**A Comprehensive Review on Hydrogen Absorption Behaviour of Metal Alloys Prepared through Mechanical Alloying**

**Thabang Ronny Somo 1,2, Thabiso Carol Maponya 1, Moegamat Wafeeq Davids 2, Mpitloane Joseph Hato 1,\*, Mykhaylo Volodymyrovich Lototskyy <sup>2</sup> and Kwena Desmond Modibane 1,\***


Received: 21 March 2020; Accepted: 18 April 2020; Published: 26 April 2020

**Abstract:** Hydride-forming alloys are currently considered reliable and suitable hydrogen storage materials because of their relatively high volumetric densities, and reversible H2 absorption/desorption kinetics, with high storage capacity. Nonetheless, their practical use is obstructed by several factors, including deterioration and slow hydrogen absorption/desorption kinetics resulting from the surface chemical action of gas impurities. Lately, common strategies, such as spark plasma sintering, mechanical alloying, melt spinning, surface modification and alloying with other elements have been exploited, in order to overcome kinetic barriers. Through these techniques, improvements in hydriding kinetics has been achieved, however, it is still far from that required in practical application. In this review, we provide a critical overview on the effect of mechanical alloying of various metal hydrides (MHs), ranging from binary hydrides (CaH2, MgH2, etc) to ternary hydrides (examples being Ti-Mn-N and Ca-La-Mg-based systems), that are used in solid-state hydrogen storage, while we also deliver comparative study on how the aforementioned alloy preparation techniques affect H2 absorption/desorption kinetics of different MHs. Comparisons have been made on the resultant material phases attained by mechanical alloying with those of melt spinning and spark plasma sintering techniques. The reaction mechanism, surface modification techniques and hydrogen storage properties of these various MHs were discussed in detail. We also discussed the remaining challenges and proposed some suggestions to the emerging research of MHs. Based on the findings obtained in this review, the combination of two or more compatible techniques, e.g., synthesis of metal alloy materials through mechanical alloying followed by surface modification (metal deposition, metal-metal co-deposition or fluorination), may provide better hydriding kinetics.

**Keywords:** metal hydrides; hydrogen storage; hydriding kinetics; surface modification; mechanical alloying

## **1. Introduction**

The physical storage of hydrogen through compressed H2 tanks, liquid H2 tanks and cryo-compressed tanks has been condemned, due to several limitations, including the high pressure required for compression of the hydrogen gas and a large amount of energy for liquefaction of hydrogen [1]. A convenient way to store hydrogen is through the absorption of hydrogen onto solid nanostructured materials such as metal alloys, chemical sorbents (metal organic frameworks (MOFs)) and carbon-based materials) or chemical hydrides [2]. Out of the above-mentioned chemical storage materials, metal hydrides offer a safe, volume-efficient technology aimed at hydrogen storage for reversible on-board applications due to their large hydrogen uptake capacity (7.6 wt. % by Mg-type alloys) [3,4]. There are excellent reviews, regarding MHs, prepared by equilibrium procedure for hydrogen storage application that have been published [4–6]. However, there are not a lot of reviews on the MHs prepared by non-equilibrium procedure.

It is well-documented that metallic hydrides are chemically formed from the reaction of hydrogen gas with metal or alloy according to Equation (1) [1]:

$$\text{Metal (M)} + \text{H}\_2\text{ (g)} \rightarrow \text{metal hydroxide (M-H)}.\tag{1}$$

The formation of metal hydride undergoes six distinct states of hydrogen absorption in metals: Hydrogen to metal; physisorption, which involves physical attachment of hydrogen to a metal through van der Waals interaction; chemisorption, the chemical attachment of hydrogen to metal by forming a chemical bond; subsurface hydrogen; solid solution (α-phase) and hydride (β-phase) [7]. The formation of α- and β-phases may be described by Equations (2) and (3), respectively [8]. An α-phase is formed when small amount of hydrogen occupies the interstitial sites of the host metal M, producing an H-poor solid solution, while β-phase involves saturation of the solid solution to generate a second phase:

$$\text{M} + \frac{1}{2}y\text{H}\_2 \quad \Leftrightarrow \text{ MH}\_Y \tag{2}$$

$$\text{MH}\_{\text{Y}} + \frac{1}{2} \left( \mathbf{x} - \mathbf{y} \right) \text{H}\_{2} \quad \leftrightarrow \text{ MH}\_{\text{x}}.\tag{3}$$

