Single-Atom Mn Active Site in a Triol-Stabilized β-Anderson Manganohexamolybdate for Enhanced Catalytic Activity towards Adipic Acid Production

Adipic acid is an important raw chemical for the commercial production of polyamides and polyesters. The traditional industrial adipic acid production utilizes nitric acid to oxidize KA oil (mixtures of cyclohexanone and cyclohexanol), leading to the emission of N2O and thus causing ozone depletion, global warming, and acid rain. Herein, we reported an organically functionalized β-isomer of Anderson polyoxometalates (POMs) nanocluster with single-atom Mn, β-{[H3NC(CH2O)3]2MnMo6O18} (1), as a highly active catalyst to selectively catalyze the oxidation of cyclohexanone, cyclohexanol, or KA oil with atom economy use of 30% H2O2 for the eco-friendly synthesis of adipic acid. The catalyst has been characterized by single crystal and powder XRD, XPS, ESI-MS, FT-IR, and NMR. A cyclohexanone (cyclohexanol) conversion of >99.9% with an adipic acid selectivity of ~97.1% (~85.3%) could be achieved over catalyst 1 with high turnover frequency of 2427.5 h−1 (2132.5 h−1). It has been demonstrated that the existence of Mn3+ atom active site in catalyst 1 and the special butterfly-shaped topology of POMs both play vital roles in the enhancement of catalytic activity.


Introduction
Adipic acid (AA) is one of the most important industrial chemicals for the production of polyamides, polyesters, and polyurethane resins [1,2].Additionally, AA is widely used as an approved additive in cosmetics, gelatins, lubricants, fertilizers, adhesives, insecticides, paper, and waxes [3].The global AA production is more than 3.5 million of metric tons with a rapid demand growth of 4-5% annually [4].However, in traditional industrial production, AA is mainly generated by the oxidization of KA oil (i.e., a mixture of cyclohexanol and cyclohexanone) with an excess of 50-60% nitric acid at 60-80 • C catalyzed with copper(II) and ammonium metavanadate.This process will suffer from the corrosion of reaction vessels and inevitable emission of N 2 O (300 kg of N 2 O per ton of adipic acid), leading to global warming and ozone depletion [3,[5][6][7].Various technologies have been implemented to avoid the N 2 O emissions, such as thermally decomposing it into O 2 and N 2 , recovering N 2 O to prepare cyclododecanone via oxidation of cyclododecene, or to help hydroxylation of benzene to phenol [8][9][10].
Taking into account the atom economy and sustainable development as well as the compatibility of the downstream approaches and industrial practice, many efforts have been devoted to developing more efficient and eco-friendly processes for the industrial production of adipic acid that avoid the use of nitric acid [11][12][13][14].Recently, alternative substrates for the green production of AA have been well investigated, including cyclohexane [11,13,[15][16][17], cyclohexene [18][19][20], cyclohexanone, and cyclohexanol [21][22][23][24][25][26][27].In 2014, Hwang and Sagadevan have reported a one-pot protocol to convert cyclohexane to adipic acid at room temperature by using ozone and UV light.However, the reaction is time-consuming (15 h) and the overall adipic acid yield (~75%) still needs to be improved [13].Sato et al. have developed a biphasic system for the oxidation of cyclohexene to AA using 30% H 2 O 2 in the presence of Na 2 WO 4 as a homogeneous catalyst and [CH 3 (n-C 8 H 17 ) 3 N]HSO 4 as a phase-transfer catalyst [14].The overall adipic acid yield (93%) is very high, but the industrial applicability of the phase-transfer catalyst is expensive, and the quaternary ammoniums cation also have a negative environmental impact [28].In 2003, Usui and Sato subsequently discovered that 78% yield of AA could be achieved by oxidizing cyclohexanol (91% yield for cyclohexanone) with 30% H 2 O 2 over H 2 WO 4 at 90 • C for 20 h [27].While some positive outcomes were obtained, the drastic reaction conditions, the demand of phase transfer catalyst, high-energy consumption, and the low yields of AA restricted their industrial applications.Thus, an efficient, eco-friendly, and low energy consumption route to AA from cyclohexanone with solvent free is highly desirable.
