Polyoxometalate-Based Catalysts for CO2 Conversion

Polyoxometalates (POMs) are a diverse class of anionic metal-oxo clusters with intriguing chemical and physical properties. Owing to unrivaled versatility and structural variation, POMs have been extensively utilized for catalysis for a plethora of reactions. In this focused review, the applications of POMs as promising catalysts or co-catalysts for CO2 conversion, including CO2 photo/electro reduction and CO2 as a carbonyl source for the carbonylation process are summarized. A brief perspective on the potentiality in this field is proposed.


Introduction
As the main greenhouse gas produced by human activity, about 10 ± 0.5 gigatons CO 2 were released in 2018, bringing the atmospheric CO 2 concentration level over the threshold of 400 ppm [1]. The large increase of global CO 2 levels has led to serious environment problem. Meanwhile, it is also the most abundant and nontoxic carbon resource for preparing useful compounds [2]. From a practical point of view, its catalytic conversion is of significance to the supply of renewable energy, chemicals, and mitigation of global warming.
Polyoxometalates with {MO x } (x = 5, 6) as basic construction units have derived an enormous fraternity of inorganic molecular complexes [3][4][5][6]. To obtain various composite materials with specific function, the modification and decoration of POM can be attained by partially substituting {MO x } units with different transition metal moieties or attaching organometallic complexes onto POM. Based on geometrical morphology, most of POMs can be classified as Keggin, Wells-Dawson, Anderson-Evans, Silverton, Waugh, Strandberg, Lindqvist, and Peacock-Weakley type structures, which are named after their corresponding discoverers [7]. Among these types, Keggin-type POMs ([XM 12 O 40 ] n− ) are comparatively well-studied. POMs discussed in the following sections are mostly Keggin type [8]. The original Keggin structure is designated α-, which consists of a tetrahedron central ion, [XO 4 ] n − 8 , caged by twelve MO 6 octahedron. The Keggin structure includes four additional isomers (β-, γ-, δ-, and ε-), each resulting from 60 • rotations of the four {M 3 O 13 } units [9]. The lacunary species of Keggin-type POMs stem from removing a variable number of {MO 6 } octahedra from the plenary polyanion, resulting in metal-oxide clusters with vacant addenda metal sites. Such vacant sites can be treated as inorganic multidentate ligands (Figure 1) [10], For example, sandwich-type POMs are usually synthesized by the reaction of transition metal ions with appropriate lacunary POM precursors [5,8]. The extraordinary diversity and synthetic accessibility of POMs have led to a wide spectrum of applications, ranging from catalysis, biochemistry/medicinal chemistry, to materials science [11][12][13][14]. In the field of catalysis, the development of POMs as oxidation and acid catalysts was flourish in the past few decades, with As early as 1988, Kozik and co-workers first reported the coordination of CO2 with several POM derivatives [16]. Afterwards more researches about the interaction of POMs with CO2 emerged [17][18][19][20][21][22]. Since the interaction between CO2 and POM is one of the crucial steps for catalytic transformations of CO2, it is necessary to clarify the mode of such interaction. However, considering the facts that CO2 can be transformed to CO [2][3] or HCO -3 with water and this transformation is both reversible and temperature-dependent, the explicit "real" form of dissolved CO2 to interact with POM catalyst, was challenging to be justified. POMs displayed variable interacting modes. 13 C NMR, UV/vis, IR with isotope-labelled CO2 ( 13 CO2 and C 18 O2), and X-ray crystallography are useful tools to solve this issue. As pointed in the seminal report of Kozik et al. [16], the IR spectra and thermochromic behavior of POM α-[SiW11O39Co] 6− in the presence of either CO [2][3] or HCO -3 was in contrast to the scenario with CO2. On 13 C NMR spectra, the 13 C chemical shifts of aqueous CO2, HCO -3 , and CO [2][3] are respectively at 125, 160, and 162 ppm. After bubbling CO2 into hydrous toluene solution of α-[SiW11O39Co] 6-, the appearance of two signals at 792 and 596 ppm demonstrated the presence of two different kinds of paramagnetic CO2 species. In their report, two patterns were suggested on the basis of all 13 C NMR, IR observations: (i) the complexation between α-[SiW11O39Co] 6-and CO2 was either via CO2 complexes with a direct η 1 metal-carbon bond or bicarbonate complexes; (ii) the existence of H-bonding in the CO2 complexes was plausible. Other modes of POM-CO2 or POM-CO [2][3] interaction were also discovered. Hill [20]. [SiMo11CoO38(CO2)]n polymeric chains reported by Xu et al. displayed the μ-η 1 ,η 1 -OCO linear coordination mode of POM-CO2 interaction [21]. Two kinds of POM-CO2 interaction modes were observed on [(n-C8H17)4N]8[α2-P2W17O61Zn-(CO2)] [21]. The stronger "side-on" binding of CO2 by is predominant at higher temperatures (room temperature down to ca. 250 K) and the more weakly coordinated "end-on" POM-Zn-O-C-O structures were observable at lower temperatures (  As early as 1988, Kozik and co-workers first reported the coordination of CO 2 with several POM derivatives [16]. Afterwards more researches about the interaction of POMs with CO 2 emerged [17][18][19][20][21][22]. Since the interaction between CO 2 and POM is one of the crucial steps for catalytic transformations of CO 2 , it is necessary to clarify the mode of such interaction. However, considering the facts that CO 2 can be transformed to CO 2− 3 or HCO − 3 with water and this transformation is both reversible and temperature-dependent, the explicit "real" form of dissolved CO 2 to interact with POM catalyst, was challenging to be justified. POMs displayed variable interacting modes. 13 C NMR, UV/vis, IR with isotope-labelled CO 2 ( 13 CO 2 and C 18 O 2 ), and X-ray crystallography are useful tools to solve this issue. As pointed in the seminal report of Kozik et al. [16], the IR spectra and thermochromic behavior of POM α-[SiW 11 O 39 Co] 6− in the presence of either CO 2− 3 or HCO − 3 was in contrast to the scenario with CO 2 . On 13 C NMR spectra, the 13 [20]. [SiMo 11 CoO 38 (CO 2 )] n polymeric chains reported by Xu et al. displayed the µ-η 1 ,η 1 -OCO linear coordination mode of POM-CO 2 interaction [21]. Two kinds of POM-CO 2 interaction modes were observed on [(n-C 8 H 17 ) 4 N] 8 [α 2 -P 2 W 17 O 61 Zn-(CO 2 )] [21]. The stronger "side-on" binding of CO 2 by is predominant at higher temperatures (room temperature down to ca. 250 K) and the more weakly coordinated "end-on" POM-Zn-O-C-O structures were observable at lower temperatures ( Figure 2). One particular case was uptaking 30 2 } 2+ linkers to generate CO 2− 3 at pH 7. Then the uptake of 30 CO 2 molecules was realized via the fast exchange of acetate by carbonate ligands [22]. Above cases indicated the abundance of POM-CO 2 interaction mode.
Molecules 2019, 24, x FOR PEER REVIEW 3 of 26 molecules was realized via the fast exchange of acetate by carbonate ligands [22]. Above cases indicated the abundance of POM-CO2 interaction mode. In this context, we have witnessed the leap in developing POMs-based catalysts for the photoor electrocatalytic CO2 reduction and CO2 as C1 synthon in organic synthesis. Herein, in this review, we focused on the recent experimental and theoretical advances in CO2 transformation with POMsbased catalyst or co-catalyst. The key factors for high-efficiency POMs-catalyzed CO2 conversion are highlighted.

