1. Discharge capacity

Figure 2 compares the discharge capacity of TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes synthesized by MA. These electrodes were prepared from the milled products that showed only trace amount of oxygen. Before obtaining high purity as-milled products, some failed milled alloys that were exposed to oxygen during their milling were subjected to XRD and electrochemical tests; the results were expectedly way inferior to those of Fig. 2. Not only the XRD patterns were full of crystallized oxidized compounds but also their corresponding discharge capacity were too inferior barely even reaching one tenth of what is presented in Fig. 2.

FIG. 2.

Evolution of discharge capacity of TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes with cycle number.

As can be seen, all electrodes require activation before reaching their maximum discharge capacities. The base binary TiNi intermetallic compound (x = 0) shows a maximum discharge capacity of over 176 mA h/g. Addition of Nb increases the maximum discharge capacity peaking at 197.5 mA h/g for the x = 0.1 intermetallic compound. However, the discharge capacities of the electrodes with higher Nb contents decreased owing to insolubility of this element in the whole compound as previously noticed in the XRD examinations [Fig. 1(a)]. Accordingly, the presence of slight amount of elemental Nb in the electrode accelerates the formation of oxides on the surface of powders in the corrosive alkaline environment. These oxides impede hydrogen transfer during electrochemical cycling. The maximum discharge capacity together with the discharge capacity in the 10th cycle for TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) intermetallic compounds are presented in Table I.

TABLE I.

Electrochemical discharge properties of TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) intermetallic compounds.

The activation behavior of all the electrodes was found to be very good, indicating that the high density of clean surfaces constitutes the electrodes after the milling process. The milled products and their associated amorphous/nanocrystalline structures have very high activity to absorb a small atom like hydrogen. Oxygen, on the other hand, is adsorbed very rapidly during the ball milling process and produces oxide layers on the particles’ surfaces which certainly hinder the hydrogen intake. These thin oxide layers formed during the synthesis process are responsible for the activation issue.^{Reference Lee, Lee, Yu, Fateev and Lee26} Nevertheless, upon hydrogen absorption/desorption cycles by the galvanostatic regime, these thin film oxides are gradually dissolved in the highly corrosive potassium hydroxide solution as the alloy powders undergo expansion and contraction making cracks in the oxides, consequently, new clean surfaces emerge. In this way, the oxide layers are broken away and the electrodes are activated revealing their highest discharge capacities. This issue is said to be marked for AB-type electrodes especially for TiFe that naturally has much more difficult activation process which has brought problems for its commercial use.^{Reference Abrashev, Spassov, Bliznakov and Popov27} However, ball milling has proved to be a very promising technique for improving the activation properties of hydrogen storage materials due to production of surface-modified amorphous/nanostructured materials having superior kinetics of hydrogen transfer.^{Reference Jurczyk, Smardz and Szajek18,Reference Lee, Lee, Yu, Fateev and Lee26}

Figure 3 shows the discharge curves of the electrodes using the current density of 80 mA/g in their maximum discharge capacity. As it is seen, the curves are characterized by a sloping plateau to potentials more negative than −0.8 V (versus Hg/HgO) after which the abrupt potential drop is observed. In other words, hydrogen desorption does not occur at a fixed potential and no discernible plateau is found for these electrodes. This behavior may be attributed to the wide distribution of energy levels of hydrogen absorption in the structure. It is mentioned that for milled powders, due to high degree of lattice distortion and high amount of amorphous/nanocrystalline fraction, there are different occupation sites with different energy levels for absorption of hydrogen atoms leading to release of the absorbed hydrogen in a range of potentials when discharging.^{Reference Rousselot, Guay and Roué24} After the depletion of hydrogen from the interior structure of the electrodes, the potential rapidly drops. It can also be deduced that higher fraction of the amorphous phase may intensify this effect as the sites with different energies are increased as a result of disorder found in amorphous/nanocrystalline structures. This phenomenon also happens when hydrogen is absorbed by making solid solution rather than forming hydrides which has been reported for the crystalline austenitic TiNi compound.^{Reference Cuevas, Latroche and Ochin12} Martensitic TiNi-based compounds, however, were clearly shown to desorb hydrogen through distinctive plateaus implying their hydride formation and multiphase nature upon hydrogenation.^{Reference Emami, Cuevas and Latroche4,Reference Cuevas, Latroche and Ochin12} In the present case, the structure of the studied intermetallic-based electrodes was found to be utterly amorphous/nanocrystalline; consequently, the energy levels for hydrogen absorption in these materials are unspecific resulting in a sloping plateau for hydrogen release. It is worth mentioning that the study of this phenomenon systematically requires derivation of pressure-composition isotherms (PCIs) by solid–gas hydrogenation or equivalent electrochemical PCI curves where hydrogen desorption occurs in an almost equilibrium potential with minimum polarization resistance. Nevertheless, for a rough approximation of dehydriding behavior of the present intermetallic compounds, polarized discharge curves were represented alternatively.

