Characterization of the As-Milled Powders
As one can observe from the low-magnification BSE images, the as-milled MgH
2-FeTi composites are dominated by powder agglomerates and some debris (
Figure 1). The individual powder particles are usually surrounded by flat surfaces, which were formed by powder-to-ball and powder-to-vial collisions during the milling process, when the relatively soft hydride particles were pressed onto the internal walls of the stainless steel vial. In addition, some of these particles are also characterized by sharp edges and cracks, which are the consequences of SPD occurring during the HEBM. The brighter areas observed for all powder composites correspond to the FeTi particles that are evenly and homogeneously distributed after milling for 3 h with MgH
2. As evident from the images, the catalyst agglomerates are considerably smaller than the MgH
2 ones. During the intensive plastic deformation, the relatively hard FeTi particles of higher mass density were embedded in the soft hydride agglomerates, which led to the formation of the observed matrix (MgH
2)-reinforcement particle (FeTi) composite structure.
Quantitative analysis of the MgH
2 agglomerate sizes was obtained from the SEM BSE images by measuring the individual length through their geometrical center in the same direction for all particles. Presented in
Figure 1, the obtained particle size histograms can satisfactorily be fitted by a lognormal-distribution function that has a maximum of around 2 microns for all the composites. This value is in agreement with other Mg/MgH
2 powders milled with metal or metal oxide catalysts [
32,
52]. Similar values have been obtained for other MgH
2-FeTi composites processed in the planetary mill for up to 36 h of milling time [
54].
The fitted lognormal particle-size-distribution functions can provide the median (mSEM) and variance (σSEM) as well as the area-averaged particle size (〈d〉area). As observed on the histograms, the 〈d〉area values slightly increase with increasing FeTi content, ranging from 3.58 μm to 5.02 μm for the MgH2-40 wt.% FeTi and MgH2-60 wt.% FeTi powders, respectively.
In our opinion, this dependency on the FeTi content may be explained by the increased amount of mechanical contact between the pre-milled FeTi particles and MgH
2 aggregates during the co-milling step. For the MgH
2-40 wt.% FeTi sample, the MgH
2 agglomerates were predominantly formed around the individual FeTi particles (
Figure 1a); however, in the MgH
2-60 wt.% FeTi composite, the MgH
2 particles tended to aggregate around the FeTi ones (
Figure 1c). This behavior is also supported by the variation in the MgH
2-particle size distribution. Namely, the smallest value (σ
SEM = 0.45) for the MgH
2-40 wt.% FeTi composite corresponds to a relatively sharper and more homogeneous distribution, while the value of σ
SEM = 0.58 for MgH
2-60 wt.% FeTi describes a wider distribution.
The corresponding EDS spectra, shown as insets in
Figure 1, provide quantitative information on the composition of the different composites. The obtained values for Mg, Ti and Fe are listed in
Table 1. As one can observe, the Fe:Ti concentration ratio is in perfect agreement with the 1:1 nominal value; nevertheless, the overall FeTi catalyst concentration is lower than the nominal value for all the composites (31.9 at.% for MgH
2-40 wt.% FeTi, 38.5 at.% for MgH
2-50 wt.% FeTi and 46.7 at.% for MgH
2-60 wt.% FeTi). Similar features have been obtained for MgH
2 mixed with significantly lower FeTi content [
52]. The discrepancy between the nominal and experimental values is related to the considerably different mechanical properties of the MgH
2 and FeTi powders, i.e., the FeTi particles can stick and/or adhere more easily to the internal walls and edges of the milling vial.
A high-resolution image of the MgH
2-40 wt.% FeTi sample captured around an FeTi particle (denoted by A) reveals that MgH
2 agglomerates and plates or layers with different thicknesses adhere to the surface of the catalyst (
Figure 2).
Supplementary point-like EDS analysis revealed significant compositional differences between the brighter and darker spots. As listed in
Table 2, heavier Fe and Ti elements mostly dominate the brighter areas (particle A), while the darkest region (denoted by B) is enriched in Mg, corresponding to a MgH
2 agglomerate or layer surrounding the catalyst particle. Regions C and D exhibit intermediate concentrations, presumably due to the overlapping of FeTi and MgH
2 particles within the range of the electron penetration depth.
The XRD patterns of the milled MgH
2-FeTi powder composites are presented in
Figure 3. As one can observe from the diffractograms, each powder exhibits the reflections of tetragonal α-MgH
2; however, the high-pressure metastable γ-MgH
2 phase is also present as a minor component. All peaks are broader, which is indicative of extensive crystallite size reduction [
32]. It is also noted that peak intensities significantly decrease in height with increasing FeTi content, and, at the same time, no Bragg peaks of any Fe-Ti phase are visible, which could correspond to a solid state amorphization of the catalyst, as can be confirmed by the broad halo developed at around 2θ = 42 deg [
69]. In addition, a reflection centered at 2θ = 44 deg is present for all the powders, corresponding to the formation of some MgO.