Photocatalytic CO2 Reduction
Only high-energy vacuum ultraviolet laser (<200 nm) is able to directly excite and split CO2 into CO and O fragments [23], therefore highly efficient photocatalyst is vital to motivate and accelerate photocatalytic CO2 reductions with low-energy common visible light (>380 nm). This reduction process usually involves multi-electron transfer and the final products are carbon monoxide (CO), formic acid (HCOOH), formaldehyde (HCHO), methanol (CH3OH), and methane (CH4) [24,25]. The corresponding redox potentials for the possible CO2 reduction at pH = 7 are listed in Table 1.
Since the pioneering application of Honda-Fujishima effect in TiO2-catalyzed CO2 photoreduction was reported [26], multitudinous different type of photocatalysts, including simple metal oxides [25], perovskite oxides [27,28], C3N4 [29], MOFs [30], conjugated polymers [31,32], and POMs [33,34] have been developed. Since the seminal study on the photochemistry of molybdates and tungstates for analytical purposes over half century ago [35,36], the photoredox chemistry of POMs has been verified for a long time. The metals in POMs are fully oxidized with d 0 electron configuration. Light absorption is mainly attributed to O→M ligand to-metal charge transfer (LMCT) bands in the wide range of the electronic spectra. Consequently, an electron is promoted from a spinpaired, doubly occupied bonding orbital (HOMO) to an empty, antibonding orbital (LUMO), thus an oxo-centered radical is generated. This photo-excited POMs are more reactive both in oxidation and reduction than the non-excited species [37]. In the photocatalysis by POM with LMCT mode, forming lower-energy O→M LMCT transitions is vital to more efficiently absorb visible light. Substituting addenda centers with different metals has been well demonstrated as one of the effectual tactics to tune the photochemical properties of POMs and optimize their visible light absorption via narrowing the band gap between valence band and conduction band. For instance, Streb et al. discovered that the vanadium substitution effectively improved the visible light absorption and photocatalytic activity of molybdate [Mo6O19] 2- [38]. TD-DFT calculations disclosed the corresponding contribution of O→V LMCT on absorption profile in the visible range. The progress of such strategy on POMcatalyzed selective photooxidation driven by visible light has been included in the recent review [15]. Besides altering the addenda centers on POMs, various other strategies have also been established in order to foster POMs as effective photocatalyst and most of their utilization were focused on In this context, we have witnessed the leap in developing POMs-based catalysts for the photoor electrocatalytic CO 2 reduction and CO 2 as C1 synthon in organic synthesis. Herein, in this review, we focused on the recent experimental and theoretical advances in CO 2 transformation with POMs-based catalyst or co-catalyst. The key factors for high-efficiency POMs-catalyzed CO 2 conversion are highlighted.

Photocatalytic CO 2 Reduction
Only high-energy vacuum ultraviolet laser (<200 nm) is able to directly excite and split CO 2 into CO and O fragments [23], therefore highly efficient photocatalyst is vital to motivate and accelerate photocatalytic CO 2 reductions with low-energy common visible light (>380 nm). This reduction process usually involves multi-electron transfer and the final products are carbon monoxide (CO), formic acid (HCOOH), formaldehyde (HCHO), methanol (CH 3 OH), and methane (CH 4 ) [24,25]. The corresponding redox potentials for the possible CO 2 reduction at pH = 7 are listed in Table 1. Table 1. Redox potentials for CO 2 reduction.

Products
Products Reaction E 0 (V) E 0 (V) pH = 7 Since the pioneering application of Honda-Fujishima effect in TiO 2 -catalyzed CO 2 photoreduction was reported [26], multitudinous different type of photocatalysts, including simple metal oxides [25], perovskite oxides [27,28], C 3 N 4 [29], MOFs [30], conjugated polymers [31,32], and POMs [33,34] have been developed. Since the seminal study on the photochemistry of molybdates and tungstates for analytical purposes over half century ago [35,36], the photoredox chemistry of POMs has been verified for a long time. The metals in POMs are fully oxidized with d 0 electron configuration. Light absorption is mainly attributed to O→M ligand to-metal charge transfer (LMCT) bands in the wide range of the electronic spectra. Consequently, an electron is promoted from a spin-paired, doubly occupied bonding orbital (HOMO) to an empty, antibonding orbital (LUMO), thus an oxo-centered radical is generated. This photo-excited POMs are more reactive both in oxidation and reduction than the non-excited species [37]. In the photocatalysis by POM with LMCT mode, forming lower-energy O→M LMCT transitions is vital to more efficiently absorb visible light. Substituting addenda centers with different metals has been well demonstrated as one of the effectual tactics to tune the photochemical properties of POMs and optimize their visible light absorption via narrowing the band gap between valence band and conduction band. For instance, Streb et [38]. TD-DFT calculations disclosed the corresponding contribution of O→V LMCT on absorption profile in the visible range. The progress of such strategy on POM-catalyzed selective photooxidation driven by visible light has been included in the recent review [15]. Besides altering the addenda centers on POMs, various other strategies have also been established in order to foster POMs as effective photocatalyst and most of their utilization were focused on photooxidation of water, photodegradation of organic dyes, as well as H 2 evolution [37,39]. Here the progress of photoreduction of CO 2 with POMs-based catalyst was summarized.  Figure 3 [40]. Triethyl amine acted as the sacrificial agent. CO was the major product (~50 µmol after 20 h irradiation with ca. 2% quantum yield). None of HCOOH, MeOH, and CH 4 was detected. UV/Vis, EPR, 13  photooxidation of water, photodegradation of organic dyes, as well as H2 evolution [37,39]. Here the progress of photoreduction of CO2 with POMs-based catalyst was summarized. Neumann and co-workers reported that Ru-substituted POM ([Ru III (H2O)SiW11O39] 5-) in combination with tertiary amines exhibited photoactivity for CO2 reduction Figure 3 [40]. Triethyl amine acted as the sacrificial agent. CO was the major product (~50 μmol after 20 h irradiation with ca. 2% quantum yield). None of HCOOH, MeOH, and CH4 was detected. UV/Vis, EPR, 13 C NMR, and isotope labelling demonstrated that coordinated CO2 ([Ru III (CO2)SiW11O39] 5-) was easily formed by substituting H2O in [Ru III (H2O)SiW11O39] 5-. DFT results demonstrated that CO2 tended to coordinate to Ru III by forming a Ru-O bond in an "end-on" manner. Based on these experimental and computational results, the authors concluded that the Ru III site was responsible for activating CO2 via the coordination and the lacunary [SiW11O39] 8-site acted as photocatalyst in this process. VI 10 O40 possessed the photoreductive activity for converting CO2 to CO, because its transition to excited CO2-reduction active state was allowed by absorbing visible light. CO were almost exclusively produced with turnover of 22.6% and 1.1% quantum yield. Besides CO, only trace CH4 but no further CO reduction products (e.g., CH3OH, HCHO, and HCOOH) was By employing H 2 instead of tertiary amines as the sacrificial agent, a viable CO 2 reduction approach with Pt/C and Re I (L)(CO) 3   The TD-DFT calculations disclosed a more detailed mechanism for the photoreduction of CO2 to CO with this POM-Re I complex catalyst [42]. Simulated absorption spectrum of Re I -POM complex was calculated at different functional level (i.e. X3LYP, M06X, B3LYP, and CAMB3LYP) and the computed results were found to be basically consistent with the experimental absorption spectrum (two major bands with maxima at 500 and 656 nm). In the ground state without any light irradiation, bonding CO2 with Re I center to afford the reactive POM-Re I -CO -2 complex had to overcome a high energy barrier (≥38 kcal mol −1 ), therefore, the direct CO2 reduction with Re I -POM occurred less likely ( Figure 5). In the conditions with photoexcitation, the excited Re I -POM was allowed to interact with CO2 to form POM-Re I -CO -2 . The subsequently protonation and further electron reduction process were exothermic by more than 78 kcal mol −1 . These energy profiles revealed that the role of POM as photosensitizer in the reduced state, electron "shuttles" as well as electron/proton reservoirs in this process. Plausible CO2 photoreduction mechanism consists of five steps: (1) photoexcitation and charge transfer; (2) disassociation of solvent molecule; (3) CO2 coordination; (4) proton and electron transfer; and (5) release of CO and recoordination of solvent.   The TD-DFT calculations disclosed a more detailed mechanism for the photoreduction of CO2 to CO with this POM-Re I complex catalyst [42]. Simulated absorption spectrum of Re I -POM complex was calculated at different functional level (i.e. X3LYP, M06X, B3LYP, and CAMB3LYP) and the computed results were found to be basically consistent with the experimental absorption spectrum (two major bands with maxima at 500 and 656 nm). In the ground state without any light irradiation, bonding CO2 with Re I center to afford the reactive POM-Re I -CO -2 complex had to overcome a high energy barrier (≥38 kcal mol −1 ), therefore, the direct CO2 reduction with Re I -POM occurred less likely ( Figure 5). In the conditions with photoexcitation, the excited Re I -POM was allowed to interact with CO2 to form POM-Re I -CO -2 . The subsequently protonation and further electron reduction process were exothermic by more than 78 kcal mol −1 . These energy profiles revealed that the role of POM as photosensitizer in the reduced state, electron "shuttles" as well as electron/proton reservoirs in this The TD-DFT calculations disclosed a more detailed mechanism for the photoreduction of CO 2 to CO with this POM-Re I complex catalyst [42]. Simulated absorption spectrum of Re I -POM complex was calculated at different functional level (i.e., X3LYP, M06X, B3LYP, and CAMB3LYP) and the computed results were found to be basically consistent with the experimental absorption spectrum (two major bands with maxima at 500 and 656 nm). In the ground state without any light irradiation, bonding CO 2 with Re I center to afford the reactive POM-Re I -CO − 2 complex had to overcome a high energy barrier (≥38 kcal mol −1 ), therefore, the direct CO 2 reduction with Re I -POM occurred less likely ( Figure 5). In the conditions with photoexcitation, the excited Re I -POM was allowed to interact with CO 2 to form POM-Re I -CO − 2 . The subsequently protonation and further electron reduction process were exothermic by more than 78 kcal mol −1 . These energy profiles revealed that the role of POM as photosensitizer in the reduced state, electron "shuttles" as well as electron/proton reservoirs in this process. Plausible CO 2 photoreduction mechanism consists of five steps: (1) photoexcitation and charge transfer; (2) disassociation of solvent molecule; (3) CO 2 coordination; (4) proton and electron transfer; and (5) release of CO and recoordination of solvent. Recently, by replacing crown ether-affiliated 1,10-phenanthroline with 6-hydrazinyl-2,2'bipyridine on the Re complex, the Re2(CO)6Cl2L2-H3PW12O40 (L = 6-hydrazinyl-2,2'-bipyridine) hybrid pair and related relay process for CO2 reduction by photocatalysis and electrocatalysis was developed [43]. Initially, [PW Only CO was observed in this catalytic electro-photochemical reduction process. The result in the dark conditions indicated that light facilitated the CO2 reduction. The formation of dirhenium complex-H3PW12O40 hybrid pair via acid-base interaction was indispensable to this catalytic electro/photochemical transformation.

