Microstructure and Hydrogen Sorption Kinetics of Ball-Milled Mg₆₀Ni₂₅Cu₁₀Ce₅ Nanocrystalline Powders

Abstract

High-energy ball milling for different durations was used to synthesize nanocrystalline Mg₆₀Ni₂₅Cu₁₀Ce₅ powders. The morphology and microstructure of the milled powders were investigated by scanning electron microscopy and X-ray diffraction, respectively. It was found that different milling times result in considerably different phase composition. The powder milled for 1 h is characterized by elemental Mg, Ni, Cu and Ce with some minor content of intermetallics. In total, 3 h milling promotes the intensive formation of intermetallic compounds, while 10 h of powder processing results in a partially amorphous state coupled with compound phases. Isothermal hydrogenation and dehydrogenation experiments were conducted in a Sieverts’-type apparatus. It was found that all powders absorb H₂ reversibly, while the shortest milling time provides the best overall capacity. Excellent kinetics without any activation cycle were obtained for the 3 h milled composite, releasing and absorbing 50% of the total hydrogen content within 120 s. Each kinetic measurement has satisfactorily been fitted by the Johnson–Mehl–Avrami function. X-ray diffraction analysis on the dehydrided powders confirmed the complete desorption.

1. Introduction

The world-wide demand for renewable and clean energy solutions intensifies, and now the global consensus is to seek new alternatives to traditional fossil fuels [1], which are responsible for serious environmental damage by intensive CO₂ emission [2]. Nevertheless, the integration of these clean energy sources into the current energy grid is still limited due to their strong spatial and weather condition dependence, in particular for wind and solar energy [3]. In this context, hydrogen as a secondary energy carrier stands out as a possible solution for these obstacles, owing to its very high energy density per mass (120–140 MJ/kg), environmental friendliness, abundance and harmless reactions in fuel cells when chemical energy is converted to electric energy [4]. The current application of hydrogen energy still faces several challenges, and, consequently, a hydrogen storage system has to fulfill different technical standards in order to be effectively used regularly, especially in the automotive industry [5,6,7]. Such a system, for example, should possess a high gravimetric and volumetric hydrogen capacity, enhanced sorption rates during hydrogenation and dehydrogenation at moderate temperatures and pressures, and a reliable operational cycle life [8]. Recently, intensive basic and applied research has been targeting the realization of efficient hydrogen storage in the solid state with appropriate H₂ density [9,10,11]. Among the known reversible hydrogen storage systems, metal hydrides have gained outstanding potential for storing hydrogen, as they exhibit remarkable storage capacity and reasonable cycling performance [12].

Magnesium and magnesium hydride are recognized as promising hydrogen storage materials, owing to their outstanding absorption capacity (7.6 wt.%, or 2600 Wh/kg), low mass density, abundant resource availability on Earth, low cost and non-toxic nature [13,14,15,16,17,18,19,20,21]. Unfortunately, commercial magnesium exhibits high thermodynamic stability and sluggish hydrogen-reaction kinetics, impeding its industrial applications [16]. Recently, researchers conducted intensive studies to overcome these difficulties, including nanostructuring using different techniques to improve the hydrogen storage performance of magnesium-based systems [16,22,23,24]. Materials processed by severe plastic deformation (SPD) [25] using the high energy ball milling (HEBM) technique usually exhibit an average crystallite size in the order of 10–100 nm [26,27]. These nanocrystals are separated by abundant grain boundaries and interfaces, which can act as fast and easy diffusion channels for hydrogen and promote penetration into the interior of the crystals [28,29,30]. Apart from the high-angle grain boundaries, lattice defects formed during the HEBM process in the interior of the nanocrystals can further enhance hydrogen sorption [30,31].

