Heteroatom modified polymer immobilized ionic liquid stabilized ruthenium nanoparticles: efficient catalysts for the hydrolytic evolution of hydrogen from sodium borohydride

Ruthenium nanoparticles stabilised by polymer immobilized ionic liquids catalyse the hydrolytic release of hydrogen from sodium borohydride. The composition of the polymer influences performance and ruthenium nanoparticles stabilised by an amine-decorated imidazolium-based polymer immobilised ionic liquid (RuNP@NH 2 -PIILS) was the most efficient with a maximum initial turnover frequency (TOF) of 177 mole H2 . mol Ru (cid:0) 1 .min (cid:0) 1 , obtained at 30 ◦ C with a catalyst loading of 0.08 mol%; markedly higher than that of 69 mol H2 . mol Ru (cid:0) 1 .min (cid:0) 1 obtained with 5 wt% Ru/C and one of the highest to be reported for a RuNP catalyst. The apparent activation energy (Ea) of 38.9 kJ mol (cid:0) 1 for the hydrolysis of NaBH 4 catalysed by RuNP@NH 2 -PIILS is lower than that for the other polymer immobilized ionic liquid stabilised RuNPs, which is consistent with its efficacy. Comparison of the initial rates of hydrolysis in H 2 O and D 2 O catalysed by RuNP@NH 2 -PIILS gave a primary kinetic isotope effect ( k H / k D ) of 2.3 which supports a mechanism involving rate limiting oxidative addition of one of the O-H bonds in a strongly hydrogen-bonded surface-coordinated [BH 3 H (cid:0) ] — -H 2 O ensemble. The involvement of a surface-coordinated borohydride is further supported by an inverse kinetic isotope effect of 0.65 obtained from a comparison of the initial rates for the hydrolysis of NaBH 4 and NaBD 4 under the conditions of catalysis i.e., at a high hydride/catalyst mole ratio. Interestingly though, when the comparison of the initial rates of hydrolysis of NaBH 4 and NaBD 4 was conducted in dilute solution with a hydride/catalyst mole ratio of 1 a kinetic isotope effect ( k H / k D ) of 2.72 was obtained; this would be more consistent with concerted activation of both an O-H and B-H bond in the rate limiting step, possibly via a concerted oxidative addition-hydride transfer in the surface-coordinated hydrogen-bonded ensemble. Catalyst stability and reuse studies showed that RuNP@NH 2 -PIILS retained 71% of its activity over five runs; the gradual drop in the initial TOF with run number appears to be due to passivation of the catalyst by the sodium borate by-product as well as an increase in viscosity of the reaction mixture rather than leaching of the catalyst.


Introduction
There is an increasing urgency to identify alternative energy sources to fossil fuels in order to meet the need to supply sustainable, clean energy as well as reduce greenhouse emissions to mitigate rising global temperatures, extreme and fluctuating weather patterns, and the negative impact on the earth's ecosystem [1][2][3][4].To this end, hydrogen is an attractive energy carrier as a source of clean efficient power in stationary, portable and transport applications [5] as it has a high energy density (142 MJ⋅kg − 1 vs 54 MJ⋅kg − 1 for natural gas) as well as the potential to be generated in high purity from water splitting where the only by-product is oxygen [6][7][8][9][10][11][12].However, hydrogen is a flammable gas which forms potentially explosive environments and, as such, there are significant safety concerns over its storage and transportation; moreover, compression and liquefaction of hydrogen are energy intensive processes.The use of hydrogen storage materials is one of the most promising solutions as they are stable and safe to handle and would allow for the generation of hydrogen on site [6,[13][14][15][16][17][18][19][20][21][22][23][24][25].To this end, sodium borohydride has appropriate credentials for use as a storage material as it has a high stability and a high hydrogen content (10.8 wt %) and is nontoxic, inexpensive and water soluble (Eq. 1) [6,13b,c,j,k, 25-32].
As the thermal decomposition of NaBH 4 requires temperatures in excess of 400 • C and its hydrolysis in water is slow, considerable effort has been dedicated to developing cost-effective catalysts that can achieve the rapid and controllable release of hydrogen that will be required for this technology to become commercially viable.While homogeneous catalysts have been shown to facilitate the solvolysis of hydrogen-rich boron compounds [33][34][35][36][37][38], noble metal nanoparticles (NPs) have recently attracted considerable attention as the hydrogen generation rate can be controlled through their size, morphology and environment and the catalyst can be recovered and reused in much the same manner as a conventional heterogeneous catalyst [39][40][41][42][43].While the high activity obtained with small nanoparticles is due to their high surface area to volume ratio and the large number of active sites, they are unstable with respect to aggregation to less reactive species which limits their practical applications [44][45], for example, integration into hydrogen-based fuel cells for use in vehicles and portable electronic devices [46][47][48].One potential solution to overcome aggregation under conditions of catalysis has been to stabilise the nanoparticles by encapsulation into a support such as porous carbon structures [49][50][51][52][53][54][55][56][57][58][59][60], zeolites [61][62][63][64][65], mesoporous silicas [66][67][68], porous organic polymers [69][70], metal organic frameworks [71][72][73][74][75][76][77] and, most recently, dendrimers [78][79][80].Additional benefits of this strategy include control of the growth and morphology due to the confinement [81][82][83][84][85][86][87], modification of their properties through surface-support interactions [88][89][90][91][92][93] and incorporation of functionality to affect synergy, for instance, bimetallic nanoparticles [94][95][96][97].At present, the most efficient supported NP catalyst for the hydrolysis of sodium borohydride is based on RuNPs confined in zeolite-Y; this system gave a turnover frequency of 550 mol H2 .molRu − 1 .min− 1 [64].Ionic liquids have also been used for the stabilization of nanoparticles [98][99][100][101]; however, the weak electrostatic interactions involved do not always provide sufficient stabilisation to prevent aggregation under the conditions of catalysis [102][103].One possible approach to improve nanoparticle stability has been to introduce a heteroatom donor such as a phosphine, amine, nitrile, ether, or thiol that can supplement this weak stabilization by forming a covalent interaction to the nanoparticle surface [104].This approach has proven successful with significant improvements in catalyst stability and performance; for example, palladium nanoparticles stabilised by a phosphine-functionalised imidazolium-based ionic liquid are markedly more efficient hydrogenation catalysts than their unmodified counterparts [105][106][107][108][109] while RuNPs stabilised by a phosphine-functionalised ionic liquid exhibited a solvent dependent chemoselectivity for the hydrogenation of aromatic ketones as reactions performed in ionic liquid were highly selective for reduction of the carbonyl group whereas the use of water as the solvent resulted in hydrogenation of both the carbonyl and the arene.Moreover, the phosphine was shown to exert a marked influence on catalyst efficiency as the corresponding phosphine-free RuNP catalyst was markedly less selective in both solvents [110][111].However, even though this strategy has been shown to improve catalyst performance, functional ionic liquids are prohibitively expensive as a bulk solvent, leaching contaminates the product and recovery, and purification of the ionic liquid can be difficult, which has limited their implementation.
These issues have been addressed by grafting ionic liquids onto supports such as mesoporous silica, polymers, and MOFs on the basis that the resulting material would stabilise the nanoparticles in much the same manner as an ionic liquid, while the covalent attachment would prevent leaching of the ionic liquid, facilitate separation and recovery of the catalyst, and reduce the amount of ionic liquid, as the catalyst would be confined within the support [112][113][114][115][116][117].Polymers are particularly attractive supports as their modular construction would enable the hydrophilicity, ionic microenvironment, charge density and redox properties to be modified in a rational manner, additional functionality to be introduced and the composition and stoichiometry of the metal precursors to be defined to facilitate access to synergistic bi-and trimetallic nanoparticles.We have recently been exploring this approach and developed heteroatom donor-decorated polymer-immobilised ionic liquids, reasoning that the heteroatom donor could influence the size, size distribution and morphology of the nanoparticles as well as modify their surface electronic structure and, thereby, modulate their efficacy as catalysts.In this regard, there have been an increasing number of reports of the beneficial effect of ligands on the performance of heterogeneous nano-catalysts, which have been attributed to steric, electronic and solubility factors [118].Our early studies showed that palladium nanoparticles immobilized on a polyethylene glycol-modified phosphine-modified PIIL is a remarkably efficient catalyst for aqueous phase Suzuki-Miyaura cross-couplings [119], the chemoselective hydrogenation of α,β-unsaturated ketones, nitriles and esters, [120] and the hydrogenation of nitroarenes [121].Moreover, gold nanoparticles stabilized by a phosphine oxide-modified polymer immobilised ionic liquid catalyses the highly selective reduction of nitroarenes to afford N-arylhydroxylamines and azoxyarenes [122] and the corresponding ruthenium nanoparticles catalyse the aqueous phase hydrogenation of aryl and heteroaryl ketones and levulinic acid with remarkable efficacy and selectivity [123].
While support-grafted ionic liquids have been used to stabilise catalysts for a wide range of transformations, there appear to be only two reports of their use to support nanoparticle catalysts for the hydrolytic evolution of hydrogen from hydrogen-rich boron derivatives, which is somewhat surprising as polymer immobilised ionic liquids are functional and tuneable supports for molecular and nanoparticle catalysts.An imidazolium-based organic polymer has recently been used to prepare highly dispersed ultrafine AuPd alloy NPs for the hydrolytic release of hydrogen from ammonia borane which outperformed both its monometallic counterparts [124] and we have recently reported that phosphine decorated polymer immobilized ionic liquid stabilized PtNPs are highly efficient catalysts for the hydrolytic generation of hydrogen from NaBH 4 [125].This study has now been extended to investigate the efficacy of phosphine oxide and amine-decorated polymer immobilised ionic liquid stabilised RuNPs as catalysts for the hydrolysis of NaBH 4 on the basis that the heteroatom donor could disrupt the key hydrogen-bonded surface-coordinated ensemble between the acidic hydrogen of water and the hydridic hydrogen of borohydride and thereby influence catalyst performance.Herein, we report the results of a comparative study to explore the influence of polymer composition on catalyst performance and reveal that that RuNPs stabilised by an amino-modified polyionic liquid outperform their phosphine oxide-decorated and unmodified counterparts.Kinetic studies in combination with deuterium isotope effects have been used to probe the mechanism and a tandem hydrogenation of 1,1-diphenylethene with hydrogen generated from the catalytic hydrolysis of NaBH 4 in D 2 O gave a mixture of isotopologues resulting from reversible β-hydride elimination/re-insertion at a surface Ru-D competing with reductive elimination.

