A step forward in solvent knitting strategies: ruthenium and gold phosphine complex polymerization results in effective heterogenized catalysts

Porous polymers based on ruthenium and gold triphenylphosphine complexes (KPhosIJRu), KPhosIJRu)Bi, KPhosIJAuCl) and KPhosIJAuNTf2)) were prepared via a cost-effective solvent knitting method with [RuHClCOIJPPh3)3] or AuXPPh3 (X = Cl, NTf2) as single monomers or combined with biphenyl, which represents a further approach to obtain heterogenized catalysts. The resulting materials mainly preserve the metal coordination environment of their parent complexes, are stable up to 350 °C and have reasonable surface areas (250–300 m g−1 for KPhosIJRu)-polymers). KPhos(Ru)s selectively catalyze the imination of alcohols in the presence of base and the results for KPhos(Au)s show they are effective for the intermolecular hydration and hydroamination of alkynes. These materials can be reused several times without significant loss of activity. This novel and simple method affords heterogenized catalysts that combine the reactivity and selectivity of their homogeneous counterparts with the stability and reusability of a heterogeneous framework.


Introduction
Heterogenized catalysts are usually prepared by immobilizing a homogeneous catalyst onto a solid support and have shown significant benefits in stability, recyclability over multiple cycles, and separation from the reactants. 1 Silica-based materials 2 and magnetic nanoparticles 3 are usually used as supports, but they have the disadvantages of a lack of homogeneity of the active sites and eventual blocking of the pores and deactivation of the catalyst. 4 Over the last few years porous hybrid organo-inorganic materials (MOFs), 5 and porous organic materials as porous organic polymers (POPs) 6 or covalent organic frameworks (COFs) 7 have been used as solid porous platforms to heterogenize homogeneous catalysts. In general, POPs show high hydrothermal and chemical stability and could be converted into catalysts by incorporation of catalytic sites as building blocks by covalent bonds and can be modified by post-synthetic methods. 8 Hyper-cross-linked polymers synthesized by a knitting polymerization method (employing AlCl 3 or FeCl 3 as a cata-lyst and dichloromethane or dimethoxymethane as a crosslinker) 9 are a relative new family of porous polymers with excellent properties, such as high thermal and chemical stability, insolubility and easy post-functionalization for use as supports for homogeneous catalysts. This method has been applied to prepare several series of porous polymers from different aromatic monomers. 6f,10 Nonetheless, the maintenance of the coordination metal environment through this strategy has not always been pursued. Lai and co-workers reported the direct knitting of palladium tetrakisĲtriphenylphosphine) and benzene where the palladium atoms were spatially isolated in the framework of the polymer. 11 Alternatively, Song and co-workers knitted 1,1′-bisĲdiphenylphosphino)ferrocene and biphenyl, affording an effective catalyst for the reduction of 4-nitrophenol, where the ferrocenyl units were not altered. 12 Recently, during the preparation of this manuscript, Huang's group achieved a highly effective catalyst for the dehydrogenation of formic acid to hydrogen by knitting a pincer ruthenium complex and benzene. 13 Here, we present an easy synthesis of new knitted metalbased polymers that maintains the molecular structure of the parent Ru-and Au-complexes (Scheme 1); these materials result in effective heterogenized metal-triphenylphosphine catalysts for the synthesis of imines, in the case of the Ru-based polymers, and for the hydration of alkynes, in the case of the Au-based catalysts.

