1.   Introduction

Metal hydride/nickel (MH/Ni) batteries have been widely studied due to the numerous advantages, such as high-power characteristics, long cycle life, and intrinsic safety [1–2]. However, the performance of MH/Ni batteries still requires improvement. In this regard, the materials of the negative electrode play an important role in battery performance, hence they are extensively investigated in recent years [3–5].

Rare earth–magnesium–nickel-based superlattice alloys have been investigated as materials for high capacity negative electrodes in MH/Ni batteries [6]. Although these alloys provide elevated capacity, they have failed to behave well when discharged at high current densities. In addition, they suffer from poor cycling stability due to more chemically reactive elements (La, Mg, etc.) and a lower proportion of B site metals (Ni, Co, etc.) than AB₅-type rare earth-based hydrogen storage alloys [7–10]. Hence, elemental substitution was performed to improve electrochemical properties, and, in previous studies, substituting Nd for La was reported effective in improving kinetics and cycling stability [11–13].

To further compensate for the loss of properties, surface modification on an elementally fixed alloy could provide better stability since the most crucial steps in electrochemical reactions occur at the interface [14–17]. Previous studies achieved surface modification through acidic or basic solution treatments, surface reduction, and metallic or alloy surface coating. These studies were devoted to reducing charge transfer resistance and suppressing electrochemical polarization to promote electrochemical properties such as high rate dischargeability (HRD) [14,18–20].

Graphene has extensively been explored in recent years due to its large specific surface area, high electron transferring rate, and excellent stability. For MH/Ni batteries, Cui et al. [21] employed ball milled MmNi_3.55Co_0.75Mn_0.4Al_0.3 with graphene nanoplates to improve the HRD from 53.0% to 68.3% at 3000 mA·g^–1. Huang et al. [22–23] prepared a graphene/Ag nanocomposite to modify the electrochemical properties of Mg₆₅Ni₂₇La₈ alloy. The corrosion resistance and cycling stability of the as-prepared electrodes were improved. Li et al. [24] reported carbon skeleton/Mg₂Ni free-standing electrodes with a 78% charge capacity retention after 50 cycles and elevated a retention rate. Lan et al. [25] prepared Ni-reduced graphene oxide (rGO) composite using the hydrothermal process and, subsequently, ball milled it with AB_3.5-type alloys. The obtained composite showed enhanced HRD. For hydrogen storage electrodes, graphene oxide (GO) has also been employed to improve the hydrogen storage materials during the hydriding/dehydriding processes due to its catalytic and conductive characteristics [26–28].

To improve the electrochemical properties of Nd–Mg–Ni-based alloys, we attempted to coat the alloy particle surface with an rGO modification layer. The microstructure and electrochemical properties of the alloys before and after modification were examined and the results were discussed.

2.   Experimental

The Nd–Mg–Ni alloy Nd_0.67Mg_0.33Ni_3.0 was prepared by induction melting of Nd, Mg, and Ni metals (>99.5% purity) by taking Mg evaporation loss into consideration. The ingots were annealed at 900°C for 10 h then cooled down naturally, crushed, ground, and sieved. The yields of the standard sieves, 200 and 400 meshes, were collected for surface treatment and subsequent tests.

GO was prepared according to the Hummer’s method [29]. During surface treatment, GO (4, 10, 20, and 30 mg) was added into 200 mL deionized (DI) water and ultrasonically stirred for 30 min to yield homogenous suspensions (parameters for the ultrasonic instrument were 240 W power and 40 kHz frequency). Afterward, Nd–Mg–Ni alloy powder (2 g) was dropped into the suspension, and vitamin C (8, 20, 40, and 80 mg) was subsequently added and stirred for 24 h until obtaining transparent solution. Here vitamin C was added in order to improve the conductivity of GO as a modifier of the electrode [30]. During stirring, the GO sheets were reduced by vitamin C, separated from the solution and encapsulated on the alloy surface. The treated alloy powders were then filtered off, rinsed, and freeze-dried. The GO coats on the alloy surfaces were estimated to be 0.2wt%, 0.5wt%, 1.0wt%, and 1.50wt%, and the GO suspensions were either achromatic color or transparent. Also, the amount of vitamin C was always set as twice that of the stoichiometry. The modification process is illustrated in Fig. 1.

