Introduction

Due to its gaseous nature and low density at room temperature, storage and transportation of hydrogen are still obstacles to the development of proton exchange membrane fuel cell (PEMFC) [1]. Hence, to solve the problem, the concept of solid hydrogen storage has been proposed and attracted intense attentions [2]. Sodium borohydride (NaBH4) has been considered as a potential candidate for hydrogen storage in the light of its high gravimetric density (10.8 wt%) [3, 4], environmental friendliness and potentially safe operation [5,6,7]. In 2001, NaBH4, a solid hydrogen storage material, was first successfully applied to Daimler Chrysler’s fuel cell concept minivan by Millennium Cell Corporation of American [8, 9].

The generation rates of hydrogen gas can be promoted by introducing a proper catalyst [10,11,12,13], such as Co–TiO2 [14], Ru/Al2O3 [15], Fe–B/Ni foam [16], Co/Al2O3 [17], Ni2P/SiO2 [18], MWCN–COOH [19], polyethyleneimine microgels [20], Hal nanotube [21] and P/boehmite [22]. Among these catalysts, Ru catalyst showed the highest efficiency of catalytic activity compared with other metal catalysts [23,24,25,26,27,28], but its practical application was restricted by the expensive price and limited resources. These limitations have inspired many efforts to develop cost-effective non-noble metal (Ni, Fe [29,30,31,32,33,34] and Co [35,36,37,38,39,40,41], etc.) catalysts as an alternative. In the past decades, considerable progress has been achieved in this area, especially in the case of Co-based catalysts. However, it is still far from satisfaction to achieve our goal.

Recently, bimetallic nanoparticles with large surface area and unique role of controlling the selectivity, activity, and stability are found to be more effective than monometallic catalysts in the field of catalysis [42,43,44,45]. In view of the above advantages of the bimetallic nanoparticles, in this study, we synthesize supported noble and non-noble bimetallic (Ru–Co) nanoparticles to catalyze H2 generation from NaBH4 methanolysis (Scheme 1) for the first time. The results illustrated that the as-synthesized Ru–Co bimetallic nanoparticles are uniformly dispersed on the support with an average diameter of 2.4 nm. Besides, hydrogen generation tests also presented that the displayed catalyst significantly enhanced the catalytic activity toward NaBH4 methanolysis, indicating that Ru–Co bimetallic nanoparticles catalyst is a promising candidate to replace pure Ru catalyst in H2 generation from NaBH4 methanolysis.

Scheme 1
scheme 1

Preparation of Ru–Co/C bimetallic nanoparticle catalyst

Full size image

Experimental section

Materials

The support material carbon black was purchased from AkzoNobel. The reagents were at the analytical grade without further purification. Ruthenium chloride (RuCl3), cobalt chloride (CoCl2·6H2O), sodium borohydride (NaBH4), methanol (CH3OH), ethylene glycol (EG) and sodium hydroxide (NaOH) were purchased from Sigma Aldrich.

Preparation of catalysts

In order to get the carboxyl group functionalized carbon black, 2.0 g carbon black was dispersed in 30 wt% HNO3 aqueous solution and the mixture was then refluxed at 140 °C for 2 h (Scheme 1i). After experiencing the cooling system, the rough carbon black was then acquired by filtering and washed by deionized water. Then the pure carbon black support can be obtained after dried for 12 h in 60 °C oven.

Carbon-black-supported Ru–Co nanoparticles (Ru–Co/C) were prepared by co-reduction of the precursors of Ru and Co by the addition of NaBH4. 2.05 ml RuCl3/EG (20 g/L) and 2.02 ml CoCl2/EG (20 g/L) and 80 mg functionalized carbon black above were added to ethylene glycol (EG) with string for 10 h (Scheme 1ii). Then, the pH of the solvents was adjusted to 9.5 by adding NaOH/EG solution (2M). Excess amount of NaBH4 were carefully added into the above flask with string vigorously, and the reaction solution was heated at 90 °C for 3 h (Scheme 1iii). After the washing and separating process, the products were dried in the oven at 60 °C for 12 h. Ru–Co nanoparticles with different molar ratio of Ru/Co were prepared by adjusting the molar concentrations of RuCl3 and CoCl2, which were denoted as RuxCoy/C (x/y = molar ratio, when the value of y was 1, it was omitted).