Polyoxometalates (POMs) have been well applied in catalysis owing to their structural diversity and fascinating properties, including tunable acidity and redox properties, inherent resistance to oxidative decomposition, high thermal stability, and impressive sensitivity to light and electricity [29][30][31].Recently, Zhong and Yin et al. have synthesized a novel hollow-structured Mn-HTS catalyst to catalyze the oxidative cleavage of cyclohexanone for AA production [22].The Mn species in Mn-HTS are presumably responsible for the production of radical intermediates to increase the reaction rate and contribute to the tautomerism between cyclohexanone and the corresponding enolic form [16,22].In addition, POMs could not only serve as co-catalysts but could stabilize the active metal sites and regulate the electronic structure of catalysts to give rise to the enhancement of catalytic activities [32][33][34][35].Over the years, our group has dedicated substantial efforts on the development of organically functionalized Anderson-type POMs [30,[36][37][38][39][40][41][42][43].Thus, we predicted that the Anderson-type POMs with manganese as the active metal site might catalyze the green oxidation of cyclohexanone, cyclohexanol, and KA oil to AA with high reactivity.Herein, a triol-functionalized butterfly-shaped β isomer of Mn(III)-Anderson POMs nanocluster, β-{[H 3 NC(CH 2 O) 3 ] 2 MnMo 6 O 18 } − (1), has been synthesized and served as a catalyst for the selective oxidation of KA oil to AA with atom economy use of 30% H 2 O 2 under solvent-free and room-temperature (without heating) conditions (Scheme 1).This route will greatly reduce the energy consumption and will provide a new perspective for the design of highly active catalysts for the green AA production.Taking into account the atom economy and sustainable development as well as the compatibility of the downstream approaches and industrial practice, many efforts have been devoted to developing more efficient and eco-friendly processes for the industrial production of adipic acid that avoid the use of nitric acid [11][12][13][14].Recently, alternative substrates for the green production of AA have been well investigated, including cyclohexane [11,13,[15][16][17], cyclohexene [18][19][20], cyclohexanone, and cyclohexanol [21][22][23][24][25][26][27].In 2014, Hwang and Sagadevan have reported a one-pot protocol to convert cyclohexane to adipic acid at room temperature by using ozone and UV light.However, the reaction is time-consuming (15 h) and the overall adipic acid yield (~75%) still needs to be improved [13].Sato et al. have developed a biphasic system for the oxidation of cyclohexene to AA using 30% H2O2 in the presence of Na2WO4 as a homogeneous catalyst and [CH3(n-C8H17)3N]HSO4 as a phase-transfer catalyst [14].The overall adipic acid yield (93%) is very high, but the industrial applicability of the phase-transfer catalyst is expensive, and the quaternary ammoniums cation also have a negative environmental impact [28].In 2003, Usui and Sato subsequently discovered that 78% yield of AA could be achieved by oxidizing cyclohexanol (91% yield for cyclohexanone) with 30% H2O2 over H2WO4 at 90 °C for 20 h [27].While some positive outcomes were obtained, the drastic reaction conditions, the demand of phase transfer catalyst, high-energy consumption, and the low yields of AA restricted their industrial applications.Thus, an efficient, eco-friendly, and low energy consumption route to AA from cyclohexanone with solvent free is highly desirable.
Polyoxometalates (POMs) have been well applied in catalysis owing to their structural diversity and fascinating properties, including tunable acidity and redox properties, inherent resistance to oxidative decomposition, high thermal stability, and impressive sensitivity to light and electricity [29][30][31].Recently, Zhong and Yin et al. have synthesized a novel hollow-structured Mn-HTS catalyst to catalyze the oxidative cleavage of cyclohexanone for AA production [22].The Mn species in Mn-HTS are presumably responsible for the production of radical intermediates to increase the reaction rate and contribute to the tautomerism between cyclohexanone and the corresponding enolic form [16,22].In addition, POMs could not only serve as co-catalysts but could stabilize the active metal sites and regulate the electronic structure of catalysts to give rise to the enhancement of catalytic activities [32][33][34][35].Over the years, our group has dedicated substantial efforts on the development of organically functionalized Anderson-type POMs [30,[36][37][38][39][40][41][42][43].Thus, we predicted that the Anderson-type POMs with manganese as the active metal site might catalyze the green oxidation of cyclohexanone, cyclohexanol, and KA oil to AA with high reactivity.Herein, a triol-functionalized butterfly-shaped β isomer of Mn(III)-Anderson POMs nanocluster, β-{[H3NC(CH2O)3]2MnMo6O18} − (1), has been synthesized and served as a catalyst for the selective oxidation of KA oil to AA with atom economy use of 30% H2O2 under solvent-free and room-temperature (without heating) conditions (Scheme 1).This route will greatly reduce the energy consumption and will provide a new perspective for the design of highly active catalysts for the green AA production.S1) and possesses a butterfly-shaped β-Anderson POM nanocluster (ca.0.88 nm) with Mn 3+ metal ion as the central heteroatom, and stabilized by two tris ligands (Figure 1a).Selected bond lengths of catalyst 1 are listed in Table S2.The anion nanocluster 1 and counter cations, NH 4 + , pack together and further assemble into a porous supermolecule structure with 4.47 Å × 11.68 Å nanopores via intermolecular hydrogen bonding and electrostatic interaction (Figure 1b,c, Table S3).

