ZnCr2S4: Highly effective photocatalyst converting nitrate into N2 without over-reduction under both UV and pure visible light

We propose several superiorities of applying some particular metal sulfides to the photocatalytic nitrate reduction in aqueous solution, including the high density of photogenerated excitons, high N2 selectivity (without over-reduction to ammonia). Indeed, ZnCr2S4 behaved as a highly efficient photocatalyst, and with the assistance of 1 wt% cocatalysts (RuOx, Ag, Au, Pd, or Pt), the efficiency was greatly improved. The simultaneous loading of Pt and Pd led to a synergistic effect. It offered the highest nitrate conversion rate of ~45 mg N/h together with the N2 selectivity of ~89%. Such a high activity remained steady after 5 cycles. The optimal apparent quantum yield at 380 nm was 15.46%. More importantly, with the assistance of the surface plasma resonance effect of Au, the visible light activity achieved 1.352 mg N/h under full arc Xe-lamp, and 0.452 mg N/h under pure visible light (λ > 400 nm). Comparing to the previous achievements in photocatalytic nitrate removal, our work on ZnCr2S4 eliminates the over-reduction problem, and possesses an extremely high and steady activity under UV-light, as well as a decent conversion rate under pure visible light.


Results and Discussion
Phase identification. ZnCr 2 S 4 crystallizes in the cubic spinel structure with the lattice parameter a ~ 10 Å. In literature, the study on ZnCr 2 S 4 mostly focused on its complicated but interesting magnetic property, for instance, two subsequent magnetic transitions were observed: the first one to an incommensurate helical antiferromagnetic order at 14 K and the second one to a coexisting commensurate spin order at 7 K 52,53 . Hence, to the best of our knowledge, it is the first report about the photocatalytic property of ZnCr 2 S 4 .
The polycrystalline sample was prepared simply by heating Cr 2 S 3 and ZnS in an evacuated tube furnace at 700 °C. As indicated by Fig. 1, the phase purity is confirmed by powder XRD. All reflections show a perfect match with the standard pattern. In addition, the recovered photocatalysts showed almost the same patterns to the initial one, which indicates ZnCr 2 S 4 was very stable during photocatalysis. The relatively broad reflection peaks suggest the possible nano-morphology. Figure 2 presents the SEM images for ZnCr 2 S 4 catalyst before and after the photocatalytic reactions. The particles are composed of numerous submicron crystallites, most of which possess obvious crystal facets as shown in the enlarged images (Fig. 2b,d). Moreover, the SEM results are consistent with the XRD experiments that all catalysts used in the photocatalytic reactions show neglectable degradation.
UV-Vis DRS spectroscopy was applied to evaluate its bandgap energy. As shown in Supplementary Fig. S1, the deep-brown powder shows a strong light harvesting ability over the visible light region. For most semiconductors, the dependence of the absorption coefficient α on the bandgap energy E g can be expressed by the following equation: α hv = A(hv -E g ) n/2 , where h, v , and A are Planck constant, light frequency, and proportionality, respectively; n is determined on the basis of the transition type ( i.e. n = 1 for direct transition, n = 4 for indirect transition).
Scientific RepoRts | 6:30992 | DOI: 10.1038/srep30992 Supplementary Fig. S1b provides the plot of (α hv ) 2 against hv (assuming it is a direct transition). The extrapolated value of hv at α = 0 gives an absorption edge energy corresponding to E g , which is 1.96 eV for ZnCr 2 S 4 .
Photocatalytic activity of ZnCr 2 S 4 . Figure 3 and Supplementary Table S1 shows the respective conversion rates for various catalysts in either 100 mL NO 3 − or NO 2 − aqueous solutions (25 ppm N, calculated by nitrogen weight). The unmodified ZnCr 2 S 4 shows substantial activities for photocatalyzing the reduction of nitrates and nitrites, ca. 0.75 and 1.03 mg N/h, respectively. In addition, the final resultant aqueous solution was checked by inductively coupled plasma-atomic emission spectrometry (ICP-AES) and no chromium was detected, further supporting the high stability of the catalyst.  Photocatalytic mechanism and activity enhancements by cocatalysts loading. With regards to the photocatalytic mechanism, we propose the reactions occur on the surface of the catalyst particles as shown in Fig. 4. Nitrate ions were first adsorbed to be NO 3 − (ads) as indicated by eq. (1). Then it can be either deeply reduced by 10 e − to N 2 (g) (see eq. (2)), or just reduced by 2 e − to NO 2 − (aq) and finally leave the surface (see eq. (3)). Apparently, the deep reduction to N 2 is quite difficult, which needs a significantly high density of electrons on the surface. It is the major reason why most catalysts cannot show high N 2 differential selectivity in literature. We also need to mention that no ammonia or H 2 was observed, no matter which cocatalyst was loaded in our study. In the following, the nitrite ions in the aqueous solution could be further adsorbed to be NO 2 − (ads) and reduced by 8 e − to the final product N 2 (g) (see eqs (4) and (5)). The corresponding h + with positive charge were consumed by oxalic ions (see eq. (6)).