Catalytic activity of Ru–Co/C

A batch-type hydrogen generation system was used to evaluate the catalytic activity of Ru–Co/C nanoparticles. In a typical measurement, the flask was loaded with NaBH4 alkaline methanol solution (20 ml, 10 wt% NaBH4 and 1 wt% NaOH). Twenty micrograms of Ru–Co/C catalyst was introduced to the flask by an injector. The generated hydrogen stream passed through a gas mass flow meter (Sevenstar CS200C) to measure the hydrogen generation rate.

Characterization of catalysts

The structure of carbon black was characterized by Fourier transform infrared radiation spectroscopy (FTIR, Nicolet 6700). The crystallization of Ru–Co/C nanoparticles was characterized by X-ray diffraction (XRD, Shimadzu XD-3A diffractometer) and thermogravimetric analysis (TG, Hengjiu HTG-2). TG was implemented from 40 to 900 °C under O2 flow. Transmission electron microscope (TEM) tests were measured by JEOL JEM-3010HR. Energy-dispersive X-ray spectroscopy (EDX) was employed to measure the elements composition of catalysts. X-ray photoelectron spectroscopy (XPS) tests were measured by Thermo Fisher LAB 250 ESCA System.

Results and discussion

Characterization of the catalysts

Initially, 37 wt% HCl and 30 wt% HNO3 are both applied to treat the carbon black. However, only HNO3-treated carbon black shows obvious changes in the FTIR (Fig. 1a). Compared with the un-treated carbon black, two obvious peaks at 1734.25 and 1210.30 cm−1 are attributed to the stretching of carboxyl groups. It is indicated that there are abundant carboxyl groups on the carbon black surface. Furthermore, these carboxyl groups could improve the hydrophilic property and the dispersion of carbon black in the ethylene glycol (EG) solvent. Meanwhile, the exposed carboxyl groups favor the absorption of Ru3+ and Co2+ precursors on the surface of the carbon black.

Figure 1
figure 1

a FTIR pattern of carbon black and HNO3-treated carbon black; b XRD pattern of Ru–Co/C and carbon black; c XRD pattern of calcined Ru–Co/C at 500 °C and standard diffraction peaks of Ru–Co alloy; d TG curve of Ru–Co/C

Full size image

The XRD pattern of Ru–Co/C is shown in Fig. 1b. A peak at 2θ = 25.00° can be seen in the spectrum, which can be attributed to the carbon black. Meanwhile, the peak at 2θ = 44° can be interpreted as the (101) plane of ruthenium of amorphous state. Notably, the peak of Co does not appear because of the lower reduction potential of Co2+/Co0 [37, 38]. To confirm the existence of Co metal, the samples are calcined at 500 °C for 2 h under N2 atmosphere and the heat-treated Ru–Co/C is characterized by XRD, which is shown in Fig. 1c. Many sharp peaks at 2θ = 40.00°, 43.40°, 45.77°, 60.50°, 72.62° and 81.33° in the XRD pattern completely match with the diffraction peaks of the standard Ru–Co alloy (JCPDS 65-8976). Besides the XRD test, TG is carried to measure the total metal loading of Ru–Co/C, as shown in Fig. 1d. Under O2 flow, carbon black decomposes above 250 °C, but metal compositions are still stable. The final weight of 26.5% indicates the actual metal loading.

In order to further determine the presence of Ru and Co metals, we selected two regions of the primary Ru–Co/C to test the EDX, which are shown in S1(a) and (b) in supplementary information. The EDX results (Table 1) suggest the atomic ratios of Ru to Co are 5.57 and 5.64 in region 1 and region 2, respectively, indicating that the Ru and Co metals exist and distribute uniformly on the carbon black support.

