**1. Introduction**

Polyoxometalates (POMs), a class of metal oxide clustered anions, have already been investigated for more than 200 years. Their history dates back to 1826, when Berzelius reported the discovery of the first POM cluster (NH4)3[PMo12 O40] nH2O [1]. However, due to di fficulties achieving insights into the POM structure, no significant progress was made until Keggin determined the structure of H3PW12 O40 in 1934 [2]. Since then, the interest of scientists in POMs has increased drastically, not only in the development of new POM structures, but also towards their use in various applications, such as catalysis, optics, magnetism, biological medicine, environmental science, life science, and technology [3–7]. In particular, their use in catalysis is one of the most examined fields because of their strong acidity, oxygen-rich surface, photoactivity, and redox capability. Despite these interesting characteristics, POMs still exhibit some drawbacks for their use in catalysis. First, POMs possess a low surface area (<10 m<sup>2</sup> g<sup>−</sup>1), which consequently hinders the accessibility of reactants and secondly, their high solubility in aqueous solutions and polar organic solvents results in a low recyclability [4]. The immobilization of POMs into/onto porous solids has been proposed to overcome these shortcomings and to achieve catalysts with a high catalytic performance. In the past few decades, many porous materials have been examined as supports for the immobilization of POMs, e.g., silica, ion-exchange resin, zeolites, and activated carbon [5–8]. Since the discovery of metal–organic frameworks (MOFs), much e ffort has been dedicated to use these porous materials as potential supports for POMs. MOFs are inorganic–organic hybrid crystalline materials that are constructed from metal ions or clusters and organic linkers through coordination bonds. These materials have attracted considerable interest in recent years due to their large surface areas, tunable pore size, and designable functionalities. So far, MOFs have shown grea<sup>t</sup> potential in gas storage and separation, catalysis, sensing, drug delivery, proton conductivity, solar cells, supercapacitors, and biomedicine [8–13]. Moreover, MOFs are regarded as an outstanding platform for introducing gues<sup>t</sup> molecules because of the high accessibility of their internal surface area and long-range ordered channels. So far, several active sites have been successfully embedded in the pores or cages of MOFs, such as noble metals, metal oxides, enzymes, and POMs [14–17].

The first report of a POM@MOF hybrid was reported in 2005 by <sup>F</sup>érey and co-workers [18]. In this seminal work, the POM, K7PW11O39 (van der Waals radius, 13.1 Å), was successfully encapsulated into the big cages of the highly stable Cr-based MOF, MIL-101, by using an impregnation method. To date, several other thermal and chemical stable MOFs have been applied as supports to host POMs for their use in catalysis, including MIL, UiO, ZIF series, NU-1000, and Cu-BTC frameworks (see Table 1). The most examined POMs that have been encapsulated into MOFs are the well-known Keggin [XM12O40]n− and Dawson [X2M18O62]n− (X = Si, P, V, Bi, etc.; M = V, Mo, W, etc.) POMs and their derivatives. These POMs are of significant interest because their structure and properties can be easily varied by removing one or more MO4+ units, leading to lacunary POMs such as [PW9O34]<sup>9</sup><sup>−</sup>, or by the substitution of X and M by different metals or a combination of two fragments of the Keggin structure, leading to sandwich-type POMs such as [Tb(PW11O39)2]<sup>11</sup><sup>−</sup>.

There are several advantages of using MOFs as a host matrix to encapsulate POMs. First of all, their exceptionally high surface areas and confined cages/channels make it possible to ensure a homogeneous distribution of the POM in the MOF host. This not only prevents the agglomeration of POMs, but also improves their stability and recyclability and ensures a fast diffusion of substrates and products. Secondly, the highly regular cages and windows of MOFs ensure a high substrate selectivity, or, in other words, only specific substrates/products are able to reach the active POM sites. Thirdly, owing to the good interaction and electron transfer between the MOF and POM, an increased synergistic catalytic performance is typically observed. Finally, the chemical environment of POMs can easily be adjusted through modification or functionalization of MOFs. Therefore, POM@MOF hybrids not only combine the interesting properties of POMs and MOFs, but also allow the aforementioned disadvantages of POMs to be tackled to afford synergistic catalysis. This review is focused on the synthetic aspects of POM@MOF hybrids, as well as their use in catalysis (organocatalysis, electrocatalysis, and photocatalysis). Alongside the POM@MOF systems discussed here, where the POMs are encapsulated inside a MOF host, POM-based MOFs have also been investigated. In these MOFs, the POMs form the actual metal nodes that are interconnected by organic linkers [19–21]. However, these fall outside the scope of this review.


