Electrochemical Preparation of Fine Powders of Nickel-Boron Alloys in Molten Chlorides for Magnetic Hydrogenation Catalysts

All of the reagents were of the analytical grade and used as received. The composites with different nickel to boron atomic ratios (R_Ni/B = 1, 2 or 3) were prepared by the reaction between Ni(NO₃)₂·6H₂O and H₃BO₃ in aqueous solution with the presence of C₆H₈O₇ (citric acid) foaming agent.³⁰ After magnetic stirring for 2–3 h, the homogenous mixture was placed in an oven at 150^°C for 12 h to obtain a dry gel, which was then calcinated at 750^°C for 4 h in a muffle furnace. Finally, the samples were cooled to room temperature with oxide composite precursors generated. These oxide composites were denoted as 2Ni₃(BO₃)₂/B₂O₃ (R_Ni/B = 1), Ni₃(BO₃)₂/NiO (R_Ni/B = 2), Ni₃(BO₃)₂/3NiO (R_Ni/B = 3) respectively.

The oxide composite was pressed at 10 MPa into a cylindrical pellet (13 mm in diameter, 1.0–1.2 mm in thickness, ca. 0.5 g in mass), which was sintered at 300^°C for 2 h. Then, the sintered pellet was wrapped with a foamed Ni layer and a Mo wire (0.18 mm in diameter) onto a Mo rod (2 mm in diameter, 1 m in length) that functioned as the current connector. This assembled electrode was used as the cathode and electrolyzed against a graphite rod anode at 700^°C. The electrolyte was the CaCl₂-NaCl mixture at equi-molar ratio. Typically, anhydrous CaCl₂ (600 g, purity:>96 wt%, content of Mg and alkali metals: ∼0.3 wt%) and NaCl (291 g) were mixed and dried at 300^°C before melting at 700^°C in a graphite crucible placed in a stainless steel reactor under argon protection. Pre-electrolysis was performed at 2.6 V for 10 h between a nickel foil cathode and a graphite rod anode.

Investigations of the catalytic activity of the prepared nickel boride powders were pursued using the reduction of 4-nitrophenol (4-NP) to 4-aminophenol (4-AP) by NaBH₄ as a model reaction at 26^°C. Aqueous 4-NP solution (50 mL, 0.266 mM) was mixed with fresh NaBH₄ (13.3 mM). The prepared nickel borides powder (20 mg) was dispersed in the mixture. UV-Vis absorption spectra (TU-1901, Beijing Purkinje General Instrument Co., Ltd.) were recorded to determine the variation of the maximum absorption intensity in the wavelength range of 250–500 nm.

Precursors and products were characterized by X-ray diffraction spectroscopy (XRD, Shimadzu X-ray 6000 with Cu Kα₁ radiation at λ = 1.5405 Å), scanning electron microscopy (SEMFEI Sirion Field Emission Gun SEM), and energy dispersive X-ray spectroscopy (EDAX, GENESIS 7000). A camera was used to record the color change of 4-NP during the catalytic reduction experiment.

Because of the low melting point of B₂O₃ (450^°C), if metal borides, such as Ni-B, are targeted by the electrolysis of solid compound precursor in molten chlorides, it is desirable to react the B₂O₃ with NiO to form the compounded oxides with higher melting points, such as Ni₃(BO₃)₂.³² Although the Ni₃(BO₃)₂ can be easily prepared by annealing NiO in molten B₂O₃ at about 1200^°C, the generated large grains of the compound may be unfavorable for the subsequent electrolytic reduction. Alternatively, here, the sol-gel method was applied to prepare the nickel borate with fine particle sizes.³⁰ In this study, since in the Ni₃(BO₃)₂ there is R_Ni/B = 1.5, to get a ratio of 1, 2 or 3, excess source of B or Ni should be added during the sol-gel deposition process, thus composites of 2Ni₃(BO₃)₂/B₂O₃, Ni₃(BO₃)₂/NiO, Ni₃(BO₃)₂/3NiO would be collected after the calcination.

