ru/g-C 3 N 4 -5% (0.6 µA cm −2 ).It is consistent with the enhanced NH 3 evolution rate on Ru/RuO 2 / g-C 3 N 4 -5% compared with that on Ru/g-C 3 N 4 -5%.Pure g-C 3 N 4 has shown negligible photocurrent density, which also maintains consistency with its negligible NH 3 synthesis performance.The interface charge transfer of all three samples was described by fitting the impedance scan plots to the equivalent circuit shown in the inset of Fig. 5b.The equivalent circuit model consists of two components: the electrode/electrolyte interface impedance (R ct and CPE 1 ) and solid-state interface impedance (R solid-state and CPE 2 ).R s is the system resistance, R ct is the electrode/electrolyte interface charge transfer resistance and R solid-state is the solid-state interface (e.g.g-C 3 N 4 / RuO 2 and g-C 3 N 4 /Ru) charge transfer resistance, respectively.The results summarized in Table 1 show that all three samples have a similar system resistance around 40 Ω cm 2 .However, charge transfer resistance (R ct ) dramatically differs among different samples.Obviously, co-catalyst (Ru/RuO 2 or Ru) loading has essentially reduced the electrode/electrolyte interface charge transfer resistance, as R ct has decreased from 2.46 × 10 6 Ω cm 2 of g-C 3 N 4 to 9.8 × 10 3 Ω cm 2 of Ru/RuO 2 /g-C 3 N 4 -5% and 3.38 × 10 4 Ω cm 2 of Ru/g-C 3 N 4 -5%.The reduced charge transfer resistance or enhanced charge transfer is attributed to the presence of metallic Ru and RuO 2 .In consequence, g-C 3 N 4 has shown negligible NH 3 production and photocurrent in part due to a high charge transfer resistance, while Ru/RuO 2 /g-C 3 N 4 -5% and Ru/g-C 3 N 4 -5% exhibit significant NH 3 synthesis activities of 13.3 μmol g −1 h −1 and 2.5 μmol g −1 h −1 , respectively.
Apart from the solid-solution interface, charge transfer at the solid-state interface is also imperative in deciding the photocatalytic activity.To be precise, the solid-state interface charge transfer resistances (R solid-state ) are 2.21 × 10 2 Ω cm 2 on g-C 3 N 4 , 1.74 × 10 3 Ω cm 2 on Ru/RuO 2 /g-C 3 N 4 5% and 1.82 × 10 4 Ω cm 2 on Ru/g-C 3 N 4 -5%.As R solid-state represents the resistance between g-C 3 N 4 /g-C 3 N 4 , RuO 2 /g-C 3 N 4 or Ru/g-C 3 N 4 particle boundaries, their smaller values compared to the solidsolution interface resistance confirmed over 5 times faster solid-solid charge transfer than the solid-solution charge transfer.One can see the increased R solid-state value on Ru/ RuO 2 /g-C 3 N 4 -5%, indicating that the enhanced activity is in part due to the improved charge transfer between the catalyst and solution instead of between C 3 N 4 particles.
Based on the above characterization and experimental results, a possible ammonia synthesis mechanism is proposed on the best sample of Ru/RuO 2 /g-C 3 N 4 -5% (Fig. 5c).When irradiated, electrons can be excited from the valence band to the conduction band of g-C 3 N 4 , which are further trapped by metallic Ru as the electrons sink.Then, these electrons are transferred from the Ru to the π antibonding orbital of N 2 , facilitating the cleavage of the NuN triple bonds, thus activating N 2 . 12,38The activated N 2 on the Ru catalyst surface further reacts with H + in water to form NH 3 , and finally forms NH 4 + in water.Meanwhile, the left holes are transported to RuO 2 and scavenged by methanol, leading to a decrease of the electronhole recombination.

Conclusions
In summary, the ternary heterostructure Ru/RuO 2 /g-C 3 N 4 system was explored and it exhibited excellent photocatalytic N 2 reduction activity and stability under full arc irradiation.It was found that g-C 3 N 4 has shown negligible activity, while Rug-C 3 N 4 and Ru/RuO 2 -g-C 3 N 4 yielded 2.5 μmol g −1 h −1 and 13.3 μmol −1 g h −1 ammonia under full arc irradiation, respectively.The resulting structure was characterized by XRD, Raman, FTIR and TEM.The electron transfer mechanism was confirmed by spectroscopy including PL and PEC.The higher activity of Ru/RuO 2 /g-C 3 N 4 over Ru-g-C 3 N 4 illustrated the

Fig. 2
Fig. 2 TEM (a), the diffraction pattern of the selected area (b) and the corresponding elemental mapping images (c) of Ru/RuO 2 /g-C 3 N 4 -5% nanosheets.The inset in (b) is the corresponding fast Fourier transform image.
and g-C 3 N 4 , indicating the least charge carrier recombination in the Ru/RuO 2 /g-C 3 N 4 sample.