Table 1 Element composition of Ru5Co/C catalyst
Full size table

The as-synthesized catalysts are then characterized by the TEM and SEM, and the test results are shown in Fig. 2a, b, e, f. Compared with the previous Ru catalyst prepared in the water solvent [28, 46], our Ru–Co nanoparticles have a higher dispersion on the carbon surface because the EG solvent reduced the agglomeration of the nanoparticles during the preparing process. Moreover, the average diameters of Ru–Co nanoparticles and Ru nanoparticles are just 2.40 and 4.32 nm, respectively. To further characterize the as-synthesized catalysts, the HRTEM tests are carried out for the catalysts, as shown in Fig. 2. As we can see from Figs. 3c and Fig. 2d, the inter-planar spacing of pure Ru and the Ru–Co nanoparticles nanoparticles is 0.2352 and 0.2123 nm, respectively, which is between the inter-planar spacing of the (100) face of Ru (0.2343 nm; Ru: JCPDS 89-4903) and the (101) face of Co (0.1915 nm; Co: JCPDS 89-7373). The above results illustrate that the Ru and Co are not monometallic nanoparticles on carbon black but alloy-structured nanoparticles.

Figure 2
figure 2

TEM images of a Ru–Co/C and b Ru/C catalyst; HRTEM images of c Ru–Co/C and d Ru/C catalyst; SEM images of e Ru–Co/C and f Ru/C

Full size image
Figure 3
figure 3

XPS spectra of Ru–Co/C catalyst

Full size image

In order to verify the alloy structure of Ru–Co nanoparticles, XPS was implemented. Figure 3a shows the XPS spectra of each element of the Ru–Co/C catalyst. The C 1s spectrum (Fig. 3b) can be fitted by three chemical states at binding energies of 284.8, 286.2 and 289.0 eV, corresponding to C in the form of C–C, C–OH and O=C–OH, respectively. The single peak at 281.5 eV is attributed to Ru 3d, which shows a little shift of 0.4 eV compared with that of pure Ru nanoparticles. Ru 3p3/2 and Ru 3p1/2 can be assigned to the binding energies of 463.0 and 485.2 eV in Fig. 3c, respectively. For Co (Fig. 3d), the signals at binding energies of 780.10 and 796.85 eV are corresponded to Co 2p3/2 and Co 2p1/2 peaks of Co0, respectively, and binding energies of 786.9 and 802.0 eV are corresponded to Co 2p3/2 and Co 2p1/2 peaks of CoO, respectively. The peak at binding energy of 783.6 eV is ascribed to Co 2p3/2 of Co(OH)2, resulting from the inevitable by-product during catalysts preparation processes [47, 48].

Correlation between catalyst composition, microstructure and catalytic activity

During the methanolysis process, electron transfers between the catalysts, reactants are combined with the active sites on the surface of the catalyst, which is, BH4− combines the positively charged metal and CH3OH combines the negatively charged metal to release a H atom, respectively. Then, two H atoms combine to produce hydrogen. As a result, the hydrogen generation rate depends on the composition of the catalyst, and the concentration of NaBH4 and NaOH in the process of methanolysis of NaBH4. Based on the electronic-structure calculations by Nørskov et al. [49], Ru–Co bimetallic alloy could show better performance than pure Ru in catalyzing hydrogen evolution. Therefore, Ru–Co/C catalysts with the Ru/Co molar ratio of 1:1, 2:1, 3:1, 4:1, 5:1 and 6:1 were prepared and applied to catalyze the H2 generation from methanolysis of NaBH4 under same condition (Fig. 4a). The faster the hydrogen generation rate is obtained, the more active the Ru–Co/C catalyst is. The hydrogen generation rate increases with Ru/Co molar ratio on the initial stage, to the maximal rate at the Ru/Co ratio of 5, and then decreases. Ru3Co/C, Ru4Co/C, Ru5Co/C, and Ru6Co/C all show higher hydrogen generation rate than Ru/C, and the maximum hydrogen generation rate of Ru5Co/C reaches up to 9.36 L min−1 g−1, which is 54.4% higher than Ru/C. Here, this phenomenon can be explained by the Sabatier principle, only the catalyst surface with optimal binding energy of intermediate can achieve the best catalyst activity [50]. In other words, low intermediates binding energies of the catalyst could not trigger the reaction, in contrast, high binding energies would prevent the product dissociating from the catalyst surface [24]. On the other hand, the Ru–Co/C catalysts show enhanced activity due to the small size and the high dispersion of the nanoparticles, which leads to a larger specific surface area than Ru/C catalyst [45]. Furthermore, BET data (Table 2) was implemented to support this point. Hence, the Ru–Co/C catalyst with an optimal Ru/Co molar ratio (Ru5Co/C) presents the best catalyst performance to generate H2 from NaBH4 methanolysis.