For heterogeneous photocatalysts, it is a common strategy to enhance the efficiency by loading cocatalysts to the particle surface. Appropriately loaded cocatalyst particles can serve as either electron or hole collectors, which could facilitate the spatial separation of photogenerated charges, and therefore allow a higher density of charges at the catalytic sites. Meanwhile, in the reduction reaction of nitrates/nitrites, the adsorption and desorption of substrate ions at catalytic sites is a dynamic equilibrium process. The loaded cocatalyst particles would also increase the number of binding sites and delay desorption of substrate ions by lowering the potential barrier. This could increase the conversion rate of NO 3 − and the selectivity of N 2 as well. We believe that the synergetic improvements on charge separation and binding ability were supposed to increase the photocatalytic efficiency of ZnCr 2 S 4 for the reduction of NO 3 − in aqueous solution. As shown in Fig. 3 and Supplementary Table S1, the use of RuO x as cocatalyst did not increase the efficiency of nitrate reduction, and only show a slight improvement to the nitrite reduction. When using Ag, Au, Pd and Pt as cocatalysts, the photocatalytic efficiency was enhanced from 0.85 to 1.53 mg N/h for NO 3 − reduction and from 1.23 to 1.85 mg N/h for NO 2 − reduction, respectively. The efficiency for NO 2 − reduction is higher than NO 3 − reduction for each studied catalyst. When loading cocatalysts, the increasing tendency of NO 3 − and NO 2 − reduction is consistent with each other, which is understandable. We applied a dual-cocatalyst loading to ZnCr 2 S 4 with 0.5 wt% Pd and 0.5 wt% Pt. The efficiency was significantly enhanced, showing almost a complete conversion of both NO 3 − and NO 2 − (See Fig. 3). It is quite interesting that the total mass of the used dual-cocatalysts was equal to 1 wt%, while it shows a significant higher activity, comparing to those of either 1 wt% Pd or 1 wt% Pt loaded samples. Accordingly, it suggested a synergetic  effect between Pd and Pt cocatalysts. XPS spectrum was collected for this Pt-Pd co-loaded photocatalyst (see Supplementary Fig. S2) and both are suggested to be metal. We speculate the synergetic effect occurs when the loaded metal particles are spatially close to each other. The intermediate product (i.e. NO 2 − ) could be promptly reduced on the metal particles nearby, rather than through desorption and re-adsorption process.
As stated above, the photocatalytic mechanism here was interpreted as surficial catalytic reactions. To support this assumption, we monitored the photocatalytic efficiency of ZnCr 2 S 4 loaded with both 1 wt% Pd and 1 wt% Pt, when adding additional ions into the aqueous solution. As shown in Supplementary Fig. S3a, the real tap water from our lab was also applied, in which the nitrate conversion rate slighted decreased from 3.42 to 3.10 mg N/h. With the rational incorporation of F − , Cl − , Br − and I − (all in the concentration of 0.001 M), the conversion rate and the N 2 selectivity decreased successively. Such a decline in efficiency is more obvious when using the large anion SO 4 2− and H 2 PO 4 − . Furthermore, the photocatalytic nitrate conversion and N 2 selectivity also decreased when increasing the concentration of NaCl (from 0.001 to 0.02 M) as shown in Supplementary Fig. S3b. All these observations imply the photocatalytic reactions occur on the catalyst surface. Similar mechanism was also proposed by previous researchers 5,11,29,32 .

Treatments of highly concentrated nitrate aqueous solution.