Figure  1.  Surface modification process for Nd–Mg–Ni alloys by rGO.

下载: 全尺寸图片 幻灯片

Electrochemical tests were performed in a standard three-electrode system, consisting of MH as the working electrode (a slurry consisting of alloy powders, fine nickel powders, and 3wt% polyvinyl alcohol aqueous solution was pasted onto two sides of a nickel foam sheet, and pressed tightly), Ni(OH)₂/NiOOH as the counter, and Hg/HgO as the reference electrode. The electrolyte was 6 M KOH aqueous solution. All electrochemical properties were tested at 298 K. Details of the testing process were described in our previous study [31]. For specific discharge capacity calculation, the mass of the rGO modified on the alloy surface was involved. In the initial activation stage, the electrodes were fully charged/discharged at 60 mA·g^–1. For high rate discharge test, discharge capacity at different mass-specific current of 60, 300, 600, 900, 1200, and 1500 mA·g^–1 was recorded. HRD was subsequently calculated according to the following equation:

where HRD denotes high rate dischargeability, %; C_i is the discharge capacity at a certain discharge current density; mAh·g^–1, C_max is the maximum discharge capacity, mAh·g^–1. Linear polarization and anodic polarization testing data were collected at 50% depth of discharge (DOD) on a CHI660B electrochemical workstation. Testing was performed at the scanning rates of 1 and 5 mV·s^–1 in the potential range of −5‒5 mV and 0‒500 mV (vs. open circuit potential), respectively. Electrochemical impedance spectrum (EIS) of the electrode was obtained in the frequency range of 1 MHz to 0.1 Hz with alternating current (AC) amplitude of 5 mV.

The X-ray diffraction pattern (XRD) of the materials was recorded on a Rigaku D/Max-2500/PC instrument with Cu K_α radiation in the 2θ range from 10° to 80° and a scanning rate of 4° min^–1. The scanning electron microscopy (SEM) images were obtained on a Hitachi S-4800 microscope equipped with an energy-dispersive X-ray operating at 20.0 kV. Transmission electron microscopy (TEM) was performed using a JEM-2010 microscope. Infrared testing was performed on an E55+FRA106 infrared spectrophotometer.

3.   Results and discussion

3.1   Microstructure

Phase structure of the alloys before and after modification is characterized by XRD and the patterns are presented in Fig. 2. The crystal structure of the alloys all included two components, Gd₂Co₇-type and Ce₂Ni₇-type phases. They indicated that the matrix alloy was composed of an A₂B₇-type structure. After modification of the surface with rGO, no remarkable changes emerged in the XRD patterns, suggesting that the mild conditions exerted little influence on the phase structure, and that the trace amount of rGO modifier did not support detection of rGO.

Figure  2.  XRD patterns of the alloys before and after rGO assembling treatment.

下载: 全尺寸图片 幻灯片

Morphologies of the alloys before and after modification were characterized by TEM and SEM. The untreated alloy presented a smooth surface as shown in Fig. 3(a). In contrast, the treatment in the thinly layered rGO suspension formed a cotton wadding modification layer on the alloy surface (Fig. 3(b)), revealing a side view of packed rGO and a successful coating process. Thickness of the coating rGO layer was estimated to be about 200–300 nm.

Figure  3.  TEM morphology of (a) untreated alloy and (b) rGO treated alloy (1.0wt% rGO).

下载: 全尺寸图片 幻灯片

Due to the large size of the alloy particles, the TEM images revealed only partial feature of the alloy particles. Therefore, SEM characterization was used to gain a panoramic view of the particle surface, and the images are collected in Fig. 4. The untreated alloy presented a smooth surface (Fig. 4(a)). Also, some fragments formed during mechanical crushing appeared on the alloy particle surface. Figs. 4(b)–4(e) show the alloy surface after modification with different amounts of rGO. At a modification concentration of 0.2wt% (Fig. 4(b)), negligible rGO appeared on the surface, in which rGO adhered to the alloy surface through a complexation effect between the functional groups (–COOH, –OH) and metal ions on the alloy surface. The complexation yielded firm modification by rGO. As the concentration of GO solution increased, the modification layer with network structure became much thicker, and encapsulation, more uniform.