Materials
All reagents were purchased from commercial suppliers and used without further purification, RuCl 3 .3H 2 O 99.9% (PGM basis) was purchased form Alfa Aesar (47182) and polymers 1a-f were prepared as previously described and their purity confirmed by 1 H and 13 C{ 1 H} NMR spectroscopy and elemental analysis.Ethanol was distilled over iodine activated magnesium with a magnesium loading of 5.0 g L − 1 and diethyl ether from Na/K alloy under an atmosphere of nitrogen.

Preparation of catalysts 2a-f 2.2.1. Synthesis of RuNP@PIILS (2a)
To a round bottom flask charged with 1a (4.0 g, 6.5 mmol) and ethanol (100 mL) was added a solution of RuCl 3 ⋅3H 2 O (1.3 g, 6.5 mmol) in ethanol (20 mL).The resulting mixture was stirred vigorously for 5 h at room temperature after which time a solution of NaBH 4 (2.0 g, 52.0 mmol) in water (10 mL) was added dropwise and the suspension stirred for an additional 18 h before concentrating to dryness under vacuo.The crude black solid was triturated with cold acetone (2 × 100 mL) then washed with water (100 mL) followed by ethanol (2 × 40 mL) to afford a black solid that was recovered from the washings via centrifugation followed by filtration through a frit.The final product was rinsed with ether until a fine black powder was obtained which was dried under vacuum to afford 2a in 87% yield (4.06 g).ICP-OES data: 5.85 wt% ruthenium and a ruthenium loading of 0.58 mmol•g − 1 .

Kinetic studies 2.3.1. Ruthenium nanoparticle-catalyzed hydrolysis of sodium borohydride
Comparative catalytic hydrolysis reactions were conducted in water at the appropriate temperature in a thermostated 50 mL round bottom flask.In a typical experiment, a flask charged with a stir bar, catalyst 2af (0.2 mol%) and NaBH 4 (0.021 g, 0.57 mmol) and fitted with a gas outlet and connected to the top of an inverted water-filled burette designed to monitor the progress of the reaction by measuring the volume of water displaced with time.The flask was stabilised at 303 K and the reaction was initiated by adding water (2 mL) and the system was immediately sealed by replacing the gas outlet; the tap to the water filled burette was then opened, the time zero volume recorded, and the water displacement monitored.The optimum activity for each catalyst was determined by varying the catalyst loadings between 0.08 and 0.32 mol % at 303 K and measuring the hydrogen produced as a function of time.Kinetic studies were also conducted according to the protocol described above using the following catalyst loadings: 0.26 mol% 2a, 0.32 mol% 2b, 0.45 mol% 2c, 0.11 mol% 2d, 0.16 mol% 2e and 0.32 mol% 2f for a range of temperatures (294 K, 298 K, 303 K, 308 K and 313 K) and the corresponding activation energies (E a ) were determined from an Arrhenius plot of the initial rate against 1/T.

Study of the catalytic efficiency as a function of the concentration of NaOH
The effect of the concentration of NaOH on catalyst efficacy was explored by conducting catalytic hydrolysis reactions at 303 K in 2 mL of alkaline 0.28 M NaBH 4 (0.021 g) across a range of sodium hydroxide concentrations (i.e.[NaOH] = 0.035, 0.07, 0.14, 0.28, 5.0, 10, 50, mM) catalyzed by 0.26 mol% 2a (0.0025 g) and monitoring the gas evolution.

Catalyst recycle studies
Recycle studies were performed at 303 K as described above using mol% 2a (0.0193 g, 0.0114 mmol) and 2e (0.0335 g, 0.0114 mmol) to catalyze the hydrolysis of sodium borohydride (0.021 g, 0.57 mmol) in water (20 mL).The progress of the reaction was monitored as described above and when the hydrolysis was complete an additional portion of fresh sodium borohydride (0.021 g, 0.57 mmol) was added, and the procedure repeated.After the 5th run samples of the catalysts were isolated and analysed by TEM.

