3.1 Preparation and characterization
CdS-Sv/Co3O4-Ov with different Co3O4-Ov mass loadings were prepared by high-temperature pyrolysis and in situ growth strategies, as shown in Fig. 1a. ZIF-67 was used as a template to obtain Co3O4-Ov nanocages by high-temperature pyrolysis under air atmosphere. Further, CdS-Sv nanoparticles were grown in-situ on the surface of Co3O4-Ov nanocages by a simple oil bath method. The scanning electron microscopy (SEM) image show that the prepared ZIF-67 is a typical orthododecahedral structure (Fig. 1b). The Co3O4-Ov nanocages obtained further by high-temperature pyrolysis retained the dodecahedral structure of ZIF-67 (Fig. 1c). In addition, we have also prepared CdS with different CdCl2 to TAA molar ratios, and all of them have nanoparticles in their morphological structures (Fig. S1). SEM images of CS/CO-x (x = 20, 40, and 60) composites show that CdS-Sv nanoparticles grow uniformly on the CO3O4-Ovnanocages surface (Fig. 1d and Fig. S2-S4). Figure 1e demonstrates the transmission electron microscopy (TEM) image of the CS/CO-40, which further confirms the uniform growth of CdS-Sv nanoparticles on the Co3O4-Ov nanocages surface. More importantly, the interface formed by CdS-Sv nanoparticles and Co3O4-Ovnanocages two-phase material was directly observed by high-resolution TEM image. The two phases of material at the interface can be aligned for the (101) lattice plane of CdS-Sv nanoparticles and the (311) lattice plane of Co3O4-Ov nanocages, respectively. Furthermore, the TEM energy dispersive spectrometer (EDS) elemental mapping results of individual CS/CO-40 (Fig. 2e) further confirmed the uniform distribution of Co, O, Cd and S elements in the CS/CO-40 composite. And the corresponding EDS spectrum are shown in Fig. S5. All these results indicate that CdS-Sv/Co3O4-Ovcomposites have been successfully prepared.
We further analyzed the phase composition of the photocatalysts by X-ray diffraction (XRD). As shown in Fig. 2a, pure hexagonal CdS (JCPDS 41-1049) and pure cubic Co3O4 (JCPDS 42-1467) have been successfully obtained. The ZIF-67 precursor has been completely converted to Co3O4-Ov by pyrolysis (Fig. S6). The diffraction peaks of the CS/CO-20, CS/CO-40 and CS/CO-60 composite photocatalysts corresponded to the CdS-Sv phase and Co3O4-Ov phase, respectively. This indicates that Co3O4-Ov is able to maintain excellent stability in the oil bath reaction for secondary in situ growth of CdS-Sv. In addition, we have also investigated the effect of different CdCl2 and TAA molar ratios on the crystal surface structure during CdS-Sv preparation. The results show that an increase in TAA content will lead to a decrease in the relative diffraction intensity of the (101) plane (Fig. S7). The molecular structure of the photocatalyst was further investigated by FTIR spectroscopy (Fig. S8). The strong absorption peaks at 659 cm− 1 and 570 cm− 1 can be observed for the Co3O4-Ov, CS/CO-20, CS/CO-40, and CS/CO-60, which are considered as the characteristic peaks of spinel Co3O4−Ov. And the absorption peaks at 659 cm− 1 and 570 cm− 1 are attributed to the stretching vibrational modes of Co2+-O and Co3+-O, where Co2+ and Co3+ are tetrahedral and octahedral coordinated, respectively [27]. Moreover, it also confirms that Co3O4-Ov maintains its structural stability in the secondary reaction. No characteristic peaks of the ZIF-67 precursor were detected in the FT-IR spectroscopy of Co3O4-Ov (Fig. S9), further indicating that ZIF-67 has been completely converted to Co3O4-Ov by pyrolysis treatment. The results of FT-IR analysis are in agreement with those of XRD analysis.