Figure  4.  SEM morphology of the alloys before and after rGO surface modification: (a) untreated alloy; (b) treated with 0.2wt% rGO; (c) treated with 0.5wt% rGO; (d) treated with 1.0wt% rGO; (e) treated with 1.5wt% rGO.

下载: 全尺寸图片 幻灯片

This network-structured modification layer was then tested for rGO by infrared spectra analysis (Fig. 5(a)). With the peak at 3430 cm⁻¹ corresponding to the stretching vibration of –OH, this peak, together with the one at 1680 cm⁻¹ (corresponding to C=O), proved the functional group of –COOH. Peaks at 1620, 1415, and 1068 cm⁻¹ were attributed to C–O–C, C–OH, and C–O stretches, respectively. It was clear that rGO was modified on the alloy surface since the functional groups mentioned above are usually characteristics of rGO. To reveal what happened when alloys, before and after modification, made to contact with electrolyte 6 M KOH, three photos were taken, shown in Figs. 5 (b)–5(d).

Figure  5.  Infrared results and wettability test for alloys by the 6 M KOH solution: (a) IR results; (b) wettability test for alloy before modification; (c) wettability test for alloy treated in GO suspension for 12 h; (d) wettability test for alloy treated in GO suspension for 24 h.

下载: 全尺寸图片 幻灯片

The three photos are of alloy before modification (Fig. 5(b)), alloy modified by soaking in GO suspension for 12 h (Fig. 5(c)), and alloy treated for 24 h (Fig. 5(d)), respectively. In Fig. 5(b), the alloy surface was shining and smooth, and the 6 M KOH was on the alloy surface, like a semi-ball, indicating poor wettability of untreated alloy by KOH solution. In Fig. 5(c), the alloy surface was coarse and gray, and the KOH solution tended to spread out on the alloy surface as a thin layer. In Fig. 5(d), the alloy surface was rather coarse and black, and the KOH dropping onto the surface was absorbed into the modification layer, with only a small proportion left on the surface. Thus, it could be concluded that the rGO modification promoted wettability of the alloy surface with 6 M KOH solution effectively. Encapsulation of the rGO modification layer with superior electron-conductivity and large specific surface area facilitated the charge transfer during charge-discharge, hence improving the electrode processes of the hydrogen storage alloys.

rGO appeared to be a good modifier on the alloy surface, and it might be more functional when kept stable on the alloy surface. To test the stability of the rGO modification layer on the alloy surface during charge‒discharge cycling, alloy powders after 100 charge‒discharge cycles were collected and then characterized by SEM (shown in Fig. 6). The untreated alloys depicted loose and needle-like morphologies. The increase in oxygen content and shape of the species identified the newly-formed species as Nd(OH)₃ [31]. Note that Nd(OH)₃ was not hydrogen‒absorbing and would scarcely provide protection to the electrode during charge‒discharge owing to its loose structure. Thus, Nd(OH)₃ brought no improvements to the electrode properties of Nd–Mg–Ni alloys. In addition to oxidation, pulverization was also induced from cracks in the middle of the particles. The cracks were due to in/out behavior of hydrogen atoms issued from the alloy bulk.

Figure  6.  SEM morphology of the alloys after charge/discharge tests: (a) untreated alloy; (b) 1.0wt% rGO treated alloy.

下载: 全尺寸图片 幻灯片

In comparison, the alloy electrode treated with 1.0wt% rGO showed neither Nd(OH)₃ nor cracks on the surface after long charge‒discharge service, ascribable to the protective function and hydrogen transferring effect. At first, rGO looked rather stable in KOH electrolyte, and its coverage of the surface undoubtedly provided a protective layer to the alloy particles, preventing direct corrosion from KOH solution. On the other hand, the alloy grain boundary was more active in the rGO modification than the other parts, and provided easy modification sites. The rGO modified at grain boundaries played an important role in hydrogen conductivity, helping the atoms to enter and exit the alloy at suitable site, and avoiding fierce crushing from hydrogen stream. Consequently, no noticeable cracks formed in the modified particles, effectively preventing the alloys from pulverization and subsequent oxidation.