**Table 1.** Physical properties of representative metal–organic frameworks (MOFs) used to encapsulate polyoxometalates (POMs) for catalysis.

#### **2. Synthesis and Design of POM@MOF**

To date, several well-known highly stable MOFs have been used to encapsulate POMs, including MIL, UiO, and ZIF series, as well as NU-1000 and Cu-BTC frameworks. One of the most commonly applied methods to embed POMs in MOFs is impregnation. Wet impregnation is a simple and straightforward method, since most of the POMs are well-soluble in polar solvents. Typically, the activated MOF powder is immersed in the POM solution to obtain the composite material. Several POM@MOF hybrids have been successfully synthesized through this wet impregnation method, such as POM@MIL, POM@ZIF, and POM@UN-1000. An important aspect allowing the use of this method is that the size of the POM must be smaller than the windows of the MOF. Moreover, for some POMs and MOFs, it was observed that the POM loading could not be enhanced by increasing the concentration of POM in aqueous solution when a certain POM loading was achieved. For example, for POMs whose size is bigger than the pentagonal windows (12 Å) of MIL-101(Cr), the POMs were only encapsulated into the large cages, while the other cages, which represent 2/3 of the total number of cages of MIL-101(Cr), were unfilled. Naseri et al. demonstrated that the loading of a sandwich-type POM [(HOSnOH)3(PW9O34)2]<sup>12</sup>− (15.2 Å × 10.4 Å) could not be enhanced by increasing the concentration of POM in the aqueous solution [29].

The impregnation method cannot be used for MOFs whose window size is smaller than the POMs, e.g., Cu-BTC, UiO, and ZIF. Therefore, for these MOFs, the one-pot (also known as bottle-around-the-ship) synthesis method has been applied to obtain POM@MOF hybrids. The one-pot method is also often used to obtain POM-encapsulated MOFs in which the anionic form of the POM acts as a structure directing agen<sup>t</sup> to ensure deprotonation of the organic carboxylate ligand. For the preparation of POM@MOF hybrids, typically, the synthesis parameters employed to obtain the parent MOF are used upon addition of the POM. The one-pot method can not only be used to synthesize POM@MOFs that cannot be obtained by impregnation, but can also confine the POMs in the MOF cages to prevent leaching if the size of the POMs is bigger than the windows of the MOFs.

Therefore, in conclusion, the synthesis approaches commonly used to incorporate POMs into MOFs are impregnation and one-pot synthesis. To choose, however, the "correct" methodology, two questions need to be addressed in advance: does the size of the POM fit into the MOF cages and can the pore window of the MOF confine the POM? If both criteria are met, one can expect that the obtained catalyst will work e fficiently at a molecular level.

As was mentioned before, the first report on the embedding of a POM into the cages of an MOF was reported by <sup>F</sup>érey's group. They showed that a Keggin-type POM, K7PW11 O40 (van der Waals radius, 13.1 Å), can be confined in MIL-101(Cr) by simple impregnation. The resulting MIL-101-Keggin solid was characterized by XRD, TGA, and N2 sorption, as well as IR and 31P solid state NMR, which confirmed the presence of Keggin ions within the pores [18]. As summarized in Table 1, MIL-101 has two types of mesoporous cages: a smaller one with an inner size of ~29 Å and pentagonal windows of ~12 Å, and a larger one with an inner size of ~34 Å and hexagonal windows of ~15 Å. Based on the size of the cage windows and the size of the POM, one can conclude that the POM can only diffuse into the largest cages.