The Structure Characterizations of Catalyst 1
The POMs-based catalyst 1 was synthesized by a reaction of flat Anderson-Evans POMs cluster, α-[Mn(OH)6Mo6O18] 3− , and triol ligands in hot (DMF) under N2 atmosphere (details see experiment part in SI).Single crystal XRD analysis revealed that catalyst 1, [NH4] β-{[H3NC(CH2O)3]2MnMo6O18}, crystallizes in the monoclinic space group C2/c (Table S1) and possesses a butterfly-shaped β-Anderson POM nanocluster (ca.0.88 nm) with Mn 3+ metal ion as the central heteroatom, and stabilized by two tris ligands (Figure 1a).Selected bond lengths of catalyst 1 are listed in Table S2.The anion nanocluster 1 and counter cations, NH4 + , pack together and further assemble into a porous supermolecule structure with 4.47 Å × 11.68 Å nanopores via intermolecular hydrogen bonding and electrostatic interaction (Figure 1b,c, Table S3).Powder XRD, IR, UV-Vis, ESI-MS, and 13 C NMR analyses were also conducted to characterize catalyst 1 (Figures 2 and S1-S3).Firstly, the phase purity of catalyst 1 was characterized by powder X-ray diffraction and IR spectroscopy.The powder XRD pattern of catalyst 1 was almost identical to the simulated powder XRD pattern from the single crystal XRD result, confirming the phase purity as well as the excellent crystallinity of catalyst 1 (Figure 2a).As shown in Figure 2b, the characteristic peaks at 937, 919, and 897 cm −1 could be assigned to the vibrations of terminal Mo=O units and those at 790, 737, and 661 cm −1 belonged to the vibrations of the Mo-O-Mo groups.The characteristic peaks at 1117, 1054, and 1027 cm −1 could be assigned to the vibration peaks of the C-O bonds which demonstrated the grafting of tris onto the surface of β isomeric Anderson cluster.Notably, the splitting of three C-O bonds could be attributed to the fact that there exist three types of C-O bonds: one type of (μ3-O)-C bonds and two types of (μ2-O)-C.
The X-ray photoelectron spectroscopy (XPS) in Figure 2c,d showed that catalyst 1 possessed binding energy (BE) around 232 eV and 235 eV belonging to Mo(VI) 3d3/2 and Mo(VI) 3d5/2 of Mo 6+ species, respectively.Moreover, the binding energies in the intervals around 641 eV can be attributed to Mn 2p3/2, indicating the presence of Mn 3+ ions on the surface of catalyst 1 [22].Since many Mn species have been demonstrated as excellent catalysts for the synthesis of AA [16,22,23,25], such a butterfly-shaped Mn(III)-Anderson POM nanocluster may potentially allow a combination of the function of Mn and POM catalysts to achieve a synergistic catalysis for the AA production.Powder XRD, IR, UV-Vis, ESI-MS, and 13 C NMR analyses were also conducted to characterize catalyst 1 (Figure 2 and Figures S1-S3).Firstly, the phase purity of catalyst 1 was characterized by powder X-ray diffraction and IR spectroscopy.The powder XRD pattern of catalyst 1 was almost identical to the simulated powder XRD pattern from the single crystal XRD result, confirming the phase purity as well as the excellent crystallinity of catalyst 1 (Figure 2a).As shown in Figure 2b, the characteristic peaks at 937, 919, and 897 cm −1 could be assigned to the vibrations of terminal Mo=O units and those at 790, 737, and 661 cm −1 belonged to the vibrations of the Mo-O-Mo groups.
The characteristic peaks at 1117, 1054, and 1027 cm −1 could be assigned to the vibration peaks of the C-O bonds which demonstrated the grafting of tris onto the surface of β isomeric Anderson cluster.Notably, the splitting of three C-O bonds could be attributed to the fact that there exist three types of C-O bonds: one type of (µ 3 -O)-C bonds and two types of (µ 2 -O)-C.

Cyclohexanone Oxidation Reaction
In our initial experiments, direct oxidation of cyclohexanone to AA was performed as a model reaction to screen the reaction conditions, and the results are listed in Table 1.As expected, the reaction did not occur without any catalysts (entry 1 in Table 1).The catalyst 1 showed cyclohexanone conversion of >99.9% and AA selectivity of 97.1% in the presence of 0.02 mol % catalyst 1 after 2 h (entry 2 in Table 1).Moreover, the as-prepared AA could be isolated as white crystalline products from the solution after cooling at −5 °C for 12 h, whose melting point is ca.152.4 °C.The reaction solution was further analysed by ESI-MS, indicating that AA was the major product with 100% intensity at m/z = 145.04,and only a few valeric acid as by-product was formed (Figure S4).Then, 1 H and 13 C NMR spectra (in [D6] DMSO) were applied to characterize the purity of the as-prepared AA crystal products (Figure S5).Control experiments showed that altering the catalyst loading had little influence on the product yield (entries 2-6 in Table 1).Regarding the oxidant, in this cyclohexanone to AA oxidization reaction, 1 mol AA generation will consume about 3.3 mol H2O2, which is much less compared with the previous protocol developed by Noyori, in which 1 mol AA generation per 4.5 mol H2O2 consumption [13].Upon decreasing the amount of H2O2, the yield of AA dropped significantly (entry 7 in Table 1).