Fig. 1a to 1c show the XRD patterns of the prepared oxide composites. When R_Ni/B = 2 or 3, Ni₃(BO₃)₂ and NiO were the dominant phases in the composite. Fig. 1d and 1e indicate that the generated Ni₃(BO₃)₂/3NiO and Ni₃(BO₃)₂/NiO composites were in a uniform morphology, with a porous structure building from fine particles of 0.1–1 μm in size.

Zoom In Zoom Out Reset image size

For R_Ni/B = 1, boron oxide should be present in the 2Ni₃(BO₃)₂/B₂O₃ composite. Since boron oxide is in amorphous structure in nature, it cannot be detected by XRD, and Fig. 1c shows only the pattern of Ni₃(BO₃)₂. The SEM image in Fig. 1f shows a lot of particles fused by dense substance, which may be amorphous B₂O₃.

The Ni₃(BO₃)₂/3NiO cathodes were electrochemically reduced by applying a cell voltage of 3.0 V for different electrolysis times. Fig. 2 shows the typical current-time plots at 700^°C in CaCl₂-NaCl. The trend of current was very similar to those of other metal oxides in some reported work.^21,24,31,33 The inset of Fig. 2 shows that the current rose to a high value (ca. 5.66 A for one pellet with a diameter of 1.3 cm) in the first 1 min of electrolysis, and then decreased quickly through a few plateau in the next 40 minutes, indicating a fast reduction process, and at last declined gradually until it reached a background current of about 40 mA. The larger reduction current of the Ni₃(BO₃)₂/3NiO composite in comparison with other oxides^21,24,31,33 could be attributed to (1) the thermodynamic advantage for the reduction of NiO (2NiO = 2Ni + O₂, ΔG = 301.8 kJ/mol vs., for example, TiO₂ = Ti + O₂, ΔG = 767.4 kJ/mol at 700^°C) and (2) the electronic conductivity of NiO, which allows the happening of electrochemical reaction at the solid | electrolyte interface. However, the metal | oxide | electrolyte three-phase interlines (3PIs) may be still most favorable to the reduction considering the higher conductivity of the metal. The reduction of the interior oxide of the pellet could have resulted in the following current decrease due to the increasing mass transfer difficulty.

Zoom In Zoom Out Reset image size

There are about three current stages as reflected by the slope changes of the current-time curve shown in Fig. 2, indicating the reduction going through a number of mechanistic and/or kinetic steps. XRD analysis was performed to identify the respective intermediate products obtained after different electrolysis times at 3.0 V. As shown in Fig. 3, the XRD pattern of the 5-minute electrolysis product shows distinctive peaks of nickel, accompanied by small peaks of Ca₃(BO₃)₂. Both the XRD peaks of NiO and Ni₃(BO₃)₂ decreased significantly compared to the pattern of the precursor shown in Fig. 1a, suggesting the rapid reduction of NiO and Ni₃(BO₃)₂ to nickel. The reduction of borate ions should be slower, which converted into Ca₃(BO₃)₂.²⁹ Weak peaks of Ca(OH)₂ can also be found. It was suggested that a quick release of O²⁻ from the metal oxide could result in the precipitation of CaO,^34–36 which would hydrolyze to Ca(OH)₂ during water washing.

Zoom In Zoom Out Reset image size

The 5-minute electrolysis sample was further studied by the SEM and EDX analysis, and the results are shown in Fig. 4. It can be seen some large particles with sizes of 2–5 μm blending with some small particles with sizes less than 1 μm, which contain Ca, Ni, B, Pt, and O as determined by the inserted EDX spectrum. The Pt element came from the sputter treatment of the sample for SEM observation. According to the point element analysis with EDX, the large particle might be Ca₃(BO₃)₂, and the small particle might be Ni or Ni₃B.