In addition, the absorption kinetics and dehydrogenation temperature of Mg/MgH₂ can further be improved by alloying Mg with different elements. The large Mg-H bond energy is the main reason for the high thermal stability of MgH₂, which can significantly be reduced using transition metals (TM) by introducing Mg-TM-H bonds [32]. The synergetic effect between Mg and Ni is manifested in a lower enthalpy of hydrogenation of Mg₂Ni and in the reduced temperature of hydrogen sorption [33,34,35,36,37,38,39]. Partial substitution of Ni by Cu can lead to significant grain refinement and improved hydrogen kinetics [32]. To enhance the sorption rate of hydrogen, rare earth (RE) elements like Ce [40,41,42,43,44,45] La [46], Y [47,48,49,50] and Nd [51,52] have been added to the Mg-TM system. This combination is designed to further destabilize magnesium–hydrogen bonds through the formation of new phases, thereby improving the hydrogenation and dehydrogenation kinetics and thermodynamics. These elements can have a synergistic effect, accelerating these processes to optimize the alloy performance.

In addition, the modified phase composition of these systems can positively influence the overall hydrogenation behavior. For example, the hydrogen storage behavior of the Mg-Ni-Y system can be improved by the promotion of the nucleation of a ternary eutectic 14H-LPSO phase [53]. It was found in in situ X-ray synchrotron radiation experiments that during hydrogenation, the formation of a novel La₂Mg₁₇H_~1.₀ solid solution develops in the Mg-Ni-La system [54]. SPD by HEBM of a Mg-Ni-Ce alloy not only refines the nanostructured powder, but enhances the formation of an amorphous phase, which can enhance the H-storage kinetics at temperature as low as 100 °C [44]. The in situ-formed Nd₂H₅ phase remains non-decomposable but finely dispersed in the Mg_95−xNi₅Nd_x alloy synthesized by HEBM, significantly enhancing the storage properties via its catalytic activity [51].

In the present research, nanocrystalline Mg₆₀Ni₂₅Cu₁₀Ce₅ powders have been synthesized by high-energy ball milling of four different durations, including 1 h, 3 h and 10 h of ball-milling. We demonstrate that severe plastic deformation can result in significant phase evolution and varying microstructure during the milling process. The goal and novelty of this study is to explore the correlation between the microstructure, phase composition and hydrogen storage performance of the Mg₆₀Ni₂₅Cu₁₀Ce₅ powder composites.

2. Materials and Methods

2.1. Sample Preparation

Nanocrystalline Mg₆₀Ni₂₅Cu₁₀Ce₅ powders were produced by high energy ball-milling in a SPEX 8000M Mixer Mill (Metuchen, NJ, USA) operating at 1425 RPM in a stainless-steel vial (volume: 65 mL). Prior to the milling procedure, the elemental Mg, Ni, Cu and Ce powders (purity 99.9%; supplier: Sigma-Aldrich (Darmstadt, Germany)) were carefully measured in the desired composition and mixed manually in order to achieve good homogeneity. Powder handling and all processing were carried out under Ar atmosphere in a glove box. Milling times were chosen as 1 h, 3 h and 10 h, and for each milling, 1 g of powder blend and 10 stainless steel balls (1/4 in.) were applied in order to attain a 10:1 ball-to-powder mass ratio. To minimize the sticking and deposition of the powder particles onto the inner surfaces of the vial, the milling procedure was interrupted after 60 min and the vial was rotated by 90 deg along its symmetrical axis.

2.2. Microstructural Characterization

2.2.1. Scanning Electron Microscopy

Morphology studies of the HEBM Mg₆₀Ni₂₅Cu₁₀Ce₅ powder blends were systematically analyzed by a FEI QUANTA 3D dual beam scanning electron microscope (SEM) applying back-scattered electron and secondary electron detection. Quantitative elemental mapping was performed by energy dispersive X-ray spectroscopy (EDS) analysis with a relative accuracy of 1%. Quantitative analysis of the powder agglomerates was carried out by the ImageJ software (version: v1.54) using the individual length of each particle through their geometrical center in the same direction.

