Structure and hydrogen storage characteristics of as-spun Mg-Y-Ni-Cu alloys

1. Introduction

The limited fossil fuels were consumed unduly not only expedites its exhaustion but also brings about a sequence of environmental issues. Particularly, astonishing speed of global warming and increasingly severe environmental contamination have stimulated an initiative to find cleaner and renewable energy sources instead of fossil fuels. Hydrogen, of all available clean energies, is universally acknowledged to be the best fuel because of its limitless source, zero emission of greenhouse gases and high energy efficiency [[1], [2], [3]]. There is an overarching technological obstacle to explore a practical hydrogen storage system which can achieve a conventional driving range (>300 miles) for transportation and vehicles [[4], [5], [6], [7]]. That is to say, the hydrogen storage technologies will determine the hydrogen fuel cell vehicle as an extensive commercial application [8,9]. One of the most promising approaches, from the point of now, to satisfy the demands for mobile transportation is to store hydrogen in metal hydrides [[10], [11], [12]]. Researchers have discovered many metal hydrides with potential applications. Especially, the USA, Japan, and China have started large-scale commercial applications of metal hydrides, for instance, RE-based AB₅ and AB₂-type alloys. Unfortunately, the all requirements of the performance for vehicular applications, which have been put forward by U.S. Department of Energy (DOE), is not satisfied with the above-mentioned metal hydrides [13]. Mg-Ni-based metallic hydrides are applied to negative electrodes of Ni-MH batteries vehicles or hydrogen fuel cell [14] on account of some preponderances, such as the theoretical electrochemical capacity of 1000 mAh/g and gaseous hydrogenation capacity of 3.6 wt% for Mg₂NiH₄ [15,16]. However, several intrinsic issues of Mg-based alloys, for example, sluggish hydrogenation and dehydrogenation kinetics, relatively high dehydrogenation temperatures and exceedingly poor electrochemical cycle stability, severely restrict their practical application [17,18]. Many methods have been adopted to conquer these disadvantages, such as alloying with other elements [[19], [20], [21]], nanocrystallization [22,23], and doping with catalysts [24,25].

Furthermore, it was ascertained that the de-/hydrogenation kinetics of Mg-based alloys is highly sensitive to their microstructures [[26], [27], [28]]. Zhang et al. [29] reported that prolonging milling duration can enhance the dehydrogenation kinetics of Mg-Ni-based alloys. Proposed by Grosdidier et al. [26] and Siarhei et al. [30], the de-/hydrogenation properties can be improved dramatically if the grain sizes of Mg-base alloys are far below micrometer scale. The dehydrogenation temperature of the Mg₂Ni alloy with nanocrystalline structures can be decreased from 573 to 473 K [31]. Some preparation technologies, particularly ball milling [32,33] and melt spinning [34], are usefully employed to prepare the Mg-based alloys with an extremely homogeneous element distribution. The maximal discharge capacity of 580 mAh/g can be obtained for (Mg₆₀Ni₂₅)₉₀Nd₁₀ alloy with an amorphous and nanocrystalline structure preparing by melt spinning, reported by Huang et al. [35]. The hydrogen absorption kinetics of LaNi_3.8Al_1.0Mn_0.2 can be obviously improved after melt-spinning, reported by Han et al. [36]. Zhang et al. [37] reported the electrochemical performances of La_0.8-xCe_0.2Y_xMgNi_3.4Co_0.4Al_0.1 (x = 0, 0.05, 0.1, 0.15, 0.2) alloys can be markedly enhanced by melt spinning.

It was reported that substituting lanthanon (La, Nd, Y, Ce, Pr) for part of Mg [[38], [39], [40]] and Cu for part of Ni [41] in Mg-Ni-based alloys results in hydride stability decreasing and the hydrogen desorption improving. In this paper, Mg and Ni in the Mg-Ni-based alloys were partly substituted by Y and Cu, respectively, and the preparation of amorphous and nanocrystalline Mg_25-xY_xNi₉Cu (x = 0-7) alloys adopted melt spinning technique. Subsequently, the impacts of spinning rate on their thermodynamic and dynamic properties were studied in detail.