Catalyst recycle studies in the presence of buffer
A borate-buffered solution was prepared by dissolving Na 2 B 4 O 7 ⋅10H 2 O (9.53 g, 25 mmol) and NaCl (4.39 g, 75 mmol) in distilled water (900 mL) in a volumetric flask.When the borate was completely dissolved the pH of the solution was adjusted to 7.2 by gradual addition of boric acid (20.99 g, 0.34 mol); the solution was then made up to one liter.Recycle studies were conducted by adding NaBH 4 (0.021 g, 0.57 mmol) to a flask containing 1 mol% 2e (0.0165 g, 0.0056 mmol) and 20 mL of the aqueous borate buffer solution.The flask was maintained at 303 K and the progress of the reaction was monitored as described above.When the hydrolysis was complete an additional portion of fresh sodium borohydride (0.021 g, 0.57 mmol) was added, and the procedure repeated for comparison with the recycle study described above in the absence of buffer.

Hot filtration tests
Hot filtration studies were conducted at 303 K following the protocol described above using either 0.2 mol% 2a (0.0019 g) or 0.16 mol% 2e (0.0026 g) to catalyze the hydrolysis of sodium borohydride (0.021 g, 0.57 mmol) in water (2 mL).The progress of the reaction was monitored as a function of time and the mixture filtered through a 0.45 μm syringe filter when the conversion reached ca.50% (10 min for 2a and 7.75 min for 2e), after which the burette assembly was reconnected, and the gas evolution monitored for a further 30 min.In a complementary procedure, a hydrolysis reaction that had reached completion was filtered through a 0.45 μm syringe filter and an additional portion of NaBH 4 (0.021 g, 0.57 mmol) added to the filtrate and the gas evolution monitored.

Catalyst poisoning study
A flask was charged with 2 mol% catalyst (2a 0.0186 g; 2e, 0.0335 g), water (20 mL) and sodium metaborate (0.0765 g, 0.57 mmol) and the resulting mixture stirred at 303 K for the predetermined time (t = 0 min, 20 min, 40 min, 60 min) to investigate whether the pre-stirring time influences catalyst efficacy.After pre-stirring for the allocated time, the reaction was initiated by addition of the NaBH 4 (0.021 g, 0.57 mmol) and the rate of hydrogen evolution quantified by measuring the volume of water displaced with time.

Tandem hydrogenation of 1,1-diphenylethene
Tandem hydrogenations were performed using two Schlenk flasks connected through tubing.One of the flasks was charged with a stir bar, either NaBH 4 (0.042 g, 1.11 mmol) or NaBD 4 (0.046 g, 1.11 mol) and 0.26 mol% 2e (0.0025 g) and the hydrolysis started by addition of either D 2 O (2 mL) or H 2 O (2 mL).The reaction flask was immediately stoppered, isolated from the second flask by closing the stopcock and stirred for 70 min.The second Schlenk flask was charged with 1,1-diphenylethene (0.180 g, 1.00 mmol), 0.5 mol% Pd/C and either CH 3 OH (2 mL) or d 4 -methanol (2 mL).After 70 min the second flask was evacuated briefly before opening the connector to the hydrolysis flask.The reaction was allowed to stir at 303 K for 18 h before the solvent was removed and the residue analyzed by 13 C{ 1 H} NMR spectroscopy and GC-MS to establish the composition and quantify the distribution of isotopologues.