2.2.2. X-Ray Diffraction

Phase analysis and mictorstructure of the milled Mg₆₀Ni₂₅Cu₁₀Ce₅ powders were studied by X-ray powder diffraction (XRD) on a Rigaku SmartLab (Tokyo, Japan) diffractometer in Bragg–Brentano geometry using Cu-Kα radiation (λ = 0.154 nm). XRD data were detected in the range from 2θ = 20 to 120°, applying a Δ(2θ) = 0.01° step size. In order to obtain quantitative parameters of the microstructure, the XRD patterns were evaluated by the Williamson–Hall analysis [55]. In brief, the full width at half maximum (FWHM) of the scattering reflections corresponding to the same crystalline phase follows a linear relationship:

$$∆K={\displaystyle \frac{0.9}{D}}+A\sqrt{{\epsilon }^2}K,$$

(1)

where D is the coherently scattering X-ray domain (usually considered as the crystallite size), A~1 is a geometrical constant and

$$K=2{\displaystyle \frac{sin\theta }{\lambda }}$$

(2)

is the diffraction vector, while

$$∆K=2{\displaystyle \frac{cos\theta (∆\theta )}{\lambda }}$$

(3)

where $∆\theta$ is the measured FWHM of the individual Bragg peaks. In the case of overlapping reflections of the different phases, a peak deconvolution has been performed prior to the Williamson–Hall analysis.

2.3. Hydrogen Storage Experiments

The hydrogenation/dehydrogenation kinetics of the HEBM powders were performed in a home-made Sieverts’-type apparatus. Isothermal hydrogenation and dehydrogenation measurements were carried out at 573 K, with an initial hydrogen pressure of 1 MPa for absorption and 1 kPa for desorption. Linear heating up to 573 K was performed in a vacuum, followed by the introduction of hydrogen to the sample chamber (the pre-set hydrogen pressure was reached in a few seconds). For each experiment, ~100 mg of powder was used.

3. Results and Discussion

Characterization of the As-Milled Mg₆₀Ni₂₅Cu₁₀Ce₅ Powders

As inferred from the low-magnification SEM micrographs, the alloy milled for 1 h is dominated by individual particles, powder agglomerates and some debris (Figure 1a). The individual particles are typically surrounded by flat surfaces which were formed through repeated powder-to-vial and powder-to-ball collisions during the milling process. In some cases, these particles exhibit some surface crack and sharp edges, which could be related to the severe plastic deformation that occurred during HEBM. One can also notice that the harder Cu, Ni and Ce particles (brighter on the image) are pressed into the softer Mg ones (darker on the image). As evident from the image, the darker Mg particles and/or agglomerates are considerably larger than the other ones.

Successive plastic deformation results in an attrition and homogeneous mixture of the individual particles (Figure 1b); moreover, some of them stick together, forming larger and continuous agglomerates. As also can be noticed, the relative contrast between the different agglomerates and regions is less pronounced after 3 h and 10 h of HEBM time, indicating a complete mixture of the constituent elements on the (sub)micron scale. Notably, some larger bright particles and blocks can be observed on the micrograph corresponding to 10 h of milling (Figure 1c), confirming that some harder and denser regions did not take place in the attrition process, probably having stuck to internal surfaces of the milling media.

Quantitative analysis of the SEM micrographs provides the corresponding particle size histograms (see Figure 1), which can adequately be fitted by a lognormal distribution function:

$$G\left(x\right)={\displaystyle \frac{1}{\sqrt{2\pi }\sigma }}{\displaystyle \frac{1}{x}}{exp}^{{\displaystyle \frac{{-\left(ln\frac{x}{m}\right)}^2}{2{\sigma }^2}}},$$

(4)

where m and σ are the median and variance of the distribution. It is evident from the histograms that the maximum occurs below 1 micron for all the composites, similarly to other Mg-based hydrogen storage powders [31,56,57]. As it is noticed on the histograms, the fitted median value decreases from m = 0.42 μm to m = 0.19 μm when the HEBM time increased from 1 h to 3 h and 10 h. At the same time, the variance of the distribution increases from σ~0.5 to σ~0.8 when milling time is increased from 1 h to 3 h and 10 h. The larger variance value corresponds to a wider distribution, which is in line with the observation described above, i.e., milling times larger than 3 h may result in the formation of larger agglomerates. As one can notice, the 3 h and 10 h powders exhibit very similar morphological features and similar microstructural parameters, demonstrating that 3 h of milling in a SPEX 8000 HEBM with a 10:1 ball-to-powder mass ratio results in the saturation of the microstructure [56]. With the available median and variance values of the lognormal size-distribution functions, the corresponding area-averaged agglomerate size can be obtained as

$${〈d〉}_{area}=m exp\left(2.5{\sigma }^2\right).$$

(5)

As seen, the ${〈d〉}_{area}$ values slightly increase for larger milling times (up to ~1 μm from 0.8 μm); however, it is concluded that the statistical analysis did not reveal any significant differences between the different powders.