Results and discussion
The knitted ruthenium and gold-phosphine-polymers (KPhosĲRu), KPhosĲRu)Bi, KPhosĲAuCl), KPhosĲAuNTf 2 )) were synthesized in a one-step hyper-crosslinking method via an AlCl 3 -catalyzed reaction (Scheme 1). 10 In this method, dichloromethane was employed as an external methylene linker between the phenyl groups from the phosphine ligands in the parent ruthenium [RuHClCOĲPPh 3 ) 3 ] or AuXPPh 3 (X = Cl, NTf 2 ) complexes. In the case of KPhosĲRu)Bi, the polymerization occurs in the presence of biphenyl, which acts as a co-monomer in a molar ratio of 2 : 1 versus the ruthenium complex.
The 13 C-CP/MAS-NMR spectra of the polymers (Fig. 1) displayed a peak at δ = 41.2 ppm corresponding to methylene linker, from CH 2 Cl 2 , which confirms the successful C-C coupling. 15 The CO signals could not be identified because of the low sensitivity of solid-state NMR spectroscopy. The peaks at δ ∼ 139 (sh) and 131 ppm correspond to aromatic carbons from triphenylphosphine ligands. 16 The 31 P-MAS-NMR spectra ( Fig. S1 and S2 †) show peaks at 28 and 26 ppm for both KPhosĲRu) and KPhosĲRu)Bi polymers, at 29 ppm for KPhosĲAuCl) and at 26 ppm for KPhosĲAuNTf 2 ) which confirm the presence of P-Ru and P-Au bonds. Further peaks at δ ∼ 60-70 ppm are attributable to phosphorous species generated by the polymerization process (for example ionic phosphonium species such as (PPh 3 Cl)Cl). 17 It is important to note that no free phosphine (at ∼5 ppm) was observed.
FTIR spectra (Fig. 2) show bands at 2058 and 1980 cm −1 corresponding to ν(Ru-H) and ν(CO), respectively, confirming that the molecular structure is mainly maintained after the polymerization process; some other vibration bands are also present, such as ν(C-H) at 3100 cm −1 , ν(C-C) at ∼1400 cm −1 from triphenylphosphine moieties and ν(P-C) at 1100 cm −1 . The KPhosĲRu)Bi spectrum also shows an extra band at ∼1600 cm −1 corresponding to the CC vibration from  biphenyl. The FT-IR spectra of the gold-polymers are shown in Fig. S5. † Fig. S16 † shows the XPS survey scans of the polymers. Ru was analysed by its 3p state instead of the 3d spectra to avoid the overlap of the C1s and Ru3d core-levels. The Ru3p region (Fig. 3a) shows a doublet peak at 464 eV and 485 eV for Ru3p 3/2 and Ru3p 1/2 , respectively, for the RuP species, which confirms that ruthenium has the +2-oxidation state. 18 The P2p spectrum (Fig. 3b) showed the presence of only one P-containing species at 133.0 eV, assigned to the PPh 3 species. For KPhosĲAuCl), the binding energy of Au4f at 86 eV justifies the presence of the goldĲI) oxidation state (Fig. 3c); the typical P2p binding energy was observed at 134 eV, confirming that no triphenylphosphine oxide was present in the sample. These results confirm that the oxidation state of the polymer is preserved.
Thermogravimetric analyses (TGA) (Fig. 4) showed a onestep degradation pattern for all the knitted metal-phosphine polymers. KPhosĲRu)Bi exhibited higher thermal stability than KPhosĲRu) with thermal decomposition temperatures of 380°C and 300°C, respectively. Whereas knitted gold-phosphine polymers showed similar degradation temperatures around 350°C. As commented previously, the metal content was confirmed by TGA. At high temperatures, gold, ruthenium and phosphine oxides are formed, Au 2 O 3 , RuO 2 and P 2 O 5 ; however, at 200°C Au 2 O 3 decomposed into Au and O 2 . 19 Therefore, considering that the Ru-polymer residue is RuO 2 and P 2 O 5 and Au and P 2 O 5 for the Au-polymers, it was possible to estimate the gold, ruthenium and phosphorous loading by TGA analyses which were in good agreement with those obtained by TXRF and ICP-OES, as given above.
The N 2 adsorption isotherms of both KPhosĲRu)Bi and KPhosĲRu) displayed the same type II profile. The Brunauer-Emmett-Teller (BET) surface areas were 250 m 2 g −1 and 300 m 2 g −1 (Fig. 5, Table 1) and the total pore volumes were found to be 0.11 and 0.20 cm 3 g −1 , respectively. The pore distributions were calculated by the N 2 -DFT method ( Fig. S13 †). The plot shows that pore sizes were mainly distributed around 2.4 nm, indicating that mesopores were present in both polymers. Furthermore KPhosĲRu)Bi also exhibited mesopores around 40 nm. KPhosĲAuNTf 2 ) and KPhosĲAuCl) polymers showed low N 2 adsorption (Fig. S14, A and B †) affording only 20 and 49 m 2 g −1 of specific surface areas, respectively.
The surface morphology was explored by field-emission scanning electron microscopy (SEM) and transmission electron microscopy (TEM) ( Fig. 6 and S7-S12 †) showing the typical morphology described for this type of porous organic polymer, made up of aggregates of spherical structures that resemble the texture of a sponge. In addition, energydispersive X-ray spectroscopy (SEM-EDX) and elemental mapping analysis revealed coherent ratios of Ru/P and Au/P (see the ESI, † Fig. S17).