Heterogeneous Catalysts
Photocatalytic CO2 reduction by POMs-based heterogeneous catalyst has received intensive research interests. As a POM-stabilized multi-Co-oxide clusters, [Co4(PW9O34)2] 10-is comprising a Co4O4 core stabilized by two oxidatively resistant polytungstate ligands (Figure 6a). The Na10[Co4(H2O)2(PW9O34)2 @graphitic carbon nitride hybrid material (Co4@g-C3N4) was reported for the efficient photocatalytic CO2 reduction [44]. It manifested the advantages of convenient recovery, steady reuse, simple preparation and flexible composition. The Co4@g-C3N4 photocatalyst with 43 wt% Co4 content showed high CO yield (107 μmol g -1 h -1 ) and excellent selectivity (94%) (Figure 6b). After 10 h reaction, the production of CO reached 896 μmol g -1 (Figure 6c). After 5 runs, the activity still remained (Figure 6d). Experimental and characterization results revealed that the Co4 unit both facilitated the charge transfer of g-C3N4 and significantly enhanced the surface catalytic oxidative activity. Recently, by replacing crown ether-affiliated 1,10-phenanthroline with 6-hydrazinyl-2,2 -bipyridine on the Re complex, the Re 2 (CO) 6 Cl 2 L 2 -H 3 PW 12 O 40 (L = 6-hydrazinyl-2,2 -bipyridine) hybrid pair and related relay process for CO 2 reduction by photocatalysis and electrocatalysis was developed [43].  3 . This reductive Re I intermediate possessed reactivity for the selective CO 2 reduction. Only CO was observed in this catalytic electro-photochemical reduction process. The result in the dark conditions indicated that light facilitated the CO 2 reduction. The formation of dirhenium complex-H 3 PW 12 O 40 hybrid pair via acid-base interaction was indispensable to this catalytic electro/photochemical transformation.

Heterogeneous Catalysts
Photocatalytic CO 2 reduction by POMs-based heterogeneous catalyst has received intensive research interests. As a POM-stabilized multi-Co-oxide clusters, [Co 4 (PW 9 O 34 ) 2 ] 10− is comprising a Co 4 O 4 core stabilized by two oxidatively resistant polytungstate ligands ( Figure 6a). The Na 10 [Co 4 (H 2 O) 2 (PW 9 O 34 ) 2 @graphitic carbon nitride hybrid material (Co 4 @g-C 3 N 4 ) was reported for the efficient photocatalytic CO 2 reduction [44]. It manifested the advantages of convenient recovery, steady reuse, simple preparation and flexible composition. The Co 4 @g-C 3 N 4 photocatalyst with 43 wt% Co 4 content showed high CO yield (107 µmol g −1 h −1 ) and excellent selectivity (94%) (Figure 6b). After 10 h reaction, the production of CO reached 896 µmol g −1 (Figure 6c). After 5 runs, the activity still remained (Figure 6d). Experimental and characterization results revealed that the Co 4 unit both facilitated the charge transfer of g-C 3 N 4 and significantly enhanced the surface catalytic oxidative activity. The Au NPs were in-situ deposited during the assembly of NENU-10 and NENU-3 as shown in Figure 7.) was recently achieved [45]. Table 2 illustrated the catalytic performance of as-prepared hybrid in the reductive CO2 transformation to CO and CH4 under visible-light irradiation. Compared to Ti-free Au@NENU-3, Au@NENU-10 presented both higher activity and selectivity in the CO2 photoreduction. Similarly, Au/K7(PTi2W10O40) also had better performance than Au/Na3(PW12O40). Results of control experiments implied the indispensability of Au nanoparticles to the CO reduction activity and precursors was inactive. Combing these patterns, authors deduced the following steps for the overall process: Firstly, Au nanoparticles produced electrons and holes where the holes oxidized H2O to generate two electrons and two protons.  The Au NPs were in-situ deposited during the assembly of NENU-10 and NENU-3 as shown in Figure 7.) was recently achieved [45]. Table 2 illustrated the catalytic performance of as-prepared hybrid in the reductive CO 2 transformation to CO and CH 4 under visible-light irradiation. Compared to Ti-free Au@NENU-3, Au@NENU-10 presented both higher activity and selectivity in the CO 2 photoreduction. Similarly, Au/K 7 (PTi 2 W 10 O 40 ) also had better performance than Au/Na 3 (PW 12 O 40 ). Results of control experiments implied the indispensability of Au nanoparticles to the CO reduction activity and precursors was inactive. Combing these patterns, authors deduced the following steps for the overall process: Firstly, Au nanoparticles produced electrons and holes where the holes oxidized H 2 O to generate two electrons and two protons.  While only six BTC ligands are needed for three Cu2 units along the 〈111〉 direction (d). The enlarged diagram of the dashed red circle (e). The exposed surface of the octahedron shape for Au@NENU-3 ({111} plane where [PW12O40] 3− is sheltered by BTC (f). The exposed surface of the cube shape for NENU-10 ({100} plane where [PTi2W10O40] 7-is exposed wholly) (g). The SEM images for Au@NENU-3 and Au@NENU-10, respectively (h, i). Reprinted with permission from [45]. Copyright © 2018 Wiley-VCH Verlag GmbH and Co. KGaA. Au@NENU

CO2 to HCOOH
HCOOH is one of the promising hydrogen carriers and the important C1 source for organic synthesis. The generation of HCOOH in CO2 photoreduction consumes the same amount of protons and electrons as that for CO formation, but it requires slightly higher reduction potential.
The POM microions are the molecules denoted as {POM} in dilute solutions, they tend to afford suprestructural macroions denoted as {POM}n (POM macroions). Three POM macroions catalysts, While only six BTC ligands are needed for three Cu 2 units along the 111 direction (d). The enlarged diagram of the dashed red circle (e). The exposed surface of the octahedron shape for Au@NENU-3 ({111} plane where [PW 12 O 40 ] 3− is sheltered by BTC (f). The exposed surface of the cube shape for NENU-10 ({100} plane where [PTi 2 W 10 O 40 ] 7− is exposed wholly) (g). The SEM images for Au@NENU-3 and Au@NENU-10, respectively (h, i). Reprinted with permission from [45]. Copyright © 2018 Wiley-VCH Verlag GmbH and Co. KGaA.