To understand the photocatalytic process (especially the selectivity of N 2 ), we performed an extended reduction experiment in a highly concentrated aqueous solution of NO 3 − (100 ppm N). In this experiment, 0.25 g of ZnCr 2 S 4 loaded with 0.5 wt% Pd and 0.5 wt% Pt was used. The NO 3 − reduction occurred in two stages (see Fig. 5). In the first stage (roughly the first two hours), only some of the NO 3 − was reduced, and the remaining NO 3 − coexisted with the reduction product NO 2 − . In this stage, production of NO 2 − and N 2 occurred simultaneously. These reactions were apparently zero-order reactions. The observed conversion rate of NO 3 − was 4.48 mg N/h. The production rates of NO 2 − and N 2 were 1.56 and 2.91 mg N/h, respectively. The calculated differential selectivity of N 2 was 65%.
When all the NO 3 − was reduced, the second stage began, which only comprises the reduction of NO 2 − to N 2 . The rates for reduction of NO 2 − and production of N 2 were similar at 1.98 and 2.04 mg N/h, respectively. This reaction was apparently still a zero-order reaction. It should be noted that the N 2 production rate in the second stage was smaller than that in the first stage. This difference arises because N 2 in the first stage is produced from the reduction of both NO 3 − and NO 2 − , while in the second stage N 2 is only produced from NO 2 − . In other words, in the first stage, some of the NO 3 − species were strongly bound to the catalytic sites and therefore can be deeply reduced to N·. Apparently, the final selectivity of N 2 could be optimized to 100% simply by extending the irradiation time. The recovered photocatalyst loaded with both Pd and Pt was checked by powder XRD (see Fig. 1) to remain intact after 4 hours UV irradiation.
Evaluation of apparent quantum yields. In the research field of photocatalytic water splitting, people prefer to use apparent quantum yields (AQYs) to compare the intrinsic activity of a photocatalyst. In fact, AQY is also a semi-quantitative parameter, because people cannot measure the exact number of the photons absorbed by catalyst powder in an aqueous solution. Here we present the AQY study using ZnCr 2 S 4 in the photocatalytic nitrate reduction.
Under the monochromatic irradiation at 380 nm, the observed AQYs along with the change of the incident beam intensity for ZnCr 2 S 4 loaded with 0.5 wt% Pd and 0.5 wt% Pt were presented in Fig. 6 and Supplementary  Table S2. With the decreasing of the beam light intensity from 1.77 to 0.92 W, the AQY increased from 0.54 to 2.15% (using 0.1 g photocatalyst) or from 1.23 to 3.73% (using 0.25 g photocatalyst). The opposite changing tendency of light intensity and AQY is understandable, because a higher density of the beam light leads to a higher ratio of the scattered photons. Therefore, when decreasing the irradiated photon density, the AQY value would become closer to the absolute quantum yield. As shown in Supplementary Table S2, with regard to the N 2 selectivity, a higher beam intensity would lead to a higher N 2 selectivity when other experimental conditions remained unchanged. This characteristic can be explained as below. The reduction of nitrate contains two successive reactions, from NO 3 − to NO 2 − and then from NO 2 − to N·. This mechanism is quite different with that of photocatalytic water reduction, where only one electron was needed from H + to H. Herein, the nitrate reduction obviously requires a much higher density of photogenerated electrons in the catalytic sites, and apparently its deep conversion to N 2 was favored when there was a high density of incident beam.
Achieve the highest activity under UV-light irradiation. As stated above, the combination of Pt and Pd loading would generate a very high activity probably due to the synergetic effect. In this section, we thus increase the loading content up to 3 wt% for each metal to achieve the highest activity under UV-light irradiation as indicated in Supplementary Fig. S4. The differential conversion and N 2 selectivity both increase accordingly as we expected. From the industrial view of point, the usage of the noble metal should be as low as possible for the economic purpose. In our case, we did not further increase the content of the cocatalysts, which probably would give an even better performance.
Here, we need to check the stability of this particular metal sulfide (with 3 wt% Pd and 3 wt% Pt loaded) in both sodium oxalate and acidic buffer solution. We first performed the cycling experiments. After each cycle, the resultant supernatant solution was poured out and the catalyst was washed by water in the reaction vessel twice. New NO 3 − aqueous solution was added into the reaction vessel and after a half-hour dark reaction for adsorption-desorption equilibrium, the UV-light was switched on. As shown in Fig. 7a, after 5 cycles (7.5 hours irradiation in total) the conversion rate slightly decrease from 5.1 to 4.8 mg N/h (about 6% decreasing). 2.7 mmol NO 3 − in total was reduced to either NO 2 − or N 2 , and only 0.34 mmol photocatalyst (ZnCr 2 S 4 ) was used. Moreover, the powder XRD pattern after long-term experiments is consistent with the initial one, indicating that it remained as a sulfide. Accordingly, we believe the strategy of using sodium oxalate as the sacrificial agent is effective to prevent the photo-corrosion of ZnCr 2 S 4 during photocatalytic nitrate reduction. The very slightly decreasing of the activity was attributed to the loss of the catalyst powder during the cycling experiments or the mechanical loss of the loading cocatalysts by stirring.