2. Experimental

To prevent Mg from volatilizing, a vacuum induction furnace was utilized to prepare Mg_25-xY_xNi₉Cu (x = 0, 1, 3, 5, 7) alloys under a helium (purity of 99.9%) pressure of 0.04 MPa. The purities of the experimental samples of Mg, Ni, Cu, and Y are at least 99.99%, which were provided by CISRI Corporation. To simply express, the different Y contents in the alloys were expressed as Y₀, Y₁, Y₃, Y₅, and Y₇, respectively. In the experiment, the cast ingot can be got with the molten alloy injected into copper cooled mold and melt-spinning with a rotating copper roller were used to re-melt and spin a part of as-cast alloys. Because of the difficulty of measurement of an accurate spinning rate, viz. cooling rate of the specimen during spinning, the linear velocity of copper roller can be adopted to approximatively stand for the spinning rate, which was set to be 15, 20, 25 and 30 m/s, respectively.

X-ray diffraction (XRD) (D/max/2400) conducted with CuKα1 radiation filtered by graphite was used to observe the phase structures of the as-cast and spun alloys and the experimental parameters are 40 kV, 160 mA, and 10 °/min, respectively.

HRTEM equipped with an electron diffraction (ED) system was applied for observing the microtopography of the as-spun alloys and confirming their crystalline states. The instrument type is JEM-2100 F (200 kV).

The measurements of de-/hydrogenation kinetics and the P-C-T curves of samples were performed by using a Sieverts apparatus with a furnace (an accuracy of ±2 K) which can be automatically controlled by a program. The hydrogenation was carried out at 3 MPa (the initial pressure for hydrogenation) and 553, 573 and 593 K, while the same temperatures and a pressure of 1 × 10^-4 MPa for hydrogen desorption. Each test requires 300 mg sample. The DSC and TGA (SDT-Q600) were employed to measure the dehydrogenation properties with the heating rates of 5, 10, 15, and 20 K/min, respectively.

3. Results and discussion

3.1. Microstructure characteristics

At the same spinning rate, Y₁ and Y₇ alloys show completely different diffraction peaks, as shown in the XRD patterns (Fig. 1). The patterns show that the sample alloys contain Mg₂Ni (PDF#35-1225), Mg (PDF#35-0821) and YNi₃ (PDF#25-0589) phases. The Y₁ alloy displays the sharp and narrow diffraction peaks, which indicates an entire crystal structure. The strain energy was stored, and grains were refined with spinning rate increasing, and this makes the diffraction packs of Y₁ alloy visibly broaden. Meanwhile, the as-spun Y₇ alloy exhibits broad and flat diffraction packs indicating an amorphous structure has been formed in Y₇ alloy. It indicates that the amorphous formation ability of Mg-Ni-based alloys could be promoted by the replacement of Mg by Y.

View original graphic| Download| PPT slide

Fig. 1.   XRD profiles of as-cast and spun Mg_25-xY_xNi₉Cu (x = 0-7) alloys: (a) Y₁ alloy, (b) Y₇ alloy.

ED was used to determine crystal states of the as-spun alloys and HRTEM was used to observe the morphologies, as presented in Fig. 2. Notably, melt spinning enables the crystalline alloy to be more nanostructured and disordered, and the average grain size is in a range of 20-30 nm measured by the linear intercept method. Meanwhile, crystal defects can be seen clearly, for instance, grain boundaries and subgrains. Carefully observing, it can find that the ED patterns present piercing multi-haloes for as-spun Y₁ alloy, which suggests the formation of an entire nanocrystalline structure. However, the Y₇ alloy shows the typical amorphous and nanocrystalline structures and the quantity of amorphous phase clearly augments with spinning rate increasing, which accords well with the XRD observation.

View original graphic| Download| PPT slide

Fig. 2.   HRTEM micrographs and ED patterns the as-spun alloys: (a), (b) and (c) Y₁ alloy spun at 15, 20 and 30 m/s; (d), (e) and (f) Y₇ alloy spun at 15, 20 and 30 m/s.