The corresponding EDS spectra presented in the right column of Figure 1 provide quantitative information on the composition of the as-milled Mg₆₀Ni₂₅Cu₁₀Ce₅ powders. The obtained values for Mg, Ni, Cu and Ce are listed in

Table 1. Elemental content of the Mg₆₀Ni₂₅Cu₁₀Ce₅ composites obtained from the EDS analysis.

Elements (at.%)	1 h	3 h	10 h
Mg	85.82	67.14	62.93
Ni	8.50	19.31	23.72
Cu	2.16	9.91	8.96
Ce	3.51	3.64	4.39

. As one can notice, the obtained values are in a good coincidence with the nominal value for the powder milled for 10 h; nevertheless, the shortest HEBM time (1 h) exhibits an anomalously larger Mg concentration, while the transition elements, i.e., Ni and Cu, are below the nominal value. The discrepancy between the experimental and the nominal values at shorter milling times may relate to the different mechanical properties of the individual constituents, and consequently to their different adherences to the internal walls and edges of the milling vial. For longer millings, this feature is less pronounced, since the milling container was rotated along its symmetrical axis after every 1 h of milling, as was mentioned in the experimental section.

A higher-resolution electron micrograph of the Mg₆₀Ni₂₅Cu₁₀Ce₅ 3 h milled powder reveals a relatively homogeneous morphology, where multiface particles and elongated agglomerates are also present (Figure 2). The supplementary point-like EDS analysis can provide information on the compositional differences between the different particles and regions. As realized from

Table 2. Compositional variation in the powder milled for 3 h. A, B, C, and D denote different particles or regions in Figure 2.

Elements (at.%)	A	B	C	D
Mg	42.41	79.21	77.73	80.03
Ni	53.68	9.33	12.16	7.06
Cu	2.91	7.99	7.48	5.26
Ce	1.01	3.47	2.63	7.65

, the bright particle (denoted by A) in the center of the image is enriched in Ni, while the Cu and Ce concentration is well below the nominal value. Regions B and C are abundant in Mg; their Cu concentration is close to the average, but they are nevertheless depleted in Ni. Rather similar concentrations can be obtained for particle D, which is slightly enriched in Ce.

Local area elemental EDS-mapping has been carried out on the Mg₆₀Ni₂₅Cu₁₀Ce₅ powder milled for the longest duration. As seen in Figure 3, the compositional homogeneity is almost fulfilled in the entire micrograph, i.e., Mg, Ni and Cu are distributed evenly in the entire area. On the other hand, the presence of some Ce-rich regions can be observed on the respective concentration map, confirming the non-perfect mixing of this element with the other constituents.

In order to monitor the phase evolution and microstructural variation due to the severe plastic deformation, powder XRD has been carried out on the initial blend of the Mg, Ni, Cu and Ce elements and on the as-milled Mg₆₀Ni₂₅Cu₁₀Ce₅ powders for different HEBM durations. As one can recognize, the diffractogram of un-milled powder blends presents sharp Bragg peaks of hexagonal Mg (JCPDS 35-0821), fcc Ni (JCPDS 04-0850), fcc Cu (JCPDS 04-0836) and orthorhombic Ce (JCPDS 38-0765); see Figure 4. Intensive milling results in significant peak broadening, indicative of extensive size reduction in coherently scattering crystalline domains [31,37]. It is evidenced from the diffractograms that, besides nanocrystallization, solid-state reactions also take place during the HEBM process. The diffractogram of the powder milled for 1 h already contains the reflections of intermetallic phases, like Mg₂Ni (JCPDS 35-1225), Mg₂Cu (JCPDS 02-1315), MgNi₂ (JCPDS 03-1027), Cu₃.₈Ni (JCPDS 09-0205) and CeMg (JCPDS 02-1453); however, a significant fraction of the volume of the original elements are still present at this stage. As the ball-milling process continues, the relative Bragg-peak intensity of the elemental powders decreases gradually, while these reflections almost diminish after 10 h of HEBM. At the same time, a broad diffraction halo develops at 2θ~42 deg, indicative of a possible solid-state amorphization process, which is further supported by the noticeable upwards bend at the start of the diffraction pattern [33]. In addition, the amount of MgO (JCPDS 45-0946) seems to be very low throughout the entire milling process. As a summary of the XRD analysis, we can conclude that the 1 h milled sample is dominated by the original elements, with some minor content of freshly formed intermetallics. The powder which underwent HEBM for 3 h is mainly characterized by the intermetallic compounds, with some traces of the initial elemental metallic phases. The longest milling time results in a partially amorphous state coupled with some unreacted Ni and compound phases.