The highest activity could be achieved by increasing the amount of powder catalyst, using the inner irradiation setup and in acidic buffer solution as shown in Fig. 7b. The activity reaches as high as ~45 mg N/h with the N 2 selectivity ~89%. This remains steady and is very close to the record of Ag/TiO 2 (ca. ~50 mg N/h) conducted by Zhang and his co-workers (by applying very similar conditions), while the latter show an apparent degradation problem due to the instability of Ag nanoparticles. Moreover, the apparent quantum yields (AQY) at 380 nm for this particular catalyst under the optimal conditions was estimated to be 15.46% (with the irradiation bean density of 0.63 W).

Visible light activity for nitrate reduction.
Although the ZnCr 2 S 4 catalyst exhibited high photocatalytic activities under UV irradiation, the majority of the energy in sunlight comes from visible light. For visible light irradiation, it is advantageous that metal sulfides have a narrow bandgap because of the high potential of the valence band arising from their S 2p orbitals. ZnCr 2 S 4 could show visible light activity, and we performed a preliminary experiment to investigate this. Photocatalytic reduction of NO 3 − (20 ppm N) was attempted with irradiation from a full arc Xe lamp. The conversion rate was 0.036 mg N/h for ZnCr 2 S 4 loaded with 0.5 wt% Pd and 0.5 wt% Pt (see Fig. 8a). The photocatalytic activity was significantly lower than that achieved with UV irradiation even though ZnCr 2 S 4 has a narrow bandgap energy (1.96 eV). This decreasing in the conversion rate could have caused by the low density of electrons when irradiated by the Xe lamp.
It is therefore necessary to enhance the light harvesting ability of our catalyst. In literature, Au nanoparticles can serve as the cocatalyst to improve the photocatalytic performance of TiO 2 according to the surface plasma resonance (SPR) theory [54][55][56] , which provide a substantial increase to the absorption of visible light photons. Here in our case, the loading of Au cocatalyst alone on the surface only show a slight increase of the activity compared to the host, when irradiated by UV light (see Fig. 3). As shown in Fig. 8, the catalysts loaded with Pd, Pt and Au (all in the mass fraction of 0.5%) show a much enhanced photocatalytic efficiency (~5 times higher) when irradiated by a full arc Xe-lamp. We believe this great enhancement in activity was due to the SPR effect of Au nanoparticles, which increase the visible light absorption ability of the catalyst, and the catalytic reduction of nitrate ions may still mostly occurs at Pd/Pt sites. The differential conversion rate could further increase up to 1.352 mg N/h, by increasing the amount of the catalyst, the cocatalysts, and using acidic buffer solution as sacrificial agents (see Fig. 8a).
As is known, the full arc Xe lamp may provide some part of UV-light irradiation, thus we have to apply a 400 nm cut-off filter to investigate its pure visible light activity. As shown in Fig. 8b, the conversion rate of nitrate is 0.064 and 0.179 mg N/h for lightly and heavily loaded catalysts, respectively. By increasing the mass of catalyst, the final optimal conversion rate is 0.452 mg N/h in buffer solution (pH ~ 4). Note that the conversion rate of 0.09 mg N/h was recently reported by Y. Kamiya under pure visible light condition, using a dual-catalyst system of Pt/SrTiO 3 :Rh and SnPd/Al 2 O 3 10 .

Potential of metal sulfides for photocatalytic nitrate reduction.
Comparing to the traditional photocatalyst TiO 2 , metal sulfides (including ZnCr 2 S 4 , CuInS 2 49 , CuFe 1−x Cr x S 2 50,51 ) all possess an intrinsic narrow bandgap energy, allowing the absorption ability to visible light. The apparent advantage for narrow bandgap is the possible higher density of the charge-excitons, which is important for nitrate reduction reactions as it requires 5 electrons in total from NO 3 − to N·. The selectivity of N 2 could be improved by technically optimizing the loading of cocatalysts. Efforts are still needed to generate possible heterojunctions to further facilitating the charge separation and light harvesting ability, which usually show superior performances than single-phase catalysts 57,58 .
