Palladium-scavenging self-assembled hybrid hydrogels – reusable highly-active green catalysts for Suzuki–Miyaura cross-coupling reactions

From waste to wealth – a self-assembled hydrogel remediates palladium from solution down to sub-ppm levels, and the resulting gel, which has embedded Pd nanoparticles, acts as a green and efficient catalyst for Suzuki–Miyaura cross-coupling reactions.


Introduction
Supramolecular gels are colloidal so materials based on the self-assembly of carefully-designed low-molecular-weight gelators (LMWGs) into a nanoscale solid-like network that spans a liquid-like phase. 1 They have applications based on their rheological properties, and also have transformative potential in high-tech applications. 2 Given that less than 1% of additive is typically able to immobilise bulk solvent, supramolecular gels are highly solvent compatible, porous materials, and small molecules can rapidly diffuse within them. Such gels therefore have signicant potential in environmental remediationpolluted water can readily diffuse through hydrogels, with interactions between gel nanobres and pollutants leading to effective pollutant immobilisation and hence removal. 3 Furthermore, supramolecular gels have potential applications in catalysis. 4 Reagents can diffuse into gels, products can diffuse out, and if the catalyst is immobilised within the network, it can be potentially removed and reused. As such, these materials are both homogeneous (solvent-compatible) and heterogeneous (solid-like), combining the advantages of both. Supramolecular gels can provide catalyst longevity, protection from trace product contamination and ease of catalyst separation/re-use. 4 There is considerable potential to combine applications, and develop gels that can both remediate waste, and subsequently act as active catalystsas yet, this has not been demonstrated.
Perhaps surprisingly, LMWGs have attracted relatively limited attention in precious metal-based catalysis. Most commonly, a palladium-binding ligand is incorporated into the gelator structure. Early work described gelators containing a pyridine moiety capable of binding Pd(OAc) 2 , with the resulting metallogels being used for the oxidation of benzylic alcohols. 5 Later work used related systems to catalyse carboncarbon bond forming reactions such as Suzuki-Miyaura cross coupling. 6 James and co-workers prepared metal-organic gels using 5-diphenylphosphanylisophthalic acid as a palladium ligand, with the subsequent functionalized Pd-xerogel showing high catalytic activity in Suzuki cross-coupling. 7 Another gel, based on palladium CNC pincer bis(carbene) complexes was reported by Dötz and co-workers and successfully used in Michael additions. 8 Taking a different approach, Maitra and Maity used an external reducing agent to form palladium nanoparticles (PdNPs) in non-ligating calcium-cholate hydrogels, 9 and showed that the dried xerogel could be used in Suzuki cross-coupling reactions. Unfortunately, the presence of K 2 CO 3 partially destroyed the xerogel network. In terms of other catalytic metal NPs, Banerjee and co-workers have embedded AgNPs and AuNPs in gels to catalyse the reduction of nitroarenes to aminoarenes, but not PdNPs. 10 Palladium has been a transformative catalytic metal, mediating a wide range of coupling reactions, yet is a nite resource, used in many applications. 11 The recovery and reuse of this precious resource is of high value, in both economic and environmental terms. Scavenging Pd from waste is therefore of key importance. 12 Furthermore, Pd is considered an unacceptable contaminant in pharmaceutical productsits use in synthesis must be carefully considered, and there is a need to sequester Pd contaminants. 13 There has also been intense interest in developing 'green' syntheses using Pd to minimise environmental costs. 14 With such issues in mind, we reasoned that supramolecular gels could scavenge Pd waste and the resulting gels could then act as heterogeneous supports in environmentally-friendly Pd-mediated coupling reactionsan integrated 'waste-to-wealth' 15 approach going signicantly beyond other approaches to heterogeneous Pd catalysis. 16 Our novel hydrogelator, DBS-CONHNH 2 , 17 based on the commercially-relevant, low-cost 1,3:2,4-dibenzylidene sorbitol (DBS) framework, 18 has been shown to extract precious metals from model waste and immobilise them within the gel as a result of in situ reduction into precious metal NPs. 19 In the case of gold extraction, the resulting gels exhibited unique conductance properties and could be further used for the modication of electrode surfaces and in electrocatalysis. 19 We reasoned that in the case of palladium, the resulting materials may have the ability to be used in catalysis for organic synthesis. We therefore set out to fully understand the ability of DBS-CONHNH 2 to scavenge palladium, and to use the resulting materials in organic synthesisfrom waste to wealth (Fig. 1).

