When Gold Cations Meet Polyoxometalates

: Merging gold(I) cations with polyoxometalate anions results in various interclusters and complexes. Herein, the syntheses of these newly emerging gold(I)/polyoxometalate materials are reviewed. The applications of these promising hybrids in organic catalysis are also summarized and evaluated in terms of the advantages and limitations of the catalysts including efficiency, synergistic effects and recyclability. metal NHC complexes for catalysis purpose, with a special interest on gold chemistry. [{Au(PR 3 )} 4 ( µ 4 - O)] 2+ versus [{{Au(PR 3 )} 2 ( µ- OH)} 2 ] 2+ ; 4/ the substituents on the phosphanes influence the aggregations of dimers (e.g. cross-edge arrangement versus parallel-edge arrangement). Very soon, there is no doubt that the chemistry of gold SICCs will be extended to ligands other than phosphanes and to clusters containing elements other than oxygen or nitrogen, thus providing exciting self-assembly structures. On the other hand, phosphane or NHC gold/POM complexes are even more emerging compounds than gold POM SICCs. In solution, their chemistry is dominated by various equilibria involving gold fragments and protons redistribution around the POM anions. Thus, crystallization trends to favor the formation of species with high contents of gold, likely associated with better solubility in organic solvents. These highly stable complexes denote excellent abilities as pre-activated, efficient and recyclable catalysts in various gold-catalyzed reactions. Moreover, the boundary between formation of mononuclear complexes or SICCs appears quite thin. Hopefully, as these hybrids complexes will become more popular, their chemistry will be better understood, giving rise to valuable catalytic and multifunctional systems under both homogeneous and heterogeneous conditions.


Introduction
In the periodic table of elements, gold (Au) is located in the sixth period (horizontal row) and belongs to group eleven (vertical column) with copper and silver (coinage metals). This specific position (Z = 79) is obviously due to its electronic configuration ([Xe]4f 14 5d 10 6s 1 ) but provides to gold atom peculiar properties. Gold is indeed the most impacted metal by relativistic effects causing the contraction of the atomic 6s orbital and an expansion of the atomic 5d and 4f orbitals considering the increased shielding effect. [1] As a direct consequence, cationic gold(I/III) complexes possesses a strong carbophilic Lewis acid character and have thus excellent catalytic activities for different reactions. [ 2 ] Moreover, gold catalysts are often tolerant toward air, moisture and to various organic functions. They exhibit a limited toxicity compared to other late-transition metals. Another property arising from the relativistic effects is the propensity of gold(I) to form Au•••Au bonding. [ 3 ] Theses (d 10 -d 10 ) interactions with strength comparable to hydrogen bonds are attributed to dispersive forces and virtual charge transfers. This phenomenon called "aurophilicity" is at the origin of various supramolecular aggregates and gold(I) clusters with fascinating coordination numbers and geometries. [4] On the other hand, polyoxometalates (POMs) refer to polyanionic clusters formed by the assembly of early transitionmetal oxide building blocks [MOx] (M = W, Mo, V…). Since their discovery two centuries ago, thousands of structurally different POMs with various properties have been reported, impacting all the areas of chemistry and notably catalysis. [5 ] Among them, Keggin [XM12O40] nor Wells-Dawson [X2M18O62] nheteropolyanions, where X is a p-block element of the periodic table (usually P or Si), are the most studied considering their well-known structures, their high stability and their simple preparation. POMs are versatile solids which can exhibit a strong acidity when their negative charges are compensated by protons or oxidative properties. [6] Despite the limitations induced by their low surface area, POMs possess a strong potential for performing molecular design of hybrid materials associated with organic or inorganic molecules. [7] In this area of research, POMs have notably been used to support and stabilize gold nanoparticles for various applications and have been recently reviewed in 2016. [8] However, in the continuously growing field of POMs, new types of hybrid materials are emerging. This minireview is thus focused on the synthesis of interclusters and complexes merging gold(I) cations and POM anions and their applications in gold catalysis.