With respect to active centers, detailed comparisons (Table 1, entries 9 and 10) with other catalysts such as Na2MoO4, (NH4)6Mo7O24 leave no doubt that the active sites in our highly active catalyst 1 is the special central Mn 3+ species.Furthermore, only few AA product could be obtained using simple manganese salts as catalyst, Mn(CH3COO)2 (entry 11 in Table 1).The catalytic activity of common flat Anderson-type POMs with Mn(III) as central heteroatom is higher than that of (NH4)6Mo7O24 and Na2MoO4, indicating that Mn 3+ species are more active than Mo 6+ species (entry 12 in Table 1).Furthermore, the catalytic property of catalyst 1 was better than that of common flat The X-ray photoelectron spectroscopy (XPS) in Figure 2c,d showed that catalyst 1 possessed binding energy (BE) around 232 eV and 235 eV belonging to Mo(VI) 3d3/2 and Mo(VI) 3d5/2 of Mo 6+ species, respectively.Moreover, the binding energies in the intervals around 641 eV can be attributed to Mn 2p3/2, indicating the presence of Mn 3+ ions on the surface of catalyst 1 [22].Since many Mn species have been demonstrated as excellent catalysts for the synthesis of AA [16,22,23,25], such a butterfly-shaped Mn(III)-Anderson POM nanocluster may potentially allow a combination of the function of Mn and POM catalysts to achieve a synergistic catalysis for the AA production.

Cyclohexanone Oxidation Reaction
In our initial experiments, direct oxidation of cyclohexanone to AA was performed as a model reaction to screen the reaction conditions, and the results are listed in Table 1.As expected, the reaction did not occur without any catalysts (entry 1 in Table 1).The catalyst 1 showed cyclohexanone conversion of >99.9% and AA selectivity of 97.1% in the presence of 0.02 mol % catalyst 1 after 2 h (entry 2 in Table 1).Moreover, the as-prepared AA could be isolated as white crystalline products from the solution after cooling at −5 • C for 12 h, whose melting point is ca.152.4 • C. The reaction solution was further analysed by ESI-MS, indicating that AA was the major product with 100% intensity at m/z = 145.04,and only a few valeric acid as by-product was formed (Figure S4).Then, 1 H and 13 C NMR spectra (in [D 6 ] DMSO) were applied to characterize the purity of the as-prepared AA crystal products (Figure S5).Control experiments showed that altering the catalyst loading had little influence on the product yield (entries 2-6 in Table 1).Regarding the oxidant, in this cyclohexanone to AA oxidization reaction, 1 mol AA generation will consume about 3.3 mol H 2 O 2 , which is much less compared with the previous protocol developed by Noyori, in which 1 mol AA generation per 4.5 mol H 2 O 2 consumption [13].Upon decreasing the amount of H 2 O 2 , the yield of AA dropped significantly (entry 7 in Table 1).catalyst 1, since the butterfly-shaped topology of β-{[H3NC(CH2O)3]2MnMo6O18} makes the Mn 3+ central atom more "uncovered" as the active site by the comparison with the flat topology of α-[Mn(OH)6Mo6O18] (Figure S6).For the organoimidization functionalized POM derivatives, the aromatic segment was directly linked on the POM cluster through the Mo≡N bond, thus the conjugated electron of organic ligands obviously enriched the POM nanocluster through delocalization [30].However, for the alkoxylation modification of the Anderson nanocluster, organic ligand electronic-driven effect was not obvious and can be ignored, as only slight electronic depletion that cause hypsochromic shift was observed, as we investigated before [38].Hence, the triol ligand just plays an important role in stabilizing β-{[H3NC(CH2O)3]2MnMo6O18} as the specific butterfly-shaped topology according to the DFT calculations we reported before [37].With respect to active centers, detailed comparisons (Table 1, entries 9 and 10) with other catalysts such as Na 2 MoO 4 , (NH 4 ) 6 Mo 7 O 24 leave no doubt that the active sites in our highly active catalyst 1 is the special central Mn 3+ species.Furthermore, only few AA product could be obtained using simple manganese salts as catalyst, Mn(CH 3 COO) 2 (entry 11 in Table 1).The catalytic activity of common flat Anderson-type POMs with Mn(III) as central heteroatom is higher than that of (NH 4 ) 6 Mo 7 O 24 and Na 2 MoO 4 , indicating that Mn 3+ species are more active than Mo 6+ species (entry 12 in Table 1).Furthermore, the catalytic property of catalyst 1 was better than that of common flat Anderson-type POMs with Mn(III) as the central heteroatom (entries 2 and 12 in Table 1).Based on the above results, it can be inferred that both the special butterfly-shaped topology and the chemical environment of Mn 3+ central atom in catalyst 1 are key factors to the high performance of catalytic oxidation