Formulation of C, Mn-ZnO CTSHSs. The core-triple shell hollow spheres are obtained via a facile one-pot solvothermal preparation of Zn, Mn CP spheres (Zn, Mn-CPSs) followed by controlled air-annealing as depicted in Fig. 1. The CPSs as precursor were firstly formulated by the coordination interaction between Zn- and Mn- acetate and salicylate ligands (Fig. 1a). The carboxylate groups of salicylic acid play a significant role that they can coordinate to metal ions, enabling a delicate control over the composition. Moreover, this method is universal to most of the transition metals because non-selectivity is witnessed for salicylic acid.26
Fourier-transform infrared spectroscopy (FTIR) spectra of the Zn, Mn-CPSs (Supplementary Information Figure S1) evidenced the interaction between deprotonated carboxylate groups of salicylate ligands and metal ions. Such interaction is assured after distinctive intensity drop of the corresponding C = O peak centered at 1657 cm-1 and a slight shift to 1654 cm-1.26 The deprotonation of the carboxylate groups is very critical to boost such coordination interaction. As a strong conjugated base, acetate anions can deprotonate the carboxylate groups of salicylic acid and thus enable reliable coordination between salicylate ligand and metal cations. To confirm this assumption, we replaced Zn- and Mn-acetates by their sulfates, nitrates or chlorides and followed the same preparation recipe. All of them fail to fabricate such spherical architecture. We attribute such failure to the inability of the corresponding anions to deprotonate carboxylate groups of salicylic acid. After formation of Zn, Mn-CPSs, the generation of hollow structures derives from the calcination process. In this step, the Mn, Zn-CPSs precursor is subjected to air calcination at 550 ᵒC, a step that culminates the unique CTSHSs morphology by heterogeneous contraction process (Fig. 1b,c).
Transmittance electron microscopy (TEM) (Fig. 2) and field emission scanning electron microscopy (FESEM) (Supplementary Information Figure S2) images show Zn, Mn-CPSs as high uniform solid spheres with an average diameter of about 1.3 µm and smooth surface texture. After calcination for 3 h, unexpectedly, the solid Zn, Mn-CPSs were converted into unique concentric C, Mn-ZnO CTSHSs, as noticed from the high contrast between the shell edges and centered hollow regions (Fig. 2b and Fig. 3e). The CTSHSs retain the spherical shape of CPSs precursor but the diameter is dramatically shrunk to 650 nm due to the contraction upon calcination. The surface turns rough, pervaded with crystallites with invasive pores. The size of the crystallites was measured to be 25 nm from high-resolution TEM. Its porous structure was also confirmed by N2 adsorption-desorption isotherm (Supplementary Information Figure S3 and Table S1). Additionally, uniform distribution of the constituent elements ca. Zn, O, Mn and C are observed from EDS elemental mapping of the calcined sample (as taken for sample calcined for 1 h, Fig. 1d-g). The elemental composition distribution is presented in Supplementary Information Figure S6.
To elucidate the evolution mechanism of the CTSHSs morphology, FESEM and TEM images of the intermediates at different calcination stages were recorded (Fig. 3 and Supplementary Information Figure S2). At the initial stages of the calcination process (after 0.5 h), a significant temperature gradient (∆T) along the radial direction of the CPSs was generated where the outermost building units are subjected to abrupt heating. Meanwhile, the carbon entities vanish and ZnO nanoparticles were crystallized at the outer surface. Thus, a core-shell structure (the first ZnO shell conjoined to the entire core) was generated (Fig. 1b, Fig. 3b). Once forming, the rigid ZnO shell plays an indispensable role in preventing further contraction of the outer diameter, even if the entire core suffers from continuous shrinkage upon the disintegration of organic moieties by further annealing, known as heterogeneous contraction.27–30
With extended calcination time, both the core and the emerged new shell experienced two opposite forces. Namely, inward cohesion force (Fc) originated from the sharp contraction of the interior core upon continuous disintegration of the skeletal carbon and outward adhesion force (Fa) emanated from the cling of adjacent crystallites (Fig. 1c). The balance between both forces culminated in varied structures. When the CPSs were calcined for 1 h, the temperature gradient gets smaller. And Fc exceeds Fa, which leads to separation of the entire core from the outer shell, forming a yolk-shell structure (Fig. 1b, Fig. 3c, and Supplementary Information Figure S2b). With calcination continued for 2 or 3 h, the entire yolk underwent similar processes described before, forming the unique second (core-double shell, Fig. 3d and Supplementary Information Figure S2c) and the third shell (core-triple shell, Fig. 3e and Supplementary Information Figure S2d).