Gold-POM Interclusters
The first gold-POM supramolecular intercluster compound (SICC) was synthesized and named this way by Jansen et al., in 2006. [ 9 ] The term SICC describes a self-assembly material comprised of two different inorganic clusters with diameters of 1 nm or larger. In the solid-state, nanoparticles aggregate through superlattices generally built from non-directional interactions between more or less spherical particles and displaying limited translational order. By contrast, SICCs made with clusters of different shapes and ionic charges exhibit a lattice with directional interactions and higher translational order, providing crystalline materials. Interestingly, their studies by single crystal X-ray diffraction (SC-XRD) give a direct observation of the different intermolecular interactions responsible for their selfassembly into supramolecular edifices.
All gold POM interclusters reported to date are made under air/moisture conditions. The reaction of 1 equiv. of the freshly prepared [Au9 ( (Figure 1). [9] The first isomer has a center crown shape featuring a D4h symmetry. The second isomer has a butterfly shape featuring a D2h symmetry. They represent the first successful structural characterization of the [Au9(PPh3)8] 3+ cluster by SC-XRD. In this case, the presence of POM anions in the crystal lattice seems to freeze the geometry of the [Au9(PPh3)8] 3+ isomers, suppressing the refinement issues related to heavy positional disorders. The second crystal structure less dense (C2/c) has voids filled with solvent molecules, and shatters easily by desolvation. Finally, using [Au8(PPh3)8] 2+ [NO3] -2 also produces 1 via cluster rearrangement occurring during the crystallization step.
Following a similar synthetic protocol than for 1,  (7). [11] Its crystal symmetry is defined by the Cc space group symmetry. The [Au9(P(p-tol)3)8] 3+ fragment has a D4h symmetric skeletal structure. Contrary to 4-5, the crystal packing is less dense and the anionic charge of the POM is compensated by an ammonium cation. It also contains large voids (24% of cell volume) filled with 16 molecules of acetonitrile, 6 molecules of DCM, and likely 4 other solvent molecules undetected, per unit cell. So, 7 is best described by To obtain SICCs with cationic gold clusters of higher nuclearity, Jansen et al. suggested to use the chelating dpph ligand (dpph = 1,6-bis(diphenylphosphanyl)-hexane), taking into consideration the successful synthesis of [Au11(dppp)5] 3+ [SCN] -3 and [Au13(dppm)6] 4+ [NO3] -4 (dppp = 1,3-bis(diphenylphosphanyl)propane, dppm = bis(diphenylphosphanyl)-methane). [ 12 , 13 ]  3in ethanol yields a red precipitate which is isolated but not characterized. The latter is dissolved in DMF and layered with toluene to afford, after a week, the SICC [Au9(dpph)4] 3+ [α-PW12O40] 3-(9) as crystals, with the P space group symmetry. [14] Similarly to 8, the cationic cluster features a crown shape. Voids account for 28% of the crystal cell volume, and are filled with solvent. As the ddph ligand failed to provide a phosphane gold cluster with higher nuclearity, [Au11(PPh3)8Cl2] + [PF6]was made according to known procedure. [15] The latter is dissolved in DCM and layered with an acetone solution containing 0.5 equiv. 2-(10) as crystals, sensitive to desolvation, with the P space group symmetry. The [Au11(PPh3)8Cl2] + 2 core displays a geometry similar to the known neutral clusters [Au11(PPh3)7X3] (X = halide, pseudo-halide). [16] Important voids filled with solvent molecules (24.1 % of the cell volume) are found. (Figure 2 (12).The dppm ligand promotes the cleavage of the gold cluster instead of increasing its nuclearity. [17] as crystals with the P , P , P21/m, P21/n, or P21/n space group symmetries (Figure 3). [17] As expected, different [Au2]:[POM] ratios set fundamentally different self-assembly edifices in the solid-state. Since 2010, Nomiya's group well known in POM chemistry, has been steadily publishing on gold POM SICCs, and the next part of this review will cover their work. It is essential to acknowledge that all their syntheses are rationalized and associated to SICCs fully characterized by elemental analysis (EA), thermogravimetric and differential thermal analysis (DTA), infrared spectroscopy (FTIR), solid-state and liquid NMR spectroscopies and finally SC-XRD. Their first SICC was unexpectedly synthesized during the course of an acid-base reaction between a free acid form of POM and a gold(I) phosphane complex bearing a labile pyrrolidinone-carboxylate ligand. [18], [19] Thus following this methodology, solutions of 12, 8, or 7 equiv. of triphenylphosphane gold(I) (S,R)-2-pyrrolidinone- [20] (23) with approximately 50 % yields, as crystals. 