Catalyst synthesis, characterisation and RuNP-catalyzed hydrolysis of sodium borohydride
The polymers required for this study were prepared via radical polymerisation of the corresponding imidazolium-based ionic liquid monomer, either styrene, (4-vinylphenyl)methanamine or diphenyl(4-vinylphenyl)phosphine oxide and the corresponding imidazoliumbased ionic liquid cross-linker in the ratio x = 1.84, y = 1.0 and z = 0.16, as previously described [119][120][121][122][123]. Catalysts 2a-f were prepared by the wet impregnation of the polymer support with ruthenium trichloride to afford precursors with a 1:1 ratio of ruthenium to neutral monomer, followed by in-situ reduction of the ruthenium with NaBH 4; to afford the product as a fine black powder in high yield; the synthesis and composition of the polymers and the catalysts is shown in Fig. 1.The composition and purity of polymers 1a-f was determined using a combination of solution and solid state 13 C{H} and 31 P{H} NMR spectroscopy and elemental analysis while the loaded RuNP catalysts were characterised by solid state 13 C{H} and 31 P{H} NMR spectroscopy, infra-red (IR) spectroscopy, high resolution transmission electron microscopy (HRTEM), SEM, X-ray photoelectron spectroscopy (XPS) and inductively coupled plasma-optical emission spectroscopy (ICP-OES) (See Fig. 2 and the supporting information for full details).The ruthenium loadings in 2a-f were determined to be 0.18− 0.75 mmolg − 1 using ICP-OES.
The solid state 13 C{ 1 H} NMR spectra of 1a-f and 2a-f each contain resonances from δ 121 to 149 ppm, which correspond to the aromatic carbon atoms of polystyrene and the carbon atoms of the imidazolium ring, as well as signals between δ 10 and 51 ppm which belong to the methylene carbon atoms of the polystyrene backbone and the methyl group attached to the imidazolium ring.Additional signals at δ 71 and 59 ppm for 2b, 2d and 2f belong to the carbon atoms of the polyethylene glycol (PEG) chain and the terminal OMe, respectively, and a signal at δ 49 ppm for 2e and 2f is associated with the CH 2 NH 2 .The surface of the RuNP catalysts was characterised by X-ray photoelectron spectroscopy by analysing the Ru 3p region as the C 1s and Ru 3d region overlapped.For catalyst 2a, stabilised by unmodified imidazolium-based polymer, a Ru 3p 3/2 peak at 463.19 eV was assigned to RuO 2, and satellite features were fitted at 465.97 eV (Table S2 and Fig. 2a).The presence of RuO species is most likely due to surface oxidation of the pre-formed metallic Ru nanoparticles.The corresponding Ru 3p 3/2 peak for catalysts containing the phosphine oxide (2c and 2d) or amine (2e and 2f) was shifted to lower binding energy (462.56 and 461.37 eV for 2c and 2d, respectively and 462.83 and 462.89 eV for 2e and 2f, respectively) compared to the Ru 3p 3/2 binding energy of 463.19 eV for catalyst 2a (Table S2 and Fig. 2b-f).A shift to lower binding energy may be indicative of electron transfer from the heteroatom of the phosphine oxide or amine to the RuNPs.Catalyst 2d containing O=PPh 2 and PEG heteroatom donors gave the largest shift (-1.72 eV) in binding energy of the Ru 3p 3/2 peak (461.37 eV for 2d) relative to 2a.TEM micrographs of 2a-f revealed that the ruthenium nanoparticles were ultrafine and near monodisperse with average diameters between 1.6 and 2.8 nm; representative micrographs and the corresponding distribution histograms based on the sizing of >100 particles for 2a-f are shown in Fig. 2. SEM images revealed that the catalyst materials were far more granular than their polymeric counterparts, which appeared largely smooth.
The hydrolysis of sodium borohydride was identified to investigate the efficacy of catalyst 2a-f on the basis that PEG-modified 'click'-dendrimer stabilised noble and bimetallic metal nanoparticles catalyse this reaction with promising initial TOFs and as such would provide a formative benchmark for comparative evaluation.Preliminary catalyst testing was conducted using recent literature protocols as a lead [75,78]; reactions were initially performed at 303 K using 0.2 mol% of 2a-f to catalyse the hydrolysis of a 0.28 M solution of sodium borohydride (Fig. 3a, b).The reaction was monitored by quantifying the amount of hydrogen liberated as a function of time using water displacement from an inverted burette assembly and all data were corrected by subtracting the background hydrogen generated over the same time under identical conditions.Hydrogen evolution started immediately with no induction period which is consistent with the metallic state of the ruthenium.Under these conditions, RuNP@NH 2 -PIILS (2e) gave the highest initial TOF of 135 mole H2 .molRu − 1 .min− 1 and reached 92% conversion after min, whereas its PEGylated counterpart RuNP@NH 2 -PEGPIILS (2f) was less active with a slightly lower TOF of 117 mole H2 .molRu − 1 .min− 1 .
Removal of the amino-group from either of these systems resulted in a reduction in the activity with RuNP@PIILS (2a) and RuNP@PEGPIILS min − 1 , respectively (Fig. 3b).For comparison, 0. .min− 1 when the reaction was performed in dilute solution (10 mL) with a reduced catalyst loading of 0.08 mol%.A series of baseline hydrolysis reactions conducted by substituting catalysts 2a-f with their corresponding polymers 1a-f confirmed that the RuNPs were essential for catalysis as the gas evolution did not exceed the background reaction under the same conditions.
As there is no clear correlation between the efficacy of catalysts 2a-f and the nanoparticle size, further studies will be conducted to explore the surface electron density of the RuNPs as a function of the support and to investigate whether the amine influences the hydrogen bonded surface ensemble responsible for substrate activation or improves the dispersibility of the catalyst in the reaction mixture and thereby access to the active site.To this end, amine-modified supports have previously been reported to improve the performance of nanoparticle catalysts compared with the corresponding unmodified catalyst.For example, ruthenium nanoparticles stabilised within the pores of amine-modified MIL-53 (MIL-53(Al)-NH 2 ) is a significantly more active catalyst for the dehydrogenation of amine-borane than its unmodified counterpart, MIL (Al)-53; this was attributed to the formation and stabilization of ultrasmall RuNPs [76].There are also numerous additional reports of the beneficial effect on catalyst performance of incorporating an amine onto the surface of a support.For instance, a marked improvement in the activity and selectivity of platinum nanowires for the partial hydrogenation of nitroarenes to N-phenylhydroxylamine [126][127], an enhancement in the activity of RuNPs for the hydrogenation of levulinic acid to γ-valerolactone [128], an improvement in activity for the transfer hydrogenation of nitroarenes catalysed by RuNP confined in an amine-modified porous organic polymer [129], an increase in activity for the PtNP-catalysed hydrogenation of quinoline [130], improvements in activity and selectivity for the Pt/Co and PdNP catalysed semi-hydrogenation of alkynes [131][132][133], and highly selective reduction of the carbonyl in cinnamaldehyde with MOF-confined Pt nanoclusters [134].
Although a comparison of the efficacy of 2a-f with literature reports of other supported ruthenium nanoparticles should be treated with caution because of the vastly disparate experimental conditions and protocols employed to collect data, the initial TOF of 177 mole H2 .molRu − 1 .
As the highest TOF was obtained with 2e, a thorough study of the reaction kinetics together with deuterium isotope effects, recycle experiments and a tandem reaction using the liberated hydrogen for the tandem hydrogenation of 1,1-diphenylethene with deuterium labelling was undertaken, details of which are discussed herein; for comparison, full details of the corresponding experiments with catalysts 2a-d and 2f are provided in the supporting information and discussed in context where appropriate.There have been numerous reports of an enhancement in activity for the metal nanoparticle catalysed hydrolytic evolution of hydrogen from sodium borohydride and amine borane in the presence of added base.For example, Astruc has reported a marked increase in the initial TOF for the hydrolysis of NaBH 4 catalysed by click dendrimer-supported RuNPs from 80 mole H2 .molRu − 1 .min− 1 to mole H2 .molRu − 1 .min − 1 in the presence of 0.2 M NaOH; an increase in TOF was also observed for a host of other catalysts including Rh, Au, Pd, Co, Ni, Fe and Co nanoparticles with the exception of PtNPs which experienced a strong negative effect [78].Significant enhancements in TOF were also obtained for the hydrolysis of hydrogen-rich boron compounds with MNP@ZIF-8 (M = Ni, Co), NiPtNP@ZIF-8 and CoPtNP@dendrimer nanocatalysts in the presence of NaOH [72,73,75,80].This enhancement has been attributed to coordination of the hydroxide to the nanoparticle surface which increases the electron density and facilitates activation of the O-H bond; in contrast, Pt is an electron-rich metal and highly reactive towards oxidative addition and as such the hydroxide ions occupy surface active sites and prevent substrate coordination.Such a large enhancement in activity for a dendrimer-stabilised RuN-P-based catalyst prompted us to study the efficiency of 2a for the catalytic hydrolysis of NaBH 4 as a function of the concentration of sodium hydroxide; reactions were conducted using 0.26 mol% of 2a to catalyse the hydrolysis of alkaline solutions of 0.28 M NaBH 4 with sodium hydroxide concentrations ranging between 0.035 mM to 100 mM (Fig. 4).There was no apparent variation in the initial TOF at low concentrations of NaOH (< 0.035 mM) while the TOFs decreased gradually at concentrations above 0.07 mM; this decrease became more dramatic when the sodium hydroxide concentration reached 5 mM and the initial TOF eventually dropped from 136 mole H2 .molRu − 1 .min− 1 in the absence of sodium hydroxide to 39 mole H2 .molRu − 1 .min− 1 in a 100 mM NaOH solution of NaBH 4 .To this end, there have been several reports of a decrease in the hydrogen generation activity with increasing NaOH concentration (1-10 wt% NaOH) for the ruthenium-catalysed hydrolysis of NaBH [138][139][140][141][142], which were attributed to strong interactions between the hydroxide ions and water decreasing the available free water needed for the hydrolysis of NaBH 4 [138].However, it is interesting to note that Kinetic studies were subsequently undertaken to determine the temperature dependence of the rate and obtain activation parameters for the hydrolytic release of hydrogen from NaBH 4 for a comparison with related systems reported in the literature.A set of reactions were conducted to monitor the hydrolysis of a 0.28 M solution of NaBH 4 as a function of time to determine the initial rates across a range of temperatures from 294 K to 313 K.The apparent activation energies (Ea) for the hydrolysis catalysed by 2a-f, determined from an Arrhenius plot of lnk against 1/T (lnk = lnA -Ea/RT) using the initial rates calculated from the linear slope of the graph, ranged from 38.9 kJ mol − 1 to 51.8 kJ mol − 1 (Fig. 5a-b and Fig. S1 in the supporting information).These values lie within the range reported for the hydrolysis of NaBH 4 with other RuNP catalysts including 35 kJ mol − 1 for RuNPs stabilised in the framework of Zeolite-Y [64], 41 kJ mol − 1 for water-dispersible, acetate-stabilized RuNPs [149], 36 kJ mol − 1 for RuNPs confined in ZIF-67 [77], 47 kJ mol − 1 for RuNPs immobilised by the anion exchange resin IRA-400 [150] and 41.8 kJ mol − 1 for ruthenium immobilised on Al 2 O pellets [151], but slightly lower than 61.1 kJ mol − 1 for RuNPs supported on amine-modified graphite [139], 56.0 kJ mol − 1 for RuNP@IRA-400 [138], 58.2 kJ mol − 1 for Ru(acac) 3 [152] and 66.9 kJ mol − 1 for ruthenium supported on carbon [153].There does not appear to be a correlation between the activation energies and the initial rates which may be attributed to variations in the number of active sites or their availability as this determines the pre-exponential factor (A) [76,154].
The hydrogen release was next investigated as a function of the concentration of 2e across a range of catalyst loadings from 0.12 mol% to 0.28 mol% in 0.28 M NaBH 4 (Fig. 6a) and the logarithmic plot of the initial hydrogen generation rate versus catalyst concentration gave a straight line with a slope of 1.04 (Fig. 6b), indicating that the hydrolysis of NaBH 4 is first order with respect to the catalyst.Similarly, the corresponding slopes for the logarithmic plots obtained with catalyst 2ad and 2f varied between 0.70 and 1.04, which are all consistent with first order kinetics; full details are presented in Fig. S2 in the supporting information.This data is also consistent with recent reports of noble metal nanoparticle-catalysed hydrogen generation from hydrogen-rich boron derivatives including a slope of 0.73 for RuNPs confined in Zeolite-Y [64], 0.94 for RuNPs stabilized by polyvinylpyrrolidinone [155], 1.06 for Ru(acac) 3 [152], 1.17 for porphyrin-stabilised RuNPs [156], 0.85 for PtCoNP@dendrimer [78], and 0.82 for Ni 2 Pt@ZIF-8 [73].The variation in the rate of hydrolysis of NaBH 4 as a function of the substrate concentration was also investigated using catalyst 2e.As the order of reaction with respect to NaBH 4 has been reported to depend on the amount of NaBH 4 in solution (i.e. the NaBH 4 :catalyst ratio), changing from 1 to 0 as the concentration of NaBH 4 increases [145], kinetic data was obtained by conducting a series of reactions with 0.026 mmol of catalyst 2e and varying the initial concentration of NaBH 4 from 0.066 mM to 0.52 mM as these amounts correspond to catalyst:hydride ratios between 2:1 and 1:4 (Fig. 7).Such low catalyst/hydride mole ratios were used to avoid the BH 4 − induced dynamic saturation of the active sites on the catalyst surface which would give zero order kinetics; under these conditions the surface is not completely covered by NaBH and there are active sites.The slope of 1.02 obtained from the logarithm plot of hydrogen generation rate versus concentration of NaBH 4 confirms that the hydrolysis is first order in substate, which undergoes rate limiting diffusion on the catalyst surface.Under the same conditions, slopes of 1.02 and 1.01 were also obtained with catalysts 2a and 2f, respectively, which are both consistent with first order kinetics; see  Conditions: 0.57 mmol NaBH 4 (0.021 g), 0.26 mol% 2a, 0.32 mol% 2b, 0.45 mol% 2c, 0.11 mol% 2d, 0.16 mol% 2e and 0.32 mol% 2f in water (2 mL).Each volume is an average of three runs.Initial rate = mol H2 ⋅min − 1 .Fig. S3 in the electronic supporting information.First order kinetics with respect to NaBH 4 have previously been reported for ruthenium on carbon [142], palladium on carbon [157] and Pd and Pt dispersed on functionalised surfaces of carbon nanotubes [158] when reactions were conducted at low concentrations of NaBH 4 .A similar study conducted with catalyst 2e at much higher catalyst/hydride mole ratios between 1:625 and 1:2500 gave a slope of 0.26 which is indicative of zero order kinetics due to saturation of the active sites on the catalyst surface during the reaction (Fig. S4 in the supporting information), as described by Patel [145].A slope of 0.17 was also obtained using catalysts 2d which is also consistent with zero order kinetics; similar kinetics have previously been described for ruthenium nanoclusters [159], Ru supported on IRA 400 [150] and ruthenium on carbon [153].