Zoom In Zoom Out Reset image size

After 15 minutes electrolysis, the XRD peaks of NiO and Ni₃(BO₃)₂ disappeared, indicating the complete metallization of the nickel ions. At the same time, both the peaks of Ni and Ca₃(BO₃)₂ weakened, suggesting the reduction of Ca₃(BO₃)₂ on Ni and the formation of nickel borides. After 30 minutes, the product only showed the XRD peaks of Ni, Ni₂B and Ni₃B, indicating the almost completion of the electrochemical reduction. This is in line with the I-t curve as shown in Fig. 2, where the current declined to a background value after the 30-minute electrolysis. The left electrolysis would reduce the trace of oxides but might mainly contribute to the homogenization of the product by the reaction between Ni and Ni₂B. It can be seen that the Ni and Ni₂B phases gradually decreased, and Ni₃B become dominant in the product after 120-minute electrolysis. The 4-h electrolysis resulted in a yield of Ni₃B higher than 95%.

Based on the above discussion, the electro-reduction of Ni₃(BO₃)₂/3NiO to Ni₃B in molten CaCl₂-NaCl should have proceeded through the following steps,

Reaction 4 might occur in cases there were local uneven distribution between Ni and Ca₃(BO₃)₂. The generated Ni₂B would convert to Ni₃B through a solid state reaction with Ni (Reaction 5). This solid reaction took about 90 minutes as reflected in Fig. 3, and finally Ni₃B was produced from the Ni₃(BO₃)₂/3NiO precursor.

Fig. 5 shows typical current-time plots for the electrolysis of 2Ni₃(BO₃)₂/B₂O₃ in the CaCl₂-NaCl melt at 3.0 V and 700^°C. The plots recorded from the electrolysis of Ni₃(BO₃)₂/3NiO were also displayed for comparison. Compared to Ni₃(BO₃)₂/3NiO, there was a long and low current plateau (from 15 min to 50 min) emerging during the electrolysis of 2Ni₃(BO₃)₂/B₂O₃. This plateau current can be attributed to slow reduction of Ca₃(BO₃)₂, considering the large amount of borate ions in the 2Ni₃(BO₃)₂/B₂O₃ composite.

Zoom In Zoom Out Reset image size

Fig. 6a shows the phase transformation during the electrolysis of 2Ni₃(BO₃)₂/B₂O₃. It can be seen that similar to the electrolysis of Ni₃(BO₃)₂/3NiO, both Ni and Ca₃(BO₃)₂ formed in the first 5 min electrolysis, suggesting again the relatively fast reduction of Ni₃(BO₃)₂ to Ni and borate ions. However, the 15-min electrolysis product still showed the distinct XRD pattern of Ni₃(BO₃)₂, probably indicating that the reduction of Ni₃(BO₃)₂ was slightly slower than the reduction of NiO if compared to Fig. 3. At the same time, the amount of Ca₃(BO₃)₂ increased due to the reaction between the calcium ions in the melt and the released borate ions from the Ni₃(BO₃)₂. Another argument might be that there was more Ni generated initially from the Ni₃(BO₃)₂/3NiO composite, which would be favorable for the reduction of both Ni₃(BO₃)₂ and Ca₃(BO₃)₂.

Zoom In Zoom Out Reset image size

The slower reduction kinetics of Ca₃(BO₃)₂ was further confirmed by the experimental results from electrolysis at different cell voltages. As shown in Fig. 6b, after electrolysis at 2.0 V for 2 hours, the XRD peaks of Ni₃(BO₃)₂ completely disappeared, and the XRD pattern shows mainly metallic Ni and Ca₃(BO₃)₂. It seems that increasing the cell voltage to 2.4 V had little help to the reduction of Ca₃(BO₃)₂, considering that Ni was still the predominant product and only a small amount of Ni-B alloys formed. The effective reduction of Ca₃(BO₃)₂ could occur at 2.8 V, and the Ni-B alloys became predominant after the 2 h electrolysis as evidenced by the XRD analysis (Fig. 6b). At the same time, the XRD peaks of Ni weakened significantly. However, there were still small XRD peaks of Ca₃(BO₃)₂ in the 2 h electrolysis product, indicating again the great difficulty in the reduction of Ca₃(BO₃)₂ even at a cell voltage of 2.8 V.