Where M represents metal and H represents hydrogen atoms. During the formation of β-phase, the equilibrium pressure is fixed at any given temperature. According to Gibbs' phase rule (*F* = *C–P* + 2, in which *F* is degrees of freedom, *C* represent the components of a reaction and *P* is the number of phases), the number of degrees of freedom, *F*, is one during the interaction of metal (M) and hydrogen (H) atoms since three phases P (two solid and one gaseous) and two components *C* (H and M) are available [8]. The constant pressure level is referred to as the plateau pressure. During the formation, the metal atoms stretch apart to accommodate the hydrogen atoms and the physical structure of the metal atoms may also change to form a metal hydride. For example, Table 1 shows different types of metal alloys ranging from AB to AB5, demonstrating that hydrogenation of metal is accompanied by increase in lattice parameters of the parent alloy [9–13]. The A2B alloys were reported to possess higher hydrogen storage capacity of 3.75 and 3.04 wt. % for P6222, and Fddd phase, respectively [11]. On the other hand, AB5 showed to have low absorption capacity of 1.43 wt. % [10]. Moreover, peaks from X-ray diffractogram may broaden and shift towards smaller 2θ values as a result of larger lattice parameters [9].

Conversely, metal hydrides release H2 gas when heat is applied on the hydride. This process is called desorption. Hydrogen atoms (H) move from the metal to the surface of the material, combine into hydrogen molecules (H2) and flow away as hydrogen gas. The metal atoms contract to form its original metal crystal structure [14]. The metal alloys in Table 1 can easily be synthesized by a well-known equilibrium procedure called arc or induction melting [15,16]. However, alloys prepared through this method suffer from poor hydrogenation performances and poisoning intolerance [17]. Recent findings reported that a convenient way to overcome some of these challenges is to prepare metal alloys using mechanical alloying (MA) through high-energy ball milling technique [15,18]. In this review, we provide the recent developments on hydrogenation performances of metal alloys prepared by mechanical alloying technique, comparison of MA with other non-equilibrium techniques. Moreover, we also consider the effect of combining MA with some surface modification technologies. We anticipate that this work can provide an inspiring perspective for further research on MHs with good hydrogen storage properties.


**Table 1.** Lattice parameter, a, of different types of alloys before and after hydrogenation and their hydrogen storage capacities.

#### **2. Non-Equilibrium Preparation Techniques**

Mechanical, kinetic and thermodynamic performances related to the formation of metal hydride are affected by structure and phases of metal or metal alloy of interest. It is currently well-known that the structure and phase of alloy materials can be controlled by synthesizing them under non-equilibrium conditions [19]. There are several processes/techniques, which are in commercial use for the synthesis of alloy materials under such conditions. Amongst such techniques, much attention has been given to mechanical alloying, melt spinning and spark plasma sintering techniques [20]. According to Scudino et al. [21], the general underlying idea in all these techniques is to synthesize alloy materials in a non-equilibrium state by "energizing and quenching". The energizing process involves the use of external dynamical force, such as melting; evaporation; irradiation; application of pressure; or storing of mechanical energy by plastic deformation on solid-state materials to put them onto a non-equilibrium state [22]. During this process, the solid-state materials change in phase to gas or liquid. In "quenching", a subsequent step after "energization", the material of interest is extinguished into a frozen state, and it is at this point where the desired structure or phase can be easily controlled [23,24]. The alloys synthesized in this sequence possess particular structural characteristics that are difficult or rare to attain by equilibrium preparation methods: Nano-crystalline grain size with a high density of grain boundaries and even lacking of long-range order (amorphous state). These structures alloy quick hydrogen uptake and excellent cycle-life behaviour [25]. Among all preparation methods that follow "energizing and quenching" mechanism, MA has been given much attention due to its user-friendly, comparatively less costly machinery and applicability to various metal alloys [19]. In this regard, mechanical alloying experimental technique is discussed with the point of interest being its effect on hydrogenation performances of metal alloy materials.

