Large polyoxometalates (POMs) are of particular interest with fascinating structures combined with their rich electronic properties of oxygen-rich surfaces and manifold molecular characteristics have received much attention in catalysis, medicine, and material science [1-3]. A particularly interesting direction in this field is the introduction of transition metal (TM) centres form heterometallic 3d-4f clusters due to their unique properties arising from the interactions of 3d and 4f metal ions [4-6]. Currently, ruthenium as 3d metal is prominent among the elements with superior catalytic activity, with a broad substrates ranging from organic products to water [7]. Until now, a variety of TM cluster incorporated large POMs have been obtained and numerous large-cluster-shaped POMs are also now well-established such as Cs₂₀[{Nb₄O₆}(Nb₃SiW₉O₄₀)₄] [8], Na₂₂Rb₆[{Co₄(OH)₃PO₄}₄(PW₉O₃₄)₄] [9], Na₃₄[{Mn₁₉(OH)₁₂}(SiW₁₀ O₃₇)₆] [10], Na₁₉H₂₆[{Ti(OH)₃}₄Cl(P₂W₁₅Ti₃O₆₂)₄] [11], K₁₂Na₁₆[H₅₆ Fe₂₈P₈W₄₈O₂₄₈] [12], K₁₂Li₁₃[{Cu₂₀Cl(OH)₂₄(H₂O)₁₂}(P₈W₄₈O₁₈₄)] [13], K₅₆Li₇₄H₁₄ [{(P₂W₁₄Mn₄^IIIO₆₀)(P₂W₁₅Mn₃^IIIO₅₈)₂}₄(P₈W₄₈O₁₈₄)] [14], (NH₄)₂₀[{(Fe^III(H₂O))₃₀}{(W)W₅O₂₁(SO₄)}₁₂(SO₄)₁₃(H₂O)₃₄] [15] and [{Ni₆^II(Tris)(en)₃(BTC)_1.5(PW₉O₃₄)}₈] [16]; however, isolable large Ru-containing complexes are rare, though the Ru-based POMs possess valuable photocatalytic performance [17, 18]. Since the first mono-Ru-substituted POM [(C₆H₁₃)4N]₅[Ru(H₂O)SiW₁₁O₃₉] was made in 1989 [19], so far the largest structure of Ru-containing POMs is organic-inorganic hybrid K₁₆Li₁₁[{K(H₂O)}₃{Ru(p-cymene)(H₂O)}₄{P₈W₄₉O₁₈₆(H₂O)₂}] [20]. Thus, exploring giant Ru-containing POMs and investigating their superior catalytic properties still meet a severe challenge. To explore the optimal strategy for whether the large Ru-cluster can be tightly incorporated to specific POM frameworks form target compounds is very fascinating and challenging topic.

The above mentioned large-cluster-shaped POMs are generally obtained by reaction of TM with lacunary precursors elements under the conventional solutions conditions. However, the high-success hydrothermal synthesis strategy has been frequently used as the most promising method to make TM-containing POMs in the reaction systems of simple materials and trivalent Ru ions leading to a series of structural novel Ru-cluster bridged POMs [21]. The introduction of trivalence Ru ions with potential performance into this reaction system have greatly arouses our attention. In addition, P and Si are frequently used as the central atoms in the many of the reported Ru-containing POMs, but only few As be seclected as the central atom. Considering all the above reasons, Na₂WO₄, NaAsO₂ and RuCl₃ were chosen as the starting material to make the desired Ru-containing POMs (Fig. S1 in Supporting information).

Fortunately, a novel Ru₂W₁₂-cluster bridged tungstoarsenate, NaH₁₁[(RuO₆)(AsW₉O₃₃)₃{(W₆O₃)(H₂O)₆}]₂ 53H₂O (NaH₁₁·1·53H₂O), have been synthesized. NaH₁₁·1·53H₂O have the large Ru₂W₁₂ cluster and show the largest structure in all the Ru-containing POMs up to date, in which the Ru₂W₁₂-cluster was tightly trapped by six same units of B-β-AsW₉O₃₃ (Figs. S2 and S3 in Supporting information). Most interestingly, the photocatalysis oxygenation coupling reaction of anilines be selected as model substrates due to the important application prospects in chemical industry [22, 23]. Results indicate that NaH₁₁·1·53H₂O is an effective photocatalyst for the preparation of azoxybenzenes from anilines with excellent yields and high selectivity. We infer the inherent redox property of oxygen-rich POMs surfaces and high photocatalytic activity of the Ru-containing metal cluster imbedded in NaH₁₁·1·53H₂O provide a sufficient driving force for the photocatalytic transformation from anilines to azoxybenzenes, which can be verified by our experiments results and the fact that hardly target products are rapidly oxidized under light irradiation in the absence of photocatalysts.

