Monodisperse AgPd alloy nanoparticles as a highly active catalyst towards the methanolysis of ammonia borane for hydrogen generation

Monodisperse AgPd nanoparticles (NPs) were successfully synthesized by coreducing silver nitrate and palladium chloride in the presence of oleylamine. Characterization techniques such as transmission electron microscopy, energy-dispersive X-ray spectroscopy, and X-ray diffraction were employed to verify the alloy nature of the obtained NPs. These AgPd NPs, supported on carbon, were found to be active catalysts for the generation of hydrogen from the methanolysis of ammonia borane at room temperature. The catalytic activity depended on the alloy composition, and the Ag30Pd70 NPs displayed the greatest activity among the tested catalysts, with a very high total turnover frequency value of 366.4 min 1 and activation energy of 37.5 kJ mol . In a reusability test, the Ag30Pd70/C catalyst retained 67% of its initial activity and 100% conversion after eight cycles.


Introduction
Storage and transport of hydrogen are the widely accepted technological barriers to the development of the so-called hydrogen economy. 1 As the lightest element in the periodic table, hydrogen in the gas form is extremely difficult to store. Considering economically viable and safe practices, efforts have been directed towards the development of chemical hydrides as a potential onboard storage form of hydrogen. 2,3 Ammonia borane (AB) has been considered to be the most promising candidate among the chemical hydrides. 4,5 This molecule is stable in the solid state at ambient temperatures and has shown an impressive gravimetric capacity of 19.6 wt% H 2 , potentially capable of meeting the 2015 target of the U.S. Department of Energy (9 wt% hydrogen for a material to be practically applicable). 6,7 Hydrogen stored in AB can be produced and obtained by pure thermal decomposition, hydrolysis and methanolysis using a suitable catalyst. Considering the mild reaction conditions and favorable kinetics, solvolysis is preferred. 8 The methanolysis or hydrolysis of AB can release 3.0 equivalents of hydrogen in the presence of a suitable catalyst. However, the hydrolysis of AB in concentrated solutions can cause liberation of ammonia gas, which can poison Pt-based fuel cells, and the hydrolysis product ammonium metaborate is difficult to recycle, which can pose enormous challenges to practical applications. 9,10 In a previous study by Ramachandran, when methanol was used instead of water, the catalytic methanolysis of AB yielded hydrogen gas without ammonia evolution, and yielded the recyclable methanolysis product ammonium tetramethaoxyborate. 6 In addition, the methanolysis reaction can even be initiated at temperatures below 0 C through use of a suitable catalyst, which satises applications in cold weather. 11 Since methanolysis of AB can achieve a complete release of hydrogen in the presence of a suitable catalyst, nding the best catalyst for this process is very important. Heterogeneous catalysts were the rst ones to be tested for the methanolysis of AB. RuCl 3 and RhCl 3 have been shown to exhibit excellent catalytic activity, but they are difficult to isolate and they display low stability, features that seriously restrict their practical application. 6 Cu@Cu 2 O, Cu 2 O, 12 Co-Co 2 B, Ni-Ni 2 B, Co-Ni-B, 13 and Cu-Cu 2 O-CuO/C 14 are cheaper transition metal-based catalysts, but suffer from low activity. The performance of the catalyst in the methanolysis of AB appears to depend on the identity of the metal. Thus, recent studies have been directed toward bimetallic alloy nanoparticles (NPs), as their catalytic properties have been found to be superior to those of either component metal alone due to synergistic effects. Our previous studies have also indicated monodisperse CoPd NPs 15 and CuPd NPs 16 to be more active than Pd-rich NPs in catalyzing the AB methanolysis reaction, with total turnover frequency (TOF) values of 27.7 and 53.2 min À1 , respectively, having been attained.
Despite the signicant progress made in the synthesis of bimetallic alloy NPs, 17,18 size-controlled fabrication of nanostructures has become one of the most important and challenging aspects of modern nanotechnology. In the widely used method, the metal NPs are generated from the reduction of metal ions in solution in the presence of a stabilizer or supporting materials. 19 To obtain uniformly sized, shapecontrolled, and highly dispersed monodisperse NPs, "hot injection" techniques, which utilize homogeneous nucleation to synthesize monodisperse nanocrystals in organic solutions, are commonly used. For example, when using this technique, Sun 20 prepared CoPd NPs with diameters of 5-12 nm and with a standard deviation of the diameter of less than 7%. Hyeon 21 even succeeded in the incremental 1 nm size controlled synthesis of iron oxide NPs ranging from 6 to 13 nm. To allow precise control over size and monodispersity, capping agents such as oleic acid and organic metal salt precursor (e.g., acetylacetonate) have usually been employed. [22][23][24] These agents, however, increase the costs of the catalytic process. Therefore, it is of crucial importance to develop rational cost-effective approaches to synthesize monodisperse alloy NPs.
Herein, we report the synthesis of monodisperse AgPd alloy NPs from inorganic salt silver(I) nitrate and palladium(II) chloride in oleylamine, and the evaluation of their performance in catalyzing the methanolysis of AB at ambient conditions. We demonstrated that the activity of these AgPd NPs can be enhanced by properly controlling the Ag/Pd molar ratio. In addition, these AgPd NPs showed a very high catalytic activity and prolonged lifetime.