The reduction of Ca₃(BO₃)₂ could be accelerated by imposing a higher cell voltage. As can be seen in Fig. 6a, the electrolysis after 60 minutes at 3.0 V product shows no XRD peaks of Ca₃(BO₃)₂, indicating most of the oxides in the cathode have been reduced. Similar to the electrolysis of Ni₃(BO₃)₂/3NiO, the post-electrolysis has mainly contributed to the alloying process, with the metallic Ni converting to Ni₂B gradually. After the 4 h electrolysis, the XRD peaks of Ni₂B became dominant in Fig. 6a, while those of Ni almost disappeared.

It should be pointed out that although the atomic ratio of Ni/B in the 2Ni₃(BO₃)₂/B₂O₃ was designed to be 1:1, as discussed above, the final electrolysis product was mainly Ni₂B. No NiB product was detected by the XRD analysis. The formation of NiB could have suffered from the thermodynamic difficulty in comparison to those of Ni₃B and Ni₂B considering electrode reactions (6–12), where the thermodynamic potentials were referred to the equilibrium potential of Ca/Ca²⁺. Reaction 14 indicates that the reduction of NiO to Ni can be easily realized considering a thermodynamic potential of as positive as 1.98 V. In comparison, the reduction of Ca₃(BO₃)₂ to B with E⁰ = 0.547 V would be much more difficult. However, it would be favorable for the reduction of Ca₃(BO₃)₂ on Ni with Ni-B alloys formed. Particularly, the formation potential of Ni₃B positively shifts to about 0.816 V, and the fast electrolysis of Ni₃(BO₃)₂/3NiO to Ni₃B has been observed as discussed above. The reduction potential (0.677 V) of Ca₃(BO₃)₂ on Ni to NiB is also higher, but Ni₃B or Ni₂B would form preferentially. Then, the potentials for the consequent reduction of Ca₃(BO₃)₂ on Ni₃B and Ni₂B negatively shift to 0.608 and 0.595 V respectively, suggesting increasing difficulty in the generation of NiB.

The mismatch in the atomic ratio of Ni/B between the precursor and the outcomes for the electrolysis of 2Ni₃(BO₃)₂/B₂O₃ could be due to the dissolution of B₂O₃ into the melt at 700^°C, whose melting point is only 450^°C. The product was a mixture of Ni₂B and Ni₃B with a nickel to boron atomic ratio higher than 2, suggesting some of the Ca₃(BO₃)₂ or borate ions could have also slightly dissolved. Similar phenomenon was observed during the electrolysis of Ni₃(BO₃)₂/NiO aiming to Ni₂B powders. After 4 h electrolysis at 3.0 V, although the product consisted of mainly Ni₂B, there exist clear XRD peaks of Ni₃B in Fig. 7a, whose content was higher than in the electrolysis product from 2Ni₃(BO₃)₂/B₂O₃ as expected.

Zoom In Zoom Out Reset image size

On the other hand, the yields from the electrolysis of Ni₃(BO₃)₂/NiO and 2Ni₃(BO₃)₂/B₂O₃ were about 86% and 82% respectively, probably due to the loss of B according to the XRD analysis. These observations suggested that more source of B in the oxide composite precursor would be needed for the preparation of pure Ni₂B powders through this electrolysis method.