Black block crystals of NaH₁₁·1·53H₂O were prepared by the hydrothermal synthesis method of NaWO₄·2H₂O, NaAsO₂ and RuCl₃ in water at 180 ℃ for 3 days (Fig. S1 in Supporting information). Though the NaWO₄·2H₂O, NaAsO₂ and RuCl₃ were selected as the starting material, 1 is composed by six B-β-AsW₉O₃₃ fragments (Fig. S2), suggesting that the polymerization reaction of W and As atoms to B-β-AsW₉O₃₃ units should have taken place in the process of the reaction. As shown in Fig. 1a, 1 contains six B-β-AsW₉O₃₃ fragments, two RuO₆ units, two (W₆O₃)(H₂O)₆ units. X-ray single-crystal structure analysis indicates that 1 owns a monoclinic Ru₂W₁₂-cluster hexamer [(RuO₆)₂(AsW₉O₃₃)₆{(W₆O₃)₂(H₂O)₁₂}]¹²⁻ (Figs. 1a-f and Table S1 in Supporting information), which consists of two symmetric 1a trimers [(RuO₆)(AsW₉O₃₃)₃{(W₆O₃)(H₂O)₆}]⁶⁻ (Figs. 1c and f) via six µ₂-oxo bridges. Twelve coordination water molecules O(63, 65, 82, 103, 104, 123, 144, 185, 204, 208, 215, 217) (Fig. S3 in Supporting information) in 1 can further locked by bond valence sum (BVS) calculations (Table S3 in Supporting information) [24, 25]. 1 displays two conspicuous features: the coresidency of six B-β-AsW₉O₃₃ fragments (Figs. 1g-l), and the occurrence of an undiscovered hexagon-shaped Ru₂W₁₂ cluster (Fig. S3a) that is incorporated to the framework of 1 and firmly stabilized by six trilacunary B-β-AsW₉O₃₃ fragments. In the Ru₂W₁₂ cluster, interestingly, eight octahedral distribution fashion of Ru(1, 2) and W(1, 28, 50, 51, 52, 53) ions by corner-sharing form intriguing cubane structural Ru₂W₆ cluster {(RuO₆)₂(WO₃)₆(H₂O)₆} (Figs. 1m and n), and another six W(4, 10, 27, 54, 55, 56) ions are absorbed on both sides of the cubane structure (Fig. S3). Another interesting feature of 1 is that the coexistence of six B-β-AsW₁₀O₃₇ units (Figs. 1g-l), and the occurrence of an undiscovered cubane-shaped Ru₂W₆ cluster that is incorporated to the framework of 1 and tightly wrapped by six lacunary B-β-AsW₁₀O₃₇ fragments. As to B-β-AsW₉O₃₃ and B-β-AsW₁₀O₃₇, the latter is easy to aggregate by bridging oxygen formed [(RuO₆)(AsW₁₀O₃₄)₃(H₂O)₆]²⁴⁻ trimers, which has been confirmed by the high-resolution electrospray ionization mass spectrometry (HR-ESI-MS), and two prominent peaks at m/z = 1627.34 and 2040.44 were assigned to the [H₁₉(RuO₆)(AsW₁₀O₃₄)₃(H₂O)₁₈(CH₃OH)₇]⁵⁻ (simulated 1627.35) and [H₂₀(RuO₆)(AsW₁₀O₃₄)₃(H₂O)₃₀(CH₃OH)}]⁴⁻ (simulated 2040.42), clearly proves the stability of the trimers 1a anions groups in methanol (Table S4 and Fig. S4 in Supporting information). The rational reason is that the B-β-AsW₁₀O₃₇ owns more exposed surface O atoms than the B-β-AsW₉O₃₃ at vacant sites, which results in that the B-β-AsW₁₀O₃₇ as multidentate ligand can better enhance the stability of the products when they chelated to the in situ formed the larger Ru-containing cluster [21]. The B-β-AsW₁₀O₃₇ segment derived from the polymerization of lacunary Keggin unit B-β-AsW₉O₃₃ by chelating a WO₆ group (Figs. 1g-l). Bond valence sum (BVS) calculations of 1 indicate that the oxidation states of the W, As, and Ru centers are +6, +3, and +3, respectively (Tables S3 in Supporting information), which were supported by the X-ray photoelectron spectra (Fig. 2a and Fig. S8 in Supporting information).