18 (Figure 4)  2+ [BF4] -2 which has a gold oxonium cluster with a regular tetrahedral geometry (Td symmetry), compounds 16, 20, 22 and 23 feature some gold clusters with distorted tetrahedral geometries (C3v symmetry). Such differences likely arise from the large size and high anionic charge of the POMs. Interestingly, three monomeric phosphane gold(I) moieties bound to oxygen atoms from the edge-shared WO6 octahedrons of the POMs are present in compounds 22 and 23. They might be considered as intermediates in the formation of [{Au(PPh3)}4(µ4-O)] 2+ clusters. [19] SC-XRD and DTA confirm the presence of co-crystallized solvents. In the solid-state, the SICCs are best described by (16)  The clusterization also proceeds with POMs free of proton, in the presence of water. Thus, 7 equiv. of [Au(PPh3)(R,S)pyrrld)] dissolved in DCM and layered with a solution of ethanolwater containing 1 equiv. of (24), with approximately 50.5 % yield, as crystals with the P21/n space group symmetry. [22] The heptagold cluster is formed by the selfassembly of oxonium tetra-and tri-gold clusters having respectively distorted tetrahedral-and triangular-planar geometries. Co-crystallized ethanol is found in the crystal lattice, and this compound is best described by (24) • EtOH. All [Au(PPh3)] + fragments, chemically equivalent in (solid/liquid) 31 [ 23 ] 3-2 (31), [24] 3-2 (32), [25] [ 26 ] [24] For 25, 31, 32, 34, a parallel-edge arrangement leads to rectangular arrays of gold atoms capped by an oxygen atom (D2h symmetry). This geometry is only encountered with thiolate-bridged tetragold clusters [{{(AuPR3)}2(µ-SZ)}2] 2+ (R = alkyl, aryl; Z = CMe3, aniline, sugar…). [24]  The structures of 33-34 comprised of two mononuclear phosphane gold cations bound to POM oxygen atoms, from edge-and corner-sharing MoO6 octahedra, suggest that clusterization is likely occurring at the POM surface. As expected, by 31 (35), with 14.0 % yield, as crystals with the P space group symmetry. [  Recently, Nomiya et al. extended the chemistry of gold POM SICCs with two electronic deficient species based on gold ammonium clusters. The reactions of 7.5 equiv. of [Au(PPh3)(R,S)-pyrrld)] solubilized in a mixture of DCMmethanol, with 1.5 equiv. of aqueous ammonia, and 1 equiv. of 3-2 (40) with 32.3 % or 30.2 % yields, as crystals with the P space group symmetry (Figure 7). [27] Both compounds are isomorph and best described by (39-40) • 2 DCM. Their unit cells contain two crystallographically independent [{Au(PPh3)}5(µ5-N)] 2+ clusters. One has a trigonal bipyramid shape with D3h symmetry, the other has a distorted square planar pyramid shape derived from C4v symmetry which is likely related to electron deficient species (bond order Au-N ≤ 1). Both geometries relate to the compound Schmidbaur et al. which has ammonium gold clusters with D3h or C4v symmetry, depending on the presence of co-crystallized molecules of DCM. [27] The 31 P NMR spectra of 39-40 are virtually identical to those of SICCs with gold oxonium clusters. They exhibit a singlet around 25 ppm for the gold ammonium cluster and a singlet around -14 ppm for the POM. DFT calculations support the preferential formation of [{Au(PPh3)}5(µ5-N)] 2+ compared to [{Au(PPh3)}4(µ4-N)] + or [{Au(PPh3)}6(µ6-N)] 3+ . They also demonstrate that clusters with D3h symmetry are only 1.6 kcal.mol -1 more stable than clusters with C4v symmetry. In that sense, it is not surprising to find both geometries coexisting in the same crystal packing.