**Figure 1.** Process flowsheet and the microstructures developed during double mechanical alloying of an Al–5wt. %Fe–4wt. %Mn powder mixture, reproduced from [19], with permission from Elsevier, 2001.

### **3. Mechanical Alloying**

#### *3.1. Background*

Mechanical alloying, also known as ball milling, is a solid-state powder processing technique involving repeated welding, fracturing, and rewelding of powder particles in a high-energy ball mill [19]. Repetitions of welding, as well as fracturing of particles, generate a lamellar structure within the powders. MA method is usually applied when synthesizing highly immiscible phases in a non-equilibrium state [26]. The immiscible phases that are prepared, include super-saturated solid solutions, non-equilibrium alloys that are usually amorphous in nature, metastable crystalline and quasicrystalline phases. The procedure of MA begins with blending of the powders in the correct extent and stacking the powder blend into the mill along with the grinding medium. This blend is then milled for the ideal time allotment until a consistent state is achieved when the arrangement of each powder particle is equivalent to the extent of the elements in the initial powder blend. The milled powder is then united into a mass shape and exposed to heat, in order to get the ideal phase and structure. Thus, the vital components of MA procedure are the raw materials, mill and procedure parameters. This procedure was described by Shang et al. [27]. Ares et al. [25] observed that diffraction peaks from the X-ray diffractogram (XRD) of Mm (Ni, Mn, Al, Co)5.2 became broader with milling time. An example of MA process is depicted in Figure 1 (synthesis of Al–5wt. %Fe–4wt. %Mn blend), with scanning electron microscopic (SEM) images provided for all the steps. The double mechanical milling, as observed in the figure is necessary to refine the intermetallics and decrease the particle size of the blend [19]. MA has also been utilized to synthesize AB (FeTi), A2B (Mg2Ni) and AB5 (binary LaNi5) alloys [28]. In most cases, the rate of hydrogenation has been greatly enhanced when compared with that of materials prepared through conventional equilibrium methods, but due to the amorphous nature of milled alloys, a loss of hydrogen content was observed [25]. MA provides various advantages, including smaller restrictions with respect to composition and works much better when applied to A2B alloys [29]. Moreover, through the use of MA method, nano-crystalline alloys are obtained, starting either with a pre-prepared alloy or with a mixture of pure elemental metals [28]. Nano-crystalline alloys that are prepared through MA technique possess better activation compared to their polycrystalline counterparts [28]. For example, Polycrystalline FeTi is activated by heating up to temperatures above 673 K for one to two hours followed by annealing at 7 bar pressure and its hydrogenation requires a high pressure of 40–70 bar [30]. However, with a mechanically-alloyed nano-crystalline FeTi alloy, a single vacuum heating at 673 K for 30 min is sufficient to obtain full hydrogen absorption cycles [30].

## *3.2. Mg-Based Alloys*

### 3.2.1. Structural Characteristics of Mechanically Milled Mg-Based Alloys

The crystal structure of a metallic Mg is hexagonal with lattice parameters a = 3.2094 Å and c = 5.2108 Å and a space group of P63/mmc [31]. Formation of MgH2 hydride consists of two steps, i.e., preparation of Mg nanoparticles, followed by hydrogen absorption to form the metal hydride. A traditional method for preparing Mg nanoparticles is gas-phase condensation method, which is much better discussed in [32]. As depicted by Figure 2a, hydrogenation of Mg nanoparticles tempers with XRD diffraction pattern of the parent nanoparticles [33]. Diffraction peaks at 2θ ≈ 48◦, 34◦ and 32◦ disappear, while several new peaks develop, forming tetragonal magnesium hydride (β-MgH2) phase [33]. To illustrate the difference between traditionally synthesized MgH2 hydride and those synthesized through mechanical/ball milling, XRD diffraction patterns of MgH2 and as-milled hydrided Mg-Al alloy are compared in Figure 2a,b, respectively. For as-milled Mg-Al alloy, Li et al. [31] reported four stable phases: the fcc solid solution of magnesium in aluminum, the hcp solid solution of aluminum in magnesium, the β-phase (Mg2Al3) and the γ-phase (Mg17Al12). Such rewarding characteristic gives rise to excellent hydriding behavior of as-milled Mg-based alloys.