Figure 5 presents the lattice parameters of the main crystalline phases determined from the Bragg-peak positions as a function of HEBM time. Regarding the hexagonal Mg phase, both the a and c lattice parameters behave similarly; namely, they are fairly close to their respective literature values (see the dashed horizontal lines) until 1 h of milling, thereafter a gradual decrease being realized as the severe plastic deformation process continues up to 3 h of milling (Figure 5a). The corresponding relative lattice shrinking is about 0.4% for both parameters. This feature is most probably due to the formation of dislocations, generated in the closed packed hexagonal structure during SPD [28,30]. Prolonged milling up to 10 h results in the complete elimination of the Mg reflections on the XRD pattern (see Figure 4). In the case of the fcc-Ni phase, a considerably different evolution is observed for its lattice parameter, i.e., it increases substantially from the literature value with milling time, reaching a relative expansion of 0.3% (Figure 5b). This phenomenon can be explained by the different mechanical properties of the individual constituents, namely that Mg (and Cu) atoms can occupy interstitial positions in the harder Ni lattice. Figure 5c presents the evolution of the lattice parameters of the main intermetallic phase (Mg₂Ni) formed by solid-state reaction. Both the a and c parameters of the hexagonal phase follow their respective literature values up to 3 h of milling; however, they behave oppositely up to 10 h of milling. As one can notice, the c parameter suffers a significant decrease, while the a parameter reveals a considerable increase. As noticed from Figure 4, the XRD pattern of the 10 h milled powder significantly differs from the other diffractograms; it corresponds to a mixed and partially disordered structure. The mechanical influence of the grinding balls during the extensive milling can promote changes even on the atomic scale [26], resulting in the observed lattice distortion of the Mg₂Ni phase. It is noted that the errors of the lattice parameter values are in the order of the symbol size.

The evolution of the coherently scattering crystallite size of the main phases obtained from the Williamson–Hall analysis can be inferred from Figure 6. A usual tendency is confirmed, i.e., both Mg and Cu crystallite size continuously decreasing with ball-milling time, reaching nanometric values with final sizes of D = 16 nm and D = 31 nm, respectively, after 3 h of HEBM, in correlation with literature data obtained on ball-milled Mg-based nanocrystalline powders [28,30,31,33,37]. Since prolonged milling results in the elimination of elemental Mg and Cu, crystallite size values are not available for that stage. It is also noticed from the plot that the microstructural refinement of Ni compared to Mg is slower (reaching D = 75 nm after 10 h of HEBM), which is probably due to Mg-enriched grain boundaries with increased reactivity to promote the formation of a disordered amorphous phase [33,34]. Nevertheless, elemental Ni is still present after 10 h, typical for the Mg-Ni system, even at higher Mg concentrations [31,37,39]. The average size of the Mg₂Ni compound phase generated by solid-state reaction is about D = 10 nm, similarly to literature values [31,37]. It is noted that the typical size of the coherently scattering nanocrystalline domains obtained from X-ray line broadening (see Figure 5) is about 15–60 times less than the average powder particle size determined from SEM image analysis (see Figure 1), which means that the individual powder particles contains thousands of crystallites surrounded by high-angle grain boundaries.

Isothermal hydrogenation and dehydrogenation measurements of the Mg₆₀Ni₂₅Cu₁₀Ce₅ powders are summarized in Figure 7. In general, all three powder composites can absorb and desorb certain amount of hydrogen; nevertheless, different milling durations and consequently the different microstructures result in rather different hydrogen storage performances.