Kinetic isotope effects
The kinetic isotope effect (KIE) is a valuable tool for elucidating information about the rate limiting step (RLS) of a reaction that has been routinely used to probe the catalytic hydrogen generation from borohydride and amine borane (AB) [160,72,79,80].While the reaction kinetics are complicated and the mechanism still not fully understood [42] it is clear that both NaBH 4 and ammonia-borane are hydride donors and provide one of the two hydrogen atoms of the derived hydrogen gas while water provides the other in the form of a proton [41,43] and that the rate determining step involves activation of one of the O-H bonds of water, as measured by the large primary KIE obtained when the hydrolysis is performed in D 2 O instead of H 2 O [78,79,80,83,161,162].Activation of an O-H bond has been proposed to occur via oxidative addition involving a hydrogen-bonded ensemble between a surface-coordinated borohydride and a water proton; the hydrogen could then be liberated either via reductive elimination between a borohydride-derived NP-H and the water-derived NP-H (Fig. 8, pathway a-c) or a concerted σ-bond metathesis-like process between a surface coordinated [BH 4 ] − and a water-derived NP-H (Fig. 8, pathway d-e), which may be facilitated by hydroxide.Alternatively, the protonic and hydridic hydrogen atoms may be transferred to the nanoparticle surface by oxidative addition of both the O-H and B-H bonds, respectively, to afford a dihydride that would generate hydrogen and BH 3 -OH via reductive elimination (Fig. 8, pathway f-g), as proposed by Astruc for the CoNP@ZIF-8 catalysed hydrolysis of NaBH 4 [75].While the pathways described in Fig. 8 are all initiated by oxidative addition of the O-H bond  of water via a hydrogen-bonded ensemble involving a surface-coordinated borohydride, Jagirdar [163] and Ma [164] have suggested that activation of the O-H bond and generation of H 2 could occur via a hydrogen-bonding interaction between a surface adsorbed water and a surface hydride generated via rapid hydride transfer from NaBH 4 to the NP surface.
The role of H 2 O in the hydrolysis of NaBH 4 catalysed by 2e was explored by conducting the reaction in D 2 O and monitoring the hydrogen evolution as a function of time to determine the KIE.Reactions were conducted under the conditions of catalysis i.e. 0.16 mol% of 2e was used to catalyse the hydrolysis of 2 mL of a 0.28 M solution of NaBH 4 at 30 • C. A comparison of the efficacy of 2e as a catalyst for the hydrolysis of NaBH 4 in H 2 O and D 2 O revealed that the reaction was more rapid in H 2 O than in D 2 O with a primary kinetic isotope effect (k H / k D ) of 2.31 (Fig. 9a); similar values of k H /k D were obtained with catalysts 2a (k H /k D = 1.76) and 2d (k H /k D = 1.53) and the corresponding data is presented in Fig. S5a-b in the supporting information.This value is comparable to the solvent isotope effect of 2.25 obtained by Astruc for the gold-ruthenium nanoalloy catalysed visible light-accelerated hydrolytic dehydrogenation of NaBH 4 and amine-borane [165] as well as 1.8 determined in a detailed kinetic analysis of the platinum-catalysed hydrolysis of NaBH 4 in alkaline media [162], 2.3 for dendrimer-stabilised RhNPs [79], 2.4 for PtCo@dendrimer [80] and 2.49 for NiNP@ZIF-8 [72] and supports a mechanism with rate limiting cleavage of an O-H bond of water in a surface-coordinated hydrogen-bonded ensemble of the type described above and shown in Fig. 8.The same comparison of initial rates between reactions conducted in H 2 O and D 2 O under stoichiometric conditions using 26 μmol of 2e for the catalytic hydrolysis of 200 mL of a 0.13 mM solution of NaBH 4 at 30 • C (catalyst:NaBH 4 ratio of 1:1) gave a primary kinetic isotope effect of 1.7 (Fig. S6d in the supporting information), which is also consistent with rate limiting oxidative addition of water.However, this KIE does not distinguish between a rate limiting step in which a surface coordinated NaBH 4 --HOH ensemble activates an O-H bond towards oxidative addition through a hydrogen-bonding interaction to afford a water-derived metal hydride and a surface-coordinated borohydride, such as that shown in Fig. 8 pathway a, and concerted activation of both the B-H and O-H bonds in a similar hydrogen-bonded ensemble; the latter process would most likely occur via oxidative addition of the O-H bond and rapid hydride transfer from the borohydride (Fig. 8 pathway  a-c) rather than oxidative addition of both the O-H and B-H bonds (Fig. 8, pathway f-g) as borohydrides are extremely potent transfer reagents.For the same reason, a subsequent σ-bond metathesis involving the surface-coordinated borohydride and the water-derived RuNP hydride would also be unlikely (Fig. 8, pathway e).
Thus, the mechanism was further probed by comparing the rates of hydrolysis of NaBD 4 and NaBH 4 catalysed by 2e at 30 • C. Analysis of the initial rates obtained for the hydrolysis of 200 mL of a 0.13 mM solution of NaBH 4 and NaBD 4 catalysed by 26 μmol of 2e, i.e., a substrate/ catalyst ratio of 1, gave a primary kinetic isotope effect (k H /k D ) of 2.72 (Fig. 9b).Reassuringly, comparable values were also obtained with catalysts 2a (k H /k D = 2.25) and 2d (k H /k D = 2.37), full details of which are presented in Fig. S6a-b in the supporting information.These values are comparable to that of 2.2 obtained for the visible light-accelerated H 2 evolution from NaBH 4 catalysed by a gold-ruthenium nanoalloy; which, together with a KIE of 2.5 obtained for the hydrolysis of NaBH 4 in D 2 O, was taken to indicate that both the O-H and B-H bonds were activated by the ruthenium atoms in the rate limiting step, most likely via concerted oxidative addition-hydride transfer, involving the surfacecoordinated hydrogen-bonded [BH 3 H − ]--H-OH ensemble, rather than oxidative addition of both the O-H and B-H bonds [75,165].Interestingly though, comparison of the rates obtained under the conditions of catalysis using 2e to catalyse the hydrolysis of 2 mL of 0.28 M solutions of NaBH 4 and NaBD 4 at 30 • C gave a KIE of 0.65 (Fig. 9c); similar values were also obtained with catalysts 2a (k H /k D = 0.87) and 2d (k H /k D = 0.85), full details of which are provided in Fig. S5d-f in the supporting information.These are inverse kinetic isotope effects and would be consistent with a surface-coordinated borohydride activating an O-H bond of water in the hydrogen-bonded ensemble prior to hydride transfer.