Gold(I)-POM Complexes
were characterized by SC-XRD with crystals having the P space group symmetry (Figure 8). The formation of [Au(PPh3)2] + 2 along with gold(0) nanoparticles is a well-known decomposition pathway for phosphane gold(I/III) complexes. [29] (Figure 9). [30] Analyses by 31  3-2 (18). [31] In this case, changing the solvent system has a dramatic impact on the reaction outcome, reflecting magnificently the variability of the gold POM hybrid structures toward experimental conditions. Once formed, the complex 52 does not undergo clusterization in solutions of MeCN or DMSO.  3-(60), with yields above 75%, and water as by-product. [ 32 ] The complexes, especially 55 with its high content of carbene gold, are soluble in wet acetonitrile or DMSO. The presence of [Au(NHC)(MeCN)] + fragment is confirmed by 13

Gold complexes supported by POMs
In 2012, a single publication from Lacôte et al. described the synthesis of hybrids with a covalent bond between neutral phosphane gold complexes and the functionalized surface of a POM anion. [33]  (65), with yields above 90% (Figure 11). [33] The triethylammonium formed during the coupling is removed by filtration through TBA-loaded cation exchange resin. The compounds are characterized by 31 P NMR spectroscopy with two signals at -12.1 and -6.6 ppm (63-65) for the POM moiety, plus a signal at 28.3 ppm (63), 26.6 ppm (64), or 26.3 ppm (65) for the phosphane gold chloride moiety. Other analyses include FTIR spectroscopy and MS. No SC-XRD studies were made. Interestingly, the addition of 1 equiv. of silver(I) hexafluoroantimonate [Ag] + [SbF6]to 63 in acetonitrile or a mixture of water-acetonitrile, promotes the formation of cationic gold fragments which are likely coordinated to acetonitrile and/or the POM surface. Those species, active in homogeneous catalysis, are detected by 31 P NMR spectroscopy with two singlets at 18.6 ppm and 27.1 ppm in acetonitrile, or a singlet at 22.1 ppm in the mixture of water-acetonitrile. The synthesis of gold complexes supported by POM anions remains a virtually unexplored field of research which might lead to robust catalysts easier to recycle. Indeed, having a POM covalently bond to the organogold fragment should preclude any undesired ion exchanges altering the catalyst formulation and solubility. Moreover, it might also, via chelating effect, promote the stabilization on the cationic gold fragments by enhancing electrostatic interactions with the POM surface. [26]

Gold Cations inside Polyoxometalate Structures
In 2010, Kortz et al., reported the first example of POM made of gold, via a straightforward protocol in open-beaker. [ 22.0 % yield, upon crystallization of the crude reaction mixture. [35] The pH has to be carefully controlled to minimize the formation of gold(III) hydroxide precipitate and colloidal gold(0). SC-XRD studies unveil the presence of two [Au4O4(AsO4  The concept of polyoxo-noble-metalates was then extended to mixed gold-palladium POMs following a slightly modified experimental protocol used for 66. Indeed, the addition of 0.8 equiv. of palladium(II) nitrate to an aqueous solution containing 1.0 equiv. of POM 66, made in situ, at pH comprised between 7 and 8, affords the POM [NaAu4Pd8O8(AsO4)8] 11 [37] SC-XRD analyses reveal a POM with a cuboid shape already encountered with {MetPd12L8} polyoxometalates. [38] The packing crystal, represented by the space group, is highly symmetric. It prevents from differencing the palladium and gold atoms in the structure, as they all come out as a single residual density peak in the asymmetric unit (which generates later the cuboid shape by symmetry operations). They all have a square planar molecular geometry as expected for d 8 (Figure 13). The chemistry of gold-based POMs, which is in its infancy, offers the promises of exciting materials as precursors of noble metal nanoparticles or heterogeneous catalysts. Moreover, the encapsulation of the gold(III) cations in the POM structures might allow less conventional octahedral [6] or square pyramidal [5] geometry of coordination. 2y (70) (with x + 2 y = 9). [39] First the triflate anions are quantitatively exchanged by the [AuCl4]anions. Then, the bulkier Lindqvist anions replace partially the [AuCl4]anions, in the MOF area they can access. The MOF network was very sensitive to desolvation, and therefore kept in methanol solution, precluding EA or DTA. However, 70 was characterized by IR and SC-XRD. The crystal packing features the space group symmetry, with very large cavities represented 70% of volume filled with highly disordered solvents molecules and POM or gold anions. It is important to note that 70 is not a "true gold POM hybrid". Indeed, there is no ionic or covalent bonds between the POM and gold fragments present in the MOF.