The characteristic vibration patterns of the typical Keggin-type polyanions dominate the IR spectra of NaH₁₁·1·53H₂O (Fig. S5 in Supporting information). Several strong characteristic bands were observed at 971, 906, 840 and 689 cm⁻¹, which can be attributed to the characteristic ν_as(As–O_a), ν_as(W–O_t), ν_as(W–O_b), and ν_as(W–O_c) [26, 27]. Two strong absorption bands at about 3445 cm⁻¹ and 1620 cm⁻¹ are assigned to the stretching vibration of lattice water and OH group, respectively [28]. The experimental powder X-ray diffraction (PXRD) pattern of NaH₁₁·1·53H₂O is in good agreement with the simulated patterns, suggesting the good phase purity of the sample (Fig. 2b) [29, 30]. The UV/vis spectra of NaH₁₁·1·53H₂O showed a characteristic absorption band around 480 nm (Fig. S6 in Supporting information), which is consistent with the presence of Ru causes the dominant oxygen-Ru responsive charge transfer [31]. The TG curves of NaH₁₁·1·53H₂O indicate that the frameworks of NaH₁₁·1·53H₂O can remain stable until 480 ℃. (Fig. S7 in Supporting information) [32].

Azoxybenzene compounds, as important chemical intermediate, have found tremendous applications in organic synthesis industry as organic dyes, photochemical switches, pharmaceuticals and food additives [33-35]. Aiming to prepare the valuable azoxybenzene derivatives, phenylamine was preferred as a model substrate to exploring ideal reaction conditions for the Ru-containing POM photocatalyzed oxidative coupling of aromatic amines in an environmentally greener processes. By employing 0.05 mol% of NaH₁₁·1·53H₂O in the presence of 2.0 mmol of H₂O₂ (30% in H₂O₂) in methanol for 48 h under 10 W LED exposure, we were surprised to found the formation of azoxybenzene as the target product with 80% yield (Table S6 in Supporting information). To produce more valuable target product with high selectivity, it is necessary to investigated various additives, such as Na₂CO₃, NaHCO₃, K₃PO₄, Na₂SO₄, Na₂SO₃, Na₂S₂O₄, Na₂S₂O₈, As₂O₃, NaAsO₂ and Na₂HAsO₄ (Table S6). Obviously, when 0.05 mmol of Na₂HAsO₄ was used, a significant amount of azoxybenzene (94.2% yield) was observed (Table 1, entry 1). This oxidative coupling reaction was accelerated slightly by increasing the molar weight of photocatalyst, then the ideal dosage could be locked in 0.05 mol% of NaH₁₁·1·53H₂O (Table S6). The photocatalyzed oxidative reaction worked in a variety of organic solvents (toluene, acetonitrile, dimethyl formamide, dichloromethane, methyl tert‑butyl ether, ethanol, methanol), in which methanol giving the highest yield and selectivity (Table S7 in Supporting information). The HR-ESI-MS analyses were used to study catalyst stability, and three prominent peaks at m/z = 1828.94, 2057.67, and 2369.76 were assigned to the [H₃(RuO₆)₂(AsW₉O₃₃)₆{(W₆O₃)₂(H₂O)₁₂}]⁹⁻ (simulated 1828.92), [H₄(RuO₆)₂(AsW₉O₃₃)₆{(W₆O₃)₂(H₂O)₁₂}]⁸⁻ (simulated 2057.66) and [H₅(RuO₆)₂(AsW₉O₃₃)₆{(W₆O₃)₂(H₂O)₁₉}]⁷⁻ (simulated 2369.77), clearly proves the stability of the complete anions groups of NaH₁₁·1·53H₂O in methanol (Fig. 2c). Interestingly, performing the single catalyst, such as Na₂WO₄, As₂O₃ and RuCl₃, was also observe to promote the oxidative coupling reactions (Table 1, entries 2–4), however with pessimistic efficiency. No desired target product was detected in the absence of catalyst NaH₁₁·1·53H₂O under light irradiation (Table 1, entry 5). Besides, the photocatalytic coupling reaction was thoroughly lose efficacy when O₂ or N₂ were chosen as the replacement of H₂O₂, which indicates the hydrogen peroxide play a significant role in this reaction system (Table 1, entries 6 and 7). Furthermore, omitting the light from the reaction system led to a significant change in the reaction outcome (Table 1, entry 8). This result clearly confirmed that the transformation of phenylamine to azoxybenzene proceeded via a photocatalytic process. However, the photocatalytic activity was no considerable improvement in the transformation of the substrate as the temperature change from 30 ℃ to 50 ℃ (Fig. S9 in Supporting information). It further proved that the reaction was implemented by the light-driven rather than thermally driven. After comprehensive optimization with the relevant reaction parameters (Tables S6 and S7), we were delighted to found that 0.05 mol% NaH₁₁·1·53H₂O, with 0.05 mmol of Na₂HAsO₄ and 2.0 mmol of H₂O₂ as oxidant in 1 mL methanol under light irradiation was the ideal photocatalytic reaction condition (Table 1, entry 1).