Tandem hydrogenation and deuterium labelling studies
The hydrogen liberated from the catalytic hydrolysis of NaBH 4 was used for the hydrogenation of 1,1-diphenylethene with various labelling experiments to determine the fate of the liberated hydrogen.In the first of these, the tandem reaction was conducted using 0.26 mol% 2a to generate hydrogen from a 0.28 M solution of NaBH 4 in D 2 O at 30 • C in a sealed tube; after 70 min the connector was opened to the second flask which contained 1,1-diphenylethene and 0.5 mol% Pd/C in d 4 -methanol and the resulting mixture was stirred for 18 h.Interestingly, analysis of the crude mixture by 1 H, 2 H and 13 C NMR spectroscopy and mass spectrometry revealed that a mixture of all eight isotopologues of 1,1diphenylethane had been generated (Scheme 1).Analysis of the methine region (δ 44.5 ppm) of the 13 C{ 1 H} NMR spectrum was used to identify and assign each of the isotopologues, which appear as a set of four singlets at δ 44.88, 44.81, 44.73, and 44.66 ppm corresponding to I, II, III, and IV, respectively, while V, VI, VII and VIII appear as a set of four 1:1:1 triplets at δ 44.46, 44.39, 44.31 and 44.24 ppm, respectively, resulting from a J CD of 19.5 Hz due to the deuterium atom attached to the methine carbon; the methyl group of these isotopologues has either zero, one, two, or three deuterium atoms.The experimental spectrum of the reaction mixture and the summed simulated spectrum of each isotopologue are shown in Fig. 10 (see Fig. S71 in the supporting information for full details of the simulated spectrum for each isotopologue).The summed simulated spectrum is remarkably similar to the experimental spectrum, which supports the assignment of the isotopologues and their relative proportions and confirms that the coupling constants, chemical shifts and line intensities and widths have been correctly determined.On the basis that the hydrogen generated from the hydrolysis of NaBH 4 in D 2 O should result from a water-derived proton and a borohydride-derived hydride, the deuterium incorporation for all isotopologues II-VIII should be one.To this end, the total deuterium incorporation of 1.3 is slightly higher than expected and could be due to H/D exchange either with the d 4 -MeOH on the Pd/C during the hydrogenation or from the generation of a mixture of HD and D 2 by exchange at the NP surface after O-D bond activation.A complementary experiment using hydrogen liberated from NaBH 4 /H 2 O for the hydrogenation of 1,1-diphenylethene in d 4 -methanol gave a total deuterium incorporation of 0.3, which confirms that H/D exchange occurs on the surface of the Pd/C; moreover, this deuterium incorporation corresponds to the excess of 0.3 above the total deuterium incorporation of one that was expected when the hydrogenation was performed in d 4 -MeOH with hydrogen generated from NaBH 4 /D 2 O.The hydrogenation was also performed in toluene with hydrogen generated from NaBH 4 in D 2 O to investigate exchange at the NP surface.Under these conditions, the total deuterium incorporation of 0.93 was close to one, indicating that H/D exchange at the NP surface is slow; a total deuterium incorporation of 1.76 was also obtained when the hydrogenation was performed in toluene using hydrogen generated from NaBD 4 in D 2 O, which is reassuringly close to the predicted value of two.Finally, the generation of minor amount of isotopologues containing -CHD 2 and -CD 3 (III, IV, VII and VIII) from each of these deuterium labelling experiments is consistent with H/D scrambling via facile reversible β-hydride elimination from a surface M-CPh 2 CH 2 D species, reinsertion of the resulting Ph 2 C=CHD into a surface M-D followed by reductive elimination from (D)HPd-CPh 2 CH 3-n D n (n = 2, 3); full details of the relative proportions of each isotopologue obtained from these labelling studies are summarised in the supporting information.A higher than stoichiometric incorporation of deuterium recently reported for the hydrogenation of styrene using 'HD' generated from the hydrolysis of tetrahydroxydiboron with D 2 O using quantum dot stabilised PtNPs was also attributed to facile reversible alkene insertion-extrusion involving metal-hydride/deuteride species [166].