The light absorption capacity of the photocatalysts was analyzed by UV-vis diffuse reflectance spectroscopy (UV-Vis DRS). The absorption edge of CdS-Sv nanoparticles is located at 550 nm (Fig. S10). Co3O4-Ov and CS/CO-40 showed strong light absorption in both UV and visible regions. According to the plot of transformed Kubelka–Munk function versus the energy of exciting light, the band gap values of CdS-Sv and Co3O4-Ov were estimated to be 2.7 eV and 2.4 eV, respectively (Fig. S10). Furthermore, based on the Mott-Schottky plots (Fig. S11), the flat band potentials of CdS-Sv and Co3O4-Ov are − 0.34 V and 0.51 V versus Ag/AgCl, respectively, which are equivalent to -0.14 V and 0.71 V versus the normal hydrogen electrode (NHE), respectively. In general, the conduction band (CB) potential of n-type semiconductors and the valence band (VB) potential of p-type semiconductors are more negative and more positive, respectively, than the flat-band potential [28, 29]. Thus, the CB of CdS-Sv and the VB of Co3O4-Ov are − 0.34 V and 0.91 V versus NHE, respectively. According to the band gap value, the VB of CdS-Sv and CB of Co3O4-Ov can be further calculated as 2.02 V and − 0.58 V respectively.
In general, the larger the specific surface area and the richer the pore structure, the more adsorption sites are found in the photocatalysts. Therefore, the physical properties of specific surface area and pore structure of CdS-Sv, Co3O4-Ov, and CS/CO-x (x = 20, 40, and 60) photocatalysts were further investigated (Fig. S12a). For the CdS photocatalysts, no significant hysteresis loops were observed on the N2 adsorption-desorption isotherm. In contrast, Co3O4-Ov, CS/CO-20, CS/CO-40, and CS/CO-60 all exhibited typical type-IV adsorption-desorption isotherms, suggesting the possible presence of meso- and macropores (Fig. S12b). The rich pore structure of Co3O4-Ov will be more favorable for the in situ confined growth of CdS-Sv nanoparticles, which will lead to the formation of tightly packed heterostructures. The BET surface area of Co3O4-Ov, CS/ CO-20, CS/CO-40, and CS/CO-60 had BET surface areas of 133.3464, 12.2690, 28.6469, and 44.9606 m2/g, respectively, which are much higher than that of the simple CdS-Sv photocatalyst (2.6295 m2/g).
We confirmed the presence of oxygen defects and sulfur defects in the prepared photocatalysts by electron paramagnetic resonance (EPR) tests. As shown in Fig. 2b, the strong EPR signal detected at g = 2.004 is able to indicate the presence of oxygen and sulfur vacancies in Co3O4-Ov and CdS-Sv, respectively [30, 31]. And stronger EPR signals for Co3O4-Ov, CdS-Sv and CS/CO-40 indicate the presence of a higher number of unpaired electrons, which is favourable for the generation of photogenerated carriers [30–32]. In addition, the EPR signal intensity of CS/CO-40 is slightly lower than that of CdS-Sv and Co3O4-Ov, suggesting that CS/CO has a lower proportion of unpaired electrons, which may be due to the bonding effect between the interfaces that allows the oxygen/sulfur vacancies to be compensated.
X-ray photoelectron spectroscopy (XPS) full and high-resolution spectra demonstrate the surface elemental composition and chemical state of CdS-Sv, Co3O4-Ov and CS/CO-40, respectively, as shown in Fig. S13 and Fig. 2(c-f). The Cd 3d3/2 and Cd 3d5/2 spin-orbit splitting peaks of CS/CO-40 are located at 411.60 and 404.80 eV, respectively (Fig. 2c), and are attributed to the presence of Cd-S bond. Interestingly, CS/CO-40 shows additional Cd 3d signal peaks at low binding energies compared to CdS-Sv, which may be attributed to the Cd-O bonds formed at the heterogeneous interface. The peaks located at 162.58 and 161.28 eV can correspond to S 2p1/2 and S 2p3/2 of CS/CO-40, respectively (Fig. 2d). Compared to the S 2p high-resolution spectrum of CdS-Sv, additional S species were also detected at the low binding energy (160.08 eV) of CS/CO-40. Combined with the EPR results, the presence of oxygen vacancies and sulfur vacancies would contribute to the formation of chemical bonds at the interface of CdS-Sv and Co3O4-Ov [33, 34]. Based on this, the peaks at low binding energy of the high-resolution spectra of Cd 3d and S 2p of CS/CO-40 can be attributed to the presence of Cd-O and S-Co bonds, respectively [35–37]. The Co 2p high-resolution spectra of Co3O4-Ov and CS/CO-40 (Fig. 2e) show typical fitted peaks corresponding to Co2+ 2p1/2, Co3+ 2p1/2, Co2+ 2p3/2, and Co3+ 2p3/2, respectively. The O1s spectrum of Co3O4-Ov was fitted to three peaks, which were attributed to lattice oxygen (~ 529.48 eV, OL), oxygen vacancy (~ 530.88 eV, OV) and chemisorbed oxygen (~ 532.48 eV, Oabs), respectively (Fig. 2f). In contrast, the O1s spectrum of CS/CO-40 can identify four fitted peaks for OL (O-Cd), OL (O-Co), OV, and Oabs. More importantly, the binding energies of both Cd 3d and S 2p in CS/CO-40 are negatively shifted compared with that of CdS-Sv. On the contrary, the binding energies of both Co 2p and O1s in the composite photocatalyst are positively shifted compared with that of Co3O4-Ov. This result indicates that when CdS-Sv is in close contact with Co3O4-Ov, the electrons on Co3O4-Ov are transferred to CdS-Sv up to the Fermi energy level (EF) equilibrium, which induces energy band bending, and thus induces the formation of a IEF pointing from Co3O4-Ov to CdS-Sv at the interface.