In general, the morphology and particle size distribution are also important structural parameters, carrying a huge responsibility for surface physisorption of molecular hydrogen (H2), surface dissociative chemisorption, surface absorption, hydrogen diffusion and phase transformation [34]. These two parameters are well-interpreted using scanning electron microscopic (SEM) images. For Mg-based alloys, mechanically milled alloys show faster hydrogen absorption kinetics and larger uptake capacity, compared to their unmilled counterparts and one of the effective factors contributing to this, is the

reduction of particle sizes of the parent materials during mechanical alloying [35]. According to the literature, small particle sizes allow easy mobility of molecular hydrogen and quick dissociation into atomic hydrogen on the surface of a material [33]. As shown in Figure 3, all the five SEM images of milled Mg-based alloys constitute of smaller particles than that of traditionally sysnthesized MgH2, [33]. Therefore, their hydrogen uptake performance is expected to be superior to that of MgH2. This is further discussed in the next section.

**Figure 2.** XRD diffraction patterns of (**a**) nanocrystalline Mg and MgH2, and (**b**) Mg-Al alloy in the hydrogenated state (top) and dehydrogenated state (bottom). Figures reproduced from [33], and [31], respectively, [31] with permission from Elsevier, 2008; [33] with permission from Elsevier, 2007.

**Figure 3.** The SEM images of the (**a**) MgH2, (**b**) as-milled MgH2, (**c**) as-milled MgH2 + 10wt. % CeCl3, (**d**) as-milled Mg–10 FeTiO3, (**e**) as-milled Mg:Al (90:10) and (**f**) as-milled MgH–5 at. % V. Figures (**a–c**) were reprinted with permission from [36], with permission from Elsevier, 2016 (**d**) reproduced from [37], with permission from Elsevier, 2015; (**e**) reproduced from [38], with permission from Elsevier, 2000 and (**f**) reproduced from [39], with permission from Elsevier, 1999.

### 3.2.2. Hydrogen Absorption Behaviour of Mechanically Alloyed Mg-Based Alloys

Mg-based alloys are reported to possess energy density of 9 MJ/kg and the highest maximum hydrogen storage capacity (7.6 and 3.6 wt. % for MgH2, and Mg2NiH4, respectively) compared to all hydride-forming materials [40,41]. The hydride stability of this alloy is very high, and therefore, a high dissociation temperature of around 561 K is required to transform MgH2 hydride into Mg and H2 gas [42]. Furthermore, the hydriding kinetics of MgH2 hydride is relatively slow because the dissociation of molecular hydrogen into hydrogen atoms is difficult on the surface of Mg metal [43]. Several reports are found in the literature concerning synthesis and hydrogen properties of Mg-based alloys synthesized by mechanical alloying [3,44–46]. Another interesting study is Andreasen [31]. The authors'work was based on the comparison of hydrogen kinetics, presented in Figure 4a, of un-milled Mg and those of ball milled Mg-Al compounds with varying Al content [31]. The kinetics increased gradually with an increase in Al content due to easy dissociation of hydrogen molecules on the surface of Al, as a result of negligible activation barrier for hydrogen dissociation for this metal. It is, therefore, understood that hydrogen molecules split into atomic hydrogen on the surface of Al before diffusing and attaching on the active sites of Mg, which has higher hydrogen sticking ability compared to Al [31]. Lototskyy et al. [37] reported the effect of synthesizing different Mg-based composites varying the amounts of multi-walled carbon nanotubes, activated carbon, FeTiO3 and graphite using mechanical alloying at a rotation speed of 500 rpm for a period of 120 min on hydrogenation behaviour of Mg, Figure 4b. The authors discovered that hydrogenation behaviour (both the uptake capacity and kinetics) of Mg mechanically alloyed with carbon-based materials are exceptional. Ref. [46] reveals that the better hydrogenation behaviour possessed by milled alloys can be attributed to their combination of tetragonal (a = 4.52 Å, c = 3.02 Å) and orthorhombic (a = 4.48 Å, b = 5.40 Å, c = 4.90 Å) phases as major phases, since these two phases are associated with small particle size and a presence of strain in the particles. Similar trends of hydrogen capacity and kinetics were observed by Lototskyy et al. [47] and Williams et al. [48] when they assessed Mg-10(FeV)-5MWCNT and Mg-5Pd composites through mechanical alloying, respectively. Tables 2 and 3 represent a compiled data of literature based on hydrogenation behaviour of Mg-based alloy materials synthesized through ball milling and mechanical alloying.