There still remains an unsolved problem. A small amount of sacrificial agents (in the level of tens of ppm) is needed to prevent the hydrolysis or self-oxidation of metal sulfides. In photocatalytic water splitting, this is unacceptable because it is uneconomic comparing to the photovoltaic industry. While for the environmental protection, we first need to reduce nitrate ions, which is very stable in aqueous solution. Thereafter, the residual sacrificial agent, for example, oxalic acid or formic acid, is quite easy to remove, especially in a low concentration. Alternatively, people tend to use other un-harmful molecules to consuming the photogenerated holes, like sucrose or glucose 12,25 . Overall, metal sulfides have shown a great success in photocatalytic water reduction, where people devoted all efforts to the band structure and morphology engineering 59 . We believe that the future of metal sulfides in the application of water purification would be also promising.

Conclusions
In conclusion, we systematically investigated the performance of spinel ZnCr 2 S 4 in photocatalytic nitrate reduction. The intrinsic narrow bandgap energy (1.96 eV) allow a strong light harvesting ability and indeed the as-prepared ZnCr 2 S 4 behaved as an efficient catalyst for nitrate reduction under UV light irradiation. Cocatalysts, including RuO x , Ag, Au, Pd, Pt, were loaded to further improve the reduction efficiency. A synergic effect was observed when loading Pd and Pt, which offered a very high activity. Increasing the loading content would enhance the activity accordingly. The highest nitrate conversion rate of 45 mg N/h together with the N 2 selectivity of 89% was achieved upon ZnCr 2 S 4 loaded with 3 wt% Pd and 3 wt% Pt, under the following experimental conditions: inner UV-irradiation (125 W), sodium-formate/formic acid buffer solution. The AQY at 380 nm for this particular catalyst was estimated to be 15.46% (with the irradiation beam density of 0.63 W). Most importantly, the visible light activity was explored for ZnCr 2 S 4 loaded with three cocatalysts simultaneously, including Pd, Pt and Au nanoparticles. With the assistance of the SPR effect of Au nanoparticles, the optimal conversion rate of nitrate is 1.352 mg N/h under full arc Xe-lamp, and 0.452 mg N/h under pure visible light (λ > 400 nm) irradiation. Our work proved that metal sulfides with appropriate modifications are good candidates for photocatalytic nitrate reduction.

Methods
Preparations of the catalysts. Cr 2 S 3 used in our study is not from commercial source but from a decomposition of a home-made Cr-containing compound. In detail, 2 mmol of CrCl 3 ·2H 2 O and 60 mmol of H 2 NCSNH 2 were mixed with 1.5 g of C 2 H 4 O 4 ·2H 2 O evenly and then the mixture was transferred into a 50 mL Teflon in a stainless-steel autoclave and sealed. After reacting at 230 °C for 3 days, a black powder sample was obtained by washing away soluble residuals, and it was further converted into Cr 2 S 3 by annealing under vacuum at 600 °C for 2 h. Thereafter, a stoichiometric mixture of Cr 2 S 3 and ZnS (Alfa Aesar, 99.9%) was homogenized using an agate mortar and followed by a heating at 700 °C for 2 h in an evacuated tube furnace. The resultant brown powder was checked to be phase-pure polycrystalline ZnCr 2 S 4 .
The noble metal or metal oxide cocatalyst loading to ZnCr 2 S 4 was performed using the procedure described below. Typically, 0.20 g of ZnCr 2 S 4 , 1.4 mL of H 2 PtCl 6 ·6H 2 O (1.48 mg Pt/mL), and 30 mL of distilled water were placed in a 100-mL beaker. This solution in a 100 mL beaker was processed with an ultrasonic treatment for 20 minutes. An appropriate amount of diluted KBH 4 aqueous solution was added into the beaker very slowly with constant stirring. Finally, the obtained powder sample was extensively washed by water and dried at 60 °C. For other metal loading, the used sources are RuCl 3 , AgNO 3, HAuCl 4 ·4H 2 O, and PdCl 2 , respectively. It is assumed that most cocatalyst ions in aqueous solution were successfully loaded. The accurate amount of the loading cocatalyst is in fact difficult to determine as the insoluble nature of most noble metal.