Applications in Gold Catalysis
In the last 15 years, there has been a strong resurgence of interest in organogold chemistry and the associated homogeneous catalysis. The strong Lewis acidity and still softness of the gold(I) cation has rendered possible a broad array of catalytic transformations. [ 40 ] Nowadays, gold(I) complexes are largely recognized as catalysts of choice to activate carbon−carbon multiple bonds toward intraor intermolecular nucleophilic additions. In these catalytic processes, the first step is often the in-situ formation of the reactive entity "[Au I (L)] + " from a gold pre-catalyst. [ 41 ] This is achieved by replacing a strongly coordinating ligand (halogen Xor alkyl R -) by a weakly bound anion in a way to liberate a coordination site for the unsaturated substrate. Two activation methods are generally employed: one by halogen abstraction of [Au(L)] + [X]complexes using the driving force of the silver halide [AgX] formation and precipitation and the other by hydrolysis of organogold complexes (usually methylgold phosphanes) with strong acids such as heteropolyacids (HPAs, acid form of POMs).
In line with the latter strategy, Hayashi and Tanaka have combined, for the first time, an organogold complex and an HPA in a catalytic system dedicated to the hydration of alkynes. [42] Indeed, triphenylphosphane methylgold(I) complex (0.01 mol%), activated by an excess of phosphotungstic acid (0.05 mol%), catalyzed the hydration of oct-1-yne 71 leading to 80% of octan-2-one 72 (Scheme 1). Despite the good yield so-obtained with acidic POM, trifluoromethanesulfonic acid was a better activator and was thereafter used for the scope of the reaction. It is noteworthy to mention that control experiments using HPA alone do not catalyse the rearrangements/reactions reported in this section. Scheme 1. Activation of triphenylphosphane methylgold(I) complex by phosphotungstic acid for the hydration of oct-1-yne.
However, in 2003, the same authors reported that HPAs were the best acidic promoters in the intermolecular hydroamination of alkynes with anilines. [ 43 ] During the acid activators screening of [Au(PPh3)(Me)] for the addition of 4bromoaniline 73 onto phenylacetylene 74, HPAs were particularly effective affording ketimines 75 in near quantitative yields (Scheme 2). Scheme 2. Hydroammination of phenylacetylene with 4-bromoaniline catalyzed by gold(I) pre-catalyst/heteropolyacid mixture. This methodology was applied to various aromatic and aliphatic alkynes 76 and substituted anilines 77 using low catalyst loadings: 0.1-0.2 mol% of gold pre-catalyst and 0.05-0.10 mol% of [H] +
In 2009, Shi et al. reported the same transformation but mostly applied to internal alkynes 79 using 1 mol% of a triphenylphosphane benzotriazole gold(I) complex [Au(PPh3)(Btz)] activated by 2 mol% of acidic POM ( Table 2). [44] In this activation mode, the protonation of the benzotriazole part, by HPA, generates an active and thermally stable gold cation associated to POM anions. Such reaction conditions proved to be applicable to various anilines 77 and internal alkynes 78. It produces, in excellent yields (83-91%), various reduced amines after treating the sensitive imine intermediates with BH3, in THF. Starting from dissymmetric alkynes, the hydroamination furnished a mixture of amines 80/81 in ratios of 6:1 to 2:1 always in favour of the Markovnikov adduct.   4). [ 45 ] Unfortunately, a maximum yield of 29 % was obtained with silicotungstic acid while the desired product was obtained in almost quantitative yield using trifluoromethanesulfonic acid (50 mol%) as activator. POM and organogold complex mixtures have also been employed by Echavarren et al. [ 46 ] to promote the skeleton rearrangement of 1,6-enynes. Various substituted enynes 87 were engaged in alkoxycyclization in the presence of methanol affording 5-exo-dig products 88, in good to excellent yields, using only 1 mol% of [Au(PPh3)Me] pre-catalyst (67-96%, Table  3). For less reactive substrates, the reactions were carried out, at reflux, in the unexplained presence of 3 mol% of the bulky and electron rich tricyclohexylphosphane (Entries 4 and 5).
