Mg-Ni-Nb₂O₅ Composite Produced by High-Pressure Torsion

Abstract

A Mg-based composite material has been produced by the consolidation at room temperature of a Mg-5wt.% Ni-2wt.% Nb₂O₅ powder mixture subjected to high-pressure torsion (HPT), one of the processing methods to induce severe plastic deformations. The microstructure, density, and microhardness of the consolidated disks were characterized after the application of up to 30 revolutions in torsion under compression stresses of 3 and 5 GPa. According to the density measurements, the composite was consolidated in full after the application of five revolutions, although disks subjected to only one revolution exhibited densities close to the maximum measured value. On the other hand, grain size and microhardness measurements showed that differences existed at locations near the center and the periphery of the HPT-processed disks. Under the stress of 5 GPa, the grain size in the central regions stabilized at about 0.35 μm after five revolutions, while at the peripherical regions it gradually decreased with an increasing number of revolutions down to about 0.15 μm after 30 revolutions. In turn, the microhardness measured along a diametral cross section steadily increased with the number of revolutions between 1 and 10 revolutions, maintaining a gradient from the center to the periphery in all cases. With the application of 20 and 30 revolutions, only the peripheral regions increased considerably in hardness. It was discovered that the magnesium particles in the initial powder mixture had formed an oxide—hydroxide surface layer, which changed the expected final density of the consolidated material by about 2 to 4.5%. This superficial contamination of the Mg powders did not prevent the material from achieving full consolidation.

1. Introduction

The production of Mg-based alloys with ultrafine grain size by application of SPD processing has been reported on multiple occasions [1,2,3,4,5]. Among many possible applications, magnesium alloys and composites are interesting as hydrogen storage materials [6,7,8,9,10,11,12] due to their high absorption capacity (7.6% by mass for pure Mg) as well as the low cost, abundance, and low density of their main component. Mg-based alloys with ultrafine or nanometric grain size are known to possess markedly improved kinetics of hydride formation and decomposition [6,7,8,9,11]. Different methods have been developed to control grain size, including a variety of plastic deformation processes. Among the latter, processes of severe plastic deformation (SPD) impose enormous plastic deformations without changing, in principle, the dimensions of the processed sample. The conservation of the original shape of the sample is achieved through special tool geometries in these processes, which prevent the free flow of the material and therefore, introduce a considerable hydrostatic stress component, making high deformations possible without fracture [1,13,14]. Proper selection of SPD conditions provides a fine microstructure in a macroscopic volume of a workpiece. Consequently, the interest in the production of bulk ultrafine grain or nanostructured materials using SPD techniques has increased significantly in the last decades.

Among the large variety of SPD processes, high-pressure torsion (HPT) is particularly useful to subject small quantities of a solid or powder material to very high plastic deformations [1]. Although the effect of HPT processing on bulk solid materials has been extensively investigated, interest in using HPT to consolidate powders as starting materials has grown in recent years [15,16,17]. The present investigation focuses on studying the structural evolution of a Mg-based powder mixture that undergoes HPT processing, resulting in a Mg-based composite of interest as a hydrogen storage material [6]. In this particular case, submicron Ni and Nb₂O₅ powders (5wt.% and 2 wt.%, respectively, have been added to relatively large, commercially pure Mg powders in order to produce a Mg-based alloy with attractive hydrogen storage capacity and hydrogenation kinetics [6]. The addition of nanometric particles of Ni and Nb₂O₅ is known to greatly improve the hydrogen storage properties of the Mg–MgH₂ system [7,8]. Furthermore, refinement of the grain/crystallite size of the magnesium matrix is also essential to obtain enhanced hydrogenation kinetics. Most frequently, the Mg-based composites for hydrogen storage applications are produced as a fine powder mixture by milling/mechanical alloying processing. One disadvantage of such an approach is that the fine Mg powders can easily ignite in contact with the atmosphere. As an alternative route to obtain a refined-grained Mg matrix with a dispersion of fine catalyst particles, we have used HPT processing of a mild-milled Mg-Ni-Nb₂O₅ powder mixture (which did not render highly reactive Mg powders) to fabricate an ultrafine grain Mg-matrix composite, which possessed the desired microstructure needed for hydrogen storage applications in a bulk material of easy and safe manipulation.

2. Materials and Methods

2.1. Sample Processing

Fine Ni and Nb₂O₅ powders were used as additions to commercially pure Mg powders, in nominal amounts of 5 and 2% by weight, respectively. All materials were purchased in powder form: Mg, ~325 mesh (Riedel de Haen, catalog #S030195); Ni, <1 μm (Sigma-Aldrich, catalog #268283); Nb₂O₅, <1 μm (Sigma-Aldrich, Milwaukee, WI, USA, catalog #203920).