Catalyst recycle and poisoning studies
Recycle studies were conducted with 2 mol% loading of 2e to investigate its activity profile during reuse and thereby its stability and longevity and potential for use in a scale-up system.The practical issues associated with separating and recovering a small amount of catalyst by filtration without loss of material after each run meant that it was not possible to perform a conventional recycle experiment.As such, a reuse experiment was undertaken by monitoring the hydrolysis until gas evolution was complete, the aqueous reaction mixture was then charged with a further portion of NaBH 4 and the gas evolution monitored; this sequence was repeated to map the catalyst efficacy against reaction time and reuse number.While the comparative conversions and TOFs shown in Fig. 11a, b were obtained during the first 2 min of the hydrolysis to enable a meaningful comparison between runs, complete conversions were obtained for each run within 4 min.The resulting gas evolutiontime profile and corresponding conversion-cycle number profile in Fig. 10.Plots of the methine region of the experimental 13 C{ 1 H} NMR spectrum and summed simulated 13 C{ 1 H} NMR spectrum for the eight isotopologues generated from the hydrogenation of 1,1-diphenylethene in d 4 -methanol using hydrogen generated from the catalytic hydrolysis of NaBH 4 in D 2 O using 0.16 mol % 2e.Fig. 11a, b shows a minor but gradual drop in conversion across five reuses, from 89% after 2 min in the first run to 78% after the same time in the 5 th run.The drop in catalyst activity in successive runs, defined as the percentage reduction in the initial TOF, shows that 2e retains 71% of its activity across five reuses (Fig. 11b, red); this is comparable to recycle studies reported for other noble metal nanoparticle catalysts including; RuNPs immobilised in ZIF-67 [77], PtCoNPs supported on carbon nanospheres [167], ruthenium nanoparticles immobilised within the pores of amine-functionalised MIL-53 [76], ruthenium supported on graphite [139], RuCo nanoclusters incorporated in PEDOT/PSS polymer [168], RuNP stabilized by polyvinylpyrrolidone, zeolite-confined RuNPs [64], click dendrimer-stabilized PtCo, Rh and Pt nanoparticles and gold-transition metal nanoalloys [72,73,78,79,80,165] and Ru-RuO 2 /C [141].
Sneddon et al. previously reported that the use of a borate buffered solution for the rhodium-catalysed release of hydrogen from ammonia triborane extended the catalyst lifetime such that Rh/Al 2 O 3 showed little change in the hydrogen release rate over 11 cycles [169].Following this lead, a preliminary comparative recycle hydrolysis conducted in freshly prepared aqueous borate buffer (pH maintained between 7. .min− 1 in the final run (Fig. 11d); even though this represents a 26% reduction in activity over the 5 cycles, it remains higher than the TOFs obtained in water under the same conditions.Interestingly, the data in Fig. 11c, d also shows that the conversion-time profile changes quite dramatically in successive cycles such that the conversion increases from 54% after min in the first run to 80% at the same time interval in the final run; in contrast, for reactions conducted in the absence of buffer, conversions decreased gradually in successive runs (Fig. 11b).A hydrolysis catalysed by 1 mol% 2e was also conducted in 0.34 M boric acid to provide a benchmark as the borate buffer solution was prepared with this concentration of boric acid and, under otherwise identical conditions, the initial TOF of 66 mole H2 .molRu − 1 .min− 1 was significantly lower than that obtained in the aqueous borate buffer solution (See Fig. S7 in the supporting information).Further studies are currently underway to identify an optimum buffer for this reaction and to develop an understanding of the changes in the conversion-time profile in consecutive runs as well as the origin of the enhancement in activity obtained when the catalysis is conducted in aqueous buffer.ICP-OES analysis of the aqueous reaction mixture recovered after the fifth run revealed that the ruthenium content was below the detection limit, suggesting that the reduction in activity was unlikely to be due to leaching of the ruthenium to generate a homogeneous species that was less active.Hot filtration studies were also conducted to explore whether soluble ruthenium species might be responsible for the gas evolution.Following a typical protocol, a hydrolysis reaction catalysed by 2 mol% 2e was filtered through a 45-micron syringe filter at ca. 50% conversion.The hydrogen liberated from the filtrate was monitored and corresponded to the background hydrolysis in the absence of catalyst (Fig. 12, blue line), indicating that the active species had been removed in the filtration i.e. it is heterogeneous, and that leaching does not generate active soluble ruthenium species.In a complementary hot filtration study a catalytic hydrolysis that reached completion was filtered through a syringe filter (0.45 μm) and a fresh portion of NaBH 4 added to the filtrate.The hydrogen liberated also corresponded to the uncatalyzed hydrolysis providing further support that the active species is heterogeneous (Fig. 12, orange line).TEM analysis of the catalyst isolated after the fifth run revealed that the ruthenium nanoparticles remained essentially monodisperse with a mean diameter of 1.8 ± 0.5 nm compared with 1.8 ± 0.6 nm for the freshly prepared catalyst (Fig. 12b) which suggests that agglomeration is not responsible for the drop in conversion with increasing use.
There have been several reports that the sodium metaborate tetrahydrate by-product generated during the hydrolysis of NaBH 4 deactivates the catalyst by adsorption on the surface [67,71,76,80,135,[170][171][172], although Wie has demonstrated that the activity of deactivated Ru on nickel foam catalyst can be partially replenished by washing the catalyst with deionised water and completely replenished by washing with HCl to remove the NaBO 2 [135].As such, a series of poisoning studies were undertaken to examine the influence of the by-product on catalyst performance; this involved pre-stirring an aqueous suspension of 2e with 100 equivalents of sodium metaborate prior to addition of NaBH 4 and monitoring the progress of the reaction as a function of the pre-stirring time.A 11 B NMR spectrum of a typical reaction solution confirmed that the tetrahydroxyborate anion B(OH) 4 was the sole by-product as the spectrum contained a single sharp resonance at δ 2.2 ppm [162,173]; no other species such as partially hydrolysed intermediates were detected.A comparison of the hydrogen evolution in the absence of NaBO 2 against the corresponding reaction with added NaBO 2 as a function of the pre-stirring time (Fig. 13a,b) confirms that the addition of metaborate passivates the catalyst.The conversions obtained after a reaction time of 2 min and the corresponding initial TOFs as a function of pre-stirring time reveal that the passivation is instantaneous as the TOF drops from 84 mole H2 .molRu .min− 1 as the pre-stirring time was increased to 60 min.
Finally, the formation of NaBO 2 can also be monitored by measuring the pH of the reaction solution as a function of time for the catalytic hydrolysis of a 0.028 M solution of NaBH 4 using 2 mol% of 2e.Fig. 14 shows that the pH of the reaction solution clearly maps to the conversion with a gradual increase from pH 8.3 at time = 0 min, recorded immediately after addition of the NaBH 4 , to pH = 11.1 after ca.2.5 min when the gas evolution had finished; for comparison a 0.028 M solution of NaBO 2 in the absence of catalyst or NaBH 4 has a pH of 11.30, which correlates with the pH of a hydrolysis reaction at high conversion.