Results and discussion
Synthesis of DBS-CONHNH 2 hybrid hydrogels with PdNPs Hydrogelator DBS-CONHNH 2 was prepared according to the previously published two-step procedure, 17a via acid-catalysed condensation of D-sorbitol with two equivalents of methyl-4formylbenzoate and subsequent reaction of the methyl ester with hydrazine monohydrate. To improve the mechanical properties of these gels and facilitate handling, we formulated DBS-CONHNH 2 with another component, agarose polymer gel (PG)this hybrid PG/LMWG strategy, is an effective way of enhancing materials performance of rheologically weak supramolecular gels. 20 These two components assemble independently of one another, 17d with DBS-CONHNH 2 providing the required functionality for precious metal capture and agarose providing mechanical robustness. We have previously performed rheological characterisation of this type of hybrid gel, with the studies indicating that the presence of agarose increases the G 0 value by an order of magnitude. 17d The hybrid hydrogel was simply formed using a simple heatcool cycle ( Fig. 2a and b) as both LMWG and PG are thermally responsive. Initially, to incorporate PdNPs inside the gel, blocks of the hybrid hydrogel (formed from 2.00 mg of DBS-CONHNH 2 , 2.50 mg of agarose and 0.55 mL of deionised water) were immersed in an aqueous solution of PdCl 2 (5 mM) and allowed to stand for 48 hours at room temperature (Fig. 2c). The diffusion of Pd 2+ into the gel and the formation of PdNPs within the gel were clearly visualised by a colour change from transparent to yellow/brown (Fig. 2d). Rheology indicated that for hybrid gels at this loading, the presence of PdNPs made the materials slightly stiffer (the G 0 value roughly doubled, Fig. S5 and S6 †). The gels were slightly more resistant to shear strain (ca. 2%) in the presence of PdNPs. Nanoparticles are known to enhance polymer rheological performance, 21 but the effects seen here are Fig. 1 Schematic of the 'waste-to-wealth' approach using DBS-CONHNH 2 /agarose hybrid hydrogels to remediate waste, generating PdNPs in situ and then using the resulting material to catalyse Suzuki cross-coupling reactions. only small. Most importantly, the gel-like nature of the materials in the presence of PdNPs was clearly maintained.
To quantify the amount of Pd inside the gels and gain insight into the rate of uptake, we studied extraction by UV-Vis spectroscopy ( Fig. S1-S4, Tables S1-S3 †). PdCl 2 displays a strong absorption peak in the visible region at approx. l max ¼ 425 nm that decreased over time during extraction. From the resulting concentration of Pd 2+ in solution we could determine the amount of Pd loaded into the gel, which was typically 7-8 mmol. This implies that one equivalent of DBS-CONHNH 2 is capable of reducing two equivalents of palladium, suggesting each acylhydrazide is responsible for the reduction of one equivalent of palladium. Two further experiments supported the existence of a 1 : 1 relationship between Pd and the acyl hydrazide. Doubling the concentration of PdCl 2 from 5 mM to 10 mM led to only very small increases of Pd within the gel (aprox. 10.5 mmol), suggesting that the system was effectively saturated. On the other hand, halving the loading of DBS-CONHNH 2 hydrogelator (from 2.00 mg to 1.00 mg) resulted in ca. half the amount of Pd inside the gel (5 mmol). To determine the inuence of the temperature on Pd uptake, the extraction experiment was also performed at 50 C. The total loading of Pd within the gel remained the same, but uptake was ca. two times faster (see ESI †). At room temperature, aer one hour, the loading was 2.2 mmol, whereas at 50 C it was 3.9 mmol.