A typical attempt at applying the CMWP algorithm on a measured XRD profile can be seen in
Figure 4. Apart from the fitted profile, the difference plot is also presented for the MgH
2-40 wt.% FeTi composite. It is clear from the image that the numerical algorithm satisfactorily converged after ~100 iteration steps and that the analysis can serve as a useful tool to characterize the microstructure of HEBM Mg-based hydrogen-storage alloys in detail [
28,
30,
37,
52,
65,
68].
The microstructural parameters, such as the median (m
CMWP) and variance (σ
CMWP), of the coherently scattering α-MgH
2 crystallites obtained from the CMWP analysis are listed in
Table 3. As can be seen, for the high-FeTi-containing MgH
2-60 wt.% FeTi alloy, the m
CMWP value is larger by a factor of ~2 compared with the other two composites, while the σ
CMWP value is about half that of the other two alloys. The smaller σ
CMWP value corresponds to a more homogeneous nanostructure. It is noted that on a micron scale, the powder-particle size distribution obtained from SEM images exhibits the highest σ value (see
Figure 1); therefore, it is concluded that different mechanisms take place on these two length scales. As was described above, the individual MgH
2 powder particles tend to agglomerate in the vicinity of the FeTi particles at a higher catalyst content. At the same time, the entire volume of these micron-scale MgH
2 particles is attrited by the formation of coherently scattering nanocrystals that are separated by large-angle grain boundaries. Apparently, this process is more effective in the presence of a higher FeTi content.
By applying m
CMWP and σ
CMWP via Equation (1), the G(x) lognormal size-distribution functions can be constructed for all the composites (see
Figure 5). Due to the largest m
CMWP value being for the MgH
2-60 wt.% FeTi alloy, its histogram is shifted toward higher values, and its maximum occurs at around 7 nm. The corresponding 〈D〉
area area-averaged particle size and 〈D〉
vol volume-averaged crystallite sizes were also determined (see Equations (2) and (3)); these values are also listed in
Table 3 and represented by symbols in
Figure 4. In accordance with the broad Bragg peaks observed in
Figure 3, all the characteristic values are in the nanometric range (8–19 nm), which correspond to an intensive nanocrystallization during the HEBM process and are in agreement with other MgH
2-based hydrogen-storage alloys processed by HEBM [
30,
32,
52]. In particular, a 9 nm final crystallite size of MgH
2 was achieved after prolonged milling time (24 h) [
55], confirming our previous statement that there exists an optimal milling time for achieving the best overall microstructure from the point of view of hydrogenation performance [
52,
65,
68]. As a consequence of the nanocrystallization process, the relative volume fraction of the grain boundaries inside the individual powder particles increases significantly, and these grain boundaries, with distorted atomic bonds and smaller atomic density [
29], can act as fast diffusion channels for hydrogen. In a recent research study, it was determined that FeTi catalyst powder particles can not only induce the nanostructural refinement of MgH
2 via HEBM, but they can also promote the abundant formation of lattice defects like dislocations [
52]. These lattice defects can further enhance the hydrogen kinetic performance of MgH
2 [
28,
30].
Figure 6 presents the continuous-heating DSC scans obtained at 40 Kmin
−1 for the MgH
2-FeTi composites. In general, each thermogram exhibits an overlapping two-step endothermic reaction, which corresponds to the dehydrogenation of the metastable γ-MgH
2, stable α-MgH
2 and hydrides at T
1 and T
2 temperatures, respectively. This feature was also confirmed previously by a temperature-dependent XRD study on the dehydrogenation of MgH
2 [
52]. The notably lower T
1 temperature characterizing the γ-MgH
2 → Mg + H
2 reaction is attributed to the weaker Mg-H bond in the high-pressure hydride phase [
52]. The T
onset onset temperature of the first reaction as well as the T
1 and T
2 peak maxima are listed in
Table 4. As one can observe, the variation in these characteristic temperatures has only a slight FeTi-content dependence; however, the overlap between the two reactions is more significant for the MgH
2-60 wt.% FeTi composite. The lowest T
onset and T
1 temperatures of this alloy assumes that it has the lowest dehydrogenation stability among all the powders, most probably due to the enhanced synergetic effect between MgH
2 and FeTi. The total dehydrogenation enthalpy release of MgH
2 (∆H
tot) clearly shows a decreasing tendency with increasing FeTi content (see
Table 4).