FIG. 3.

Discharge curves of TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes synthesized by MA at their maximum discharge capacity.

2. Cycle stability

Figure 4 demonstrates the cycle stability of TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes produced by MA after their activation cycle in terms of capacity retention (S _n). The cycle stability of the negative electrode material is universally shown with capacity retention (S _n) as

(1)$${S_n} = {{{C_n}} \over {{C_{{\rm{max}}}}}} \times 100\% \quad ,$$

where C _n denotes the discharge capacity at nth cycle and C _max denotes the maximum discharge capacity obtained for the electrode.

FIG. 4.

Capacity retention of TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes with cycle number.

Another parameter can also be defined which is indicative of capacity decay per cycle (R _n) as

(2)$${R_n} = {{{C_{n - 1}} - {C_n}} \over {{C_{n - 1}}}} \times 100\% \quad .$$

This parameter represents the capacity loss during one cycle. For the sake of comparison between the electrodes, S ₁₀ and R ₁₀ are presented in Table I.

As Fig. 4 and Table I suggest, slight substitution of Nb enhances the cycle stability but further addition impairs it in comparison with that of the binary (x = 0) base intermetallic compound. This could be ascribed to the relatively lower Ni content of the ternary intermetallic compound as Nb content increases that might eventually lead to higher oxidation of elements in KOH solution. However, better cycle stability of the ternary x = 0.2 intermetallic compound than that of x = 0.1 cannot be justified based on the Ni content. In fact, the presence of slight amount of nanostructured Ni homogeneously combined with the amorphous TiNiNb in x = 0.2 induces the “coating effect” on particles creating a nanocomposite which can thus act as a surface modifier or a catalyst in the whole compound suspending the oxidation process of the active material and, therefore, improving cycle stability.

Comparing these results with those presented by Wang et al.^{Reference Wang, Lei and Wang21} indicates somewhat different outcomes where the Ti₅₀Ni₄₁Nb₉ compound which corresponds rather closely to the x = 0.2 alloy in this study possessed higher electrochemical properties compared to x = 0 in terms of discharge capacity and cycle stability. This discrepancy could be associated with the totally different synthesis method of the ternary TiNiNb in both studies resulting in different structures of the powder electrodes, i.e., an amorphous/nanocrystalline structure in this investigation versus the fully crystalline structure of the studied alloy by Wang et al.^{Reference Wang, Lei and Wang21}

It is interesting to point that for all electrodes, the capacity loss after the cycle at which they reach their maximum discharge capacities is much higher than these at other cycles (Fig. 4). This is because after activation, the whole surface of particles is oxide-free and highly prone to oxidation by the corrosive environment; after being oxidized to some extent, the passive oxide layer protects the particles’ surface from further corrosion decelerating the oxidization kinetics, consequently, slowing the capacity drop.

The reason for capacity degradation during electrochemical charge/discharge cycling of TiNi-based intermetallic compounds is the gradual loss of the active material with oxidation by the formation of the oxide thin film on the surface of powders. These oxide layers impede hydrogen transfer by obstructing the process of hydrogen intake into the bulk of powders lowering the discharge capacity.^{Reference Guiose, Cuevas, Décamps, Leroy and Percheron-Guégan7} It should be stressed that the oxide layer formed on powders in the alkaline media must not be mistaken with the atmospheric oxide layer formed on TiNi-based intermetallic compounds during ball milling which causes the activation issue, as these two oxides must have different compositions, otherwise the activation stage would have no contribution to the discharge capacity revival. Yet, distinguishing between these two oxide thin films requires advanced surface microscopy and spectroscopy.