It is worth noting here that agarose alone can also be used as a gel support/ligand for PdNPs, but external reducing agents such as citric acid are necessary for NP formation. 22 Furthermore, the literature reports indicate that a maximum palladium loading of just 0.1 mmol g À1 onto agarose can be achieved. This compares with the 1.8 mmol g À1 onto agarose/DBS-CONHNH 2 achieved by us here. This 18-fold enhancement clearly demonstrates the active role played by DBS-CONHNH 2 in enabling efficient loading of precious metal NPs into the hybrid hydrogel.
We then tested the ability of these gels to scavenge Pd from solution at lower concentrationsas this demonstrates their potential use in the clean-up of waste streams. The resulting solution aer uptake was analysed by Atomic Absorption Spectroscopy (AAS). For example, when a 0.83 mM PdCl 2 solution (3.6 mL) was brought into contact with the gel block, the resulting concentration of Pd in solution aer 48 hours was below the AAS detection limit (<0.04 ppm). This suggests that >99.97% of Pd became embedded within the gel, indicating an outstanding ability of this hybrid hydrogel to scavenge Pd II from waste.
Transmission electron microscopy (TEM) indicated that the PdNPs formed during loading of the gel are not randomly distributed within the gel, but appear to be in close proximity to the gel bres (Fig. 3). This is in agreement with a mechanism in which the acyl hydrazides are oxidised 19,23 and hence mediate reduction of Pd 2+ to Pd 0 , with the resulting nanoparticles having limited mobility within the gel network. We suggest that electrostatic interactions between the hydrazide functionalised gel bres and electron-decient Pd 2+ ions lead to reduction on the periphery of the gel nanobers. TEM indicates that the PdNPs are mostly spherical and the average size is <5 nm. We propose that the PdNPs generated in this way will be effectively "naked" un-capped particles (no explicit stabilising ligand is added), which should make them highly catalytically active (see below). The network of the gel should also prevent NP aggregation, limit problems such as catalyst leaching and hence improve catalyst recyclability (see below).

Cross-coupling reactions using PdNPs/DBS-CONHNH 2 as a catalyst
We then went on to demonstrate that these Pd-loaded gels were catalytically procient in a key synthetic methodology -Suzuki-Miyaura cross-coupling. In this way, high-value-added applications can be achieved using materials that have been scavenging for palladium waste. Cross-coupling reactions are most usually performed in organic solvents under an inert atmosphere with various ligands that can increase catalyst activity, 11 although there has been signicant work on developing more environmentally-friendly approaches 14 and some ligand-free cross-coupling reactions are also known. 24 Due to the fact that in our hybrid hydrogels the PdNPs are stabilised by the gel network, we carried out cross-coupling reactions without any additional ligands.