With the best reaction conditions, the evaluation of general applicability was subsequently performed (Scheme 1). A variety of substituted anilines with both electron-withdrawing and electron-donating were fleetly transformed into the corresponding azoxybenzene in excellent isolated yields (Scheme 1, 3b–o). Initially, the oxidation was performed with aniline leading to azoxybenzene with the highest isolated yield (Fig. 2d and Scheme 1, 2a). However, anilines substituted with electron-donating groups (Me, Et, ⁱPr, n-Bu, C(Me)₃, OMe, OEt) led to lower yields of the isolated products (Scheme 1, 2b–2l). Most impressively, various anilines substituted with electron-withdrawting (F, CF₃, Cl, Br) led to corresponding products 2m–2r in high isolated yields. In summary, it showed that the electron-deficient group reduces the electron cloud density around the benzene ring and further accelerates the photocatalysis reaction process. Moreover, we also investigated the sensitivity of steric hindrance for the substituents position on the phenyl ring. Substituents at the para position products for electron-donating groups formed the relevant azoxy compounds in better yields than that of ortho and meta position products due to steric effects (Scheme 1, 2b–2d and 2i–2k). The same experimental results were obtained for the electron-withdrawting groups at different position products (Scheme 1, 2m–2o). Most interestingly, a para-C(Me)₃ group as electron-donating substituent and a para-CF₃ group as electron-withdrawing substituent converted the target products with a good yields (Scheme 1, 2h and 2p).

Additionally, the scalability of the photocatalysis experiment was examined by oxidizing anilines on a ten-times scale, affording the desired azoxybenzene about 90% isolated yields, which further validated the good catalytic performance of NaH₁₁·1·53H₂O (Fig. 3a). The product of azoxybenzene is soluble in the reaction system, but can successfully be separated by extraction with normal hexane without interfering with the homogeneous catalyst. The reusability was tested by ten consecutive times for model substrate after the initial reaction (Fig. 3b). The experimental procedure was as follows: after the reaction, the solution was extracted to separate the product, the catalyst was remain in methanol phase, addition of 1 mmol anilines and 2.0 mmol 30% H₂O₂, restart directly for the next cycle. This extraction method shows that the soluble products also be effectively separated from the catalyst and to subsequently recycle the latter in the homogeneous reaction system. Simultaneously, the good time-resolved stability of polyanion 1 in methanol was proved for continuous usability (Fig. 3c).

In summary, we have demonstrated the capacity of largest Ru-containing POM catalyst NaH₁₁·1·53H₂O for the oxidation of anilines to achieve various azoxybenzene compounds. This protocol works with a wide range of substrates and tolerates a series of functional groups on the anilines. This work would enrich the POM family and provide an opportunity for the expansion of the potential applications of Ru-containing POM materials.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This work was supported by the National Natural Science Foundation of China (Nos. 22171071, 22071044, 21771054, 21571050).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2021.12.081