CO 2 to HCOOH
HCOOH is one of the promising hydrogen carriers and the important C1 source for organic synthesis. The generation of HCOOH in CO 2 photoreduction consumes the same amount of protons and electrons as that for CO formation, but it requires slightly higher reduction potential.
The  72 ]@RGO hybrid ({Mo 132 } 1064 @RGO; RGO = reduced graphene oxide) featured with peculiar structures, were respectively employed in the photocatalytic CO 2 reduction coupling with water oxidation [46]. The enormous spherical superstructures in these three catalysts in dilute dispersion ({Mo 154 } 1156 , {Mn 6 P 3 W 24 } 931 , {Mo 132 } 1064 @RGO) are respectively made up by wheel-shaped {Mo 154 } rings, bent rod {Mn 6 P 3 W 24 } units and "Keplerate" {Mo 132 } spheres (Figure 8). The numbers of metal oxide cluster units in these superstructures were calculated by following equation: n = (4πR 2 )/(72.006σ 2 ) × 60. (R, representing the hydrodynamic radius of POM vesicles formed in the dispersion, was obtained by dynamic light scattering analysis. σ, representing the diameter of isolated single cluster, was determined by the van der Waals radii of the constituent atoms.) The results showed that although HCOOH is the main product of CO 2 reduction with these three catalysts, a considerable amount of HCHO was also obtained when {Mo 154 } 1156 and {Mo 132 } 1064 @RGO catalysts were applied. In the case of {Mn 6 P 3 W 24 } 931 , CO 2 was exclusively reduced to HCOOH. In terms of formic acid production, {Mo 154 } 1156 and {Mo 132 } 1064 @RGO displayed better performance than {Mn 6 P 3 W 24 } 931 ( {Mo132}1064@RGO displayed better performance than {Mn6P3W24}931 (Table 3). This phenomenon can be rationally explained by that the simultaneous excitation of multitudinous photoactive clusters on the surface of {Mo154} and {Mo132} vesicles under light irradiation. For {Mo132}1064@RGO, good conductivity of RGO may help the facile electrons transfer.  [48]. In the former case, the max TON per mole was 613. Both FT-IR and Raman spectroscopies showed that the structure remained integral after reaction and the activity was kept after ten cycles. In the latter one, {Mo368}, which consisted of a central ball shaped unit {Mo288} and two {Mo40} capping units, showed excellent selectivity for HCOOH (95.73%) with impressive TON (27666) and TOF (4419 h −1 ). Its external quantum efficiency reached 0.6%. It is worth noting that in all above cases external photosensitizer was unessential. Because intervalence charge transfers (IVCT) of Mo V to Mo VI and W V to W VI have appropriate gap between conduction band and valence band, which can give rise to absorbance maxima in the region of visible light.   [48]. In the former case, the max TON per mole was 613. Both FT-IR and Raman spectroscopies showed that the structure remained integral after reaction and the activity was kept after ten cycles. In the latter one, {Mo 368 }, which consisted of a central ball shaped unit {Mo 288 } and two {Mo 40 } capping units, showed excellent selectivity for HCOOH (95.73%) with impressive TON (27666) and TOF (4419 h −1 ). Its external quantum efficiency reached 0.6%. It is worth noting that in all above cases external photosensitizer was unessential. Because intervalence charge transfers (IVCT) of Mo V to Mo VI and W V to W VI have appropriate gap between conduction band and valence band, which can give rise to absorbance maxima in the region of visible light.

CO 2 to CH 4
Compared to the photoreductive conversion of CO 2 to CO or HCOOH, photocatalytic methanation of CO 2 involving an eight-electron transfer process was much more challenging. [PTi 2 W 10 O 40 ] 7− was reported as the first POM-based photocatalyst for reducing CO 2 to CH 4 with CH 3 OH as electron donator [49]. Although [PTi 2 W 10 O 40 ] 7− had poor activity and limited efficiency in photoreduction of CO 2 , this result suggested that POM could be an active catalyst in the CO 2 photoreduction.
Since this time, few works focusing on photocatalytic CO 2 reduction to CH 4 during past decades. Recently, a remarkable achievement in photocatalytic CO 2 reduction to CH 4 [50]. These two water-insoluble POMs showed good structural stability and extended solar spectrum absorption range in aqueous solutions.
Under CO 2 atmosphere with triethanolamine (TEOA) as sacrificial agent and [Ru(bpy) 3 ]Cl as photosensitizer, CH 4 and CO were the main gaseous photoreduction products, only a trace amount of HCOOH was detected in the aqueous phase. The productivity of CH 4 for NENU-605 and NENU-606 reached up to 170 nmol (894.7 nmol g −1 h −1 ) and 402 nmol (1747.8 nmol g −1 h −1 ) respectively. Good CH 4 selectivity of 76.6% (NENU-605) and 85.5% (NENU-606) was achieved ( Table 4, entry 1 and 2). Moreover, H 2 evolution as the side reaction was not detected. Regarding to the higher CH 4 selectivity of NENU-606 than NENU-605, the heterometallic Mn II /Co II ions in NENU-606 might be more favorable to the adsorption and activation of CO 2 than the homometallic Mn II ions in NENU-605. In contrast, Mn[Mo 6 O 12 (OH) 3 (HPO 4 ) 3 (PO 4 )] 2 (NENU-607), the dimer containing only one Mn II atom sandwiched between two P 4 Mo 6 unit and displaying a similar connection mode to NENU-605 and NENU-606 (Figure 9), had much lower activity (Table 4, entry 3). Outcomes from control experiments demonstrated the necessities of both photosensitizer and light irradiation as well as the contribution from solvent effect ( Table 4, entries 4-6). 13 C-isotope labelling verified that CO 2 was the carbon source of CO and CH 4 . Deduced from these patterns, a plausible multi-step mechanism for the photocatalytic CO 2 methanation was proposed ( Figure 10). Initially, the photosensitizer absorbs light to produce photo-excited electrons from its HOMO and then transfers electrons to the P 4 Mo VI 6 unit through the matched LUMO positions (P 4 Mo VI 6 + 6e − → P 4 Mo V 6 ). Simultaneously, the electron holes produced in the valence band of ruthenium complex was consumed by TEOA (TEOA + h + → TEOA + ). Subsequently, strongly reductive P 4

CO2 to CH4
Compared to the photoreductive conversion of CO2 to CO or HCOOH, photocatalytic methanation of CO2 involving an eight-electron transfer process was much more challenging.
[PTi2W10O40] 7-was reported as the first POM-based photocatalyst for reducing CO2 to CH4 with CH3OH as electron donator [49]. Although [PTi2W10O40] 7-had poor activity and limited efficiency in photoreduction of CO2, this result suggested that POM could be an active catalyst in the CO2 photoreduction.
Since  [50]. These two water-insoluble POMs showed good structural stability and extended solar spectrum absorption range in aqueous solutions.
Under CO2 atmosphere with triethanolamine (TEOA) as sacrificial agent and [Ru(bpy)3]Cl as photosensitizer, CH4 and CO were the main gaseous photoreduction products, only a trace amount of HCOOH was detected in the aqueous phase. The productivity of CH4 for NENU-605 and NENU-606 reached up to 170 nmol (894.7 nmol g -1 h -1 ) and 402 nmol (1747.8 nmol g -1 h -1 ) respectively. Good CH4 selectivity of 76.6% (NENU-605) and 85.5% (NENU-606) was achieved ( Table 4, entry 1 and 2). Moreover, H2 evolution as the side reaction was not detected. Regarding to the higher CH4 selectivity of NENU-606 than NENU-605, the heterometallic Mn II /Co II ions in NENU-606 might be more favorable to the adsorption and activation of CO2 than the homometallic Mn II ions in NENU-605. In contrast, Mn[Mo6O12(OH)3(HPO4)3(PO4)]2 (NENU-607), the dimer containing only one Mn II atom sandwiched between two P4Mo6 unit and displaying a similar connection mode to NENU-605 and NENU-606 (Figure 9), had much lower activity (Table 4, entry 3). Outcomes from control experiments demonstrated the necessities of both photosensitizer and light irradiation as well as the contribution from solvent effect ( Table 4, entries 4-6). 13 C-isotope labelling verified that CO2 was the carbon source of CO and CH4. Deduced from these patterns, a plausible multi-step mechanism for the photocatalytic CO2 methanation was proposed ( Figure 10). Initially, the photosensitizer absorbs light to produce photo-excited electrons from its HOMO and then transfers electrons to the P4Mo

Electrocatalytic CO2 Reduction
Electrocatalytic CO2 reduction to produce various C1 and C2 molecules, for instance, CO, HCOOH, HCHO, CH3OH, CH4, oxalic acid, ethylene, and ethanol, is regarded as a promising proctol for the production of liquid fuels or bulk chemicals. The corresponding reactions and standard potentials were list in Table 5 [51].

Electrocatalytic CO 2 Reduction
Electrocatalytic CO 2 reduction to produce various C1 and C2 molecules, for instance, CO, HCOOH, HCHO, CH 3 OH, CH 4 , oxalic acid, ethylene, and ethanol, is regarded as a promising proctol for the production of liquid fuels or bulk chemicals. The corresponding reactions and standard potentials were list in Table 5 [51].