3.2 Photocatalytic performance and charge separation
The reactivity of the photocatalysts was evaluated under simulated sunlight using RCSP as a biomass reaction substrate. In contrast to some of the typical current photocatalytic research work on the use of raw biomass and biomass polymers as reaction substrates (Table S1), we have innovatively achieved the one-step photocatalytic conversion of RCSP to high value-added C1/C2 gases. The results of the photocatalytic reaction showed that CO, CH4, C2H6, and C2H4 were the main C1 and C2 gas products, and almost no CO2 was produced (Fig. S14). Firstly, we investigated the photocatalytic properties of CdS nanoparticles prepared with different CdCl2 and TAA ratios (1:1, 1:2, and 1:3). In the normal reaction, CO, CH4, C2H6 and C2H4 were detected in the photocatalytic reaction products of all CdS nanoparticles (Fig. S15). Further we also prepared CdS/Co3O4 with different Cd2+/S2− molar ratios and evaluated the photocatalytic properties. The results show that CS/CO-40 (Cd2+/S2− = 1:1) has a higher C1/C2 gas yield (Fig. S16). Based on these results, we further prepared CS/CO-x (x = 20, 40, and 60) photocatalysts to evaluate the photocatalytic activity. As shown in Fig. 3a, the C1/C2 gas products evolution rates of CS/CO-x (x = 20, 40, and 60) are significantly higher than those of CdS-Sv and Co3O4-Ov, and this result further confirms that the heterostructure building can effectively improve the photocatalytic performance. CS/CO-40 exhibited the best photocatalytic activity with CO, CH4, C2H6 and C2H4 evolution rates of 691.99, 2057.69, 202.93 and 187.29 µmol/g, respectively. And it reveals that the CO yield was increased by about 2.3 and 4.7 times, the CH4 yield was increased by 5.4 and 7.2 times, the C2H6 yield was increased by 3.1 and 4.0 times, and the C2H4 yield was increased by 2.9 and 3.3 times compared with CdS-Sv and Co3O4-Ov, respectively. And among all the prepared photocatalysts, CS/CO-40 showed the highest CH4 selectivity and total hydrocarbon (CxHy) gas selectivity of 65.53% and 77.96%, respectively (Fig. 3b). The results of time-dependent photocatalytic product yields of CdS-Sv, Co3O4-Ov and CS/CO-x (x = 20, 40, and 60) are displayed in Fig. 3c and Fig. S (17–20).
We further explored the effect of photocatalytic reaction conditions on the activity. As shown in Fig. 3d, when only DMSO was not added to the reaction system, only CO and CH4 were detected in the photocatalytic products. Experimental group-3 demonstrated that small amounts of CO, CH4, C2H6 and C2H4 could also be produced in the photoreactive system without a catalyst. Experimental group-4 further confirmed that small amounts of CO and CH4 as well as trace amounts of C2H6 and C2H4 could be detected in the absence of catalyst and NaOH. However, only a small amount of CO could be detected in the photoreactive system in the presence of only NaOH without catalyst and DMSO. All experimental control group results illustrate the important auxiliary roles of DMSO and NaOH for promoting the photocatalytic conversion of structurally complex natural polymers to C1/C2 gases.