Chemicals
AgNO 3 , PdCl 2 , isopropanol, hexane and methanol were of analytical levels and used as received without any further purication. The borane-ammonia complex (90%) and oleylamine (80-90%) were purchased from Sigma-Aldrich and Aladdin, respectively. All pieces of glassware were washed several times with aqua regia and ordinary distilled water.

Preparation of AgPd NPs
Typically, under the protection of a constant nitrogen ow, 0.2 mmol AgNO 3 and 0.2 mmol PdCl 2 were added into a fournecked ask with 15 mL of oleylamine. The mixture was heated to 110 C gradually and kept at this temperature for 0.5 h in order to remove oxygen and moisture from the ask, and then the solution was heated to 210 C at a heating rate of 4 C min À1 and maintained at this temperature for 1 h before being cooled to room temperature. Aerwards, the mixture and 40 mL of isopropanol were transferred to centrifuge tubes, and the product was separated by centrifugation at 10 000 rpm for 10 min. Finally, the product was dispersed in hexane.
For comparison, Ag NPs and Pd NPs were also, separately, synthesized and obtained. The Ag NPs were prepared by carrying out thermal decomposition of AgNO 3 in the presence of oleylamine. The Pd NPs were prepared by carrying out thermal decomposition of PdCl 2 in oleylamine. The procedures used to synthesize the Ag NPs and Pd NPs were similar to that used to synthesize AgPd NPs.

Preparation of the catalyst
Typically, accurately weighed 10 mg of AgPd nanoparticles were redispersed in 5 mL hexane and 5 mL acetone in a vial with a volume of 20.0 mL, and then a mass of 30 mg of Ketjen carbon ($750 m 2 g À1 ) was added. The resulting mixture was sonicated for 2 h to ensure that the nanoparticles were completely loaded onto the support. Next, the black solid residue AgPd/C catalyst was obtained aer the hexane and acetone were vaporized as a result of a constant ow of nitrogen.

Evaluation of the catalytic activity
To evaluate the catalytic activity of the AgPd/C catalyst for methanolysis of AB, the classic water-displacement method involving determining the rate of hydrogen generation was performed. 15 Typically, a volume of 10 mL of a methanol suspension of the AgPd/C catalyst was transferred into a reaction ask (25 mL) containing a Teon-coated stir bar placed on a magnetic stirrer. The reaction temperature was maintained at 25 C using a constant temperature bath during the process, and the reaction ask was connected to a burette lled up with water to measure the volume of hydrogen. Weighed AB was then added into the reaction ask quickly with stirring at a rate of 800 rpm. The volume of hydrogen generated from methanolysis of AB in the ask was monitored and measured by recording the displacement of the water level at various intervals of time. The reaction was considered to be complete when the generation of more gas was no longer observed.

Characterizations
Samples used for TEM and HRTEM analyses were prepared in advance by sonicating AgPd in hexane at a proper concentration and then depositing a drop of the resulting dispersion on a carbon-coated copper grid. The morphologies of the various types of NPs were visualized by using a TEM instrument operating at 300 kV, and an elemental analysis of the AgPd NPs was performed by carrying out an energy-dispersive X-ray spectroscopy (EDX) experiment. The dried and powdered samples were analyzed via X-ray diffraction (XRD) by using an X-ray diffractometer (Phillips, Netherlands) with Cu Ka radiation (40 kV, 30 mA). The composition and metallic content of the NPs were determined by performing inductively coupled plasma mass spectrometry (ICP-MS) (Agilent 7700x). The dried NPs were dissolved in HNO 3 to assure the complete dissolution of the metal into the acid, and the obtained solution was then diluted with 2% HNO 3 . X-ray photoelectron spectroscopy (XPS) measurements were taken with an X-ray photoelectric spectrophotometer, and all binding energies were referenced to the C 1s peak, at 284.8 eV.