Mixing of the powders was carried out in a planetary mill under nitrogen atmosphere using the following mild milling conditions: steel balls of 10 mm diameter; milling time, 45 min; rotation speed, 250 rpm; ball/powder mass ratio, 5/1; 1 mL of methanol added as a process control agent per 30 g of powder. The milling operation was aimed at dispersing the original Ni and Nb₂O₅ fine powders over the much larger Mg particles, as illustrated in Figure 1a,b. The as-milled powder mixtures were stored in closed containers (for extended periods of time) without additional measures to prevent them from coming into contact with the atmosphere. The effect of the oxide/hydroxide layer formed on the Mg particles under these conditions was of interest in determining the hydrogen storage properties of the composite, but this issue will not be discussed in this paper.

Figure 2 is a schematic diagram of the processing route of the as-milled powders. In the first stage, the mixture was compacted by uniaxial compression under a pressure of 1.05 GPa maintained for 30 s, to produce disks with a diameter of 10 mm and a thickness of ~2 mm. Processing by HPT of the pre-compacted disks was performed at Toyohashi University of Technology, Japan. The pre-compacted samples were subjected to 1, 5, 10, 20, or 30 revolutions (N) in torsion under quasi-constrained conditions at a speed of 0.2 rev/min and under pressures of 3 or 5 GPa. In the following, most of the results correspond to disks processed under 5 GPa.

2.2. Characterization

Phase analysis of the HPT-consolidated products was carried out by X-ray diffraction (XRD) using a Rigaku model Ultima IV diffractometer (Toyohashi, Japan) with Cu-Kα radiation under the following operating conditions: 40 kV voltage, 15–90° 2θ sweep with 0.02° step size, and 1s period.

The microstructure of the disks consolidated by the HPT process was examined by optical microscopy (OM) using a Keyence digital microscope VHX-6000, (Saltillo, Mexico), scanning electron microscopy (SEM) with a HITACHI SU5000 (Toyohashi-Japan), and transmission electron microscopy (TEM) by JEOL brand high-resolution transmission electron microscope JEM 2200FS+Cs (Chihuahua, Mexico). Elemental chemical composition was determined in both SEM and TEM by energy-dispersive spectrometry (EDS) using an Oxford Aztec system (Chihuahua, Mexico). For TEM examination, thin lamellae from locations near the center and periphery of the disks were prepared by focused ion beam (FIB), FEI model Scios (Chihuahua, Mexico), using a beam of Ga ions. Grain size measurements were made manually on dark-field TEM micrographs from these lamellae. The number of measured grains for each specimen and location varied from a minimum of 30 to a maximum of 150.

Determination of the density of the samples was carried out according to the ASTM B962-17 standard [18] using an analytical balance with a precision of 0.001 mg and Tridecane oil as the immersion medium. For microhardness measurements, a diametral cross section of the disks was metallographically prepared, and measurements were made starting from the center at locations separated by 1.5 mm, using a microhardness tester with a load of 98.07 mN (HV 0.01) maintained for 10 s. Three indentations were made at each location and the average of the calculated hardness is reported in the following.

3. Results and Discussion

3.1. Microstructural Characteristics after HPT

The X-ray diffraction patterns from samples subjected to 0 (only precompacted), 5, and 30 revolutions under a pressure of 5 GPa are shown in Figure 3. Similar results were obtained when applying 3 GPa (not shown). The phases expected from the composition of the original powder mixture are identified in all the XRD patterns, although the Nb oxide exhibits only one clear diffraction peak due to its low content. However, it should be noted that there are no recognizable peaks from magnesium oxide or hydroxide phases.

For a solid sample undergoing HPT processing, the applied strain depends on the number of revolutions and the location along the disk radius [1]. It will be shown below that the initial powder mixture was almost fully densified on the application of only one revolution. Therefore, the applied plastic deformation of the Mg-Ni-Nb₂O₅ composite is expected to follow the general pattern found in the HPT processing of solid samples. Variations of the microstructure from the center to the edge of the disks, with an increasing number of revolutions (N), are documented by the optical micrographs in Figure 4 for specimens subjected to 5 GPa pressure during HPT.

In general, these optical micrographs indicate a gradual refinement of the microstructure, with changes occurring more rapidly at the periphery of the disks. It should be noticed that most of the dark areas in the micrographs do not correspond to pores in the samples, as will be discussed in the following.