Post-funtionalization of triphenylphosphine knitted polymers (KPhosBiĲM), M: Ru, Au)
For comparative purposes, we have tried to prepare polymers following a post-functionalization strategy (Scheme 2), using
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

View Article Online
a similar approach to that previously reported with benzene as a co-linker. 20 Triphenylphosphine was reacted with biphenyl (molar ratio 1 : 1) in the presence of AlCl 3 and dimethoxymethane as an external linker in dichloroethane (DCE) as a solvent, yielding a polymer denoted KPhosBi. It showed a BET surface area of 704 m 2 g −1 (Table 1) (N2 adsorption/desorption isotherm, Fig. S15 †). Its NMR solid state spectra are shown in Fig. S3. † Then, KPhosBi was reacted with RuCl 3 . nH 2 O, following a similar method to the one used to synthe-size the precursor RuHClCOĲPPh 3 ) 3 . However, TXRF and XPS analyses of the resulting solid revealed that the ruthenium loading in KPhosBiĲRu) is only 0.02 mmol g −1 (Fig. S16 †). Therefore, we discarded this approach to heterogenizing the Ru-phosphine complex. Alternatively, KPhosBi was reacted with AuCl(tht) in dichloromethane, affording the corresponding KPhosBiĲAu) polymer. ICP-OES analysis revealed that the gold loading was 1.01 mmol g −1 . The 31 P-NMR spectrum shows that P-Au and phosphine moieties are present in the polymer (Fig. S4 †). XPS analysis shows a doublet corresponding to Au4f at around 85 eV. In both spectra, the characteristic phosphine peak of P2p was observed at around 136 eV (Fig. S16 †). In this case, it seems that the corresponding supported Au-phosphine complex has been formed.