Figure 4
figure 4

a Effect of Ru/Co molar ratio on hydrogen generation (\( c_{{^{\text{NaBH}}_{4} }} = 10\; {\text{wt}}\% \), cNaOH = 1 wt%, T = 25 °C); b effect of NaBH4 concentration on hydrogen generation (cNaOH = 1 wt%, T = 25 °C, Ru5Co/C); c effect of NaOH concentration on hydrogen generation (\( c_{{^{\text{NaBH}}_{4} }} = 10\; {\text{wt}}\% \), T = 25 °C, Ru5Co/C); d effect of NaOH concentration on initial lag time

Full size image
Table 2 BET surface area of samples
Full size table

Correlation between reaction conditions and catalytic behavior

Considering that the catalytic performances closely depending on the concentration of NaBH4, the effects of NaBH4 concentration on the H2 generation rate are investigated using Ru5Co/C as the catalyst, which is shown in Fig. 4b. It can be seen that the hydrogen generation rate is accelerated with increasing NaBH4 concentration from 5 to 10 wt%. However, further increasing the NaBH4 concentration (15 and 20 wt%) led to the significant decreasing of H2 generation rate. The average H2 generation rate of 2.25 L min−1 g−1 is achieved with the NaBH4 concentration of 10 wt%. These phenomena are consistent with the previous results of Patel et al. [36]. and Liang et al. [28]. The above phenomena are attributed to the NaBH4 methanolysis. In fact, the NaBH4 methanolysis process includes the adsorption of BH4− on catalyst surface and the reaction of BH4− with solvent. At a low NaBH4 concentration, the adsorption of BH4− on catalyst surface is the leading role, which caused that the H2 generation rate increases with NaBH4 concentration increasing. However, at a high NaBH4 concentration, the BH4− is saturated on the active sites of catalyst surface so that the reaction of BH4− with solvent became the key step. In this situation, the excess BH4− causes a considerable solution viscosity, which could limit the mass transport and reduce the methanolysis efficiency [36].

Apart from the effects of NaBH4 concentrations, hydrogen generation experiments are carried out with various concentrations of NaOH solutions. The results (Fig. 4c) show that the average hydrogen generation rate increases from 1.14 to 1.91 L min−1 g−1 as the concentration of NaOH rises from 1 to 5 wt%, which means the activity of the Ru–Co/C is weakly dependent on the concentration of NaOH. On the other hand, the initial lag time rises from 1 s to 30 min (Fig. 4d) when the NaOH concentration increases from 1 to 5 wt% because of the inhibiting effect of OH− anions, which indicates that NaOH could act as a stabilizer in the methanolysis [36]. Much more OH− anions exist in the high concentration of NaOH, and these OH− anions will compete with the BH4− anions for occupation on the surfaces of the catalyst.

Kinetic study of the catalyzed methanolysis

Kinetic studies at various temperatures in the methanolysis of NaBH4 are carried out. The effects of temperature on the H2 generation rate with Ru5Co/C catalyst at the different temperatures are investigated, and the result is shown in Fig. 5a. As expected, H2 generation rate increases with the increase of the temperature. The hydrogen generation rate of NaBH4 methanolysis with catalyst can be described by an Arrhenius relation (Eq. 1):

$$ r = k_{r} = A \cdot \exp \left( { - E_{\text{a}} /RT} \right) $$
(1)

where r is the reaction rate (mol min−1 g−1), A is the reaction constant (mol min−1 g−1), Ea is the reaction activation energy (kJ mol−1), R is the gas constant and T is the reaction temperature (K).