The powder composite milled for 1 h requires a full activation cycle to achieve fresh surfaces, which then can absorb a significant amount of hydrogen (see Figure 7a). This feature is probably due to the presence of a thin oxide layer developed during the HEBM process on the surface of unreacted elemental Mg particles. According to

Table 3. Hydrogenation and dehydrogenation data obtained from the soprtion measurements.

Milling Time (h)		Hydrogen Capacity (wt.%)	Total Time (s)	Time to Reach 50% Capacity (s)
1	ABS1	1.18	1942	312
	ABS2	4.43	3273	546
	DES1	–1.38	2950	726
	DES2	–4.25	450	68
3	ABS1	0.69	1920	255
	ABS2	0.79	2402	275
	DES1	–0.75	3245	127
	DES2	–0.71	1001	126
10	ABS1	0.26	1610	99
	ABS2	–	–	–
	DES1	–0.29	1420	87
	DES2	–0.27	1250	102

, the capacity of this sample exceeds 4.2 wt.% both in absorption and desorption. Taking into account the theoretical capacity of Mg (7.6 wt.% H₂) and its weight percent in the powder mixture (34 wt.%), the pure Mg should absorb only 2.6 wt.% hydrogen in total. Therefore, it is evident that other phases, i.e., the intermetallics (typically Mg₂Ni) formed by solid-state reaction, can readily absorb and release hydrogen as well. It is also realized from Figure 7a and Table 3 that the dehydrogenation of the 1 h milled powder after the activation cycle presents excellent kinetics, releasing 50% of the total H₂ in approximately one minute. These values confirm better hydrogen capacity as well as faster kinetics at 573 K when compared to the hydrogen storage performance of nanocrystalline Mg₂₀Ni_10−xCu_x (x = 0–4) alloys prepared by melt spinning [32]. The superior kinetics of our Mg₆₀Ni₂₅Cu₁₀Ce₅ powder relate not just to the presence of nanocrystallites (see Figure 6), however, but to enhanced lattice defects as well that are created by the intense plastic straining during the HEBM process [22,30,31]. It is also noted that such capacity values can be achieved by fast absorption and desorption in ball-milled Ce₅Mg_95-xNi_x alloy series, especially for X = 15 at.% [43]. When Ni is replaced by Nd, the obtained capacity can further be increased up to 5 wt.% [51].

As the HEBM time increased up to 3 h, the hydrogenation and dehydrogenation performance of the Mg₆₀Ni₂₅Cu₁₀Ce₅ powder show significant changes, which can directly be related to the different phase content (see Figure 4) as well as the different microstructure (see Figure 6) of this powder. As seen in Figure 7b, this powder mixture of intermetallics and some unreacted metallic elements is ready to store hydrogen during the first absorption. The similar hydrogenation and dehydrogenation kinetic curves corresponding to the successive cycles presumes that surface oxidation of the Mg particles has negligible contribution due to its low volume fraction. As one can realize from Table 3, the kinetics are excellent, absorbing half of the maximum amount of hydrogen in approximately 250 s, and releasing it within ~120 s [46]. The enhanced kinetics can be the consequence of the reduced average crystallite sizes (see Figure 6), and the larger relative volume fraction of grain boundaries, which can act as fast diffusion channels for hydrogen. At the same time, the increased amount of Mg₂Ni with lower hydrogen activation energy can promote the easier dissociation of H₂ molecules at the surface [22]. Despite the exceptional absorption kinetics of this composite, its hydrogenation performance significantly deteriorates; the obtained capacity for absorption (0.8 wt.%) and for desorption (0.75 wt.%) is well below the respective values corresponding to the 1 h milled powder. Thus, it is concluded that the formation of nanocrystalline intermetallics has a positive effect on the cyclic stability and kinetics of Mg₆₀Ni₂₅Cu₁₀Ce₅; nevertheless, among Mg₂Ni, Mg₂Cu, MgNi₂, Cu_3.8Ni and CeMg, only the first compound exhibits significant hydrogen sorption capacity [23,31]. It is also apparent from Figure 7c and Table 3 that the longest milling time (10 h) results in the poorest overall sorption characteristics. The amount of absorbed and desorbed H₂ of the partially amorphous alloy is only ~0.25%, which is in line with previous observations [49]. The observed non-continuous and sluggish kinetics for the second absorption might correspond to the formation of a significant amorphous component (see Figure 4) that can contain non-stoichiometric oxygen within its excess-free volume. These oxygen-rich amorphous regions are not detectable by powder diffraction, and more precisely cannot be distinguished from the primary amorphous phase, especially when crystalline phases are also present. The regions with different oxide content may be intact to hydrogen to different extents, resulting in the alteration of the hydrogen absorption rate during the entire hydrogenation reaction.