of cyclohexanone to AA.Moreover, analyses of the H 2 O 2 degradation over these different catalysts revealed that simple manganese salts, Mn(CH 3 COO) 2 , could result in the rapid decomposition of H 2 O 2 , and thus lead to low catalytic activity, while for catalyst 1, Na 2 MoO 4 , (NH 4 ) 6 Mo 7 O 24 , and [NH 4 ] 3 •α-[Mn(OH) 6 Mo 6 O 18 ] catalysts, the H 2 O 2 degradation rate could be ignored.It should be noted that in such a green AA production, both catalyst 1 and H 2 O 2 were essential to the high performance of oxidative catalytic activity.Of note, the valance state of reference catalysts including Mn(CH 3 COO) 2 , Na 2 MoO 4 , (NH 4 ) 6 Mo 7 O 24 , and [NH4] 3 •α-[Mn(OH) 6 Mo 6 O 18 ] were clear.The valance state of Mo in these reference catalysts was also Mo 6+ and the valance state of Mn in [NH4] 3 •α-[Mn(OH) 6 Mo 6 O 18 ]  was also Mn 3+ .It is likely that such high catalytic oxidation performance may be attributed to the special butterfly-shaped topology of β-{[H 3 NC(CH 2 O) 3 ] 2 MnMo 6 O 18 }, leading to the specific chemical environment of Mn 3+ central atom in catalyst 1, since the butterfly-shaped topology of β-{[H 3 NC(CH 2 O) 3 ] 2 MnMo 6 O 18 } makes the Mn 3+ central atom more "uncovered" as the active site by the comparison with the flat topology of α-[Mn(OH) 6 Mo 6 O 18 ] (Figure S6).For the organoimidization functionalized POM derivatives, the aromatic segment was directly linked on the POM cluster through the Mo≡N bond, thus the conjugated electron of organic ligands obviously enriched the POM nanocluster through delocalization [30].However, for the alkoxylation modification of the Anderson nanocluster, organic ligand electronic-driven effect was not obvious and can be ignored, as only slight electronic depletion that cause hypsochromic shift was observed, as we investigated before [38].Hence, the triol ligand just plays an important role in stabilizing β-{[H 3 NC(CH 2 O) 3 ] 2 MnMo 6 O 18 } as the specific butterfly-shaped topology according to the DFT calculations we reported before [37].
In order to further understand the relationship between the reaction selectivity, conversion, and reaction conditions in such catalytic oxidation, we investigated the effect of reaction parameters.The reaction selectivity and conversion versus reaction temperature, time, and catalyst amounts were further investigated to optimize the reaction condition.The catalyst shows a low activity with 100% AA selectivity at 5 • C, but with increasing temperature the conversion of cyclohexanone increased.At 25 • C (room temperature), we found that the formation of AA was optimum with both desirable selectivity and conversion.However, with continuous increase of temperature to 100 • C, the conversion of cyclohexanone slightly increased but the AA selectivity significantly decreased.This may be due to the fact that the butterfly-shaped Mn(III)-polyoxometalate nanocluster was a high reactivity catalysis and the H 2 O 2 was also a strong oxidant, leading to the continuous and over-oxidization of AA towards other products under the elevated temperatures.The catalyst shows a low cyclohexanone conversion with 100% AA selectivity at the very beginning of the reaction.With the increasing time, the conversion of cyclohexanone increased but adipic acid selectivity decreased slightly.At about 1.7 h, we found that the formation of AA was optimum with both desirable selectivity and conversion of 97.8%.However, after 2 h, the selectivity significantly decreased and the prolongation of reaction time will cause the over-oxidization and the decomposition of AA. (Figure S7).In addition, the optimum reaction temperature and time were determined to be 2 h at 25 • C. Based on the optimum reaction time, all the Turnover Frequency(TOF) in the entry of Tables 1 and 2, which indicated the intrinsic catalytic activities of these catalysts was further determined in the beginning of the reaction time (0.1 h) with low conversion under 10%.AA selectivity at 5 °C, but with increasing temperature the conversion of cyclohexanone increased.At 25 °C (room temperature), we found that the formation of AA was optimum with both desirable selectivity and conversion.However, with continuous increase of temperature to 100 °C, the conversion of cyclohexanone slightly increased but the AA selectivity significantly decreased.This may be due to the fact that the butterfly-shaped Mn(III)-polyoxometalate nanocluster was a high reactivity catalysis and the H2O2 was also a strong oxidant, leading to the continuous and over-oxidization of AA towards other products under the elevated temperatures.The catalyst shows a low cyclohexanone conversion with 100% AA selectivity at the very beginning of the reaction.