Catalytic applications
KPhosĲRu): imination of alcohols. This reaction, from unreactive alcohols without an oxidant, is an important nonpolluting route to provide aldehydes and related products. 21 On the other hand, imines are important compounds in modern organic synthesis because they are key intermediates in the synthesis of nitrogen-containing heterocycles. 22 Generally, imines are formed by condensation of ketones or aldehydes and amines, catalysed by an acid. Due to the instability of the carbonyl group, the synthesis of imines from more stable and low-cost reactants, such as alcohols, has attracted considerable attention, with the aerobic oxidative coupling of alcohols and amines the most promising strategy. 23 It is well known that ruthenium is a widely used and versatile catalyst for the dehydrogenation of alcohols. 24 However, it is hardly used for the imination reaction. 25 Ru-parent complex 1 is an efficient homogeneous catalyst in reactions such as the hydrogenation of aldehydes and ketones, transfer hydrogenation or transfer hydrogenative C-C bond forming reactions. 26 Thus, these data prompted us to study a possible imination reaction from alcohols and amines with knitted Ru-polymers as heterogeneous catalysts and the results were compared with those obtained from a homogeneous 1-complex. Since KPhosĲRu) and KPhosĲRu)Bi have similar characteristics, we chose KPhosĲRu) as the reference catalyst. Reaction of 1.2 mmol of benzyl alcohol, 1.0 mmol of aniline and 2 mol% of catalyst in toluene at 100°C for 2 h in the presence of a base, quantitatively afforded N-benzylidenebenzylamine (96% selectivity) (Table 2, entry 1). Previous experiments revealed that the reaction in the presence of KOH resulted in longer reaction times (entry 4) than using potassium tert-butoxide (entry 3), to achieve a similar conversion with the same substrate. Moreover, it was tested that the reaction did not occur without a base (entry 5). The reaction was not efficient at room temperature and the best results were obtained in toluene at 100°C. In the absence of a catalyst, only traces of product were obtained (entry 6).
The substrate scope was examined under optimized conditions. Electron-rich 4-methoxyaniline, benzylamine and 1-octylamine gave excellent conversions and selectivity into a At P/P 0 = 0.99.  (Table 2, entries 3, 7, 8), and 4-bromoaniline afforded 90% conversion after 20 h of reaction (entry 9). The possibility of recycling and reusing the catalyst, along with the occurrence of leaching processes, were examined under the conditions in entry 1. Recycling experiments show that the activity of KPhosĲRu) does partially decrease after two cycles, probably due to some loss of catalyst during filtration, and more time is needed to achieve >90% conversion (Fig. 7). In order to evaluate whether the structure of the catalyst had changed after recycling, we undertook XPS analysis of the recovered catalyst (Fig. S21 †). The XPS spectrum in the region of Ru3p and P2p is very similar to that of fresh Rupolymer. FT-IR after recycling (see Fig. S6 †) shows that ν(Ru-H) appears as weak signal at 2055 cm −1 and ν(CO) at 1982 cm −1 which indicates that the coordination environment of ruthenium in the polymer is mainly maintained. Gratifyingly, no leaching of active ruthenium species occurred. Upon removal of the catalyst by hot filtration after 30 minutes of reaction, the conversion remained unaltered for 3 h (Fig.  S18b †).
KPhosĲRu) was also investigated for the transfer hydrogenation of acetophenone in i-PrOH at reflux, resulting in total conversion after 16 h of reaction. The reaction between ben-zyl alcohol and acetophenone leads to a mixture of condensation products ketone/alcohol (80/20).
Based on our results and previous reports, the proposed mechanism for the alkylation of amine with alcohol 27 is that a Ru-alkoxide intermediate is formed under basic conditions, which catalyses the dehydrogenation of alcohol to a carbonyl compound, and subsequently, the base promotes the condensation of carbonyl compound with amine to imine.
KPhosĲAuX): hydration of alkynes. Hydration of alkynes is a very environmentally benign method to form carbon-oxygen bonds. Furthermore, it has been extensively studied that the efficiency of this reaction lies in the coordination environment of the gold centre, being far more active when it forms a cation, although the nature of the corresponding contra-anion can also have an effect. The hydration of phenylacetylene was selected as a model reaction to test the catalytic activity of KPhosĲAuX) (X = Cl, NTf 2 ) polymers in the presence of different co-catalysts.
First, we carried out reactions with KPhosĲAuCl) polymer in the presence of different silver-containing cocatalysts ( Table 3, entries 1-3) in order to remove the halide coordinated to the goldĲI) centres and generate catalytically suitable species for substrate activation ([PPh 3 Au] + species). It was found that the best results were obtained in the presence of silver triflimide because the triflimide anion stabilizes the cation [PPh 3 Au] + , and in all cases, the ketone was obtained. The reaction evolution in the presence of silver co-catalysts is shown in Fig. S19. † Therefore, the system KPhosĲAuCl)/ AgNTf 2 was employed with different terminal alkynes affording total conversion after 1 h of reaction (entries [11][12][13]. However, when the internal alkyne methylphenylacetylene was employed, the reaction did not occur even after 20 hours (entry 14).
Some other silver-free co-catalysts were used, such as: CH 3 CN/NH 4 BF 4 , TfOH and LiNTf 2 (entries 4-6). Again, the best results were obtained in the presence of lithium triflimide (entry 6). However, longer reaction times were necessary to achieve 100% yield. Rather, when the postfunctionalized KPhosBiĲAuCl)/LiNTf 2 system was applied as a  catalyst, only traces of hydrated product were detected after 3 hours of reaction, and longer reaction times were necessary to achieve good conversion (entry 7). The use of silver salts as cocatalysts significantly affected the reusability of the catalyst, since after three cycles the catalyst lost its efficiency. This might be a consequence of the socalled 'silver effect', reported by Shi and coworkers, 28 which suggests that silver cations not only activate gold moieties but also modify them, affording different inactive complexes. Subsequently, the possibility of recycling the catalyst, along with the occurrence of leaching processes, was examined with the KPhosĲAuCl)/LiNTf 2 system. Even though the catalyst life has been increased, it can clearly be observed that its activity decreases in each cycle (Fig. 8). Therefore, we tried to regenerate the catalyst by treating it with LiNTf 2 in methanol overnight, and we observed that the activity increased slightly up to 70%.
With the previous results in hand, we explored the catalytic activity of KPhosĲAuNTf 2 ) under standard conditions (Table 4). It was found that 4-methoxy phenylacetylene and 1-octyne quantitatively yields the ketone after 1 h in the presence of 20 μl of water (entries 1 & 3); when the reaction was performed without water the ketone was not obtained (entries 2 & 4). When the same conditions were applied to internal alkynes, it was found that methyl phenylacetylene led selectively to the ketone after 3 h of reaction with a higher amount of water (entry 6); however, diphenylacetylene only yields the enol ether (Table 3, product A) after 22 h (entries 7, 8) and 1,4-dichlorobut-2-yne does not react (entry 9). These findings suggest the following: (A) the ketone is only formed after hydration of ketal, as was previously reported by Corma et al. 29 (B) The ketal of the terminal alkyne undergoes hydration to ketone faster than those of the internal alkynes.
In order to verify the heterogeneity of the gold-catalyst, a hot filtration experiment was carried out. After 20 minutes, the solid was separated from the reaction media and the filtrate was stirred under the same conditions. Since the conversion did not increase after hot filtration, this experiment confirmed that no leaching occurs during the process (see Fig. S20 †).
Finally, recycling of KPhosĲAuNTf 2 ) using 4-methoxy phenylacetylene as a substrate was performed under entry 1 conditions. As can be observed in Fig. 9, the catalyst maintains its activity for ten cycles; however, the selectivity decreases after the third run and decreases to 40% in the sixth cycle. It is therefore necessary to regenerate the catalyst, A u C l ĲPPh 3 )/AgBF 4 10 45 10 AuClĲPPh  which is carried out by stirring it in methanol in the presence of LiNTf 2 overnight. This reactivation resulted in an increase in the selectivity to 100% for four more cycles. KPhosĲAuX): hydroamination of alkynes. The catalytic performance of the KPhosĲAu)-complexes was also tested in the hydroamination of phenyl acetylene with aniline as a model reaction in the presence of a cocatalyst (silver salts or CH 3 CN/ NH 4 BF 4 ). 30 Screening of cocatalysts was performed at 70°C and, as can be seen in Table 5, the performance of KPhosĲAuCl) depends on the silver salt anion. Anions of tetra-fluoroborate were found to be detrimental to the process, whereas triflate enabled a much better performance and triflimide stood out as the best choice, as with the hydration of alkynes. It is important to note that when the reaction was carried out with KPhosĲAuNTf 2 ) (without any cocatalyst), the product was obtained selectively in 2 h with high yield (entry 1).