Synthesis and characterization of the monodisperse AgPd NPs
Zhang has reported the synthesis of AgPd alloy NPs with dimensions of 2.2 nm by coreducing Ag(Ac) and Pd(acac) 2 in oleylamine, oleic acid and 1-octadecene at 180 C. 22 We replaced the above precursors with inorganic metal salts to yield a procedure that is more cost-effective but for which obtaining monodisperse nanoparticles is more challenging. Specically, AgPd NPs were synthesized from AgNO 3 and PdCl 2 in oleylamine at 220 C under the protection of a constant ow of nitrogen. The morphology and structure of the as-synthesized AgPd NPs were observed by performing TEM. For a molar ratio of Ag to Pd of about 1 : 1, the composition of the assynthesized NPs was determined using ICP-MS to be Ag 36 Pd 64 .
The as-synthesized NPs were observed to be spherical, with a mean diameter of 6.2 nm (Fig. 1(a)). These NPs were monodisperse, with a standard deviation of $5%. STEM-EDX images of the Ag 36 Pd 64 NPs are shown in Fig. 1(b)-(e). Close inspection of several as-synthesized NPs suggested the Ag (Fig. 1(d)) and Pd  ( Fig. 1(e)) atoms to be evenly distributed throughout the NP. Moreover, EDX elemental line scanning of two NPs ( Fig. 1(b)) also veried the presence of the alloy structure in these NPs. The composition and size of AgPd bimetallic NPs could be facilely controlled by changing the ratio of AgNO 3 to PdCl 2 . In our study, the total amounts of AgNO 3 and PdCl 2 were each set at 0.4 mmol, and ICP analyses indicated that as-synthesized NPs with the formulas Ag 71 Pd 29 , Ag 44 Pd 56 , Ag 36 Pd 64 , Ag 30 Pd 70 and Ag 12 Pd 88 were obtained when using the molar ratios of AgNO 3 to PdCl 2 of 9 : 1, 7 : 3, 5 : 5, 3 : 7, and 1 : 9, respectively. The TEM images of each of the above ve NPs are shown in Fig. 2(b)-(f), with an image of the Ag NPs shown in Fig. 2(a) and that of the Pd NPs in Fig. 2(g). All of the as-synthesized NPs were observed to be spherical in shape and well-dispersed, except for the Ag 12 Pd 88 and Pd NPs. The results may be attributed to the smaller particles more easily agglomerating during the synthetic process. As shown in Fig. 2(h), when the mole fraction of Ag was increased from 30% to 100%, the dimensions of the resulting particle increased on average from 5.4 nm to 12.3 nm. The inuence of the molar ratio of the metals on the size of the bimetallic alloy NPs has also been reported in earlier studies. 25,26 The dependence of the size of the alloy NPs on the composition has been indicated to be related to the collision energy, sticking coefficient, rate of nucleation and growth. 27 XRD patterns of the monometallic and bimetallic NPs are shown in Fig. 3(b). Each pattern exhibited several diffraction peaks, which can be indexed to the (111), (200), (220), (311) and (222) planes of the fcc structure of Ag and Pd, representing ve diffraction rings in the selected area electron diffraction (SAED) pattern ( Fig. 3(a)). Fig. 3(c) shows a plot of the diffraction angle versus the Pd molar fraction. All of these XRD peaks gradually shied to larger angles as the Pd content was increased. As shown in the inset of Fig. 3(c), an intense peak resulted from each of the AgPd NPs produced using AgNO 3 to PdCl 2 ratios of  This journal is © The Royal Society of Chemistry 2016 9 : 1, 7 : 3, 5 : 5, 3 : 7, 1 : 9, respectively. The curves in this gure indicated that the positions of the (111) diffraction peaks from the AgPd NPs were indeed between that from a pure Ag crystal at 38.23 and that from a pure Pd crystal at 39.23 . These results further conrmed the alloy nature of the products synthesized from the various initial molar ratios of AgNO 3 to PdCl 2 .
The Ag 3d and Pd 3d regions of the XPS spectra of three representative AgPd alloy NPs are shown in Fig. 4(c) and (d), respectively. For comparison, Fig. 4(a) and (b) show the XPS spectra of the Ag 3d and Pd 3d regions of monometallic Ag NPs and Pd NPs, respectively. The differences (D values) between the positions of the 3d 5/2 and 3d 3/2 peaks of Ag were observed to be 6.0 eV, and the differences between the positions of the 3d 5/2 and 3d 3/2 peaks of Pd were observed to be 5.3 eV. These values were the same as the reported values for zero-valent Ag and Pd, respectively. 28,29 Hence, Ag and Pd atoms existed in the zerovalent state in all of the prepared alloy samples. The positions of the XPS peaks are listed in Table 1 for further reference. The binding energy of Ag 3d was clearly shied to lower energy and the binding energy of Pd 3d shied to higher energy when the Pd content was increased, indicating that Ag and Pd formed an alloy cluster. 30 The formation of interfacial metal-metal bonds for a bimetallic alloy can produce large perturbations to the Hamiltonian, binding energy density, and hence the catalytic properties. 31 Sun has mentioned that the polarization of AgPd shis the valence density of state to generate excessive electrons, which makes the AgPd alloy serve as a charge donor in the catalytic reaction. 32