3.2. Hydrogen absorption and desorption kinetics

Fig. 3 exhibits the relationships between hydrogenation capacity and reaction time at 573 K and 3 MPa for the as-cast and spun Mg_25-xY_xNi₉Cu (x = 0-7) alloys. In the initial stage, there is a highly rapid hydrogen absorption rate for the sample alloys, and after that, the amount of hydrogen is nearly saturated during a long period of hydrogenation. It is well known that de-/hydrogenation rates are crucial factors for hydrogen storage materials applied to a vehicle-mounted hydrogen storage system. In this work, the hydrogenation kinetics is characterized by a hydrogenation saturation ratio (Rta) (a ratio of hydrogenation capacity at a fixed time to the saturated hydrogenation capacity), and it is defined as Rta=Cta/C100a×100%, where C100a and Cta are hydrogenation capacities at 100 and t min, respectively. It is found that C100a values are above 98% of the saturated hydrogenation capacities in according to the results of all experimental alloys, so the saturated hydrogenation capacity can be approximately expressed by C100a value. Taking hydrogenation time of 5 min as a measuring stick, the relationships between spinning rate and the values of R5a (t = 5) can be built, as inserted in Fig. 3. It could be seen the R5a value visibly decreases as spinning rate increasing. Specifically, the R5a value is lowered from 87.6%-80.2% for Y₁ alloy and from 91.5% to 77.6% for Y₇ alloy with spinning rate rises from 0 to 30 m/s. Likewise, the dehydrogenation ratio (Rtd) is used to signify the dehydrogenation kinetics of samples, and its definition is Rtd=Ctd/C100a×100%, where Ctd stands for dehydrogenation capacity at t minutes and C100a is the same as the previous definition. Choosing hydrogenation time of 10 min for comparison, the relationships between R10d values of Y₁ and Y₇ samples and spinning rate could be built, as presented in Fig. 4. Apparently, the dehydriding rates markedly enhance with spinning rate increasing. Specifically, the R10d values at 573 K is raised to 69.3% from 56.7% for Y₁ alloy and from 68.5% to 75.6% for Y₇ alloy when spinning rate augments from 0 to 30 m/s.

View original graphic| Download| PPT slide

Fig. 3.   Hydrogen absorption kinetic curves of the as-cast and spun Mg_25-xY_xNi₉Cu (x = 0-7) alloys at 573 K: (a) Y₁ alloy, (b) Y₇ alloy.

View original graphic| Download| PPT slide

Fig. 4.   Evolution of the R10d values of the as-cast and spun Mg_25-xY_xNi₉Cu (x = 0-7) alloys with spinning rate: (a) Y₁ alloy, (b) Y₇ alloy.

Based on the structural identification, it can be known that melt spinning engenders obvious implications for de-/hydrogenation kinetics. The factors that influence the hydrogenation process of alloys include several aspects as follows: the dissociate rate of hydrogen molecular on the alloy surface, the ability of H atoms penetrating the metal from the oxide layer surface, the diffusion rate of H atoms inside the bulk metal and pass through the newly formed hydride. The results show that the increasing spinning rate engenders distinct negative impacts on hydrogen storage kinetics, which is due to the glass forming facilitated by replacing Mg by Y and increasing spinning rate on a basis of the fact that the diffusion capacity of H atoms in an amorphous phase is lower in comparison with that in a nanocrystalline phase. Moreover, the negative effects of the stored strain energy on hydrogenation kinetics caused by melt spinning cannot be ignored as well. Regarding the impacts of the hydrogen desorption kinetics improvement, it can be thought to be due to the increased internal strain and decreased grain size caused by melt spinning because the dehydrogenation performances of Mg-base alloys could be significantly promoted by decreasing grain size far below the micrometer scale [42]. Besides, the stored internal strain enhances with spinning rate growing and consequentially gives rise to a marked decrease in metal hydride stability, which promotes the dehydrogenation, reported by Northwood et al. [43].