In order to investigate the photogenerated charge separation efficiency of the photocatalysts, transient photocurrent and electrochemical impedance spectroscopy tests were performed. By continuously recording the transient photocurrent response for several on/off cycles under light irradiation, it was confirmed that the photocurrent of the photocatalysts exhibited high reproducibility and stability over multiple cycles (Fig. 4a). The CdS-Sv/Co3O4-Ov composite photocatalysts showed stronger photocurrent intensity compared with CdS-Sv and Co3O4-Ov, which further confirmed that the composite photocatalysts possessed higher photogenerated electron-hole separation efficiency. Similarly, electrochemical impedance spectroscopy (EIS) demonstrated that the composite photocatalysts have smaller charge transfer resistance and thus more efficient photogenerated charge separation efficiency (Fig. 4b). In addition, the effect of heterogeneous structure building on photogenerated charge separation was further revealed by PL spectra. It is generally believed that a slower radiative recombination between photogenerated carriers will lead to a lower PL emission intensity [38, 39]. The PL emission intensity of the composite photocatalysts are weaker than that of CdS-Sv (Fig. S21), which indicates that the successful construction of the heterojunction can effectively promote the charge separation efficiency. All of the above photoelectrochemical test results show that CS/CO-40 has a superior photogenerated charge separation ability, which is consistent with the photocatalytic reaction results.
Reactive radicals are essential for a deeper understanding of the photocatalytic reaction mechanism and play an important role in photocatalytic reactions. To further identify the main oxidatively active species during the photocatalytic reaction, we performed free radical trapping experiments using benzoquinone (BQ), isopropanol (IPA) and ethylenediaminetetraacetic acid (EDTA) as superoxide radicals (·O2−), hydroxyl radicals (·OH) and hole (h+) trapping agents, respectively. As shown in Fig. S22, the addition of all scavengers had a significant effect on product generation. In the experimental group with the addition of BQ, the production of C2H4 and C2H6 was suppressed while the evolution rate of CH4 was significantly promoted. Interestingly, scavenging of ·OH by isopropanol inhibited methane production while having less effect on the evolution rates of CO, C2H4 and C2H6. When the h+ were captured by EDTA, the evolution rate of all products decreased. The analysis of the above results indicates that ·O2− and ·OH are the two major oxidatively active radicals, which play a key role in the production of C2 products and CH4, respectively.
The ·O2− and ·OH enerated during the reaction were determined using ESR tests with DMPO as the trapping agent, and the results are shown in Fig. 4(c, d). Under dark conditions, neither DMPO-·O2− nor DMPO-·OH ESR signals were detected. Under light, the photocatalyst exhibited strong DMPO-·O2− and DMPO-·OH ESR signal peaks. And the DMPO-·O2− and DMPO-·OH ESR signal intensities are enhanced with the increase of irradiation time. The ability of the VB of CdS-Sv and the CB of Co3O4-Ov to satisfy the standard potentials of H2O/·OH (1.99 V vs. NHE) and O2/·O2− (− 0.33 V vs. NHE), respectively, demonstrates that the charge transfer between CdS-Sv and Co3O4-Ov follows the Z-scheme pathway [40, 41]. The above results also indicate that ·O2− and ·OH are the main active radicals produced during the photocatalytic reaction.
3.3 DFT calculation and photocatalytic mechanism
Interfacial interactions and built-in electric fields between CdS-Sv and Co3O4-Ov interfaces in CdS-Sv/Co3O4-Ov heterojunctions are crucial to promote efficient photogenerated charge transfer. It is well known that when two semiconductors with different EF and work functions (Φ) are in close contact, a IEF will be formed at the interface [42, 43]. The interfacial interactions and charge transfer mechanisms in CdS-Sv/Co3O4-Ov heterojunctions are further revealed by DFT calculations. The structures of CdS-Sv, Co3O4-Ov, and CdS-Sv/Co3O4-Ov heterojunctions model were successfully established (Fig. S23). The crystal structure models of CdS-Sv and Co3O4-Ov from different views are shown in Fig. S24 and S25. The Φ and EF of CdS-Sv and Co3O4-Ov were simulated with first principles (Fig. 5a and 5b). The Φ of CdS is significantly larger than that of Co3O4-Ov, which implies that the electrons of the latter are more likely to escape. When the heterogeneous interface is in contact, the difference in EF will result in the transfer of electrons from Co3O4-Ov to CdS-Sv. The electron density distribution at the heterojunction interface was further simulated with charge density difference to visualize the inter-interface electron transfer (Fig. 5c and 5d) [44, 45]. The results confirm the coupled connection of chemical bonds between the interfaces of the two materials and the presence of strong interfacial interactions. A spontaneous interfacial charge transfer pathway from Co3O4-Ov to CdS-Sv is theorized to exist when the two materials are in close contact, and the interfacial chemical bonding can act as a channel to facilitate inter-interfacial charge transfer. The results of the above analysis are consistent with the XPS analysis.