In the previously reported examples, the nature of the catalysts obtained by mixing gold pre-catalysts and POMs (structure, stability, heterogeneity, recyclability…) as well as the impact of the inorganic polyanionic POM on the catalytic activity have not been established. However, in gold(I) catalysis, the role of the counteranion is far from innocent affecting, the rate of reaction, the selectivity as well as the reactivity. [47] In 2014, Blanc et al. [28] have started to investigate the synthesis, the characterization (see  [30,32] These highly stable and pre-activated Au/POM hybrids have been tested in a classical transformation catalyzed by gold(I), i.e. the rearrangement of enyl acetate 89, [ 48 ] under heterogeneous conditions ( Table 4). They all showed efficient catalytic activity affording selectively the cyclopentenone 90 in wet dichloromethane (65-94%) or in mixture with the nonhydrolyzed product 91 in dry toluene. Regardless of the ratio gold cation/POM anion, the formation of 90 was obtained in comparable yields even if the acid free catalyst 44 exhibited a longer reaction time (Entries 1 vs 2-4). The reactivity was obviously influenced by the electronic nature of gold ligand (for example phosphane vs phosphite, entry 4 vs 8). However, the nature of the POM also impacted the efficiency of the reaction (Entry 6 vs 7 and 10 vs 11).  Table 5). [28] This unique and synergistic catalyst complex promoted the rearrangement more efficiently than a large range of homogeneous gold catalysts. Moreover, this transformation was successfully applied to various propargylic gem-diesters 92 with a good to excellent yields (34-95%). Taking benefit of residual Brønsted acidities remaining on hybrids, Blanc and coworker have also demonstrated the capacity of Au(I)/POM compounds to act as dual catalysts ( Table 6). [30] In a two steps sequence, the cyclic enol ether 95 was generated from N,O-acetal 94 via an aza-Prins cyclization using 10 mol% of the strongly electron withdrawing tri(pentafluorophenyl)phosphane gold(I) complex, and was subsequently hydrolyzed to furnish the ketone 96 ( Table 6, entry 1). [49] With Au(I)/H + /POM, the product 96 was obtained in a onestep procedure with only 2.5 mol% of catalyst loading in the same reaction conditions (67-89%, Table 6, entries 2-7). Interestingly, similar and even better yields were achieved with the Au(I)/POM compared to the homogeneous version. 3-47 was also applied to various other gold-catalyzed transformations confirming its polyvalence, multifunctionality as well as its competitive activity compared to other homogenous catalysts. [30] Additionally, the Au/POM hybrids have been evaluated in the oldest gold-catalyzed reaction: the hydration of alkynes. [ 50 ] Diphenylacetylene 97 was efficiently converted into 1,2diphenylethan-1-one 98 in an ether/water mixture (88-98%, Table 7) using phosphane or NHC gold(I) cations associated with acid-free POMs. [31,32]  3-56 (1) 16 98 [b] [a] HPLC Yields. [b] Yield was calculated by 1 H NMR analysis relative to an internal standard (dimethyl terephthalate) from the crude mixture. Gold(I) complexes tethered to a Dawson α1-organotinsubstituted polyoxotungstate 63-65 (see Section 3) prepared by Lâcote et coworkers [33] have been developed in order to cyclize very sensitive allenol derivatives and facilitate recycling of the gold catalysts.