Products
Half  showed the electrocatalytic for CO 2 reduction. However, the final CO 2 reduction product was unstated.
More than decade later, Proust et al. reinvestigated this reaction in more detail by using [α-SiW 11 O 39 Co] 6− in the electro-assisted reduction of CO 2 ( Figure 11) [53]. The [α-SiW 11 O 39 Co] 6− contained a Co II in place of a W VI . The square-pyramidal CoO 5 with a vacant site was generated by losing a coordinated water molecule from CoO 5 (H 2 O) octahedral when the POM was extracted from aqueous to organic media. Except CO and HCHO, neither H 2 nor other CO 2 reduction products were detected. This indicated that the unique selectivity of [α-SiW 11 O 39 Co] 6− POM catalyst in the electroreduction of CO 2 . The turnover of CO 2 to CO reached to 3.7 with the faradic efficiency of 13%. HCHO was the other detectable product. Its amount varied from 2.1 × 10 −7 to 2.2 × 10 −6 mol with the faradic yield varying from 25% (high HCHO content with low electrolysis charge) to 0.8% (low HCHO content with high electrolysis charge).   Electrocatalytic reduction of CO2 with POM was first reported by Kozik et al. [52]. Cyclic voltammograms (CV) of several POMs with or without the presence of CO2 in various nonpolar solvents were compared. For α-[SiW11O39Co] 6-, a large change on its CV was recorded after CO2 was bubbled. Meanwhile for α-[P2W18O62] 6-, almost no change was observed before and after bubbling CO2. The evidence from CV hinted that α-[SiW11O39Co] 6-might be active to CO2 reduction. Further investigations on reduced α-[SiW11O39Co] 6-with CV indeed confirmed that α-[SiW11O39Co] 6-showed the electrocatalytic for CO2 reduction. However, the final CO2 reduction product was unstated.

Products Half-Electrochemical Thermodynamic Reactions V vs. SHE
More than decade later, Proust et al. reinvestigated this reaction in more detail by using [α-SiW11O39Co] 6-in the electro-assisted reduction of CO2 ( Figure 11) [53]. The [α-SiW11O39Co] 6-contained a Co II in place of a W VI . The square-pyramidal CoO5 with a vacant site was generated by losing a coordinated water molecule from CoO5(H2O) octahedral when the POM was extracted from aqueous to organic media. Except CO and HCHO, neither H2 nor other CO2 reduction products were detected. This indicated that the unique selectivity of [α-SiW11O39Co] 6-POM catalyst in the electroreduction of CO2. The turnover of CO2 to CO reached to 3.7 with the faradic efficiency of 13%. HCHO was the other detectable product. Its amount varied from 2.1 × 10 -7 to 2.2 × 10 -6 mol with the faradic yield varying from 25% (high HCHO content with low electrolysis charge) to 0.8% (low HCHO content with high electrolysis charge). Inspired by organometallic CO2 reduction catalyst [Cp*Rh III (bpy)Cl] + (Cp* = pentamethylcyclopentadienyl anion; bpy = 2,2'-bipyridine), [α-H2PW11O39{Rh III Cp*(OH2)}] 3-with analogous coordination structure via grafting a {Cp*Rh III } 2+ fragment on the monovacant [PW11O39] 7-anion was prepared ( Figure 12) and tested as the electrocatalyst for CO2 reduction [54]. Compared to previously  Figure 12) and tested as the electrocatalyst for CO 2 reduction [54]. Compared to previously reported [CoSiW 11 O 39 ] 6− catalyst, its electrochemical behavior in the presence of CO 2 exhibited a clear improvement, strongly suggesting some interaction with the POM derivative despite the presence of a coordinating solvent. But H 2 was still as major product (68% faradic yield) with HCOO − as minor reduction product (4.5% faradic yield). The remarkable capability of storing and donating electrons endows POMs the feature of reversibly transferring multi-electrons. Hence, selecting proper co-catalyst to establish synergistic catalytic systems containing POM catalyst is reasonable to achieve the efficient CO2 reduction. On one side, by taking the advantage of POMs on reversible transfer of multi-electrons or protons, cocatalysts can efficiently overcome the barrier of CO2 activation. On the other side, co-catalysts with good conductivity and adsorbing CO2 capability can facilitate the electroreduction and improve the activity of catalytic system. Poor electrical conductivity and electron-donating capability are the major drawbacks for MOFs being as efficient electrocatalysts. Therefore, the integrated use of POM and MOF would combine the strengths of each other and generate electrocatalysts with excellent activity [55][56][57].
Following the above intention, Lan et al. prepared various Polyoxometalate-Metalloporphyrin Organic Frameworks (PMOFs) by assembling Zn-ε-Keggin cluster ε-PMo  (Table 6) [58]. In the skeleton of these M-PMOFs, the moieties of Zn-ε-Keggin cluster, as electron reservoirs, potently helped the electron transfer to CO2 reduction catalyst M-TCPP and improve its performance of CO2 reduction. The best results were obtained on Co-PMOF with lower onset potential, tafel slope, and electrochemical impedance as well as higher partial CO current density (jCO), electrochemical active surface area (ECSA), faradic efficiency for CO (FECO) and turnover frequency (TOF). Co-PMOF converted CO2 to CO with a superior faradic efficiency of 99%. The value of TOF was elevated to 1656 h −1 at -0.8 V. No evident activity attenuation was observed during the 36 h stability test, which indicated the robustness of Co-PMOF. The remarkable capability of storing and donating electrons endows POMs the feature of reversibly transferring multi-electrons. Hence, selecting proper co-catalyst to establish synergistic catalytic systems containing POM catalyst is reasonable to achieve the efficient CO 2 reduction. On one side, by taking the advantage of POMs on reversible transfer of multi-electrons or protons, co-catalysts can efficiently overcome the barrier of CO 2 activation. On the other side, co-catalysts with good conductivity and adsorbing CO 2 capability can facilitate the electroreduction and improve the activity of catalytic system. Poor electrical conductivity and electron-donating capability are the major drawbacks for MOFs being as efficient electrocatalysts. Therefore, the integrated use of POM and MOF would combine the strengths of each other and generate electrocatalysts with excellent activity [55][56][57].  (Table 6) [58]. In the skeleton of these M-PMOFs, the moieties of Zn-ε-Keggin cluster, as electron reservoirs, potently helped the electron transfer to CO 2 reduction catalyst M-TCPP and improve its performance of CO 2 reduction. The best results were obtained on Co-PMOF with lower onset potential, tafel slope, and electrochemical impedance as well as higher partial CO current density (j CO ), electrochemical active surface area (ECSA), faradic efficiency for CO (FE CO ) and turnover frequency (TOF) . Co-PMOF converted CO 2 to CO with a superior faradic efficiency of 99%. The value of TOF was elevated to 1656 h −1 at −0.8 V. No evident activity attenuation was observed during the 36 h stability test, which indicated the robustness of Co-PMOF.   DFT calculations was employed to explicitly understand the excellent performance of Co-MPOF and synergism between Zn-ε-Keggin cluster and Co-TCPP. Owing to rather high Gibbs free energies of 0.96 eV (ΔG1 in Figure 14a), the formation of adsorbed intermediates * COOH was regarded as the rate-determining step (RDS) for CO2 reduction on Zn-ε-Keggin cluster. On the contrary, the formation of adsorbed intermediates * CO on Co-TCPP was RDS with high Gibbs free energies of 0.53 eV (ΔG2 in Figure 14a). In accordance with expectation, remarkable drops of both ΔG1 and ΔG2, particularly the much lower ΔG1 = 0.34 eV, were revealed. For Co-PMOF, the more favorable CO2 reduction active site is Co on Co-TCPP instead of Zn on POM. In fact, this Zn POM was inactive in CO2 reduction, and the synergistic effect was derived from the intramolecular electron transfer between the electron mediator of POM and Co-TCPP. The effect of different metal center in porphyrin was also computed (Figure 14b). The energy profile of reaction on different metal centers was well consistent with the outcomes of experiments. Based on the experimental results and theoretical calculations, the possible mechanism about reducing CO2 to CO on Co-PMOF was suggested. Firstly, the POM captures and transfers electrons from the electrode to the Co II center. Subsequently, Co II centers was reduced to Co I . Then Co I interacts with CO2 to afford Co II -* COO -, which was converted to Co II -* CO via the proton-coupled electron transfer. Finally, CO is desorbed and released (Figure 14c,d). DFT calculations was employed to explicitly understand the excellent performance of Co-MPOF and synergism between Zn-ε-Keggin cluster and Co-TCPP. Owing to rather high Gibbs free energies of 0.96 eV (∆G 1 in Figure 14a), the formation of adsorbed intermediates * COOH was regarded as the rate-determining step (RDS) for CO 2 reduction on Zn-ε-Keggin cluster. On the contrary, the formation of adsorbed intermediates * CO on Co-TCPP was RDS with high Gibbs free energies of 0.53 eV (∆G 2 in Figure 14a). In accordance with expectation, remarkable drops of both ∆G 1 and ∆G 2 , particularly the much lower ∆G 1 = 0.34 eV, were revealed. For Co-PMOF, the more favorable CO 2 reduction active site is Co on Co-TCPP instead of Zn on POM. In fact, this Zn POM was inactive in CO 2 reduction, and the synergistic effect was derived from the intramolecular electron transfer between the electron mediator of POM and Co-TCPP. The effect of different metal center in porphyrin was also computed (Figure 14b). The energy profile of reaction on different metal centers was well consistent with the outcomes of experiments. Based on the experimental results and theoretical calculations, the possible mechanism about reducing CO 2 to CO on Co-PMOF was suggested. Firstly, the POM captures and transfers electrons from the electrode to the Co II center. Subsequently, Co II centers was reduced to Co I . Then Co I interacts with CO 2 to afford Co II -* COO − , which was converted to Co II -* CO via the proton-coupled electron transfer. Finally, CO is desorbed and released (Figure 14c,d).
The design and application of [α-SiW 12 O 40 ] 4− -modified AgNC@BSA catalyst (AgNC@BSA = silver nanoclusters capped with bovine serum albumin; BSA is both the stabilizer and reducing reagent for the synthesis of AgNC) in the electroreduction of CO 2 to CO also exhibited the same concept of synergism [59]. Similar to the case of Co-PMOF, although [α-SiW 12 O 40 ] 4− had little CO 2 reduction activity, it played the role of electron transfer mediator and assisted the CO 2 reduction through its strong interaction with CO 2 on the surface of AgNC. It had excellent faradic efficiency (>75%) in DMF containing 1% (v/v) H 2 O. The overpotential of [α-SiW 12 O 40 ] 4− -modified AgNC@BSA electrode was about 0.7 V. The onset potential for this POM-decorated AgNC electrocatalyst was about 400 mV more positive than that on bulk Ag. Figure 14. The DFT calculation and proposed reaction mechanism. The free energy diagrams CO2 reduction reaction (CO2RR) for the Zn-ε-Keggin POM, the Co-TCPP metalloporphyrin and the Co-PMOF (a); comparison of the free energy of each elementary reaction (ΔG1, ΔG2, and ΔG3 represent the free energy of * COOH formation, * CO formation, and CO desorption process, respectively) in CO2RR for Co-PMOF, Fe-PMOF, Ni-PMOF, and Zn-PMOF (b); proposed mechanistic scheme for the CO2RR on Co-PMOF (c, d); color code: MoO6 octahedra, bottle green; Zn, light blue ball; Co, magenta ball, N, blue ball; C, grey ball; O, red ball . Reprinted with permission from [58]. Copyright © 2015 Attribution 4.0 International (CC BY 4.0).
The design and application of [α-SiW12O40] 4--modified AgNC@BSA catalyst (AgNC@BSA = silver nanoclusters capped with bovine serum albumin; BSA is both the stabilizer and reducing reagent for the synthesis of AgNC) in the electroreduction of CO2 to CO also exhibited the same concept of synergism [59]. Similar to the case of Co-PMOF, although [α-SiW12O40] 4-had little CO2 reduction activity, it played the role of electron transfer mediator and assisted the CO2 reduction through its strong interaction with CO2 on the surface of AgNC. It had excellent faradic efficiency (>75%) in DMF containing 1% (v/v) H2O. The overpotential of [α-SiW12O40] 4--modified AgNC@BSA electrode was about 0.7 V. The onset potential for this POM-decorated AgNC electrocatalyst was about 400 mV more positive than that on bulk Ag.