Based on these results, a charge transfer mechanism for heterojunction photocatalysts is proposed. The energy band structure was determined by UV-vis DRS and Mott Schottky tests. The energy band structure of the CdS-Sv /Co3O4-Ov system before and after contact is shown in Fig. 5e. When CdS-Sv is in contact with Co3O4-Ov, the electrons in Co3O4-Ov are transferred to CdS-Sv until the EF are balanced, resulting in the formation of a unique IEF. Under illumination, the IEF promotes the transfer of photoexcited electrons from CdS-Sv to the VB in Co3O4-Ov, and the composite is realized through Z-scheme charge transfer. The photogenerated electrons in the CB of Co3O4-Ov reduce O2 to ·O2−, while the holes generated in the VB of CdS-Sv oxidize H2O to form ·OH. The ESR results and DFT calculations further indicate that the current system should be a Z-scheme heterojunction rather than type-II heterojunction. In addition, the chemical bonds formed at the heterointerface between CdS and Co3O4-Ov can act as interfacial charge transfer channels and also play an important role in accelerating the charge transfer between CdS-Sv and Co3O4-Ov. Photocatalytic conversion of RCSP to high value-added C1/C2 gases under the synergistic effect of photoactive radicals, DMSO and NaOH (Fig. 6a). In a world, the CdS-Sv/Co3O4-Ov heterojunction has a high performance photocatalytic RCSP conversion performance due to its strong light absorption ability, follows a Z-scheme charge transfer pathway in order to retain a strong redox capacity, and is able to promote charge transfer through interfacial chemical bonding.
For further in-depth analysis of the reaction mechanisms, glucose, fructose and xylose were further investigated as typical lignocellulose-derived monomers. The results of photocatalytic experiments using glucose, fructose and xylose as reaction substrates showed that gas-phase products such as CO, CH4, C2H4 and C2H6 were detected, which is in agreement with the results of the RCSP photocatalytic experiments (Fig. S26). And liquid phase product analysis showed that organic acids such as lactic acid, acetic acid, propionic acid and formic acid were the main products, with lactic acid having the highest yield (Fig. S27). Combined with previous studies [11, 46], we propose possible reaction pathways for the conversion of glucose, fructose, and xylose to organic acids, as shown in Fig. S28 and Fig. S29. Glucose is oxidised by α- and β-oxidation to formic acid, intermediate-I, ethanedioic acid, and intermediate-II. Ethanedioic acid can produce acetic acid and formic acid by dehydration and C-C bond breaking, respectively. Formic acid can also be produced when the C-C bond of intermediate-I is broken. The isomerisation of glucose to fructose is further followed by retro-aldol condensation reaction to produce glyceraldehyde and 1,3-dihydroxyacetone. Lactic acid is then further produced by dehydration and 1,2-hydride shift. During the oxidation of xylose, xylose is first isomerised into intermediate I and intermediate II. And then Intermediate-I generates glyceraldehyde and ethanedioic acid by retro-aldol condensation reaction. Intermediate-II generates acetic acid and ethanedioic acid by α- and β-oxidation. The above products were further converted to lactic acid and formic acid, respectively. Lactic acid dehydrates to form propionic acid. We speculate that the production of products such as CO, CH4, C2H4 and C2H6 may be attributed to the further oxidation of organic acids during photocatalysis.
In order to verify the conjecture, the product distribution of photocatalysis was further evaluated using organic acids such as formic acid, lactic acid, propionic acid and acetic acid as reaction substrates. The results were as expected (Fig. S26), lactic acid, acetic acid, formic acid, and propionic acid showed high selectivity for the conversion of CH4, CO, C2H4, and C2H6, respectively. Based on the above findings and previous studies, we propose a possible mechanism of transformation, as shown in Fig. 6b. Typical lignocellulosic monomers such as glucose and xylose are oxidised in the presence of oxidatively active species to produce organic acids such as lactic, propionic and acetic acids as well as formic acid through cascade-by-cascade oxidation and C-C bond breaking, which are ultimately oxidised completely to form methane.