In 2010, Gascon and co-workers prepared PW12@MIL-101 (PW12 = [PW12 O40] <sup>3</sup>−) composites by using a one-pot and wet impregnation method [30]. The authors observed a homogeneous distribution of PW12 when the one-pot synthesis was applied under stirring conditions. By using the wet impregnation method, high loadings of PW12 in MIL-101 resulted in a drastic decrease in the surface area and pore volume. However, this decrease in surface area and pore volume was smaller for the one-pot synthesis method in comparison to the impregnation method using the same PW12 loading. The authors stated that in the one-pot synthesis, both the large- and medium-sized cavities were occupied, while, when using the impregnation method, only the larger cavities were accessible.

Canioni and co-workers compared di fferent synthesis methods for encapsulating POMs in MIL-100(Fe) [31]. The authors observed a good agreemen<sup>t</sup> between the experimentally obtained POM loading and the maximum theoretical loading for the PMo12@MIL-100 (PMo12 = [PMo12 O40] <sup>3</sup>−) obtained by a one-pot solvothermal synthesis. In addition to this, the solvothermally obtained

PMo12@MIL-100 showed a good stability in aqueous solution and no POM leaching was observed after 2 months. On the contrary, the PMo12@MIL-100 material prepared through impregnation exhibited significant POM leaching after 2 months.

Based on the above examples, POM leaching was observed for the POM@MIL-101 and POM@MIL-100 obtained by impregnation, since immobilization is based on an adsorption equilibrium. One way to circumvent this leaching is to use amino-functionalized MOF structures, e.g., UiO-66-NH2 and MIL-53-NH2, which can ensure a better interaction with the polyanions [32–34]. The formation of complexes such as −NH3 +[H2PW12 O40] − between primary amines, ammonia, or pyridine and PW12 is well-documented [35]. In 2012, Gascon and co-workers used a microwave-assisted one-pot synthesis to obtain PW12@MIL-101-NH2(Al) as their attempts to synthesize MIL-101-NH2(Al) containing PW12 by one-pot solvothermal synthesis were not successful [36]. One year later, Bromberg et al. examined the encapsulation of POMs in amino-functionalized MOFs (NH2-MIL-101(Al) and NH2-MIL-53(Al)) by immobilization. They concluded that POMs electrostatically interact with the MOF surface to form a stable composite. The thermal stability of the composites PW12@NH2-MIL-53(Al) and PW12@NH2-MIL-101(Al) was similar to the stability of the parent MOFs [37].

Besides MIL-101, Cu-BTC (namely HKUST-1 or MOF-199) has also been used as a host material to encapsulate Keggin- and Dawson-type POMs [38,39]. As shown in Table 1, the larger cages of Cu-BTC have an inner diameter of 1.3 nm and a pore window of 0.9 nm, which perfectly ensures the stable entrapment of POMs. For example, PW12 with a diameter of approximately 1.06 nm was used as a structure directing agen<sup>t</sup> for the self-assembly of Cu-BTC at room temperature [38]. The authors observed an enhanced chemical and thermal stability after the embedding of POM and no POM leaching was noted during catalysis in several studies [40,41]. In one of these studies, Shuxia Liu's group prepared a series of Keggin-type POMs in Cu-BTC, denoted as NENU-n, n = 1~10, and formulated as [Cu2(BTC)4/3(H2O)2]6 [H*n*XM12 O40]·(C4H12N)2 (X = Si, Ge, P, As, V, Ti; M = W, Mo), by using a one-pot hydrothermal synthesis [42,43]. The templating effect of the POMs resulted in highly crystalline composite materials which showed an enhanced thermal stability. Moreover, as large crystals were obtained, the structures were elucidated by means of single-crystal X-ray diffraction, demonstrating that the Keggin polyanions were confined in the larger cuboctahedral cages (inner diameter of 1.3 nm) [43].