Figure 5a–d show low magnification, bright field TEM micrographs from thin lamellae prepared from locations near the center and the periphery of a disk subjected to 20 N under 5 GPa pressure. As a starting point, we can identify the second phases in the composite, namely, Ni and Nb₂O₅, as dark particles and agglomerates. This was verified by EDS analysis, as illustrated by the X-ray maps in Figure 5e. In many cases, these particles are associated with dark gray regions—more clearly noticed in Figure 5a–c—in which EDS analysis revealed the presence of magnesium and oxygen (Figure 5e) so that they have been identified as products of the reaction of the original Mg particles with oxygen and water vapor from the atmosphere, as previously reported in [6]. In other words, the original particles of Mg formed a surface layer, which is most likely composed of partially crystalline mixtures of magnesium oxides and hydroxides [19], labeled as Mg_xO_yH_z on the micrographs in Figure 5.

The detailed characterization of these layers is beyond the scope of the present work. Nevertheless, it is clear from these micrographs that the Mg_xO_yH_z layers have been fragmented by the plastic deformation occurring at the edge of the HPT disk. To a lesser extent, this also occurred in the central region of the disks, in accordance with the expected degree of plastic deformation in the HPT process. The apparent refinement of the microstructure observed in the optical micrographs in Figure 4 is therefore related to the fragmentation of the regions of the Mg_xO_yH_z component induced by plastic deformation.

The average grain size of magnesium grains after HPT processing was measured on TEM micrographs taken at higher magnifications, as illustrated in Figure 6a–c. The results of these measurements are reported in Figure 7. The data in this figure indicate that the grain size near the center of the disks processed by HPT reaches a “stable”, submicron size of 0.35 ± 0.13 μm after being subjected to 5 N, and practically remains the same with increasing revolutions up to 30 N. On the other hand, the grain size near the periphery of the disks has almost the same value (~0.35 μm) after 5 N but then decreases with increasing N until it reaches a value of 0.135 ± 0.03 μm after 30 N. Although more detailed work is needed, it is probably safe to assume that the grain size decreases gradually from the center to the peripheral regions. In addition, the grain size values are very similar in samples subjected to either 3 GPa (not shown) or 5 GPa pressure during HPT. Notice that the initial particle size of the Mg powders was about 60 μm (Figure 1a), which underlines the marked microstructural refinement produced by HPT in this Mg-based alloy. It has been argued that such refinement is brought about by both grain subdivision from dislocation accumulation and dynamic recrystallization [20]. In our material, the presence of fine particles of Ni and Nb₂O₅ probably also plays a role in the resulting grain sizes.

3.2. Density Measurements

The results of density measurements are shown in Figure 8. For both HPT pressures in use, the density of the disks increased up to a total number of 5–10 N and then remained practically without change for 20 and 30 N. However, the maximum density values were slightly larger for disks processed under 3 GPa (~1.915 g/cm³) compared to those subjected to 5 GPa (~1.890 g/cm³). The calculated density of the mixture with nominal composition Mg-5wt.% Ni-2wt.% Nb₂O₅ is 1.83 g/cm³ (rule of mixtures); this is basically the density determined for the pre-compacted disks (N = 0) in Figure 8. Two main factors may contribute to density values different from 1.83 g/cm³ in our samples, namely, remanent porosity and the existence of phases other than those considered in the nominal composition. It has already been shown that a Mg_xO_yH_z “contamination” layer existed on the surface of the Mg powders, which most likely consisted of a partially crystalline mixture of magnesium oxide and hydroxide.

According to the literature [1,21], oxide layers on magnesium surfaces only grow to a few nm in thickness under usual ambient conditions; however, layers of magnesium hydroxide can grow to several µm in thickness. Therefore, as an approximation, we have calculated the expected density of Mg-5wt.% Ni-2wt.% Nb₂O₅ mixtures in which some fraction of Mg is present as Mg(OH)₂ and plotted such density values also in Figure 8 (blue diamonds). From these calculations, we expect the amount of “contamination” existing in the original Mg powders to be of the order of 5 to 8 wt.%. Although the uncertainty in phase composition in the powder mixture prevents the calculation of remanent porosity in the HPT processed disks, it is likely that the pore volume has decreased to an insignificant amount considering the very large plastic deformation and high compression stresses applied to the disks during the HPT process. In this respect, it is relevant to mention that the thickness of the HPT disks was less than 1 mm after SPD processing.

3.3. Microhardness Evolution

The changes in microhardness along a diametral cross section of the disks processed by HPT under 5 GPa pressure are shown in Figure 9. For each specimen subjected to a given number N of revolutions, we observe in this figure a more or less symmetrical and gradual increase in hardness from the center to the periphery of the disk. The overall hardness level increases clearly with increasing N up to 10 N but, after 20 and 30 N, the increase is not significant except at the periphery of the disks. The gradient in hardness from the center to the periphery grows larger in disks subjected up to 20 N and decreases somewhat with 30 N.