3.2   Electrochemical property

Fig. 7 shows the discharge capacity of Nd–Mg–Ni alloys as a function of cycle number, at the initial charge‒discharge test (Fig. 7(a)) and for 100 cycles of service. In Fig. 7(a), rGO modification delivered a slight increase in discharge capacity. This was because the rGO modification was able to suppress the electrochemical polarization owing to its electro-catalytic activity, and to suppress the concentration difference owing to its high hydrophilicity. These led to more complete discharge, while in Fig. 7(b), rGO modification was found to be effective in suppressing capacity degradation during long cycling. After modified with rGO the discharge capacity loss decreased from 49.2 mAh·g^–1 (untreated) to only 38.2 mAh·g^–1 (1wt% rGO) after 100 cycles.

Figure  7.  Discharge capacity of Nd–Mg–Ni alloy electrodes with and without rGO modification as a function of cycle number: (a) 20 cycles; (b) 100 cycles.

下载: 全尺寸图片 幻灯片

Surface treatment was believed to promote charge transfer rate on the electrode surface, improving HRD. The SEM characterization showed that treating solution with a higher GO concentration induced more rGO modification on the alloy surface while changes in HRD did not follow this trend. HRD reached an optimum value at modification amount 1.0wt% (as seen in Fig. 8(a)), then dropped as the modification amount further increased to 1.5wt%. At a discharge current density of 1500 mA·g^–1, the HRD value increased remarkably from 53.2% to 83.9% (increased by 30.7%) when the alloy was treated with 1.0wt% rGO. Improvement in electrochemical kinetics was also found in the mid-discharge potential of the electrodes (plotted in Fig. 8(b)). Optimum treatment led to an increase of 0.073 V. The higher mid-potential was ascribed to less polarization on the electrodes.

Figure  8.  Electrochemical kinetic properties of Nd–Mg–Ni-based alloys with and without rGO modification: (a) HRD plots of the alloy electrodes; (b) discharge potential of the alloy electrodes; (c) linear polarization plots for alloy electrodes at 50% DOD with a scanning rate of 1 mV·s^–1; (d) anodic polarization plots for alloy electrodes at 50% DOD with a scanning rate of 5 mV·s^–1; (e) EIS of the alloy electrode at 50% DOD at AC amplitude of 5 mV.

下载: 全尺寸图片 幻灯片

To reveal the electrochemical kinetics mechanism changes in the charge transfer rate at the electrode surface, a diffusion process, in bulk, was tested by both linear and anodic polarization. In the linear polarization profiles, all appeared linear within the small potential scope (–5‒5 mV vs. open circuit potential) (Fig. 8(c)). Using these profiles, the exchange current density characterizing charge transfer rate can be calculated by Eq. (2) [32]. Calculated results are listed in Table 1.

Table  1.  Electrochemical parameters of alloys with and without treatment

Sample	Cmax / (mAh·g−1)a	HRD1500 / %b	Mid-potential / (V vs. Hg/HgO)c	I0 / (mA·g−1)d	IL / (mA·g−1)e
Untreated	314.6	53.2	−0.810	207.1	3353.3
0.2wt% rGO	317.7	63.2	−0.821	236.3	3453.3
0.5wt% rGO	323.4	75.8	−0.823	246.8	3486.7
1.0wt% rGO	321.9	83.9	−0.849	255.5	3766.7
1.5wt% rGO	319.2	69.4	−0.830	249.8	3653.3
Notes: a maximum discharge capacity; b high rate dischargeability at 1500 mA·g−1; c potential at 50% depth of discharge; d exchange current density; e limiting current density.

下载: 导出CSV 

| 显示表格

where I₀ is the exchange current density, mA·g^–1; I_d is polarization current density, mA·g^–1; R is the gas constant, J·mol^–1·K^–1; T is the absolute temperature, K; η is the overpotential, mV; and F is the Faraday constant, C·mol^–1. The calculated exchange current density increased as rGO modification was performed; furthermore, when the modification amount was 1.0wt%, the exchange current density appeared to be higher than that of the others. It indicated that the alloy with 1.0wt% rGO modification possessed a higher charge transferring rate at the alloy electrode surface.