Optimal Suzuki-Miyaura reaction conditions were established by monitoring the coupling reaction of 4-iodotoluene with phenylboronic acid (Table 1). Reactions were performed in a vial without stirring to avoid mechanical degradation of the hybrid hydrogel. Due to the increasing importance of green solvents in organic synthesis, we tried to avoid traditionallyused organic solvents, such as toluene or THF. Best results were obtained for reactions performed in a mixture of ethanol and water (3 : 1) that enables both efficient cross-coupling reaction and easy separation of the products. At room temperature, 95% yield was obtained aer 144 hours (entry 1). At 50 C, a similar yield (95%) was obtained aer 18 hours (entry 2). Further increasing the reaction temperature to 70 C did not result in any signicant enhancement of reaction rate (entry 3). Higher temperatures were not used because of limitations of gel stability. As can be seen from entries 2, 5 and 6, the choice of base did not have any signicant inuence on the progress of the reaction. On the other hand, if reactions were performed without any base, very low yields (16%) were obtained (entry 4). Therefore, K 2 CO 3 was selected as the most suitable base as it is low-cost and environmentally benign. We attempted to reduce the reaction time further, but this was not easily possible. We suggest diffusion in and out of the gel limits the rate. To test the inuence of the agarose on the hybrid-hydrogel, we performed the same cross-coupling reaction with just the DBS-CONHNH 2 hydrogel loaded with Pd. This gave similar results as with the hybrid-hydrogel, but the gels were not so easily handled. This suggests that role of the agarose is mostly mechanical, but small contributions to catalytic function cannot be fully ruled out. 22 As a next step, we investigated the inuence of Pd-loading on reaction progress ( Table 2). As a control, the reaction using hybrid-hydrogel without any Pd did not yield any of the desired product (entry 1). On the other hand, using 1.5 mol% Pd gave 95% yield in 18 hours (entry 2). Similar yields were obtained for reactions with lower catalyst loadings (entries 3-5). However, for very low concentrations of Pd (0.01 mol%), longer reaction times were needed and the obtained yield was only 47% (entry 6). In further studies, 1 mol% Pd in gels was used to ensure equal distribution of the PdNPs and consistent reaction conditions.
Reaction conditions were then further applied to the reactions of a broader range of functionalized aryl halides with   phenylboronic acid (Table 3) to explore the scope of this methodology. In general, both electron rich (entries 1-3) and electron poor (entries 4-9) aryl iodides gave excellent yields (>90%). This included an aromatic aminesuch functional groups are important in pharmaceutical applications. Moreover, in most cases, there was no need for any further purication except for removal of excess boronic acid, which was simply achieved by extracting the desired product into diethyl ether and washing with 1 M aq. NaOH and water. Importantly, the amount of residual palladium found in the crude product was below the AAS detection limit (<0.04 ppm), which easily meets the criteria for the oral concentration limit for medical products (<10 ppm) even without any further purication. This lack of catalyst contamination in the product demonstrates a clear advantage provided by the heterogeneous nature of the gel-based approach exemplied here. To further investigate the scope of this reaction, we tried to couple various aryl bromides and chlorides (Table 4). In the case of aryl bromides, the reactivity remained almost the same as with aryl iodides (entries 1-3). However, the reaction of aryl chlorides provided poor yields, even with prolonged reaction times (entries 4 and 5). These results are in agreement with literature data for cross-coupling reactions, 14c and in accord with the relative strengths of the C-X bond, which must be broken in the rate-determining step of the catalytic cycle. 25 We also explored several different boronic acids (data not tabulated here) and determined that couplings using an alternative aromatic boronic acid (p-methoxyphenyl boronic acid) worked well, whereas the use of an aliphatic boronic acid (butylboronic acid) was unsuccessful.
We also scaled up our standard reaction (4-iodotoluene with phenylboronic acid) to 5 mmol, in order to test the feasibility of this process at larger scale. For this purpose, a larger hybrid hydrogel was prepared (12.50 mg of DBS-CONHNH 2 , 15.63 mg of agarose, 3.44 mL of H 2 O, 1% mol of Pd). The desired product 2a was isolated in nearly identical yield (91%) to those previously described, although, a longer reaction time (24 hours) was necessary to reach full conversion. Again, we suggest diffusion of reagents into the gel limits rateone way to circumvent this would be to dose larger-scale reactions with multiple smaller gel blocks.

Catalyst recyclability
Since one of the main advantages of our hybrid hydrogels with PdNPs should be ease of recyclability, we performed a series of reactions between 4-iodotoluene and phenylboronic acid to investigate this. Aer completion of each run, the product was extracted with diethyl ether and the hybrid hydrogel was simply removed from the reaction vial with a spatula, washed with diethyl ether to ensure extraction of all the product, and then washed with deionized water so it is compatible with the aqueous solvent environment and is ready to be used directly in the subsequent reaction. As such, this gel constitutes a very easily recycled reaction-dosing formeven formal ltration is not required.