Figure 5
figure 5

a Effect of temperature on hydrogen generation; b Arrhenius plot obtained from data in a for hydrogen generation from NaBH4 solution

Full size image

From the slope of lnkr versus 1/T (Fig. 5b), the apparent activation energy (Ea) was determined to be 36.83 kJ mol−1, which is obviously lower than the reported values of NaBH4 spontaneous methanolysis (53.0 kJ mol−1 by Lo et al. [51], and 63.0 kJ mol−1 by Xu et al. [17]) and value of NaBH4 methanolysis using pure Ru supported on Al2O3 as the catalyst (51.0 kJ mol−1 by Su et al. [15]).

Reusability of the catalyst

The stability of a catalyst is vital for its application. Figure 6a shows the deterioration in catalytic activity of Ru–Co/C catalyst after reusing the catalyst. The hydrogen yield was calculated by Eq. 2:

$$ {\text{Yield}}\left( {\text{\% }} \right) = \left( {\frac{{V_{\text{r}} }}{{V_{\text{t}} }}} \right) \times 100{\text{\% }} $$
(2)

where Vt is the theoretical volume of H2 generation and Vr is the real volume of H2 generation [15].

Figure 6
figure 6

Hydrogen generation in multi-cycle test

Full size image

As shown in Fig. 6, the hydrogen yield in the 1st cycle and the 8th cycle are 99.2 and 95.6%, respectively, which indicates the recyclability of catalyst is excellent. In addition, the hydrogen generation rate decreases by 30% after eight cycles. The decrease in efficiency is attributed to the changes of the catalyst composition and the partial inactivation of the catalyst on the surface of by the NaB(OCH3)4 formation during the reaction.

To test the composition of the catalyst before and after the experiment, the final catalyst was then characterized by the EDX (S2 in supplementary information). The atomic ratio of Ru to Co became 6.43 after eight cycles (Table 3), while the initial ratio was 5.64 before the experiment. This result is in accord with our preceding work (as shown in Fig. 1e, f) which shows that Ru6Co/C has a lower activity than Ru5Co/C. The interesting result indicates that the Ru and Co metals still exist on the carbon support. However, part of Co metal was lost from the catalyst caused by erosion of the OH− in the reaction.

Table 3 Element composition of Ru–Co/C catalyst after 8 cycles
Full size table

The efficiency of our Ru–Co/C catalysts and the catalysts reported in previous literature are listed in Table 4. Compared to the reported catalysts, the efficiency of our catalysts is far better than that of many other catalysts including pure Ru/Al2O3 catalyst. It should be pointed out that the maximum H2 generation rates of our catalyst could even reach up to 9.36 L min−1 g−1.

Table 4 Comparison of the maximum H2 generation rates of the NaBH4 methanolysis between our Ru–Co/C catalyst and various catalysts reported in literature
Full size table

Conclusions

In summary, Ru–Co bimetallic nanoparticles supported on the carbon black (Ru–Co/C) have been prepared by the co-reduction method using NaBH4 as a reductant. The average diameter of the Ru–Co nanoparticles is mere 2.4 nm, and the nanoparticles are uniformly dispersed on the carbon black. The materials show an excellent catalytic performance for the H2 generation from NaBH4 methanolysis. And the maximum H2 generation rate of the as-synthesized Ru–Co bimetallic nanoparticles at 25 °C could reach up to 9.36 L min−1 g−1, which exceeds the values of Ru/C catalyst and most reported values of transition metal or noble metal contained catalysts performed on similar conditions. Due to the synergistic effect of alloyed structure and the smaller average nanoparticles size, the activity of the Ru–Co/C catalysts was greatly enhanced. The Arrhenius apparent activation energy is determined to be 36.83 kJ mol−1, which is significant lower than the values of other catalysts reported in the literature for NaBH4 methanolysis. Besides the high activity, the Ru–Co/C catalysts also exhibit excellent stability and reusability. Though the hydrogen generation rate reduces gradually, the hydrogen yield has no significant decrease after 8 cycles. The results in this study have indicated that the bimetal Ru–Co catalysts can be a relative low-cost and high efficiency catalyst for the substitution of noble metal catalysts in H2 generation from NaBH4 methanolysis and have potential application in the field of fuel cell.