As the interpretation of the hydrogen kinetic experiments, we can conclude that two effects compete against each other during the HEBM process: nanocrystallization vs. (intermetallic) phase formation. The former one is induced by the successive powder–powder, powder–ball and powder–vial collisions, while the latter one is the consequence of solid-state reactions. The 1 h milled powder still contains elemental constituents, and seemingly elemental but already nanocrystalline Mg positively affects the capacity; however, the cycling performance is worse due to the presence of a thin oxide layer. On the other hand, intermetallic phase formation reduces the overall capacity (the theoretical capacity of Mg₂Ni is 3.2 wt.% [31]); however, it significantly improves the cycling stability and kinetics due to the easier dissociation of the hydrogen molecules.

In order to explore and comprehend the mechanism of hydrogen sorption of the Mg₆₀Ni₂₅Cu₁₀Ce₅ powders milled for different durations, the isothermal sorption measurements (see Figure 7) were normalized to their respective maximum capacity. These normalized $\alpha \left(t\right)$ reaction functions were fitted by different kinetic model functions available in the literature [58,59,60,61,62]. It was found that the best fit can be achieved by the Johnson–Mehl–Avrami (JMA) model [63,64]. This model assumes that the nucleation of the new phase commences at random positions, and these embryos grow homogeneously over the entire volume of the material. The JMA kinetic function can be written as

$$\alpha \left(t\right)=1-e^{{-kt}^n},$$

(6)

where k is a temperature-dependent reaction constant and n relates to the growth dimensionality of a nucleating phase [64]. For a better visualization, Equation (4) can be transformed into the following equation:

$$\mathit{ln}(-\mathit{ln}(1-\alpha ))=n·\mathit{ln}\left(t\right)+ln(k).$$

(7)

The slope of the fitted straight line on the $\mathit{ln}(-\mathit{ln}(1-\alpha ))$ vs. $\mathit{ln}\left(t\right)$ function directly provides the value of n, while its intercept with the ordinate provides k. Figure 8 depicts the transformed 1st and 2nd absorption and 1st and 2nd desorption functions together with the fitted straight lines for all the Mg₆₀Ni₂₅Cu₁₀Ce₅ powders.

As seen in Figure 8a, the $\mathit{ln}(-\mathit{ln}(1-\alpha ))$-transformed functions obey a linear relationship for the activation as well as the subsequent absorption in the case of the powder that underwent HEBM for 1 h. Therefore, both hydrogenation processes can satisfactorily be ascribed to a single n exponent. On the other hand, a detectable deviation from the straight line is visible for the 1st absorption of the powder milled for 3 h. This feature assumes the initial stage of hydrogenation mechanism up to t~400 s (n = 1.37) is somewhat different from main part of the reaction (n = 0.79); see

Table 4. JMA exponents obtained from the linear fit of the transformed absorption and desorption kinetic curves for different milling times.

Milling Time (h)	ABS1 Exponent [n]		ABS2 Exponent [n]	DES1 Exponent [n]	DES2 Exponent [n]
1	0.91		0.83	1.09	1.34
3	1.37	0.79	0.79	1.22	1.38
10	0.75		-	0.67	0.71

. At the same time, the transformed reaction function of the 2nd dehydrogenation follows a linear relationship with an exponent n = 0.79. This feature is in correlation with the different shape of the two absorption curves; see their multiple intersections in Figure 7b. As seen from Table 4, the n parameters describing absorption are in the range of n~0.75–0.9, confirming that—despite the different microstructure of the three powder mixtures (see Figure 4 and Figure 6)—the hydrogen-absorption mechanism is akin for all composites. These relatively low exponent values presume a specific kinetics of diffusion-controlled growth when nucleation of the hydride phase(s) commence near lattice defects [65] that are abundantly generated within the powder particles during repeated collisions with the hardened milling balls. Similar values were obtained for ball-milled nanocrystalline Mg/MgH₂ powders catalyzed by FeTi [56].