The isothermal hydrogen absorption and desorption kinetic measurements of the MgH
2-FeTi powders are presented in
Figure 7. As noted from
Figure 7a, all the composites can absorb a significant amount of H
2 in a very short time, corresponding to excellent kinetics in this system. According to
Table 5, the largest capacity (5.8 wt.% H
2) is achieved for the lowest FeTi-containing composite (MgH
2-40 wt.% FeTi). Mixing 40 wt.% pre-milled FeTi with MgH
2 should decrease the theoretical capacity of Mg (7.6 wt.% H
2) down to 5.4 wt.%; therefore, it is evident that the FeTi catalyst also participates in hydrogen absorption. These capacity values are significantly higher than those obtained for the MgH
2 nanocrystalline powder that had been subjected to HEBM with FeTi for 25 h [
48]. In addition, the finding that 3.3 wt.% H
2 is absorbed in the first 100 s of hydrogenation confirms the excellent kinetics of this material, which confirms that a synergetic effect clearly develops between MgH
2 and FeTi. Noteworthily, when only 10 wt.% FeTi was mixed with MgH
2, a similar amount of hydrogen was absorbed only after several hundred seconds [
52]. Another important feature of the dehydrogenation experiments is that the MgH
2-50 wt.% FeTi powder possesses the best absorption rate by far at shorter times (4.3 wt.% absorbed H
2 after 100 s). Our values indicate a considerably better kinetic performance than that observed for the MgH
2 + (FeTi)
0.92Mn
0.08 composite (3 wt.% after 500 s at T = 573 K) [
51]. As also inferred from
Figure 7a and
Table 5, the capacity and kinetics of the MgH
2-60 wt.% FeTi composite are evidently the worst, which is a consequence of the smaller theoretical capacity of the FeTi solid solution and/or amorphous phase with respect to Mg.
Figure 7b confirms that the hydrogen desorption was complete for all powders, similarly to the dehydrogenation during the continuous DSC experiments (see
Figure 6). In general, the amount of released hydrogen was slightly low for all samples; e.g., it was 5.5 wt.% in the case of the MgH
2-40 wt.% FeTi sample, see also
Table 5. Despite the exceptional absorption kinetics of this alloy, its dehydrogenation performance is also somewhat poorer, releasing 2.8 wt.% H
2 in 120 s. It is also apparent from
Figure 7 and
Table 5 that the highest FeTi content results in the poorest overall sorption characteristics, most probably due to the highest average powder particle size (see
Figure 1). The larger agglomerate size on the micron scale clearly corresponds to fewer free surfaces per unit volume; therefore, the overall hydrogen penetration and diffusion are less favored in this powder. Another feature that might be responsible for the poorest hydrogenation kinetic performance of the MgH
2-60 wt.% FeTi composite occurs at the nanoscale. As determined from the CWMP analysis (see
Table 3), the σ
CMWP variance value of the coherent-scattering nanocrystals is the smallest among all the powders. These nanocrystals, therefore, are relatively similar in size and, henceforward, their grain boundaries at the contact surfaces do not possess a very different overall microstructure. In contrast, the other two powders exhibit a larger variance and an inhomogeneous crystal size distribution, so the grain boundaries among these different nanocrystallites can cover a wider structural diversity, which can also be responsible for their better hydrogenation kinetics.
In order to reveal the underlying mechanism of hydrogen sorption of the MgH
2-FeTi composites, the measured kinetic absorption and desorption curves were normalized to their maximum capacity. Thereafter, these
absorption and desorption normalized functions were fitted with different model functions available in the literature. Accordingly, the kinetic data can satisfactorily be fitted by the Johnson–Mehl–Avrami (JMA) function:
where
k is a temperature-dependent reaction constant and
n relates to the growth dimensionality of a nucleating phase [
70]. This model assumes that the nucleation of the new phase occurs randomly and homogeneously over the entire volume of the material. For better visualization, Equation (4) can be transformed into the following equation:
By fitting the
term as a function of
with a straight line, the slope of the fitted line determines the value n, while its intercept with the ordinate provides the value for k.
Figure 8 presents such fitted straight lines for all the MgH
2-FeTi powders. Generally, the
-transformed absorption functions follow well the linear relationship for all powders (
Figure 8a); therefore, the entire hydrogenation process can satisfactorily be described by a single n exponent. These fitted values are listed in
Table 6.
The transformed dehydrogenation functions for the MgH
2-40 wt.% FeTi sample obeys a linear relationship (
Figure 8b); however, a significant deviance from the straight line is observed for the MgH
2-50 wt.% FeTi and MgH
2-60 wt.% FeTi composites, in correlation with an extensive change in the shape of the desorption curves at t = 300 s (see
Figure 7b). This feature assumes that there is a variation in the dehydrogenation mechanism as a function of time.
As listed in
Table 6, the n parameter describing absorption is in the range of n~0.3–0.5, with only a slight difference occurring with varying catalyst content. These values are in accordance with those obtained for the sorption of MgH
2 catalyzed by a much smaller amount (10 wt.%) of nanocrystalline FeTi [
52], indicating that the H-sorption mechanism of Mg/MgH
2 is very similar for all cases, irrespective of the relative abundance of FeTi particles to the MgH
2 ones. These anomalously low n values describing a JMA-type transformation correspond to a specific type of diffusion-controlled phase growth in which precipitates of the nucleating phase (MgH
2 for absorption and Mg for desorption) commence near lattice defects (vacancies and dislocation or grain boundaries) [
70], which are abundantly generated by SPD during the HEBM process. The desorption of MgH
2-40 wt.% FeTi exhibits a single JMA exponent (n = 0.95) that corresponds to a combination of several overlapping processes, including diffusion-controlled growth with a decreasing nucleation rate and nucleation in the vicinity of lattice defects [
70]. The double-logarithm JMA plots for the other two alloys do not show a linear tendency; both curves exhibit two linear segments, corresponding to a significant variation in the dehydrogenation mechanism as a function of time.