The changes in microhardness along the radius of the disks and with the number of applied revolutions is a common feature in HPT processing [2,15,22]. Theoretically, the applied shear deformation in pure torsion is zero at the center and increases continuously with the distance from the center position [15]. HPT is performed under a large compression stress, and it is clear from the data in Figure 9 that the center of the disks is actually subjected to plastic deformation in this process, although to a lesser extent compared to locations away from the center. It is also clear that work hardening of the material is still occurring after 30 N.

The sample subjected only to the action of the 3 GPa pressure, without torsional deformation (N = 0), reached a rather uniform microhardness value of ~40 HV. Therefore, the hardening induced by the torsional deformation should be of foremost importance in determining the microstructure of the disks subjected to even only one revolution. According to the grain size measurements shown in Figure 7, the grains in the central region of the disks reached a “stable” size of 0.35 ± μm after 5 N, although the microhardness in the central region keeps increasing up to 10 HPT revolutions (Figure 9). On the other hand, near the periphery of the disks, the measured grain size is similar to that in the central regions after 5 N but decreases markedly (~180 nm) after 10 N and keeps decreasing for 20 N (~150 nm) and 30 N (~130 nm). The data in Figure 9 mostly indicate that the microhardness changes are steeper near the edge of the disks, especially for disks that underwent 20 and 30 HPT revolutions.

Figueiredo and Langdon compiled published data on the HPT processing of magnesium alloys, including measurements of grain size and hardness for various conditions of applied pressure and number of revolutions [5,23]. The minimum grain size and maximum hardness values quoted in these references for experiments with pure magnesium are 320 nm (mid-radius position, 10 mm disk, 6 GPa pressure, 10 N) and 60 kg/mm². This grain size value is remarkably close to the apparently “stable” value reached in the center region of our samples processed under 5 GPa pressure, but it is about twice the size we found in the periphery of the same disks (Figure 7) after 30 revolutions. In turn, the highest hardness values at the periphery of our samples (Figure 9), in the range of 73–92 HV, are much higher than those reported for pure magnesium; however, they are somewhat lower than those reported for commercial Mg alloys processed by HPT ([5,23]).

A number of publications have addressed the mechanisms of grain refinement by SPD processing [1,5] and more specifically by HPT [3,24]. Pippian et al. [25] propose that grain refinement during the HPT of metals is mainly the result of the subdivision of the original grains by the continuous accumulation of dislocations and migration of grain boundaries brought about by the strains and stresses imposed by the SPD process. At some point during deformation, recovery processes come to balance the generation of dislocations, and grain subdivision reaches a quasi-steady state, which manifests itself in the attainment of a refined “stable” grain size. The actual grain sizes generated at this stage depend mainly on the purity and second-phase content of the material and on the homologous temperature during deformation.

The materials in our work are composites with a magnesium matrix and second phases consisting of fine (<1 μm) particles of Ni and Nb₂O₅, as well as aggregates of a Mg_xO_yH_z component with sizes that diminished from about 10–15 μm at the start of HPT processing and reached <1 μm with increasing deformation. However, since the degree of deformation produced by HPT is not uniform over all the sample, the size of the Mg_xO_yH_z component would vary with the radial position and the number of applied revolutions. The existence of three different second phases in our material would explain the higher hardness values produced by HPT processing compared to pure Mg and the significantly smaller grain sizes found locally in the HPT-processed disks.

4. Conclusions

Processing by HPT turned out to be an effective way to fully consolidate a Mg-based powder alloy containing submicron Ni and Nb₂O₅ particles, as well as an oxide-hydroxide layer on the Mg particles, with the application of as little as one revolution.

The application of HPT to the Mg-5Ni-2Nb₂O₅ powder mixtures produced a composite with submicron grain sizes although with some heterogeneity depending on the location of the HPT disks. The grains at center locations achieved a size near 0.35 nm after five revolutions and remained practically constant up to 30 revolutions, while in the periphery of the disks the grain size decreased steadily from 0.35 nm after five revolutions to 0.135 nm after 30 revolutions.

Increasing the number of revolutions in the HPT process led to an increase in the microhardness over the whole volume of the disks up to the application of 10 revolutions, at which point the hardness at the center and peripheral locations was near 63 HV_0.01 and 73 HV_0.01, respectively. With the increasing number of revolutions, only peripheral locations showed an increase in hardness, reaching values in the range of 87–92 HV_0.01 after 30 revolutions.

The overall density of the HPT disks reached practically the rule-of-mixtures value for the Mg-5Ni-2Nb₂O₅ powder mixture after only one revolution. With the application of five revolutions, the measured density values were between 2 and 4.5% higher than the expected value for a fully densified composite. This was explained by the existence of a certain amount of a magnesium oxide-hydroxide component present on the surface of the original Mg powders.