It is noteworthy that the disintegration of the carbon moieties on both exterior and interior surfaces induces the formation of observable pores within the shell layers (Supplementary Information Figure S2e), which facilitates the mass transfer of the reactants/products.31,32
All the calcined samples together with commercial ZnO (for comparison) show diffraction peaks (Supplementary Information Figure S4) that are well indexed to Wurtzite hexagonal ZnO (JCPDS PDF Card No. 36-1451).33 Adequate purity and good crystallinity of the prepared ZnO CTSHS samples are reflected from the sharp peaks and absence of secondary peaks even with 5% Mn loading. Additionally, compared with commercial ZnO, the lattice of the prepared catalysts is mostly contracted after the doping process (see Supplementary Information for more discussion). These results suggest the successful doping of Mn and/or C within the crystal lattice of ZnO and Mn species are implemented to substitute Zn ions in the lattice.
The XPS spectra further indicate the successful doping of Mn within the crystal lattice of ZnO, originating from the distinctive Mn 2p and Mn 3 s peaks appeared in the XPS profiles and the shift of Zn 2p3/2 peak (see Fig. 6 and Supplementary Information Figure S5 for more details about the chemical state of Mn within the ZnO lattice). The actual compositional percentages of Mn and C in 2%Mn, C-ZnO sample are 1.8 and 8.3%, as derived from inductively coupled plasma-atomic emission spectroscopy (ICP-AES) measurements and elemental analysis, respectively, (Supplementary Information Figure S6). The presence of Mn is believed to promote catalytic conversions owing to their ability to share electrons with reactant molecules and/or intermediates.34–37 Besides, doping gives rise to defect states and impurity levels. They not only reduce the bandgap of the semiconductor, but also improve the light absorptivity of the photocatalyst. The defect states could support the charge separation process through obstructing the electron-hole recombination by electron trapping.38–41 Specifically, introduced transition metal ions could act as distinctive active sites or enrich present active sites, which evidently support reactants adsorption, activation and further chemical conversion.42–44
The photocatalytic performances of the prepared catalysts were evaluated for CO2 PR under simulated solar light at ambient conditions without photosensitizer, precious cocatalyst or sacrificial reagents. Methanol (CH3OH) was found to be the predominant product together with small amounts of carbon monoxide (CO) (Fig. 4a). The 2%Mn, C-ZnO CTSHSs (CZ-2 in Fig. 4a) photocatalyst affords the highest PR activity with CH3OH evolution rate 11.64 µmol g-1 h-1, two and ten times higher than C-ZnO CTSHSs and comm. ZnO, respectively. The selectivity toward CH3OH production is also dramatically improved from 38% (comm. ZnO) to 85% (CZ-2). The Zn in the comm. ZnO has low tendency to bind the CO intermediate.17 Therefore, the CO desorbs readily from the surface and emerges as a major product. By contrast, in case of Mn-doped catalyst, the Mn centers were reported to show moderate adsorption affinity for CO molecules45. This assists subsequent reduction to CH3OH as a predominant product. These results manifest Mn ions as the more workable active centers during CO2 PR over Mn-ZnO photocatalyst.
The time-course product evolution rate further affirms the effectiveness of the 2%Mn, C-ZnO CTSHSs sample to sustain almost the same activity even after prolonged testing time (Fig. 4b). These findings unambiguously disclose the promising potential of the prepared CTSHSs photocatalysts for CO2 PR.
Furthermore, control experiments manifest that no detectable products are observed (either in the dark or under light irradiation) in absence of the photocatalyst and/or CO2, suggesting that CH3OH and CO are indeed generated from CO2 photocatalytic reduction.46,47 Isotopic labeling experiment (Fig. 5a) strongly confirms that the carbon-based products are indeed derived from CO2 reduction and not from any other carbon source. The achieved CO2 PR rate of the prepared Mn and C-doped photocatalyst under sacrificial-reagent and photosensitizer-free conditions is superior to most state-of-the-art photocatalysts (Supplementary Information Table S3), which reflects its potential.