General characterizations. Powder X-ray diffraction (XRD) data were collected on a PANalytical X'pert diffractometer equipped with a PIXcel 1D detector (Cu Kα radiation, 1.5406 Å). The operation voltage and current were 45 kV and 40 mA, respectively. Scanning electron microscopy images were recorded using a JEOL JSM-7800F electron microscope at a working distance of 4.0 mm. UV-Vis diffused reflectance spectra (DRS) were recorded by Shimadzu UV-3600 spectrometer equipped with an integrating sphere attachment. BaSO 4 was used as reflectance standard. X-ray photoelectron spectra (XPS) were acquired with UK Kratos Axis Ultra spectrometer with Al Kα X-ray source operated at 15 kV and 15 mA. Electron binding energies were calibrated against the C 1 s emission at E b = 284.8 eV to correct the contract potential differences between the sample and the spectrometer.
Photocatalytic activity evaluation. The photocatalytic activities of the prepared catalysts were mostly tested in a sealed circulation system equipped with a vacuum line (Perfect Light, LabSolar-IIIAG), a 150-ml Pyrex glass reactor, and a gas sampling port that was directly connected to a gas chromatograph (Shanghai Techcomp, GC7900). A 5 °C cycling bath was applied to the reaction vessel to keep the temperature constant and cool. The gas chromatograph was equipped with a thermal conductivity detector and a column packed with 5A molecular sieves. Helium was used as the carrier gas to detect the so-produced N 2 online. 500 W Hg or 300 W Xe-lamp were used to provide UV-or visible light irradiation (CEL-M500 or CEL-HXF300, Beijing AuLight Ltd. Co.), which was applied from the top of the reaction vessel (See Supplementary Fig. S5).
In most cases, our photocatalytic experiments were performed using the above setup. In some cases, we also applied a setup with an inner irradiation lamp. The photocatalytic reaction was carried out in a double-walled quartz cell cooled by water with a 125 W Hg lamp as the light source. A schematic view of the setup can be found in SI.
KNO 3 and KNO 2 were used as the nitrate and nitrite source, respectively. The concentration was calculated by the weight of nitrogen. For example, 100 mL of NO 3 − (NO 2 − ) aqueous solution containing 50 ppm N was obtained by adding 36.1 mg of KNO 3 (or 30.4 mg of KNO 2 ) into 100 mL of distilled water. Either sodium oxalate (in the concentration of 0.026 mol/L) or an acid buffer solution (the initial concentration of sodium-formate and formic are both 0.035 mol/L) was used as the sacrificial reagent. The respective dosage of sacrificial agent is 2-and 7-times excess (assuming all the carbon in C 2 O 4 2− or COOH − were oxidized to CO 2 ). In previous reports, the common usage of sacrificial agent was about 2~15 times excess 5,6,10,40 .
During the reaction, a small amount of the solution was withdrawn periodically, the catalyst was immediately separated by centrifugation, and the upper solution was analyzed to determine the residual concentration of NO 3 − and NO 2 − with an UV-Vis spectrophotometer according to the colorimetric method 60 . Ammonia was not detected throughout our study. For detail, we mixed the Nessler's reagent (HgCl 2 -KI-KOH aqueous solution) with the sample solution. If there was any ammonia, there should be an absorption centered at 420 nm. The detection limit is 0.015 mmol/L experimentally, and no such absorption was observed in our study and thus, we conclude no ammonia can be produced if ZnCr 2 S 4 was used as the photocatalyst. In fact, ZnCr 2 S 4 is unable to photocatalyze the H 2 generation in any conditions, including using the common sacrificial agent of Na 2 S and Na 2 SO 3 . We believe this might be responsible to the absence of ammonia.
It is known that the photocatalytic activity strongly depends on the applied experimental conditions, including but not limited to the mass of the photocatalyst, the incident beam intensity, the volume and the concentration of the starting aqueous solution, and so on. Please note the photocatalytic efficiency in this work is presented as the reduced amount of nitrate (calculated according to the mass of N) per hour with the unit mg N/h. Unless further stated, the N 2 selectivity refers to the differential selectivity. In our case, the final N 2 selectivity could easily achieve 100% by simply extending a few more hours of irradiation.
In real application, the conversion rate and differential selectivity are the common criteria to evaluate the quality of a particular photocatalyst. While, from the fundamental aspect, the apparent quantum yields (AQYs) under a monochromatic irradiation is more meaningful. In our study, the number of reacted electrons can be calculated according to the nitrate reduction rate and the N 2 selectivity, and the number of incident photons can be measured by the Si-photodiode.