The desorption of the Mg₆₀Ni₂₅Cu₁₀Ce₅ powders can be associated with a single JMA exponent for all milling times, and the obtained values for 1 and 3 h of milling are somewhat larger than those obtained for absorption (n~1.1–1.4). These values refer to a combination of multiple processes, including diffusion-controlled growth with decreasing nucleation rate and nucleation of the dehydrided phases in the vicinity of the existing lattice defects and grain boundaries [65]. The slightly larger values for the second desorption may originate from the different nucleation rate and/or the reduced number of lattice defects.

The phase formation during hydrogen sorption for each powder can be followed in Figure 9. Each graph presents the XRD measurements of the as-milled powder, as reference, the cycled (dehydrided) and the fully hydrided states. In general, the patterns show that the dehydrogenation is practically complete for each powder, independent of their phase composition and in accordance with the similar absorbed and released amount of H₂; see Table 3. Taking a careful look, it can be realized that a very small amount of MgH₂ did not convert to Mg during the dehydrogenation process of the powder milled for 3 h. It is also recognized that the relative Bragg-peak intensities of the different phases characterizing the individual powders has not changed after the hydrogenation/dehydrogenation cycle, therefore the phase mixture of the cycled state is identical to that of the as-milled powder and corresponds to an almost complete absorption–desorption process. This is also supported by the negligible oxidation for the 1 h and 3 h milled samples; see the unchanged intensity of the MgO crystalline peaks. The absorbed state of the powder which underwent HEBM for 1 h reveals the formation of multiple hydride phases, including MgH₂ (JCPDS 12-0697), confirming that elemental Mg is still present in this powder blend. In addition, Mg₂NiH₄ (JCPDS 35-1235) and a solid solution Mg₂NiH₀.₃ (JCPDS 40-1204) can also be identified in the XRD pattern. As also seen, some amount of unreacted Ni is present in this state. The hydrided state of the 3 h milled sample is mainly characterized by the two hydrided states of Mg₂Ni, while MgH₂ is less dominant in accordance with the reduced H-storage capacity of this composite (see Figure 7 and Table 3). The XRD pattern of the absorbed state of the powder milled for 10 h is still dominated by an amorphous component as well as hydrides of intermetallics. It is noted that oxidation is more significant in this case; this can also contribute to the reduced capacity of this alloy.

4. Conclusions

In this research, HEBM for different durations was used to produce nanocrystalline Mg₆₀Ni₂₅Cu₁₀Ce₅ powders. Quantitative analysis of the SEM micrographs revealed that particle-size histograms follow a lognormal distribution for powders milled up to 1 h, 3 h and 10 h. The average powder particle size was ${〈d〉}_{area}$~ 1 μm for each milling time. Local area elemental EDS mapping on the powder milled for 10 h confirmed an almost complete mixture of the initial elements with compositional homogeneity.

X-ray diffraction experiments revealed that different milling times result in considerably different phase composition. The pattern of the as-milled powder for 1 h is dominated by the crystalline peaks of elemental Mg, Ni, Cu and Ce, with some minor content of freshly developed intermetallics. HEBM up to 3 h promotes the intensive formation of intermetallic compounds, such as Mg₂Ni, Mg₂Cu, MgNi₂, Cu₃.₈Ni and CeMg, while 10 h of powder processing results in the formation of a partially amorphous state with compound phases.

Isothermal hydrogenation and dehydrogenation experiments established that all powders absorb H₂ reversibly, while the shortest milling time provides the best overall capacity. Excellent kinetics without any necessary activation were obtained for the 3 h milled composite, releasing and absorbing 50% of the total hydrogen content within 120 s. Each kinetic measurement has been satisfactorily fitted by the Johnson–Mehl–Avrami function. X-ray diffraction analysis on the dehydrided powders confirmed the complete desorption. The anomalously low JMA exponent corresponds to a diffusion-controlled growth when precipitates of the nucleating phase occur near lattice defects.