Any factor causing higher oxidation of the active material influences the cycle stability. For instance, pulverization is an important factor in which the powders are cracked during the electrochemical cycling and repetitive hydriding/dehydriding due to expansion and contraction of the crystal structure upon hydrogen absorption/desorption; the cracked particles with new clean surfaces are rapidly oxidized and in this way, the fraction of the active material is further reduced.^{Reference Gao, Zhang, Miao, Liu and Pan28} The less ductile an intermetallic-based powder or/and with excessive lattice volume change upon hydrogen transfer, the more prone it is to pulverization. The austenitic or amorphous TiNi compound is relatively a ductile intermetallic compound. In the present case, this was also confirmed by the excessive adherence of powders to balls and vial after milling. The superior electrochemical stability of TiNi-based compounds, especially the austenitic binary one, has been attributed to the remarkable passivating effect of oxide layers such that upon decrepitating of the active material, the newly formed oxides decelerate the oxidation process, consequently, lowering the detrimental effect of pulverization.^{Reference Cuevas, Latroche and Ochin12} By contrast, Zr-substituted TiNi compounds with monoclinic structure have been found to greatly suffer from pulverization mainly due to the loss of electrical contact between particles upon oxidation as a result of severe pulverization of the active material in these compounds.^{Reference Guiose, Cuevas, Décamps, Leroy and Percheron-Guégan7}

3. Kinetic properties

In this part, the kinetic properties of TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes are investigated using various electrochemical techniques. To directly evaluate the kinetic performance of these electrodes, the HRD of the activated electrodes is shown in Fig. 5 at the current densities of 160, 320, 480, 800, and 1200 mA h/g. HRD is calculated using the following equation^{Reference Rousselot, Guay and Roué24}:

(3)$${\rm{HRD}}\left( \% \right) = {{{C_y}} \over {{C_y} + {C_{80}}}} \times 100\% \quad ,$$

where C _y is the discharge capacity at y mA/g current density and C ₈₀ is the remaining capacity at a current density of 80 mA/g after the electrode has been discharged at y current density. The cut-off potential was −0.6 V versus the reference electrode.

FIG. 5.

HRD% for TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes at various current densities.

HRDs % of the electrodes at the current density of 1200 mA/g are listed in Table II. As can be seen in Fig. 5, all the electrodes preserve more than 60% of their capacities even at the high current density of 1200 mA/g, indicating their excellent catalytic properties. As it is obvious, the ternary x = 0.05 electrode has the highest HRD followed by the binary (x = 0) base intermetallic compound. The results indicate that slight addition of Nb improves the HRD by a small amount and further addition reduces it.

TABLE II.

Characteristic kinetic parameters of TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes.

In general, the HRD values for the electrodes are controlled by two factors: one is the charge transfer kinetics on the particles’ surface controlled by the activation polarization and the other is hydrogen diffusion kinetics into the bulk of particles controlled by concentration polarization.^{Reference Rousselot, Guay and Roué24,Reference Zhao and Ma29} In other words, the kinetic performance of hydrogen storage electrodes is controlled by both the kinetics of processes taking place at the electrode–electrolyte interface and the diffusion of hydrogen atoms into the particles’ lattice. As a consequence, the electrochemical kinetic parameters of an electrode-like exchange current density (I ₀), limiting diffusion current density (I _L), and hydrogen diffusion coefficient (D _H) are investigated to fully analyze the kinetic properties of TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) intermetallic compounds.

a. Determination of exchange current density

Figure 6 shows the linear polarization curve for TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes produced by MA. According to Butler–Volmer,^{Reference Zhao and Ma29} when the over potential is scanned in a small range, the amount of exchange current density (I ₀) can be calculated by

(4)$${I_0} = {{IRT} \over {F\eta }}\quad ,$$

where I is the applied current density (mA/g), R is the gas constant (J/mol K), T is the absolute temperature (K), F is the Faraday’s constant (C/mol), and η is the total over potential (mV). Therefore, the I/η ratio can be calculated from the slope of the line in Fig. 6.

FIG. 6.

Linear polarization curves for TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes at 50% DOD and 0.1 mV/s scan rate.

The calculated values of I ₀ are presented in Table II. I ₀ indicates the speed at which the oxidation or reduction reactions are taking place at equilibrium potential.^{Reference Tliha, Khaldi, Boussami, Fenineche, El-Kedim, Mathlouthi and Lamloumi1} In general, as the charge transfer increases on the hydride surfaces, I ₀ rises, lowering the activation polarization. The higher the exchange current density, the higher the electrocatalytic properties on the particles’ surface. For this reason, the particle size significantly affects this parameter such that smaller particles have higher reaction ability thus higher I ₀. The high I ₀ value of the x = 0.05 ternary intermetallic compound despite its larger particle size, especially in comparison with the x = 0 intermetallic compound, suggests its superior electrocatalytic activity among all the electrodes. In addition, the surface properties such as surface area, porosity, and surface oxides strongly influence the exchange current density.^{Reference Tliha, Khaldi, Boussami, Fenineche, El-Kedim, Mathlouthi and Lamloumi1} High reactivity of the present electrodes could also be related to the nanoporous surface morphology of the powders induced by MA. Beside surface properties, the state of charge and measurement method significantly affect the obtained value for I ₀. Consequently, comparing the I ₀ parameter from different studies may be misleading.^{Reference Tliha, Khaldi, Boussami, Fenineche, El-Kedim, Mathlouthi and Lamloumi1,Reference Zhao, Ma, Yang, Ding and Shen30} In the case of studies nearly identical to the present work, the great electrocatalytic behavior of MAed TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) intermetallic compounds can be observed comparatively.