The above relatively purer Ni₃B and Ni₂B electrolyzed from the Ni₃(BO₃)₂/3NiO and 2Ni₃(BO₃)₂/B₂O₃ were tested to be used as catalysts for the reduction hydrogenation of p-nitrophenol (4-NP). As shown in Figure 8, both the Ni₃B and Ni₂B exhibited uniform nodular particles with sizes of 200∼500 nm, which might be able to provide large surface area for the heterogeneous catalysis. Small amounts of O in the two borides can also be detected by the EDX analysis. This was usually ascribed to the formation of thin layer coating of oxides on the Ni₃B particles formed during water washing.^{25–28,37,38}

Zoom In Zoom Out Reset image size

4-Aminophenol (4-AP) has many important applications, which can be used as analgesic and antipyretic drugs, photographic developer, corrosion inhibitor, anticorrosion lubricant, and so on.³⁹ 4-AP was usually prepared by reduction of 4-NP with NaBH₄ in the presence of catalysts.^40,41 The catalytic reduction process can be represented as,

It is interesting that the synthesized Ni₃B and Ni₂B powders are ferromagnetic as shown in Fig. 9a. This will be very beneficial for the reclamation of these catalysts. In this study, during the catalytic reaction between the 4-NP and NaBH under magnetic stirring, the Ni-B alloy powders were always caught on the stirring bar. Since there will be a remarkable color change from the 4-NP solution (light yellow) to 4-AP (colorless), UV-vis analysis was used to monitor the reduction process (Fig. 9b). Aqueous solution of 4-NP shows a distinct spectra profile with an absorption maximum at 317 nm, which shifts to 400 nm in the presence of NaBH₄ due to the formation of the 4-nitrophenolate ion.⁴² Without addition of the Ni-B alloy catalysts, the color of aqueous solution kept almost unchanged during standing at room temperature. The color changed immediately after the addition of the Ni-B alloys. It can be seen from Fig. 9b that the time-dependent absorption spectra show a decrease in the intensity of the absorption peak at 400 nm and a concomitant increase of a new peak at 298 nm, which is the sign of the generation of 4-AP. After about 30 minutes, the peak due to the nitro compound was no longer observed, indicating the completion of the reduction of the 4-NP.

Zoom In Zoom Out Reset image size

The reaction rate was assumed to be independent of the concentration of sodium borohydride since this reagent was used in large excess. Therefore, the kinetic data could be fit with the first-order rate law:⁴³

Since the absorbance of 4-NP is proportional to its concentration, the ratio A₀/A_t (A₀ the initial absorbance and A_t the absorbance at time t of the solution at 400 nm) should be equal to the ratio of the corresponding concentrations of 4-NP (C₀/C_t). Indeed, a good linear relationship between ln(C₀/C_t) and reaction time was found for the catalytic reduction of 4-NP (Fig. 10). The regression analysis suggest rate constants k of about 0.12 and 0.10 min⁻¹ in the presence of Ni₂B and Ni₃B powders respectively, which are comparable to the catalytic activity of nickel nanoparticles.⁴⁴

Zoom In Zoom Out Reset image size

This ferromagnetic property of the nano-crystalline Ni₃B and Ni₂B makes them distinguished from those amorphous Ni-B alloy catalysts, which were non-magnetic and difficult to be recaptured from the reaction system. The crystalline Ni-B alloys particularly Ni₃B display good magnetic property.⁴⁵ Pure Ni₂B is non-ferromagnetic in nature. Here the electrolysis prepared Ni₂B powders showed enough ferromagnetism to be captured by the stirring bar, probably due to they were actually a mixture of Ni₂B and Ni₃B. In addition, the crystalline Ni₃B and Ni₂B alloy catalysts would be more stable than those amorphous Ni-B alloys. During repeated use as hydrogenation catalysts, spontaneous crystallization of the amorphous alloys was often found and could lead to segregation of Ni and B.⁴⁶ These suggest that the electrolysis generated nano-crystalline Ni-B alloys are promising hydrogenation catalysts. In future studies, the stability and cyclicity of these nano-crystalline Ni-B alloy catalysts need to be tested, and the correlation between the magnetic nature and the catalytic performance could be an interesting topic for the magnetic catalysts.