3.3. Hydrogen desorption activation energy

The hydrogen desorption activation energy is investigated for revealing the mechanism of hydrogen desorption kinetics affected by melt spinning operation. Generally speaking, activation energy is an important ingredient that influences gas-solid reaction kinetics, and it is looked as a total potential barrier needed to conquer for the gas-solid reaction. As dehydrogenation reaction is generally related to the total energy barrier, the driving force of hydrogenolysis can be confirmed via computing activation energy. In this paper, the Arrhenius and Kissinger methods are adopted for estimating hydrogen desorption activation energy. The dehydrogenation kinetic curves of Mg_25-xY_xNi₉Cu (x = 0-7) alloys are measured according to the computational conditions of Arrhenius method from 553 to 593 K. The dehydrogenation kinetic curves of the as-spun (20 m/s) Y₁ and Y₇ alloys are used as representatives, as shown in Fig. 5. It is evident that the higher temperature is, the faster hydrogenation kinetics will be. Usually, the Johnson-Mehl-Avrami (JMA) model is utilized to simulate the nucleation and growth processes for hydrogen desorption, and the JMA equation is [44]:

ln[-ln(1-α)]=ηlnk+ηlnt (1)

where the Avrami exponent is expressed by η, an effective kinetic parameter is represented by k, and Greek alphabet α expresses the phase transition fraction at time of t that could be identified with standard H wt% ranging from 0 to 1. Eq. (1) is used to obtain the JMA graphs of ln[-ln(1-α)] vs. lnt at temperatures of 553, 573 and 593 K, as inserted in Fig. 5, which is found to be nearly linear. It implies that the dehydrogenation reaction of samples abides by instantaneous nucleation followed by interface controlled three-dimensional growth process [45]. The values of ηlnk and η could be derived from the intercept and slope of the JMA plots at different temperatures. Consequently, it is easy to calculate the rate constant k. Thus, according to the Arrhenius equation, the activation energy (Eade) can be achieved for the dehydrogenation process, and the equation is [44]:

k=Aexp-($\frac{-E^de_a}{RT}$) (2)

where the universal gas constant (8.3145 J/mol/K) is represented by a capital letter R, T expresses the absolute temperature, A stands for a temperature independent coefficient, and the meaning of k is the same as previous definitions. The Arrhenius plots of lnk vs. 1/T of the dehydrogenation kinetics are displayed in Fig. 6. Thus, the activation energy Eade was obtained according to slopes of Arrhenius plots, and Fig. 6(a) and (b) show relationships between the Eade values and spinning rate. We note that with spinning rate increasing the value of Eade markedly declines. Specifically, the value of Eade lowers to 64.77 from 71.54 kJ/mol for Y₁ alloy and from 67.36 to 55.13 kJ/mol for Y₇ alloy when spinning rate augments from 0 to 30 m/s.

View original graphic| Download| PPT slide

Fig. 5.   Hydrogen desorption kinetic curves of the as-spun (20 m/s) Y₁ and Y₇ alloys at 553, 573 and 593 K and Avrami plots of ln[-ln(1-α)] vs. lnt: (a) Y₁ alloy, (b) Y₇ alloy.

View original graphic| Download| PPT slide

Fig. 6.   Arrhenius plots of the as-cast and spun Mg_25-xY_xNi₉Cu (x = 0-7) alloys and evolution of hydrogen desorption activation energy Eade with spinning rate: (a) Y₁ alloy, (b) Y₇ alloy.

Kissinger method is also used to compute dehydrogenation activation energy as a contrast. The Kissinger equation is [46]:

$\frac{d[ln(β/T^2_p)]}{d(1/T_p)}$= $\frac{-E^de_k}{R}$ (3)