As shown in Table 5, no signicant change of catalyst activity was observed during the rst eleven consecutive runs. Aer the 11 th run, there was a small loss of activity (run [12][13][14] most probably caused by mechanical degradation of the catalytic gel. In total, only ca. 0.8 mg of Pd produced more than 1.7 g of the desired product 2a across this series of repeated reactions.  TEM imaging of the recycled and reused catalyst aer ve runs was similar to that of the fresh catalyst (Fig. 4). However, some limited aggregation of palladium nanoparticles was observed, probably due to the inuence of elevated reaction temperatureaverage PdNP size was ca. 10 nm (Fig. S7 †). Indeed, raising the temperature of these gels was observed to cause a small amount of PdNP aggregation even in the absence of reaction (Fig. S8 †). The recyclability study and TEM images of used catalyst suggest that the amount of Pd leaching into the reaction mixture must be negligible. This was further proven by AAS analysis of the hot reaction mixture (conversion ca. 50%) which quantied the concentration of Pd in solution as only 0.7 ppmimportantly, this is well below the required level for medical products.
To further prove the heterogeneous nature of the catalyst, we performed a hot ltration test. 26 The reaction of 4-iodotoluene, phenylboronic acid, K 2 CO 3 and Pd-gel catalyst (1% mol) was le to react for 24 hours and then the hot reaction mixture was ltered using a nylon syringe lter (0.22 mm). Another reaction between 4-iodoanisole (0.60 mmol) and phenylboronic acid (0.72 mmol) was directly carried out in the ltrate. Aer 24 hours, the crude reaction mixture was analysed by 1 H NMR and a small amount (34%) of conversion was observed. This indicates that the small amount of palladium in the solution (in good general agreement with AAS studies) means the reaction can be partly homogeneously catalysed. However, since conversion is much less than when the hybrid hydrogel catalyst is present (34% vs. complete conversion), we conclude catalysis is mostly heterogeneous.

Reaction in a ow-through device
Heterogeneous catalysts are oen used in ow-through devices that offer many advantages over commonly used reaction setups, such as enhanced heat and mass transfer, possibilities of scale-up, or specied control over reaction and retention times. 27 We were therefore interested to determine whether our hydrogels with PdNPs could be used in this way. We prepared a very simple ow-through device made from a plastic syringe. In a typical experiment, a 3 mL plastic syringe was partly blocked with cotton-wool at the bottom and a hot hydrosol (made from 2.0 mg of DBS-CONHNH 2 and 0.65 mL of deionised water) was transferred on the top. The hydrosol was allowed to cool down at room temperature as the hybrid hydrogel formed (Fig. 5a). PdNPs were embedded within the gel in a similar way as described before (by the interaction with a solution of PdCl 2 , Fig. 5b). This very simple device was then directly used for the Suzuki coupling experiments in the ow-through mode illustrated (Fig. 5c) under the inuence of gravity.
We initially studied our standard cross-coupling reaction between 4-iodotoluene and phenylboronic acid ( Table 6, entry 1). The reaction conditions were almost identical to those used in the normal reaction setup. The only difference was the use of KOH as base instead of K 2 CO 3 (due to the shorter reaction times in ow-through mode). The reaction was performed in an incubator at 50 C to ensure that the temperature was the same in all parts of the gel. When the reaction was performed with a very fast ow-rate ca. 9 mL min À1 (complete diffusion through gel was nished in 20 seconds), conversion was 70%. Since these gels are relatively fragile (no agarose PG is present in these experiments), we were not able to apply any pressure and thus, the ow-rates varied signicantly, and could not be directly controlled.
This fast ow rate was not typicalmore usually ow rates 0.02-0.2 mL min À1 were observed (entries 2-7), with owthrough therefore being complete in 15-150 min. In these cases, near quantitative conversion was obtained, suggesting that with the faster ow rate described above, the gel may have been somewhat cracked, with not all of the reagent coming into contact with the immobilised catalyst. Clearly, however, our gels are highly active and can be efficiently used in a ow-through mode. As can be seen from Table 6, when these ow-through conditions were applied to a variety of substrates, in all experiments we obtained excellent yields in much shorter reaction times compared with the classical setup. Due to solubility issues with some products (entries 5-7), we changed solvent from an EtOH/H 2 O mixture to PEG 200. This resulted in a small drop in observed ow-rate, but the isolated yields were still excellent. In future work it may also be worth exploring other green solvents in this ow-through approach.