Electromicrobial Conversion of CO2
Since the introduction of lithoautotrophic bacterium Ralstonia eutropha H16 as the production host for biological formate conversion as well as the coupling between this biological formate Figure 14. The DFT calculation and proposed reaction mechanism. The free energy diagrams CO 2 reduction reaction (CO 2 RR) for the Zn-ε-Keggin POM, the Co-TCPP metalloporphyrin and the Co-PMOF (a); comparison of the free energy of each elementary reaction (∆G 1 , ∆G 2 , and ∆G 3 represent the free energy of * COOH formation, * CO formation, and CO desorption process, respectively) in CO 2 RR for Co-PMOF, Fe-PMOF, Ni-PMOF, and Zn-PMOF (b); proposed mechanistic scheme for the CO 2 RR on Co-PMOF (c,d); color code: MoO 6 octahedra, bottle green; Zn, light blue ball; Co, magenta ball, N, blue ball; C, grey ball; O, red ball. Reprinted with permission from [58]. Copyright

Electromicrobial Conversion of CO 2
Since the introduction of lithoautotrophic bacterium Ralstonia eutropha H16 as the production host for biological formate conversion as well as the coupling between this biological formate conversion method and the electroreductive conversion of CO 2 to formate on indium cathode was employed to produce fuels for internal combustion engines from CO 2 [63], the integration of electrocatalytic reduction with microbial conversion represents an edge-cutting strategy for the direct but non-photosynthetic CO 2 conversion to important molecules.
Microorganisms, represented by R. eutropha and Clostridium spp, have been well harnessed to transform CO 2 and H 2 to energy-dense liquid fuels [64]. As potent hydrogen-oxidizing bacterium, R. eutropha has exhibited its versatility in synthesize poly[R-(−)-3-hydroxybutyrate], which is one of the ingredients to manufacture biodegradable plastics [65]. On the other aspect, many POMs-based catalysts have presented excellent activity for hydrogen evolution reaction (HER) [12,66]. Searching biocompatible POM-based HER catalysts and utilizing suitable microorganism to catalytically reduce CO 2 by consuming H 2 generated from HER can effectually accomplish the electromicrobial conversion of CO 2 .
Recently, Zhang and co-workers demonstrated the feasibility of such reckoning. The combination of a heterometallic Co/Cu-containing polyoxometalate/carbon cloth (Cu 6 Co 7 /CC; Cu 6 Co 7 = [Cu(en) 2 ] 6 [((PW 9 O 34 )Co 3 (OH)(H 2 O) 2 (O 3 PC(O)(C 3 H 6 NH 3 )PO 3 )) 2 Co] 2− , en = ethanediamine; see the structure of Cu 6 Co 7 POM in Figure 15; CC = carbon cloth) precatalyst with bacterium R. eutropha H16 was developed to achieved the electricity-driven bioconversion of CO 2 to biomass in neutral water [67]. Cu 6 Co 7 /CC hybrid material possessed good biocompatibility in this CO 2 conversion system. It was evidenced that most of the H 2 produced on the Cu 6 Co 7 /CC cathode was consumed by the bacteria for living and growth. In excess of half input electrical energy was transported into biomass. The authors assumed that the electricity would be supplied by a photovoltaic device with an efficiency of 18%, the expected overall solar-to-biomass efficiency could reach 10%, which would be nearly 10 times higher than the natural photosynthesis. This Cu 6 Co 7 /CC-R. eutropha hybrid system exemplified the potential to explore non-photosynthetic but highly efficient CO 2 -fixation methods by using POMs-based catalysts and solar electricity. employed to produce fuels for internal combustion engines from CO2 [63], the integration of electrocatalytic reduction with microbial conversion represents an edge-cutting strategy for the direct but non-photosynthetic CO2 conversion to important molecules.
Microorganisms, represented by R. eutropha and Clostridium spp, have been well harnessed to transform CO2 and H2 to energy-dense liquid fuels [64]. As potent hydrogen-oxidizing bacterium, R. eutropha has exhibited its versatility in synthesize poly[R-(−)-3-hydroxybutyrate], which is one of the ingredients to manufacture biodegradable plastics [65]. On the other aspect, many POMs-based catalysts have presented excellent activity for hydrogen evolution reaction (HER) [12,66]. Searching biocompatible POM-based HER catalysts and utilizing suitable microorganism to catalytically reduce CO2 by consuming H2 generated from HER can effectually accomplish the electromicrobial conversion of CO2.
Recently, Zhang and co-workers demonstrated the feasibility of such reckoning. The combination of a heterometallic Co/Cu-containing polyoxometalate/carbon cloth (Cu6Co7/CC; Cu6Co7 = [Cu(en)2]6[((PW9O34)Co3(OH)(H2O)2(O3PC(O)(C3H6NH3)PO3))2Co] 2-, en = ethanediamine; see the structure of Cu6Co7 POM in Figure 15; CC = carbon cloth) precatalyst with bacterium R. eutropha H16 was developed to achieved the electricity-driven bioconversion of CO2 to biomass in neutral water [67]. Cu6Co7/CC hybrid material possessed good biocompatibility in this CO2 conversion system. It was evidenced that most of the H2 produced on the Cu6Co7/CC cathode was consumed by the bacteria for living and growth. In excess of half input electrical energy was transported into biomass. The authors assumed that the electricity would be supplied by a photovoltaic device with an efficiency of 18%, the expected overall solar-to-biomass efficiency could reach 10%, which would be nearly 10 times higher than the natural photosynthesis. This Cu6Co7/CC-R. eutropha hybrid system exemplified the potential to explore non-photosynthetic but highly efficient CO2-fixation methods by using POMs-based catalysts and solar electricity.