Additionally, the C-ZnO sample shows five times higher activity than the comm. ZnO sample (even with the same BET surface area, see Supplementary Information Table S1). To delve into the origin of such enhanced activity, UV-vis light absorption, photoluminescence (PL), time-resolved PL (TRPL) and, CO2 adsorption isotherm and electrochemical impedance spectroscopy (EIS) measurements as controlling factors of a photocatalyst activity toward CO2 PR were further tested and the results were depicted in Fig. 4 and Fig. 5. Appreciable enhancement of the visible light absorption was sustained by the fascinating hollow morphology, thanks to the multiple reflections of the incident light inside the interior cavities of the fascinating core-triple shell hollow structure48 (inset of Fig. 4c). Moreover, the codoping process introduces impurity levels within the forbidden gap of ZnO, which indeed contributes to narrowing the bandgap (as depicted from Tauc plots, Supplementary Information Figure S7) and enhancing the light-harvesting ability of the doped samples. A stepwise improvement of the light absorptivity of the prepared photocatalysts is observed in line with Mn-content (Fig. 4c). This finding should be correlated to the fact that more impurity states are developed within ZnO bandgap with increasing Mn doping. In addition, the doping with Mn could initiate other electronic transitions such as ligand to metal charge transfer, metal to ligand charge transfer and d-d transition.49–51 All these transitions are accounted for the whole system light harvesting, which leads to enlarged absorption response.52,53
Furthermore, the PL spectra of the prepared catalysts are step-wisely quenched after C and Mn doping. This reveals that suppressed charge recombination is achieved by doping. The 2%Mn, C-ZnO sample exhibits the lowest PL behavior (Fig. 4d), benefiting from two factors. One is (i) the defect state. The rest are (ii) the core-shell morphology and porous structure. The former could act as trapping centers, quenching the electron-hole inhalation process. Whereas the latter reduces both diffusion lengths of charges and reactants, thus minimizing charges recombination opportunities and provide accessible channels for mass/charge transfer.54 TRPL results also demonstrate that 2% Mn, C-ZnO can retard charge carriers recombination and prolong their lifetime (average charge carrier-lifetime ca. 1.1, 0.9 and 0.8 ns for 2%Mn, C-ZnO, C-ZnO and Comm. ZnO, respectively, Fig. 5b and Supplementary Information Table S4). These results indeed affirm the potential of Mn doping to delay charge recombination.
In addition, the prepared samples (C-ZnO and Mn, C-ZnO CTSHSs) attain very low resistance for charges transfer (Rct) process compared to comm. ZnO sample (as speculated from the EIS measurements, Fig. 5d, where smaller arc diameter represents lesser resistance). Obviously, the fascinating architecture favors charge carrier separation and diffusion from bulk to surface region through few nanometer-thick shells. Especially, the Mn-doped sample shows the least Rct, benefiting from charge trapping action exerted by the doping-based defect states (at optimum content) and the following interfacial charge transfer.55,56
Besides, Mn species could be beneficial for enhancing the adsorption of CO2. To authenticate such a hypothesis, the CO2 adsorption isotherms for C-ZnO and 2%Mn, C-ZnO-CTSHSs are provided in Fig. 5c. Obviously, the 2% Mn-doped sample showed much more adsorption affinity (merely twice) toward CO2 than the undoped sample does. This result coincides well with the photocatalytic activity of both samples (Fig. 4a). The higher the CO2 adsorption affinity of a photocatalyst is, the better the photoreduction activity will be expected. Therefore, it is reasonable to notice such enhanced activity for Mn-doped photocatalyst. The improvement of CO2 adsoprtion after Mn doping was also reported for Mn-In2S3 tested for CO2 electroreduction.57
Two plausible factors contributing to the high CO2 adsorption affinity are proposed. One is the higher content of surface oxygen-species (see XPS interpretation, Supplementary Information for details) than C-ZnO sample. The other is the potential of Mn species replaced Zn ions in the lattice to activate CO2 molecules.