Conclusions
This manuscript presents four new heterogenized catalysts formed by the polymerization of phosphine metal complexes via a solvent knitting method. In this way, the polymerization of RuHĲCO)ClĲPPh 3 ) 3 resulted in a single-site robust material, which maintained its catalytic activity after six cycles in the synthesis of imines. On the other hand, the highly active homogeneous catalyst AuĲPPh 3 )NTf 2 was polymerized by the same procedure, leading to a heterogeneous framework that also maintained its catalytic activity even after 10 cycles in the hydration of alkynes and proved to be active in the hydroamination of alkynes.
In conclusion, we have extended a knitting polymerization strategy to the field of transition metal complexes, affording a straightforward route to obtain metal-containing polymers and opening up the possibility of heterogenizing many other soluble catalysts.

Catalytic activity
Typical procedure for the synthesis of imines. In a 5 mL SUPELCO reactor: 0.1 mmol of amine, 0.12 mmol of benzyl alcohol and 2.0 mg of KPhosRu in 0.5 mL of toluene were mixed and heated at 100°C for the corresponding time ( Table 2). The conversion was analyzed by GC-MS chromatography.
Typical procedure for the hydration of alkynes. a) without co-catalyst: alkyne (0.2 mmol), 10 mg (7.2 μmol) of KPhosĲAuNTf 2 ), 0.4 mL of methanol and 20 μL of H 2 O were introduced into a 5 mL SUPELCO reactor and the mixture heated under stirring at 60°C for the corresponding time (Table 4); b) in the presence of co-catalyst: in a 5 mL SUPELCO reactor: alkyne (0.2 mmol), KPhosĲAuCl) (10 mg, 9.7 μmol) and cocatalyst (10.7 μmol) were mixed in 0.4 mL of methanol and 20 μL of H 2 O at 60°C for the corresponding time (Table 3).
Some representative examples of the chromatograms obtained are recorded in Fig. S21-S25. †

Conflicts of interest
There are no conflicts to declare.