Fig. 8(d) presents the anodic polarization profiles of the electrodes. In each line, the peak current value was recognized as limiting current density, which was related to hydrogen diffusion processes in the bulk of the alloys. When the applied overpotential was small, hydrogen diffusion in the bulk was fast enough to provide sufficient hydrogen for charge transfer reaction at the surface. Consequently, current density increased with applied overpotential. But as overpotential increased further, there was a point at which the hydrogen concentration at the surface dropped to nearly zero, owing to a limiting hydrogen diffusion rate in the bulk of the alloys. Subsequently, the current density decreased as overpotential increased further. Thus, the peak in the current density was related to the hydrogen diffusion rate in the alloy bulk. Limiting current density is listed in Table 1, confirming that rGO modification promoted the hydrogen diffusion rate in the bulk of the alloys [33].

Electrochemical impedance spectra of the electrode are shown in Fig. 8(e). It can be seen that each spectrum consisted of two semicircles and a straight line. It was reported that the semicircles in the high-frequency and low-frequency regions corresponded to the contact impendence and the charge transfer resistance (R_ct), respectively, and that the sloped straight line corresponded to Warburg impendence [34]. There was almost no difference between the small semicircles of each spectrum, which meant that the contact impedance differed little between electrodes. However, the radius of the large semicircle decreased as a function of rGO modification, and the 1.0wt% rGO modification induced the smallest radius, which meant that the alloy with 1.0wt% rGO modification showed the lowest charge transfer resistance.

3.3   Discussion

Several steps were included over the entire anodic process, including diffusion of hydrogen atoms from the bulk to the electrode surface, anodic oxidation, and diffusion of OH^– from the bulk of the electrolyte to the electrode (Eqs. (3)–(5)). As was stated, after rGO modification, the charge transfer resistance decreased and exchange current density increased, which indicated an easier occurrence of the reaction in Eq. (4). The fast charge transfer delivered a low concentration at the electrode surface. Then, a larger difference in hydrogen concentration between that in the bulk and that at the surface was presented. The larger difference was a greater driver of hydrogen diffusion in the alloy bulk. This, in turn, improved hydrogen diffusion in the bulk of the alloys, indicated by an increase in limiting current density. Thus, the reaction in Eq. (3) was promoted. As for the third step expressed in Eq. (5), it was a more direct consequence of rGO modification. The functional groups in rGO, such as –COOH and –OH, were hydrophilic, and increased the wettability of the electrode surface, therefore more OH^– existed around the electrode to take the resultants away from the electrode. Removal of extra charges promoted the entire reaction and the initial activation step. As was stated above, the entire anodic electrode process in Eq. (6) might get a significant promotion, owing to rGO modification.

Besides promoting reactions in the anodic process, rGO was also a good modifier for improvement of the cycling stability of the alloy electrodes. Without modification, larger amounts of Nd(OH)₃ species would remain during cycling. Note that Nd(OH)₃ was not hydrogen‒absorbing and would scarcely provide protection to the electrode during charge‒discharge, owing to its loose structure. Thus, Nd(OH)₃ was clearly detrimental to the electrochemical properties of Nd–Mg–Ni alloys. By contrast, the rGO modifier suppressed formation of a loose hydroxide layer, but induced formation of a dense metal oxide layer, which acted as a protectant of the alloy surface, thus improving cycling stability.

4.   Conclusions

The reduction process induces coverage of the alloy surface with rGO to form a net-shaped layer. As the concentration of treatment solution increases, the coating layer becomes thicker. The formed rGO layer improves the overall electrochemical properties of electrodes. The rGO layer facilitates charge transfer owing to its high conductivity and catalytic activity, consequently promoting hydrogen diffusion by exerting larger concentration difference between the bulk and the electrode surface. At a modification amount of 1.0wt%, kinetic properties are remarkably improved, with HRD at 1500 mA·g^–1 increasing by 30.7%, and mid-discharge potential rising by 0.073 V. The rGO layer remains stable during charge‒discharge cycling of the electrodes, enhancing cycling stability. Thus, the capacity loss after 100 cycles decreases from 49.2 mAh·g^–1 to only 38.2 mAh·g^–1. Modification of Nd–Mg–Ni alloys by coating rGO is a facile and effective method for improving electrochemical properties, especially electrochemical kinetics.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (NOs. 21303157 and 51771164), the Natural Science Foundation of Hebei Province (No. E2019203161), and Scientific Research Projects in Colleges and Universities in Hebei Province (No. QN2016002).