where the absolute temperature of the maximal desorption rate in the DSC curves is represented by a simplified form of T_p, β expresses the heating rate, the abbreviation of Ekde is used to express activation energy, and the letter of R stands for the ideal gas constant. Taking computational conditions of Kissinger method into consideration, DSC, with heating rates of 5, 10, 15 and 20 K/min, was utilized to survey the hydrogen desorption reactions of Y₁ and Y₇ alloys which were in the saturation level at 3 MPa and 573 K, as presented in Fig. 7. As for hydrogen desorption process, there is an obvious endothermic peak, and there are similar peak shapes for all the alloys, suggesting the similar reaction process in every reaction. Fig. 8 demonstrates the Kissinger graphs of ln(β/Tp2) vs. 1/T_p are nearly linear and built by making use of the logarithmic transform of Eq. (3). Therefore, it is easy to count the activation energy Ekde on the basis of the slopes of the Kissinger plots, and the relationships between spinning rate and Ekde value are shown in Fig. 8. The result is similar to that of the Arrhenius method, the Ekde values of alloys clearly decline when spinning rate augments. Specifically, the augment of spinning rate from 0 to 30 m/s brings on the value of Ekde decreasing to 62.82 from 73.72 kJ/mol for Y₁ alloy and from 62.79 to 48.86 kJ/mol for Y₇ alloy. In addition, replacing Mg with Y results in more amorphous structures inside the alloys. The microstructure changes caused by the addition of Y may be the main factor to enhance the hydrogen desorption kinetics. In comparison with the above-mentioned results of using Arrhenius and Kissinger methods, dehydrogenation activation energy markedly lowers with spinning rate increasing. Melt spinning can result in a nanocrystalline and amorphous structure in the alloys which provides more grain boundaries. The added grain boundaries are considered as the main factor for promoting the dehydrogenation kinetics by providing lots of diffusion paths for hydrogen atoms. Hence, the decreased dehydrogenation activation energy is confirmed as the real driving force for greatly enhancing hydrogen desorption kinetics.

View original graphic| Download| PPT slide

Fig. 7.   DSC curves of the as-cast and spun Y₁ and Y₇ alloys at various heating rates: (a) Y₁ alloy, (b) Y₇ alloy.

View original graphic| Download| PPT slide

Fig. 8.   Kissinger plots of the as-cast and spun Mg_25-xY_xNi₉Cu (x = 0-7) alloys and evolution of hydrogen desorption activation energy Ekde with spinning rate: (a) Y₁ alloy, (b) Y₇ alloy.

3.4. Hydrogen storage thermodynamics and P-C-T curves

P-C-T curves of the as-cast and spun Mg_25-xY_xNi₉Cu (x = 0-7) alloys are tested at 3 MPa and 573 K, as presented in Fig. 9, from which it is found that there are two plateaus, corresponding to the hydriding and dehydriding processes of Mg/MgH₂ (lower pressure plateau) and Mg₂Ni/Mg₂NiH₄ (higher pressure plateau), respectively. Meanwhile, the de-/hydrogenation pressure plateaus show big hysteresis (H_f = ln(P_a/P_d)) and an evident inclination. Melt spinning brings on a visible enhancement on the de-/hydrogenation pressure plateaus, which can be ascribed to the stored internal strain generated by melt spinning. Moreover, a clear effect of hydrogen storage capacity was caused by the variation of spinning rate, and the spinning rate dependence of hydrogen storage capacity is displayed in Fig. 9. Obviously, hydrogen absorption capacity exhibits different varying tendency with spinning rate rising. Namely, hydrogen absorption capacity improves at the beginning and declines later for Y₁ alloy, but that of Y₇ alloy always lessens with spinning rate increasing. Specifically, the hydrogen absorption capacity of Y₁ alloy adds to 3.97 wt.% (15 m/s) from 3.74 wt.% (0 m/s) and then lowers to 3.22 wt.% (30 m/s) and that of Y₇ alloy keeps declining from 2.82 wt.% (0 m/s) to 2.24 wt.% (30 m/s). In a general way, it is usually to evaluate the thermal stability of an alloy hydride with its thermodynamic parameters of de-/hydrogenation, which is received via the Van't Hoff equation [47]:

Ln($\frac{P_{H_2}}{P_0}$)=$\frac{\Delta{H}}{RT}-\frac{\Delta{S}}{R}$ (4)

where R stands for the gas constant, T is the sample temperature, PH2 stands for the equilibrium hydrogen gas pressure and P₀ expresses the standard atmospheric pressure. The P-C-T curves of Mg_25-xY_xNi₉Cu (x = 0-7) alloys were gauged at several sets of temperature of 553, 573 and 593 K under calculation conditions of Van't Hoff equation, and the P-C-T curves of Y₁ and Y₇ samples spun at 20 m/s were demonstrated in Fig. 10 as representatives. The Van't Hoff graphs of ln(PH2/P0) vs. 1/T of Y₁ and Y₇ alloys were drawn by using Eq. (4) as inserted in Fig. 10. According to the Van't Hoff equation of Eq. (4), it is easy to compute the thermodynamic parameters, because ln(PH2/P0) and 1/T show a good linear relationship and Fig. 11 displays the relationships between spinning rate and the ΔS and ΔH values of hydriding and dehydriding reactions of samples. We can clearly see that the ΔH and ΔS values markedly reduce with spinning rate increasing, which indicates the hydriding/dehydriding thermodynamics can be visibly affected by spinning rate, namely decreasing the thermal stability of alloy hydrides, as proposed by Tanaka et al. [48].

View original graphic| Download| PPT slide

Fig. 9.   P-C-T curves of the as-cast and spun Y₁ and Y₇ alloys at 573 K: (a) Y₁ alloy, (b) Y₇ alloy.

View original graphic| Download| PPT slide

Fig. 10.   P-C-T curves of the as-spun (20 m/s) Y₁ and Y₇ alloys in the temperature range of 553-593 K and Van't Hoff plots: (a) Y₁ alloy, (b) Y₇ alloy.

View original graphic| Download| PPT slide

Fig. 11.   Variations of the ΔH and ΔS absolute values of the hydrogen absorption and desorption reactions of the alloys with spinning rate: (a) Y₁ alloy, (b) Y₇ alloy.

The temperature programmed desorption curve, the temperature of which was increased at 5 K/min, of Mg_25-xY_xNi₉Cu (x = 0-7) samples in a saturated hydrogenation state at 3 MPa and 573 K is displayed in Fig. 12. Each sample remains the same weight to avert the influence of pressure on dehydrogenation temperature. From the chart, the initial hydrogen desorption temperatures of alloys obviously decrease with spinning rate growing, manifesting that the stability of hydride was lowered by melt spinning. To be specific, the initial temperature decreases from 518.6 to 501.1 K for Y₁ alloy and from 499.5 to 484.5 K for Y₇ alloy when spinning rate enhances from 0 to 30 m/s. The decreased starting dehydrogenation temperature caused by melt spinning can be attributed to the formation of amorphous and nanocrystalline structures, as reducing the grain size can impact the thermal stability of the alloy hydrides.

View original graphic| Download| PPT slide

Fig. 12.   Temperature programmed hydrogen desorption curves of the as-cast and spun alloys: (a) Y₁ alloy, (b) Y₇ alloy.

4. Conclusions

The hydrogen storage performances of the as-cast and spun Mg_25-xY_xNi₉Cu (x = 0-7) alloys were systematically studied and the major conclusions are drawn as follows:

1 A visible decrease of thermodynamic performances (ΔH and ΔS) is caused by melt spinning. The starting dehydrogenation temperature of alloy hydrides obviously lowers with spinning rate increasing, for which the dramatic reduction of grain size generated by melt spinning has the primary responsibility.

2 Melt spinning brings clearly decreasing on hydrogen absorption kinetics but promotes their hydrogen desorption kinetics dramatically. To be specific, when spinning rate enhances from 0 to 30 m/s, the R10d value at 573 K enhances from 56.7% to 69.3% for Y₁ alloy and from 68.5% to 75.6% for Y₇ alloy.

3 The Kissinger and Arrhenius methods were adopted to calculate the dehydrogenation activation energy of alloys. The results show that the dehydrogenation activation energy lowers with spinning rate increasing. The activation energy of dehydrogenation is regarded as the real driving force of the dehydrogenation kinetics of alloy hydrides.

Acknowledgment

It is thankful to the National Natural Science Foundations of China (Nos. 51761032, 51871125 and 51471054) for financial support of the work.

The authors have declared that no competing interests exist.