Besides MIL-101 and Cu-BTC, isostructural imidazolate frameworks, namely ZIF-8 and ZIF-67, have also been frequently used as the host matrix. The sodalite-type cavities of ZIF-8 have a size of approximately 1.1 nm, but the accessible window of the cavity is rather small (0.34 nm). Keggin-based POMs possess a relatively larger particle size up to 1.3 ~ 1.4 nm in comparison to the cavities of ZIF-8, but can fit perfectly in their anionic form (1 nm diameter of PW12) [33]. Therefore, the bottle-around-the-ship method is an ideal approach for confining POMs inside ZIF-8 or ZIF-67 [44]. For instance, Malka et al. reported a POM encapsulated in ZIF-8 for its use as an esterification catalyst. The authors were able to obtain a PW12 loading of 18 wt% by using a one-pot synthesis strategy at room temperature in aqueous solution. However, after three catalytic cycles, degradation of the MOF occurred and 9% of the POM leached out [45]. A way to overcome the POM leaching in ZIF-8 was demonstrated in the work of Jeon et al. In this study, an impregnation method was used to functionalize the surface of the ZIF-8 nanoparticles with a Keggin-type PW12, in order to obtain a core–shell MOF–POM composite. Interestingly, due to the strong interaction, the POM-decorated MOF became insoluble in hydrophilic solvents [46].

Lin and co-workers constructed a POM@MOF molecular catalytic system with a Ni-containing POM [Ni4(H2O)2(PW9O34)2]<sup>10</sup>− (namely Ni4P2) into an [Ir(ppy)2(bpy)]+-derived MOF by one-pot synthesis, and the MOF was isostructural to UiO-66, with extended ligands. Ni4P2 POMs can be encapsulated in the octahedral cages with an inner dimension of 2.2 nm [47].

In the studies mentioned above, the one-pot synthesis and wet impregnation methodology both give a high chance of success in the synthesis of POM-encapsulated MOFs. Although the impregnation method is straightforward, it is only applicable for MOF pore windows larger than the POMs. However, leaching of the POMs might happen unless precautions are taken in advance to ensure a good interaction between POM and MOF supports. The use of POMs as a template might enhance the crystallinity of the MOF framework, which makes a one-pot synthesis very attractive. However, it is important to note here that the size of the POM needs to be larger than the pore window of the MOFs to prevent leaching. In addition to this, the one-pot synthesis method is not applicable to all MOF structures. In most cases, the POM@MOFs materials obtained through a one-pot synthesis or impregnation method have a positive influence on the thermal stability in comparison to the parent non-functionalized MOF.

Besides the commonly used one-pot and impregnation method, there are some other efficient methods for constructing POM-encapsulated MOFs. In 2018, Zhong et al. synthesized NENU-3 (PW12@HKUST-1) by a liquid-assisted grinding method [48]. By using a two-step synthesis, PW12 and the Cu salt were first dissolved and evaporated to obtain the copper salt of PW12. Hereafter, the H3BTC ligand was added in the presence of a small amount of alcohol (MeOH and EtOH), which was used as the grinding liquid. The mixture was ground for 5 min and the color gradually changed to blue. After washing and drying at 60 ◦C for 24 h, the obtained nanocrystalline, NENU-3, showed a high crystallinity, and the surface area was slightly higher compared to NENU-3 obtained in one-pot solvothermal synthesis. In 2019, G. Li et al. employed an in-situ hot-pressing approach to encapsulate the Keggin-type PW12 into an indium-based MOF (MFM-300(In)) [49]. As shown in Figure 1, all the ingredients, including the POMs, were ground in the absence of a solvent, after which they were packed with an aluminum foil and heated on a plate at 80 ◦C for only 10 min to obtain PW12@MFM-300(In) composites. The resulting materials exhibited a high crystallinity and stability and no PW12 aggregates were observed on the MOF surface.

**Figure 1.** The hot-pressing synthesis process of PW12@MFM-300(In). Reprinted from [49]. Copyright (2019), with permission from Elsevier.