It is noteworthy that in ow-through mode the reaction is much faster than in the standard reaction set-up described earlier (Table 1). We suggest this is the result of the ow through the device under the force of gravity ensuring contact of  reagents with the catalyst, and preventing the passive diffusion of reagents in and out of the gel from becoming rate limiting.
We then tested the agarose/DBS-CONHNH 2 hybrid gel loaded with PdNPs in ow-through mode. This gave broadly similar results to DBS-CONHNH 2 alone, however the ow rate was somewhat lower. We hoped that these gels may be strong enough to handle ow-through reactions under pressure, but unfortunately they were too fragile. Further work here will focus on optimising the PG component to combine with DBS-CONHNH 2 such that these materials are even more robust, 17c and can potentially be incorporated into columns. This should enable automation and more rapid ow-through reaction processes (ca. 1 min) that will become competitive with other literature reports. 28 Our system also offers the signicant advantage of working at relatively low temperatures.
Finally, having developed a simple ow-through device capable of quick ltration we wondered if this could also be used for Pd scavenging. A hydrogel (made from 3.1 mg of DBS-CONHNH 2 and 0.75 mL of deionised water) was prepared in a 1 mL syringe and placed in the incubator at 50 C. A solution of PdCl 2 (0.7 mL, 2.86 mM) was added portion-wise to the top of the gel and allowed to ow through it. Aer complete ltration (ca. 20 min), the gel was further washed with 0.8 mL of deionised water. Aer this experiment, the colour of the gel had changed from transparent to orange indicating embedding of PdNPs (Fig. S9 †). The collected ltrate was colourless, and was mixed with 3 mL of EtOH and then directly used as a solvent for the Suzuki coupling between 4-iodotoluene and phenylboronic acid. However, aer 24 hours we did not detect any product. This implies that effectively all of the Pd from the initial solution (500 ppm) was successfully scavenged in real time (20 min) using this very simple ow-through device. We suggest this approach has real potential for cleaning up Pd from waste streams or pharmaceutical products.

Conclusions
In conclusion, we have shown that DBS-CONHNH 2 is capable of converting waste-to-wealth by scavenging palladium from solution, converting it in situ and without external reducing agent into nanoparticle form, and that the resulting gels then have the capacity to catalyse Suzuki-Miyaura cross-coupling reactions. Scavenging was able to remove palladium from solution down to <0.04 ppm levelsimportantly far below the acceptable levels in pharmaceutical/medical products, suggesting potential uses for these gels in the clean-up of pharmaceutical processes and products. Suzuki-Miyaura reactions were also performed using green solvents (EtOH/H 2 O) with benign bases (K 2 CO 3 ) and avoided the need for inert atmosphere conditions. Excellent yields together with high stability and easy recyclability make these gels efficient catalysts for this type of reaction. Limited Pd leaching was observed, with concentrations below the tolerance for pharmaceuticals, suggesting potential applications of these gels in high-value pharmaceutical synthesis. Furthermore, the use of agarose to provide the hybrid gels with physical robustness means that these catalytically-active gels can easily be dosed into reactions, recovered at the end by 'shing out', and reused in further reactions. Moreover, DBS-CONHNH 2 gels could be simply used in ow-through mode, giving rapid full conversion of reagents into products with easy purication. In summary, this research demonstrates that DBS-CONHNH 2 is an effective way of scavenging 'waste' palladium and converting it into catalytic gelphase 'wealth' capable of efficient, environmentally-friendly Suzuki-Miyaura reactions.

Conflicts of interest
There are no conicts to declare.