b. Determination of limiting diffusion current density and hydrogen diffusion coefficient

Figure 7(a) illustrates the anodic polarization curves of TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes. Clearly, as the over potential increases, the anodic current initially increases and reaches a maximum point called the limiting diffusion current density (I _L). I _L is said to be principally controlled by the diffusion of hydrogen atoms and is less affected by the charge transfer resistance at the surface of electrodes especially at high discharge current densities.^{Reference Rousselot, Guay and Roué24,Reference Etiemble, Rousselot, Guo, Idrissi and Roué31} This parameter is listed in Table II for the electrodes. Expectedly, the values of I _L are consistent with those of I ₀. It can be concluded that both kinetic properties of mass transfer and charge transfer are in agreement for all the electrodes and that a distinctive factor may be responsible for controlling these properties which must be mainly the surface oxide layer.

FIG. 7.

(a) Anodic polarization curves for TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes at 50% DOD and 1 mV/s scan rate and (b) potentiostatic discharge curves obtained for TiNi_(1−x)Nb_x (x = 0, 0.05, 0.1, 0.2) electrodes.

The hydrogen diffusion coefficient (D _H) is estimated using the potentiostatic discharge method; the results are drawn in Fig. 7(b). Potentiostatic discharge is a very simple and convenient method commonly utilized for the study of hydrogen diffusivity in the bulk of hydrogen storage materials.^{Reference Guiose, Cuevas, Décamps, Leroy and Percheron-Guégan7} When a large step potential is applied to the hydride electrode, hydrogen atoms on the surface of particles are depleted rapidly and after a prolonged time, the current declines linearly. In this linear region, the hydrogen diffusion is said to control the electrode current density, which can be calculated by the following equation^{Reference Zhao and Ma29}:

(5)$$\log \;i = \log \left( {{{6FD} \over {d{a^2}}}\left( {{C_0} - {C_{\rm{S}}}} \right)} \right) - {{{\pi ^2}} \over {2.303}}{D \over {{a^2}}}t\quad ,$$

where i (A/g) is the current density, F is the Faraday’s constant (C/mol), D (cm²/s) is the hydrogen diffusion coefficient, d (g/cm³) is the density of the hydrogen storage material, a (cm) is the particle radius, C ₀ (mol/cm³) is the initial concentration of hydrogen in the bulk of the material, C _S is the surface concentration of hydrogen, and t (s) is the discharge time.

The curves in Fig. 7(b) can be divided into two regions. The first region is the nonlinear one in which the hydrogen atoms are rapidly consumed on the surface of powder particles by the application of a large step potential of +500 mV (versus stabilized open circuit potential). In the second region, however, the current declines slowly and linearly. In this region, the hydrogen atoms diffuse through the bulk of the powders according to the concentration gradient of hydrogen atoms with time. The D/a ² values are calculated using the above equation and listed in Table II. It is seen that the values for all Nb-substituted intermetallic compounds are all nearly identical and slightly less than the x = 0 (binary) intermetallic compound. Introducing the mean particle radius of approximately 5 and 10 µm for x = 0 binary and x = 0.05 ternary intermetallic compounds, D _H will be estimated to be 9 × 10⁻¹² and 3.2 × 10⁻¹¹ cm²/s, respectively. D _H can be estimated for all the electrodes according to their average particle size (Table II). Obviously, Nb substitution has facilitated the diffusion of hydrogen atoms in the electrodes. The hydrogen diffusion coefficient for hydrogen storage materials has been reported to be in the range of 10⁻⁹ to 10⁻¹⁴ cm²/s,^{Reference Zhao and Ma29} even though, as with I ₀, D _H is largely affected by the microstructure, state of charge (i.e., the hydrogen concentration in the electrode), measurement method, and temperature.^{Reference Tliha, Khaldi, Boussami, Fenineche, El-Kedim, Mathlouthi and Lamloumi1}

It should be emphasized that determination of the real particle size of the active electrode material for the calculation of the hydrogen diffusion coefficient is rather difficult since the galvanostatic cycling on the electrodes would cause pulverization effect on particles; this is believed to be less pronounced in the present case due to the relatively ductile nature of the studied materials.