Non-Reductive CO 2 Conversion to Carbonyl-Contained Organic Chemicals
Non-reductive chemical CO 2 conversion provides diverse alternatives to achieve both the synthesis of practical chemicals and CO 2 fixation at the same time [68]. Currently, the industrial manufacture of several chemicals demands CO 2 as starting material [69]. Predominantly, due to the basic properties of POM-based materials, the capture and transformation of CO 2 with these materials has been widely explored. Therefore, the versatility of POMs-based materials in non-reductive chemical CO 2 fixation has been fully presented. Various useful chemicals, including cyclic carbonate [19], dimethyl carbonate [70][71][72][73][74], urea [74], quinazoline-2,4-(1H,3H)-diones [74,75], 2-benzimidazolone [74], 2-oxazolidinones [18], α-methylene cyclic carbonate [74], and methacrylic acid [76,77] were able to be synthesized with POM-based catalyst (Scheme 2). Considering that several reviews have already given comprehensive summaries of representative progress on non-reductive CO 2 fixation by POM catalysts [69,78,79], herein we briefly introduced the recent advances about non-reductive CO 2 conversion to organic chemicals, which were not collected in the very recent review published in 2018 [79].  [69,78,79], herein we briefly introduced the recent advances about nonreductive CO2 conversion to organic chemicals, which were not collected in the very recent review published in 2018 [79].
Various important applications of cyclic carbonates make the research on POM or other material-catalyzed synthesis of cyclic carbonates always receive long-lasting interests [80]. By integrating POM components, chiral organic catalysts and CO2 activator in the skeleton of a single MOF, the polyoxometallate-organocatalyst-metal organic framework (POMOF) was designed by Duan group [81]. Chiral cyclic carbonates were produced efficiently from olefins and CO2 on the POMOFs tandem catalyst. By ingeniously connecting Keggin-type POM anions α-[ZnW12O40] 6-, Lproline-derived asymmetric organocatalysts pyrrolidine-2-yl-imidazole (PYI) and NH2functionalized bridging links 2-amino-4,4'-bipyridine via the six-coordinated Zn II nodes in these POMOFs (α-[ZnW12O40] 6--PYIs), a clear division of catalytic working for oxidation-coupled CO2 conversion was established that POM as an oxidation catalyst, pyrrolidine moiety as a chiral organocatalyst, Zn II ions as Lewis acid catalyst to activate the epoxide intermediate and amino groups on 4,4'-bipyridines as renewable CO2 absorption reagent in this tandem process ( Figure 16). The multi-catalytic sites were orderly distributed and spatially matched in the framework. The captured CO2 molecules are synergistically fixed and activated by well-positioned pyrrolidine and amine groups, providing further compatibility with the terminal W=O activated epoxidation intermediate and driving the tandem catalytic process in a single workup stage and an asymmetric fashion. Various important applications of cyclic carbonates make the research on POM or other materialcatalyzed synthesis of cyclic carbonates always receive long-lasting interests [80]. By integrating POM components, chiral organic catalysts and CO 2 activator in the skeleton of a single MOF, the polyoxometallate-organocatalyst-metal organic framework (POMOF) was designed by Duan group [81]. Chiral cyclic carbonates were produced efficiently from olefins and CO 2 on the POMOFs tandem catalyst. By ingeniously connecting Keggin-type POM anions α-[ZnW 12 O 40 ] 6− , l-proline-derived asymmetric organocatalysts pyrrolidine-2-yl-imidazole (PYI) and NH 2 -functionalized bridging links 2-amino-4,4 -bipyridine via the six-coordinated Zn II nodes in these POMOFs (α-[ZnW 12 O 40 ] 6− -PYIs), a clear division of catalytic working for oxidation-coupled CO 2 conversion was established that POM as an oxidation catalyst, pyrrolidine moiety as a chiral organocatalyst, Zn II ions as Lewis acid catalyst to activate the epoxide intermediate and amino groups on 4,4 -bipyridines as renewable CO 2 absorption reagent in this tandem process ( Figure 16). The multi-catalytic sites were orderly distributed and spatially matched in the framework. The captured CO 2 molecules are synergistically fixed and activated by well-positioned pyrrolidine and amine groups, providing further compatibility with the terminal W=O activated epoxidation intermediate and driving the tandem catalytic process in a single workup stage and an asymmetric fashion.  [69,78,79], herein we briefly introduced the recent advances about nonreductive CO2 conversion to organic chemicals, which were not collected in the very recent review published in 2018 [79].

Outlook
POMs are regarded as discrete metal oxide clusters with much smaller nanometric size than bulk metal oxides, therefore most of the distinctions on chemical properties, especially CO2 reductive conversion-related redox property, between bulk metal oxides and POMs were originated from quantum size effect. Quantum size effect leads to the changes of electronic configurations. These changes arise through systematic transformations in the density of electronic energy levels as a function of the size. By plotting the correlation between redox potential and electron density per volume (Figure 18a) as well as the diagram of band structure (Figure 18b), the quantum size effectresulted distinction on redox property among bulk WO3, colloid WO3 and [SiW12O40] 4-has been clearly illustrated [88]. Despite the prevalence of both POMs and bulk oxides as photocatalysts of harvesting light in recent years, the smaller size endows POMs with wider gap and lower reduction potentials. The case given by Hill and Geletii et al. involving water oxidation catalyst [Co4(H2O)2(PW9O34)2] 10-(Co4POM) also manifested the distinction between POM and bulk metal oxide on redox property. In their report, (PW9O34) 9--sandwiched Co4 unit in Co4POM was unambiguously identified as the dominant active center for water splitting and not CoOx. [89]