To further explore the functional role of Mn, in-situ irradiated XPS (ISI-XPS) analysis for Mn, C-ZnO CTSHSs sample saturated with CO2 was conducted under both dark and irradiated conditions. The high-resolution XPS peaks are demonstrated in Fig. 6 and supplementary Figure S5 together with the same peaks from bare sample (without CO2 treatment). The loading of CO2 molecules onto the surface was assured by emergence of a tiny M-CO3 peak in C 1 s XPS (Supplementary Information Figure S5b*), suggesting the successful bonding of CO2 molecules to metallic sites. To probe how Mn ions could assist activation of CO2 molecules during the PR process, we calculated the average oxidation states (AOS) of Mn under test according to the following equation:58
where ΔS is the multiplet splitting of Mn 3 s peak (Fig. 6a, b). By substituting ΔS values obtained from Fig. 6 in Eq. 1, the AOS of Mn species are ~ 5, 6 and 5.5 for untreated sample, CO2-loaded sample under dark and light irradiation, respectively. Owing to the calcination of the bare samples under open atmosphere, it is not surprising for Mn in Mn,C-ZnO CTSHSs photocatalyst (without CO2 loading) to acquire such high AOS. Because we measure the change of AOS, further investigations are required to elucidate the actual functional oxidation state involved in CO2 activation. For simplicity, we use Mn5+ and Mn6+ to indicate Mn species with AOS ~ 5 and 6, respectively.
The significant increase in AOS of Mn after CO2 loading and before illumination could be explained in terms of electron transfer from Mn to adsorbed CO2 molecules34,35 In our system, we noticed that the AOS of Mn in CO2-loaded photocatalyst can be restored again under light irradiation. This is due to that photogenerated electrons transfer from ZnO to Mn6+ centers to regenerate the primal Mn5+ again (Fig. 7). This light-switchable conversion between Mn5+/Mn6+ is very critical for restoring the original activity of the Mn-doped catalyst without external treatment or activation steps. In addition, this conversion holds a promising potential for all-in-one CO2 reduction technology. Concisely, Mn6+ species can capture the photogenerated electrons from the CB of ZnO, yielding Mn5+. Mn5+ functions as the active centers for CO2 adsorption and activation via one electron transfer process to yield CO2˙ˉ and Mn6+. At the same time, if more electrons are captured at Mn6+, the generation of CH3OH becomes possible. This finding authenticates the role of Mn species as ionized cocatalyst in the system.
In fact, the regeneration of the Mn species after the activation step is very critical to maintain the activity (Fig. 4b). Moreover, the charge trapping action is of great essence for charge carrier’s separation.Thus the optimal doping from Mn results in lowest charges recombination as revealed from PL results (Fig. 4d).
In this vein, Mn functions as Lewis base center, which supports the adsorption of acidic CO2 molecules and further transfer of one electron thus culminates activated (Fig. 7). This electron transfer process is typically mimicking the electron transfer process involved in Photosystem II during natural photosynthesis.35 Such a process is followed by multiple proton-coupled electron transfer reactions to produce the desired CH3OH molecules (Eqs. 2–6, Supplementary Information).
The adsorption, activation and subsequent reduction of CO2 molecules over Mn-doped ZnO photocatalyst is further confirmed by in-situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) results (Supplementary Information Figure S8, see supporting information for more details), which assures the adsorption, activation and successful conversion of activated CO2 molecules onto CH3OH at the photocatalyst surface (Fig. 7).
All these results unambiguously disclosed the potential of the prepared and light-switchable ionized cocatalyst (Mn5+/Mn6+) integrated C, -ZnO CTSHSs as an efficient photocatalyst for CO2 PR under ambient conditions without sacrificial reagents, precious cocatalysts or external treatment set up for activity restoration.
In summary, core-triple shell C, Mn-codoped ZnO hollow spheres were rationally fabricated via a one-pot synthetic protocol. In addition, they were tested as a highly active and selective catalyst for CO2 photoreduction under simulated sunlight and ambient conditions. The unique core-triple shell hollow structures codoped with Mn and C, holded manifold structural and functional features. These unique features culminated improved light absorption capability, enhanced charge separation and migration efficiency, suppressed e-/h+ pairs undesired recombination probability and abundant adsorption and activation active sites for effective photoredox catalysis. All of these accounted for the superior activity of the prepared photocatalyst. The Mn species played a significant role to activate CO2 molecules through an electron-transfer mechanism, assembling that occur within Photosystem II in natural photosynthesis. We have observed that the ionized Mn species can restore their primal oxidation state and hence their activity by virtue of the photogenerated electrons. Therefore, it can be operated as a light-switchable ionized cocatalytic system. These findings will support the design of efficient photocatalytic materials for sustainable clean energy production and beyond. The present study sheds light on the significance of hollow structure materials as active photocatalyst for CO2 reduction as well as the importance of light-switchable ionized cocatalysts during this process.