4. CV curves

To investigate the overall electrochemical process of the electrodes, CV curves are drawn in Figs. 8(a) and 8(b), for TiNi_(1−x)Nb_x (x = 0.05, 0.1, 0.2) ternary intermetallic compounds at the scan rates of 0.1 and 0.5 mV/s after full activation. In Fig. 8(a), for the low scan rate of 0.1 mV/s, in the cathodic direction from −0.4 V to more negative potentials, the applied current is initially very low before reaching the potentials of the hydrogen adsorption reaction. Following the potential of −0.8 V, the hydrogen atoms begin to be adsorbed on the whole electrode’s surface resulting in the current rise. As the hydrogen atoms fully cover the electrode surface, the current gradient declines until reaching a plateau. In this equilibrium state, the adsorbed hydrogen atoms are continuously absorbed into the electrode allowing more adsorbing atoms on the surface. After fully charging, the hydrogen atoms are combined into molecules and begun to evolve. In the reversal scan, however, the current is positively increased until the oxidation process occurs in positive current values and continues to rise until reaching a maximum value. The location of the anodic peak for the electrodes at around −0.84 V corresponds to the oxidation of absorbed hydrogen and the location of the cathodic peak between −0.9 and −1.0 V corresponds to the adsorption of hydrogen on the electrode surface. The area under the oxidation peak is indicative of the amount of oxidized hydrogen or namely the discharge capacity, which is consistent with the order of the discharge capacity of the electrodes described previously. The anodic peak height may also be a good indicator of the kinetics of electrodes. However, at low scan rates, low polarizations are applied to the electrodes and due to their all outstanding kinetics, their anodic peak heights are not very different. At higher scan rates like 0.5 mV/s [Fig. 8(b)], on the other hand, higher kinetics of electrodes might result in a higher peak current which is the case for the x = 0.05 ternary intermetallic compound compared to the other two Nb-contained electrodes. However, the same cannot be said about x = 0.1 and x = 0.2 ternary ones with higher kinetic properties of the latter possibly due to the higher capacity of x = 0.1 compared to that of the x = 0.2 ternary intermetallic compound. Interestingly, the base x = 0 binary intermetallic compound is superior among all the electrodes in this respect. A part of this could be ascribed to the slightly higher D/a ² value for the x = 0 binary intermetallic compound and a part is ascribed to the fact that the x = 0.05 ternary intermetallic compound has undergone the cathodic direction up to −1.1 V compared to −1.2 V of the other electrodes. Thus, a lower time was resulted for the x = 0.05 ternary intermetallic compound to fully absorb hydrogen atoms at the scan rate of 0.5 mV/s. Overall, the superiority of all the electrodes in terms of kinetics of hydriding and dehydriding processes is evident from the appearance of distinctive current plateaus in anodic and cathodic scan directions. From Figs. 8(a) and 8(b), it is noticed that at low scan rates, the peaks are more symmetric indicating the higher reversibility of hydride reactions. However, at high scan rates, the reaction turns to a total irreversible one and a transformation from hydrogen diffusion to charge transfer occurs meaning that the effect of mass transfer on process control is reduced by the scan rate increment. In fact, high scan rates induce intense polarization and hence lead to an irreversible electrode process.^{Reference Jiang, Wang, Wang, Chen, Tang, Liu and Xie32}

FIG. 8.

CV curves for TiNi_(1−x)Nb_x (x = 0.05, 0.1, 0.2) ternary electrodes at the scan rate of (a) 0.1 mV/s, (b) 0.5 mV/s, and (c) CV curves of the x = 0 binary intermetallic compound at various scan rates.

The effect of scan rate on CV curves of the x = 0 binary intermetallic compound is further depicted in Fig. 8(c). As clearly noticed, increasing the scan rate moves the anodic potential to more positive and cathodic potential to more negative values owing to higher polarization and an irreversible procedure. In other words, the anodic and cathodic peaks are more separated yet less discernible and the electrode reaction becomes more irreversible. It is noteworthy to point that in Fig. 8(c), the peak current values are almost proportional to the magnitude of the applied scan rates implying that the whole electrode reaction is controlled by surface adsorption and desorption namely the hydrogen redox reaction.^{Reference Ma, Wang, Hu, Jin and Chen33}