With the increasing time, the conversion of cyclohexanone increased but adipic acid selectivity decreased slightly.At about 1.7 h, we found that the formation of AA was optimum with both desirable selectivity and conversion of 97.8%.However, after 2 h, the selectivity significantly decreased and the prolongation of reaction time will cause the over-oxidization and the decomposition of AA. (Figure S7).In addition, the optimum reaction temperature and time were determined to be 2 h at 25 °C.Based on the optimum reaction time, all the Turnover Frequency(TOF) in the entry of Tables 1 and 2, which indicated the intrinsic catalytic activities of these catalysts was further determined in the beginning of the reaction time (0.1 h) with low conversion under 10%.(23 mmol), in an unsealed reactor with reflux condensing tube and magnetic stirring unless otherwise noted. 2 Cyclohexanol conversion based on cyclohexanol consumed. 3Product selectivity = content of this product/the cyclohexanol consumed; AA: adipic acid; COA: 6-(cyclohexyloxy)-6-oxohexanoic acid; CP: cyclohexanone peroxide; others: probably CO2, cyclohexanone or 6-oxohexanoic acid et al. 4Reaction was carried out without H2O2.5 30% H2O2 (50 mmol) was used instead. 6Reaction was carried out without DMSO. 7Reaction was carried out with 5 mmol DMSO. 8Reaction time was 0.5 h.Yield of cyclohexanone is 37.8%, determined by gas chromatography using chlorobenzene as the internal standard.

Cyclohexanol Oxidation Reaction
From the viewpoint that some practical AA production starts from cyclohexanol, herein, we have also studied the oxidization of cyclohexanol towards AA in the presence of catalyst 1 (Table 2).The reaction was conducted as the following: catalyst 1 (5.4 mg, 4.6 × 10 −6 mol), DMSO (1 mmol), 30% H2O2 (100 mmol), and cyclohexanol (23 mmol) were mixed and stirred in unsealed reactor with

Cyclohexanol Oxidation Reaction
From the viewpoint that some practical AA production starts from cyclohexanol, herein, we have also studied the oxidization of cyclohexanol towards AA in the presence of catalyst 1 (Table 2).The reaction was conducted as the following: catalyst 1 (5.4 mg, 4.6 × 10 −6 mol), DMSO (1 mmol), 30% H 2 O 2 (100 mmol), and cyclohexanol (23 mmol) were mixed and stirred in unsealed reactor with reflux condensing tube without extra heating.After 2 h, cyclohexanol conversion of >99.9% and AA selectivity of 85.3% could be obtained.However, the ESI-MS result shows that few byproduct species contained valeric acid (m/z = 101.06),6-(cyclohexyloxy)-6-oxohexanoic acid (m/z = 229.14)(Figure S8) were observed.Experiments showed that the catalyst loading had little influence on the product yield, but if without catalyst or H 2 O 2 , AA products could not be obtained (Table 2, entries 1-7 in Table 2).The yield of AA dropped significantly when the amount of H 2 O 2 decreased (entry 8 in Table 2).As shown in entry 8 of Table 1, DMSO almost has no influence on the cyclohexanone oxidation reaction towards AA.However, in the case of cyclohexanol oxidation, only low yield of AA (13.7%) could be obtained if without DMSO (entry 9 in Table 2), while an excess addition of DMSO results in the reduction of AA yield (40.0% yield, entry 10 in Table 2) since DMSO could also react with H 2 O 2 to form methyl sulfone, leading to an additional consumption of H 2 O 2 , where the remaining H 2 O 2 cannot make cyclohexanol completely converted into AA, and thus an incomplete oxidation byproduct 6-oxohexanoic acid instead (Figure S9).Furthermore, we found that 37.8% yield of cyclohexanone could be obtained after 0.5 h, indicating that cyclohexanol was firstly oxidized into cyclohexanone.From entry 9 and entry 11 in Table 2, we may consider that DMSO is the co-catalyst for the oxidation of cyclohexanol into cyclohexanone.The generated cyclohexanone could be easily converted into AA product selectively in the presence of catalyst 1 and H 2 O 2 (Table 1).
Compared with Tables 1 and 2, the decrease of AA yield from catalytic oxidation of cyclohexanol could be observed, which can be explained by the fact that the rate of alcohol oxidation to AA is lower than that of ketone.[23] With respect to active centers, detailed comparisons with other catalysts including Na 2 MoO 4 , (NH 4 ) 6 Mo 7 O 24 , and [NH 4 ] 3 •α-[Mn(OH) 6 Mo 6 O 18 ] leave no doubt that the active site is Mn 3+ center in the special butterfly-shaped topology of catalyst 1 (entries 9-12 in Table 1, entries 12-15 in Table 2).