Methanolysis of AB catalyzed by monodisperse AgPd NPs
The as-synthesized NPs supported on Ketjen carbon acted as catalysts for the generation of hydrogen from the methanolysis of the AB solution. As shown in Fig. 5, pure Ag NPs, pure Pd NPs and 5 compositions of AgPd NPs supported on Ketjen carbon exhibited different levels of catalytic activity for the methanolysis of AB. Interestingly, among the tested catalysts, Ag 30 Pd 70 /C NPs displayed the highest catalytic activity with a completion time of 15 s for 100 mM AB, obviously better than those of pure Pd NPs and Ag NPs, which indicated the existence of a synergistic effect between the Ag and Pd. The metal content of the oleylamine-coated Ag 30 Pd 70 was determined, using ICP-AES, to be 34.6 wt%. The catalytic activity of Ag 30 Pd 70 /C for the methanolysis of AB was also assessed by determining its TOF, which was found to be 366.4 mol of H 2 (mol of catalyst min) À1 . The very high activity of Ag 30 Pd 70 NPs probably resulted from their small dimensions and regular shapes.

Methanolysis kinetics of the Ag 30 Pd 70 /C catalyst
As the Ag 30 Pd 70 /C catalyst exhibited the highest activity, we chose it for further study of its kinetics. First, the methanolysis reaction was carried out using various dosages of the Ag 30 Pd 70 NPs, specically at 0.5 mg, 2 mg, 5 mg, and 10 mg of the NPs     Fig. 6(a). The slope of the linear initial portion of each plot was chosen as the rate of hydrogen generation corresponding to the respective catalyst concentration. As shown in Fig. 6(b), a plot of the natural logarithm of the rate of hydrogen generation versus the natural logarithm of the concentration of Ag 30 Pd 70 was found to be nearly linear, with a slope of 1.16. The linear nature of this plot indicated that the methanolysis of AB followed nearly rst order kinetics with respect to the concentration of Ag 30 Pd 70 /C. To obtain the activation energy of the methanolysis of AB catalyzed by AgPd/C NPs, reactions at various temperatures from 0-30 C were carried out. The values of the rate constant k at these temperatures were calculated from the slope of the linear part of each plot shown in Fig. 7(a). The Arrhenius plot, i.e., ln k vs. 1/T, for the catalyst is shown in Fig. 7(b), from which the apparent activation energy was determined to be approximately 37.5 kJ mol À1 . Table 2 summarizes performance features of the various catalysts towards the methanolysis of AB at room temperature. The Ag 30 Pd 70 /C catalyst was determined to be the  most active catalyst, even better than Ru 11 -and Rh 38,39 -based catalysts based on comparing their TOF values.

Reusability of the Ag 30 Pd 70 /C catalyst
The reusability of a catalyst is crucial for practical applications. Aer the rst run of methanolysis of AB (1 mmol) catalyzed by 3.2 mM NPs, the catalyst was kept in the solution and another equivalent amount of AB was added to the reaction system. The gas generation was monitored and the same procedure was repeated. Fig. 8(c) shows the hydrogen productivity versus reaction time for the generation of hydrogen from the AB solution catalyzed by Ag 30 Pd 70 /C at 8 recycles under ambient conditions. The activity of Ag 30 Pd 70 /C catalysts decreased with the number of recycles, and the TOF value dropped from 366.4 to 248.9 min À1 aer the eight runs. The TEM images shown in Fig. 8(a) and (b) indicated that there was no noticeable change either in NP size or morphology, so the decreased activity of the Ag 30 Pd 70 /C catalysts may be attributed to the loss of metal from the surface of the carbon or the deactivation effect of the increasing metaborate concentration during the methanolysis of AB. To examine this possibility, supernatant liquid aer the reusability tests was ltered to remove the support. ICP analysis of the solution indicated the presence of a small amount of Ag.

Conclusions
In summary, we have reported the synthesis of monodisperse AgPd alloy NPs by the coreduction of silver nitrate and palladium chloride in oleylamine. These NPs exhibited notable activity towards hydrogen generation from the methanolysis of AB at room temperature due to the uniform size of the particles on the support and the synergistic effects between Ag and Pd. The catalytic activity was found to be closely related to the alloy composition, and the Ag 30 Pd 70 NPs were shown to be the most active, with a TOF value of 366.4 min À1 . The kinetic studies that we carried out on these alloy NPs indicated the catalytic methanolysis of AB to be rst-order with respect to the catalyst concentration. The activation energy was determined to be 37.5 kJ mol À1 . Moreover, the Ag 30 Pd 70 /C catalyst retained 67% of its initial activity aer 8 catalytic cycles. Their tunable catalytic properties shown here indicate that these AgPd alloy NPs have great potential in developing AB as a hydrogen storage material for fuel cell applications.