Outlook
POMs are regarded as discrete metal oxide clusters with much smaller nanometric size than bulk metal oxides, therefore most of the distinctions on chemical properties, especially CO 2 reductive conversion-related redox property, between bulk metal oxides and POMs were originated from quantum size effect. Quantum size effect leads to the changes of electronic configurations. These changes arise through systematic transformations in the density of electronic energy levels as a function of the size. By plotting the correlation between redox potential and electron density per volume ( Figure 18a) as well as the diagram of band structure (Figure 18b), the quantum size effect-resulted distinction on redox property among bulk WO 3 , colloid WO 3 and [SiW 12 O 40 ] 4− has been clearly illustrated [88]. Despite the prevalence of both POMs and bulk oxides as photocatalysts of harvesting light in recent years, the smaller size endows POMs with wider gap and lower reduction potentials. The case given by Hill   The prospering arising of new POM structures provides vast opportunities to excavate latent catalysts for chemical CO2 conversion. Titanium oxo-clusters [90] and polyoxo-noble-metalates [91] are exemplified as two promising representatives among these new POM structures. Regarding titanium oxo-clusters (also named polyoxotitanates), there are diverse variants including titanium oxoalkoxides clusters TinOm(OR)4n-2m, titanium oxo-carboxo-alkoxides clusters TinO2n-x/2y/2(OR)x(OOCR')y, and titanium oxocarboxo clusters [TinOm(OOCR)p] (2m + p = 4n) [90]. The tunable wide polynuclearity (the number of titanium atoms n varies from 2 to 28) and facile fabrication of hybrid architectures with organic components by covalent bonding are the two major features of these clusters. MOF MIL-125 (MIL stands for Material from Insitut Lavoisier) consisting of 1,4benzenedicarboxylate-connected titanium-oxo-hydroxo clusters has been reported to possess possible photocatalytic property when alcohols are adsorbed inside its framework [92]. Based on this indication and as the extensive applications of titanium oxide in CO2 photocatalytic reduction [93][94][95], there are likely to be extensive catalytic application of hybrid materials containing titanium oxocluster cores in CO2 reductive conversion. The emergance of various novel polyoxo-noble-metalates including polyoxopalladates, -platinates, and -aurates [91] also provides different prospects for catalytic CO2 reductive conversion. On these polyoxo-noble-metalates, the noble metal atoms Pd, Pt, and Au act as "addenda" atoms rather than as heteroatoms [91]. Despite that there are no direct accessible active sites on noble metal centers with saturated coordination environments, the very recent report about the catalytic Suzuki-Miyaura cross-coupling application of polyoxopalladatebased MOF materials has implied that these polyoxo-noble-metalates may acquire catalytic activity via in situ partial reduction [96]. For polyoxoplatinate [Pt12O8(SO4)12] 4- [97], although it is insoluble in any media, its six dumbbell-shaped [Pt2] 6+ anionic units perhaps will attract electrocatalytic research interests, because they may be in-situ partially reduced to molecular mixed-valent metal clusters or platinum black-type "suboxides" [98]. Then these plausible in-situ generated species may be applied as HER catalysts for electromicrobial conversion of CO2.
More than the mode of LMCT, developing POM-based catalysts with the intrinsic function of metal-to-metal charge transfer (MMCT) will also be beneficial to look for more efficient POM-based approach to harvest visible-near infrared emission of sunlight with higher quantum yield. The MMCT transition occurs when two metal centers with different valence states are coupled by bridging ligands (M a+ -O-M' b+ ). The two metal centers are respectively reduced and oxidized (M a+ -O-M' b+ → M (a-1)+ -O-M' (b+1)+ ) during this transition. The MMCT transition often takes place in polynuclear complexes [99]. The oxo-bridged and all-inorganic heterobinuclear units such as Zr IV -O-Co II [100][101][102][103], Zr IV -O-Cu I [104], Ti IV -O-Co II [100], and Ti IV -O-Mn II [105] supported on silica material that display MMCT transitions can extend the optical absorption from UV to visible regions. Some of these materials have been applied as catalyst or cocatalyst for CO2 photoreduction [101][102][103][104]. Several reported POM-based systems, such as the anchored polynuclear charge-transfer complexes consisting of Ce III ions and Cu II -substituted Keggin-type Cu II PW11O39 [106], the metal-oxide nanoclusters consisting of Ce III or Co II ions and Keggin-type PW12O40 3- [107], [Co II W12O40] 6- [108], and The prospering arising of new POM structures provides vast opportunities to excavate latent catalysts for chemical CO 2 conversion. Titanium oxo-clusters [90] and polyoxo-noble-metalates [91] are exemplified as two promising representatives among these new POM structures. Regarding titanium oxo-clusters (also named polyoxotitanates), there are diverse variants including titanium oxoalkoxides clusters Ti n O m (OR) 4n-2m , titanium oxo-carboxo-alkoxides clusters Ti n O 2n-x/2-y/2 (OR) x (OOCR') y , and titanium oxocarboxo clusters [Ti n O m (OOCR) p ] (2m + p = 4n) [90]. The tunable wide polynuclearity (the number of titanium atoms n varies from 2 to 28) and facile fabrication of hybrid architectures with organic components by covalent bonding are the two major features of these clusters. MOF MIL-125 (MIL stands for Material from Insitut Lavoisier) consisting of 1,4-benzenedicarboxylate-connected titanium-oxo-hydroxo clusters has been reported to possess possible photocatalytic property when alcohols are adsorbed inside its framework [92]. Based on this indication and as the extensive applications of titanium oxide in CO 2 photocatalytic reduction [93][94][95], there are likely to be extensive catalytic application of hybrid materials containing titanium oxo-cluster cores in CO 2 reductive conversion. The emergance of various novel polyoxo-noble-metalates including polyoxopalladates, -platinates, and -aurates [91] also provides different prospects for catalytic CO 2 reductive conversion. On these polyoxo-noble-metalates, the noble metal atoms Pd, Pt, and Au act as "addenda" atoms rather than as heteroatoms [91]. Despite that there are no direct accessible active sites on noble metal centers with saturated coordination environments, the very recent report about the catalytic Suzuki-Miyaura cross-coupling application of polyoxopalladate-based MOF materials has implied that these polyoxo-noble-metalates may acquire catalytic activity via in situ partial reduction [96]. For polyoxoplatinate [Pt 12 O 8 (SO 4 ) 12 ] 4− [97], although it is insoluble in any media, its six dumbbell-shaped [Pt 2 ] 6+ anionic units perhaps will attract electrocatalytic research interests, because they may be in-situ partially reduced to molecular mixed-valent metal clusters or platinum black-type "suboxides" [98]. Then these plausible in-situ generated species may be applied as HER catalysts for electromicrobial conversion of CO 2 .
More than the mode of LMCT, developing POM-based catalysts with the intrinsic function of metal-to-metal charge transfer (MMCT) will also be beneficial to look for more efficient POM-based approach to harvest visible-near infrared emission of sunlight with higher quantum yield. The MMCT transition occurs when two metal centers with different valence states are coupled by bridging ligands (M a+ -O-M' b+ ). The two metal centers are respectively reduced and oxidized (M a+ -O-M' b+ → M (a−1)+ -O-M' (b+1)+ ) during this transition. The MMCT transition often takes place in polynuclear complexes [99]. The oxo-bridged and all-inorganic heterobinuclear units such as Zr IV -O-Co II [100][101][102][103], Zr IV -O-Cu I [104], Ti IV -O-Co II [100], and Ti IV -O-Mn II [105] supported on silica material that display MMCT transitions can extend the optical absorption from UV to visible regions. Some of these materials have been applied as catalyst or cocatalyst for CO 2 photoreduction [101][102][103][104]. Several reported POM-based systems, such as the anchored polynuclear charge-transfer complexes consisting of Ce III ions and Cu II -substituted Keggin-type Cu II PW 11 O 39 [106], the metal-oxide nanoclusters consisting of Ce III or Co II ions and Keggin-type PW 12  (M x OH y = V IV O, Cr III (OH 2 ), Mn II (OH 2 ), Fe III (OH 2 ), Co II (OH 2 ), Ni II (OH 2 ), Cu II (OH 2 ), Zn II (OH 2 )) [109] have sufficiently demonstrated the feasibility that such molecular inorganic MMCT (or MPCT, metal-to-polyoxometalate charge transfer) [108,109] transition enable POM-based catalyst to function as efficient visible-light-driven multielectron-transfer catalysts. Regarding the metal-oxide nanoclusters consisting of Ce III or Co II ions and Keggin-type PW 12 O 40 3− , even the straightforward video guide of preparing catalytic material has been given [110]. These will further encourage to utilize MMCT of POM-based catalytic materials in CO 2 reduction. It is noteworthy that prolonging the picosecond-scale lifetimes of photogenerated states during MMCT transition is still an important issue to be tackled [111]. The works from Lan and Hill et al. has validated the effects of heterometal location [108] and different addenda substitution [109] on the lifetime of photo-excited state.
On the other hand, the tools of advanced operando spectral characterization techniques [112], electrochemical measuring methods [113] and theoretical calculations [114], imbibing more understandings and attaining insights on catalytic mechanisms and structure-reactivity relationships of those reported POM-based catalysts will be helpful to design better POM catalysts for CO 2 conversion with explicit destination. On CO 2 photo/electroreduction, the hybridation of POM with various novel materials can provide chances to fabricate catalysts with broader visible-light absorption spectrum region or better selectivity but lower overpotential. In those systems with hybrid electro-or photocaytalytic CO 2 catalysts, once the light-harvesting/electron-storage centers and catalytically active sites are designated in an integrated system, the key issue is the bridging of the two components-charge kinetics. Exploiting time-resolved spectroscopic techniques to characterize charge kinetics is useful to understand these hybrid catalysts and thus helps to improve their catalytic efficiencies [99].
In catalytic oxidation, POMs were looked upon as inorganic analogs of porphyrin [115][116][117]. The thriving development of transition metal porphyrin or phthalocyanine derivatives as heterogeneous molecular catalysts for electrochemical CO 2 reduction [118] makes us have such confidence that this analogy can be still established in electrochemical CO 2 reduction. Therefore, the accumulated understandings about electrochemical CO 2 reduction by transition metal porphyrin or phthalocyanine may help the development of POM-based catalysts for the electroreductive CO 2 conversion. By sunlight or solar electricity, producing fuels and chemicals via CO 2 conversion is a promising and reliable option for largely reducing fossil fuel consumption in the future [119]. Thus POM-catalyzed transformation of CO 2 with photo-or electrochemical methods will retain long-lasting interest of both academic and industrial research.
To the non-reductive CO 2 conversion, searching for POMs with stronger Lewis basicity will benefit the elevation of turnover efficiency and the mitigation of harsh conditions under halogen-free condition. To enhance Lewis basicity, high negative charge density on the terminal oxygen atoms on POMs is required. Calculating and comparing natural bond orbital (NBO) charges of oxygen atoms in POMs has been demonstrated as a useful access to gain more insights into the Lewis basicity on POMs [84,120]. Making full use of both structural traits and diversity of POMs would help to break through some ceilings of practical CO 2 chemical conversion.