In homogeneous gold catalysis, β,β-diaryl β-hydroxyl allenes 99, are known to rapidly generate vinylallene derivatives by dehydration, presumably due to adventitious traces of acid brought by the catalyst. [33] Pre-catalyst diarylphosphane gold(I) chloride complex 63, covalently bound to POM and activated in dichloromethane with silver(I) hexafluoroantimonate was engaged in the catalytic cyclization of such derivatives. The oxacyclization products 100 were formed at room temperature after 2 days in excellent yields (88-97%, Table 8, entries 1-4), whatever the substitution of β,β-diaryl part. Triarylphosphane gold-POM derivatives 64 and 65 showed the same reactivity despite longer reaction times (5 days, entries 5 and 6). Control experiments using activated triphenylphosphane gold(I) chloride in the presence of buffer additives such as [TBA] +

6[P2W18O62]
6or pyridine also afforded the cyclic ethers but with significant amount of elimination products after prolonged reaction time demonstrating the advantage of POM-Au tethering strategy (Entries 7 and 8). The high selectivity observed in such cycloisomerization of sensitive allenols was attributed to the ability of the tetrabutylammonium POM-gold hybrids to act efficiently as an intramolecular buffer able to capture protons on the metal-oxide surface. Tacking advantage of a gold/POM interactions on activated hybrid 63, which might stabilize the gold cation, recycling process in the cycloisomerization of allenols 99 was performed (Scheme 5, Eq. 1). The catalyst could be recovered by precipitation and centrifugation up to 3 times without any loss of yield, but the reaction rate dropped significant in the third cycle. " complexes allowing to recover the catalysts up to 5 times (Scheme 5, Eq. 2-4). [28,30,32] In all cases, the best results were obtained using hybrids with a 1 to 1 gold:POM ratio (x = 1) which maximized the heterogeneity of the catalyst. However, the composition and the size of the POM clusters as well as the nature of the ligand are important factors to stabilize the gold(I) cation and these parameters should be tuned in function of the targeted transformation. Despite the relatively low number of recovered cycles, the results were promising, and improvements are doubtlessly expected in a near future.

Conclusion
The chemistry of gold/POM hybrids is at an early stage, but its development has accelerated during the recent years with the idea to merge the remarkable properties of both entities in the same material.
In one hand, gold/POM supramolecular intercluster compounds (SICCs) have emerged but are so far restricted to phosphane gold oxonium or phosphane gold ammonium clusters. Their chemistry is facing two challenges: 1/ find the suitable conditions (solubilization/crystallization) to have reproducible experiments which can be scaled up; 2/ understand better the formation of the gold clusters to predict in accordance their future geometries in the solid-state. From the SICCs previously described, it seems likely that: 1/ the clusterization of the gold complexes is occurring at the POM surface; 2/ the counter-ions associated with the POMs influence the nuclearity of the gold clusters formed (e.g. tetra-versus hepta-gold clusters); 3/ the bulkiness and charge of the POMs influence the formation of gold dimers (e.g. [{Au(PR3)}4(µ4-O)] 2+ versus [{{Au(PR3)}2(µ-OH)}2] 2+ ; 4/ the substituents on the phosphanes influence the aggregations of dimers (e.g. cross-edge arrangement versus parallel-edge arrangement). Very soon, there is no doubt that the chemistry of gold SICCs will be extended to ligands other than phosphanes and to clusters containing elements other than oxygen or nitrogen, thus providing exciting self-assembly structures.
On the other hand, phosphane or NHC gold/POM complexes are even more emerging compounds than gold POM SICCs. In solution, their chemistry is dominated by various equilibria involving gold fragments and protons redistribution around the POM anions. Thus, crystallization trends to favor the formation of species with high contents of gold, likely associated with better solubility in organic solvents. These highly stable complexes denote excellent abilities as pre-activated, efficient and recyclable catalysts in various gold-catalyzed reactions. Moreover, the boundary between formation of mononuclear complexes or SICCs appears quite thin. Hopefully, as these hybrids complexes will become more popular, their chemistry will be better understood, giving rise to valuable catalytic and multifunctional systems under both homogeneous and heterogeneous conditions.