Conclusions
Ruthenium nanoparticles stabilized by polymer immobilized ionic liquids catalyze the hydrolytic evolution of hydrogen from sodium borohydride; catalyst stabilized by an amino-modified imidazoliumbased polymer was the most active with an initial TOF of 171 mole H2 .mol Ru − 1 .min− 1 , this is among the highest to be reported for a RuNP-based system.Kinetic studies revealed that the reaction was first order in catalyst as well as sodium borohydride at low hydride/catalyst mole ratios but zero order with respect to NaBH 4 concentration with high hydride/catalyst mole ratios.The apparent activation energies of 38.9  kJ mol − 1 to 51.8 kJ mol − 1 are in the region commonly reported for the platinum group metal catalyzed hydrolysis of hydrogen rich boron derivatives; the apparent activation energy of 38.9 kJ mol − 1 for RuN-P@NH 2 PIILS is lower than each of the other catalysts tested and consistent with its higher initial TOF.A kinetic isotope effect (k H /k D ) of 2.3 obtained for reactions conducted in H 2 O and D 2 O and a k H /k D of 2.72 for reactions conducted with NaBH 4 and NaBD 4 at a low catalyst/ hydride mole ratio indicate that both the O-H and B-H bonds are activated by the ruthenium atoms in the rate limiting step, most likely via a concerted oxidative addition-hydride transfer involving the surfacecoordinated hydrogen-bonded [BH 3 H-]--H-OH ensemble rather than oxidative addition of both the O-H and B-H bonds.Interestingly though, the k H /k D of 0.67 obtained from comparing the initial rates of hydrolysis for NaBH 4 and NaBD 4 under conditions of catalysis, i.e. at a high catalyst/hydride mole ratio, is an inverse KIE which would be consistent with a surface-coordinated borohydride activating an O-H bond of water in the hydrogen-bonded ensemble prior to rapid hydride transfer.Reuse experiments showed that RuNP@NH 2 -PIILS retains 79% of its activity over 5 runs and poisoning studies conducted by adding NaBO 2 to a catalytic reaction suggest that the reduction in activity is most likely due to passivation of the catalyst by absorption of the metaborate by-product on the nanoparticle surface.A tandem hydrogenation of 1,1-diphenylethene in d 4 -MeOH with hydrogen generated from the catalytic hydrolysis of NaBH 4 in D 2 O gave a mixture of all eight possible isotopologues with a total deuterium incorporation greater than one while the use of toluene for the hydrogenation using NaBH 4 /D 2 O gave a total deuterium incorporation close to one.This is consistent with slow H/D exchange at the NP surface and fast H/D exchange on the surface of the Pd/C coupled with H/D scrambling via facile reversible beta hydride eliminationreinsertion during the hydrogenation.This programme is currently exploring the use of PIIL supported bimetallic nanoparticles with varying proportions of noble and earth abundant metals to establish how the composition of the NP influences catalyst performance with the aim of identifying an optimum synergism that will be suitable for use as a hydrogen generation system for portable applications of proton exchange membrane fuel cells (PEMFC).In addition, PIILs are an ideal support to investigate how polymer properties such as charge density, the number and type of heteroatom donor and functionality, porosity and hydrophilicity influences the size, morphology, and efficacy of the nanoparticles as well as to tailor catalyst-support interactions to enhance efficacy.Ultimately, this catalyst technology will be extended to include the hydrogen evolution reaction to develop stable, durable, highly active cost-effective catalysts for use in AEM based electrolysers and fuel cells.

Declaration of Competing Interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Reece Patterson, Anthony Griffiths reports financial support was provided by Engineering and Physical Sciences Research Council.

Fig. 3 .
Fig. 3. (a) Hydrolytic release of hydrogen from an aqueous solution of NaBH 4 as a function of time at 303 K catalysed by 0.2 mol% 2a-f and (b) corresponding TOFs for the catalytic reactions shown in (a).Conditions: 0.57 mmol NaBH 4 , 0.2 mol% 2a-f, water (2 mL), 30 • C. Each volume is an average of three runs.

Fig. 9 .
Fig. 9. (a) Hydrogen release from 2 mL of 0.28 M NaBH 4 in H 2 O (red line) and D 2 O (blue line) at 303 K, catalysed by 0.16 mol% 2e (0.0025 g) (b) hydrogen release from 200 mL of 0.13 mM solutions of NaBH 4 (red line) and NaBD 4 (blue line) in H 2 O at 303 K using a stoichiometric amount of catalyst 2e (0.026 mmol, 0.0764 g); (c) hydrogen release from 2 mL of a 0.28 M solution of NaBH 4 (red line) and NaBD 4 (blue line) in H 2 O at 303 K catalysed by 0.2 mol% 2e (0.0033 g) Each volume is an average of three runs.

1 . 1 . 1 .
Fig.11a, b shows a minor but gradual drop in conversion across five reuses, from 89% after 2 min in the first run to 78% after the same time in the 5 th run.The drop in catalyst activity in successive runs, defined as the percentage reduction in the initial TOF, shows that 2e retains 71% of its activity across five reuses (Fig.11b, red); this is comparable to recycle studies reported for other noble metal nanoparticle catalysts including; RuNPs immobilised in ZIF-67[77], PtCoNPs supported on carbon nanospheres[167], ruthenium nanoparticles immobilised within the pores of amine-functionalised MIL-53[76], ruthenium supported on graphite[139], RuCo nanoclusters incorporated in PEDOT/PSS polymer[168], RuNP stabilized by polyvinylpyrrolidone, zeolite-confined RuNPs[64], click dendrimer-stabilized PtCo, Rh and Pt nanoparticles and gold-transition metal nanoalloys[72,73,78,79,80,165] and Ru-RuO 2 /C[141].Sneddon et al. previously reported that the use of a borate buffered solution for the rhodium-catalysed release of hydrogen from ammonia triborane extended the catalyst lifetime such that Rh/Al 2 O 3 showed little change in the hydrogen release rate over 11 cycles[169].Following this lead, a preliminary comparative recycle hydrolysis conducted in freshly prepared aqueous borate buffer (pH maintained between 7.2 and 8) containing 0.28 M NaBH 4 and 1 mol% 2e resulted in a marked increase in activity as evidenced by the initial TOF of 133 mole H2 .molRu − 1 .min− 1 obtained for the first run compared with 95 mole H2 .molRu − 1 .min− 1 for the corresponding reaction in water.The initial TOF increased to 146 mole H2 .molRu − 1 .min− 1 in the second run but then decreased gradually in subsequent cycles to 109 mole H2 .molRu − 1

Fig. 12 .
Fig. 12.(a) Hot filtration experiment for the hydrolysis of 20 mL of a 0.028 M solution of NaBH 4 at 303 K catalysed by 0.16 mol% 2e (0.0026 g), confirming that filtration quenches the reaction.Red linehydrogen evolution in the presence of 2e; blue linehydrogen evolution in the presence of catalyst with filtration at t = 7.75 min; orange linehydrogen evolution after filtration at complete conversion and addition of a further portion of NaBH 4 ; (b) sizing histogram of RuNPs for 2e after five reuses and a TEM image of the recovered material, scale bar = 20 nm.

Fig. 13 .
Fig. 13.(a) Volume of hydrogen against time for the hydrolysis of 20 mL of a 0.028 M solution of NaBH 4 at 303 K catalysed by 2 mol% 2e (0.0335 g, 0.0114 mmol) as a function of pre-stirring time with added NaBO 2 (0.0765 g, 0.57 mmol); (b) conversions obtained after a reaction time of 2 min and the corresponding initial TOFs as a function of pre-stirring time with NaBO 2 .

Fig. 14 .
Fig. 14.A plot of pH and conversion against time for the hydrolysis of a 0.028 M solution of NaBH 4 (20 mL) at 303 K catalysed by 2 mol% 2e (0.0335 g).