KA Oil Oxidation Reaction and Catalyst Recyclability
Since compound 1 has been demonstrated as a highly active catalyst for the selective oxidation of both cyclohexanone and cyclohexanol towards AA production (Tables 1 and 2), we assumed that the catalytic oxidation of KA oil (a mixture of cyclohexanone and cyclohexanol) towards AA could be achieved in the presence of catalyst 1.As shown in Table 3, the influence of cyclohexanone (-one) and cyclohexanol (-ol) mixtures with various ratios on the AA yield has also been studied.In the absence of cyclohexanone, the AA yield is 84%.The increase of the percentage of cyclohexanone from 20 to 100% in the reaction mixture led to an increase of AA yield from ca. 88 to 98%, which is in accordance with the above results (Tables 1 and 2).According to the 2:1 ratio of cyclohexanone and cyclohexanol in the current industrial KA oil, a reaction catalyzed by compound 1 was conducted to simulate current industrial AA production line from KA oil as follows: catalyst 1 (7.0 mg, 6 × 10 −3 mmol), DMSO (1 mmol), 30% H 2 O 2 (100 mmol), cyclohexanol (10 mmol), and cyclohexanone (20 mmol) were mixed and stirred at room temperature.After 2 h, pure AA product could be obtained with 93% yield.The above results demonstrated that the selectively catalytic oxidation of KA oil could be achieved by catalyst 1.The recycle ability of catalyst 1 was further tested by reusing the reaction solution, which was rotary evaporated and concentrated to 2 mL at low temperature of 50 • C. Of note, suitable sacrificial reductant Na 2 SO 3 was added to eliminate the possible residual H 2 O 2 before the evaporation operation, though according to the almost atom economy use of H 2 O 2 catalytic reaction protocol, the residual H 2 O 2 would not be very high.With this safety precaution, the safety of the whole catalyst cycling process would be guaranteed (Figure 3).Fresh 30% H 2 O 2 (120 mmol) and substrates (20 mmol cyclohexanone and 10 mmol cyclohexanol) were added into the reaction each recycle run.AA products can still be obtained with good yield and high selectivity.However, an appreciable loss of catalytic ability of catalyst 1 (approaching to 88% AA yield) was observed after three runs, indicating fewer catalyst deactivation during catalysis.A total Turnover Number (TON) of 20,900 could be achieved over catalyst 1 after five runs.Moreover, a high TOF of 2325 h −1 could also be obtained in the first run with catalyst 1.
AA products can still be obtained with good yield and high selectivity.However, an appreciable loss of catalytic ability of catalyst 1 (approaching to 88% AA yield) was observed after three runs, indicating fewer catalyst deactivation during catalysis.A total Turnover Number (TON) of 20,900 could be achieved over catalyst 1 after five runs.Moreover, a high TOF of 2325 h −1 could also be obtained in the first run with catalyst 1.

General Methods and Materials
All syntheses and manipulations were performed in the open air.The [MnMo6O18(OH)6] 3− was synthesized according to literature methods [44].All other chemicals including solvents were commercially available as reagent grade and used as received without further purification from Adamas-beta ® (Shanghai, China).IR spectrum was measured using KBr pellets and recorded on a Perkin Elmer FT-IR spectrometer.UV-Vis spectrum was measured in acetonitrile solution by Agilent Cary 300 spectrophotometer (Agilent Technologies Inc., San Francisco, CA, USA).The mass spectrum was obtained using an ion trap mass spectrometer (Thermofisher LTQ, Waltham, MA, USA).Negative mode was chosen for the experiments (capillary voltage 33 V).Sample solution (in acetonitrile) was infused into the ESI source at a flow rate of 300 μL min −1 . 1 H and 13 C NMR spectra were obtained on a JEOL JNM-ECA400 spectrometer and are reported in ppm (JEOL Ltd., Tokyo, Japan).Elemental analyses of C, H, and N were performed by Elementar Analysensysteme GmbH (Elementar Analysensysteme GmbH, Langenselbold, Germany) while the Elemental analyses of Mn and Mo were performed by X-ray fluorescence (XRF) element analyzer PANalytical Epsilon 5 (PANalytical B.V., Almelo, The Netherlands).Thermal gravimetric analysis (TGA) measurements were performed on a Mettler Toledo TGA/SDTA851 (Mettler-Toledo group, Zurich, Switzerland) in flowing Ar 50.0 mL/min with a heating rate of 20 K/min.XPS measurements were performed under ultrahigh vacuum (UHV) with 1.0 × 10 −7 Torr, axis HS monochromatized Al Kα cathode source at 150 W, focused X-ray 100 μm beam, pass energy: 55 eV wih 0.1 ev step length, detect angle (take off): 45°

General Methods and Materials
All syntheses and manipulations were performed in the open air.The [MnMo 6 O 18 (OH) 6 ] 3− was synthesized according to literature methods [44].All other chemicals including solvents were commercially available as reagent grade and used as received without further purification from Adamas-beta ® (Shanghai, China).IR spectrum was measured using KBr pellets and recorded on a Perkin Elmer FT-IR spectrometer.UV-Vis spectrum was measured in acetonitrile solution by Agilent Cary 300 spectrophotometer (Agilent Technologies Inc., San Francisco, CA, USA).The mass spectrum was obtained using an ion trap mass spectrometer (Thermofisher LTQ, Waltham, MA, USA).Negative mode was chosen for the experiments (capillary voltage 33 V).Sample solution (in acetonitrile) was infused into the ESI source at a flow rate of 300 µL min −1 . 1 H and 13 C NMR spectra were obtained on a JEOL JNM-ECA400 spectrometer and are reported in ppm (JEOL Ltd., Tokyo, Japan).Elemental analyses of C, H, and N were performed by Elementar Analysensysteme GmbH (Elementar Analysensysteme GmbH, Langenselbold, Germany) while the Elemental analyses of Mn and Mo were performed by X-ray fluorescence (XRF) element analyzer PANalytical Epsilon 5 (PANalytical B.V., Almelo, The Netherlands).Thermal gravimetric analysis (TGA) measurements were performed on a Mettler Toledo TGA/SDTA851 (Mettler-Toledo group, Zurich, Switzerland) in flowing Ar 50.0 mL/min with a heating rate of 20 K/min.XPS measurements were performed under ultrahigh vacuum (UHV) with 1.0 × 10 −7 Torr, axis HS monochromatized Al Kα cathode source at 150 W, focused X-ray 100 µm beam, pass energy: 55 eV wih 0.1 ev step length, detect angle (take off): 45 • on X-ray microprobe (ULVAC-PHI Quantera SXM, Ulvac-Phi Ltd., Chigasaki, Kanagawa, Japan).Binding energy was calibrated with C1s = 284.8eV.The powder product was measured by PANalytical X'Pert (PANalytical B.V., Almelo, The Netherlands).Powder X-ray powder diffractometer operated at a voltage of 60 kV and current of 55 mA with CuKα radiation (λ = 1.5406Å) (PANalytical B.V., Almelo, The Netherlands).Gas chromatography (GC) (Shimadzu Ltd., Kyoto, Japan) was performed on Shimadzu GC-2010 Plus equipped with an RTX-5 (30 m × 0.25 mm × 0.25 µm) column and FID detector for liquid phase analysis (GC conditions: injector temperature: 240 • C; detector temperature: 250 • C, oven temperature: 50-180 • C with heating rate of 10 • C min −1 ; carrier: helium; column flow: 4 mL min −1 ; linear velocity: 37 cm s −1 ) or a Carboxen ® -1010 PLOT (15 m × 0.25 mm × 0.25 µm) Capillary GC Column and BID detector (Shimadzu Ltd., Kyoto, Japan) for gas phase analysis (GC conditions: injector temperature: 230 • C; detector temperature: 235 • C, oven temperature: 26 • C; carrier: helium; column flow: 5 mL min −1 ; linear velocity: 77 cm s −1 ).High performance liquid chromatograph (HPLC) (Agilent Technologies Inc., San Francisco, CA, USA.) was performed on an Agilent 1290 series HPLC instrument with a HYPERSIL BDS inverse phase C 18 column (250 mm × 4.6 mm × 5 µm), operating at room temperature.The solid mixtures were separated and dried at 70 • C in vacuum for 24 h and finally dissolved in methanol and identified by HPLC-MS with 15:75:10 (v/v) of CH 3 OH:H 2 O:KH 2 PO 4 as mobile phase, detected at 210 nm of wavelength and 1.0 mL/min of flow, and quantified by HPLC external standard calibration curve method.

The Synthesis of Compound 1
A mixture of [TBA] 3 [Mn(OH) 6 Mo 6 O 18 ] (1.748 g 1 mmol) with tris(hydroxymethyl)-aminomethane (0.242 g, 2 mmol) was dissolved in 25 mL hot DMF.After being heated for 12 h under nitrogen gas, the reaction solution was cooled down to room temperature to remove the precipitates by filtration and a dark orange solution was obtained.Then the filtrate was poured into ether, resulting in precipitation.After the solution became clear, the supernatant liquid was poured off.The product was obtained as dark orange powders (49% yield based on Mo).

Figure 2 .
Figure 2. (a) XRD patterns (inset: digital photos, crystal and powder) of catalyst 1 (black, simulated XRD pattern of single crystal XRD of catalyst 1; blue, powder XRD pattern of compound 1); (b) The IR spectrum of catalyst 1; (c,d) The XPS spectra of (c) Mo3d and (d) Mn2p for catalyst 1.

Figure 2 .
Figure 2. (a) XRD patterns (inset: digital photos, crystal and powder) of catalyst 1 (black, simulated XRD pattern of single crystal XRD of catalyst 1; blue, powder XRD pattern of compound 1); (b) The IR spectrum of catalyst 1; (c,d) The XPS spectra of (c) Mo3d and (d) Mn2p for catalyst 1.