**Table 2.** Thermodynamic and kinetic properties of ball-milled MgH2 doped with catalytic additives.

ΔHdes = Enthalpy of desorption\*SWNT = Single-walled carbon nanotube.

**Figure 4.** (**a**) Hydrogenation kinetics of pure un-milled Mg compared with Mg-Al compounds with varying Al content ball milled for 20 h. Hydrogenation performed at T = 400 ◦C and a hydrogen pressure of 38 bar. Data extracted from ref. [31], with permission from Elsevier, 2008. (**b**) Hydrogenation kinetics of Mg and its carbon-based composites (15 bar and H2/250 ◦C), reproduced from [37], with permission from Elsevier, 2015.

3.2.3. Ball-Milled MgH2 Doped with Catalytic Additives

In the previous section, much attention is given to mechanically milled Mg-based alloys where a pure milled/un-milled Mg metal is admixed with other several materials, such as FeTiO3 before their interaction with hydrogen is subsequently explored. Here, we discuss the influence of catalytic effect of doping ball milled MgH2 with different catalystic additives. Previous experiments indicate clearly that admixing, deposition or doping of catalysts and additives could greatly enhance thermodynamic property and de/rehydrogenation kinetic parameters of MgH2. In these studies, metal [49], metal oxide [50–52], metal halides [36,53], metal hydride [49], carbon materials [54] and nanosized alloys [55,56] were used as the catalysts and additives. Table 2 summarizes thermodynamic and kinetic effects of these catalysts. The materials in the table were prepared through ball milling technique.

We can conclude from the table that doped MgH2 materials, prepared through mechanical milling, enjoy superior hydrogenation kinetics, compared to undoped materials. However, this particular excellent improvement significantly reduces the maximum absorption capacity of the system. For metal halide family, particularly MgH2+10wt. %CeCl3, the loss of capacity can be related to in situ formation of CeMg alloy and MgCl2, which seem to be good catalysts for faster absorption but poor gas absorbers [53]. On the other hand, drastic decrease of absorption capacity revealed by MgH2+SWNT composite is suggested to come from structure destruction of the SWNT during milling [54]. Table 2 also shows that composites which resulted in major decrease in hydrogen capacity are those with a dopant of 10 wt. % or more. We therefore suggest preparation of MgH2 composites with dopants that are less than a concentration of 10 wt. %. Moreover, milling period should also be taken into consideration when these materials are formed, more especially when carbon materials are in abundance as they seem to be sensitive to milling conditions. The difference between enthalpy of desorption (ΔHdes) of the materials is negligible and as thus thermodynamic property of MgH2 remain unchanged upon doping.

## *3.3. AB5-Type Alloys*

## 3.3.1. Structural Characteristics of Mechanically Milled AB5-Type Alloys

Another interesting class of metal alloys is AB5. The A site in AB5 represents one or more strongly hydride-forming elements while B site represents one or more weakly hydride-forming elements, and usually helps to dissociate the H2 molecules on the surface of alloy material [58]. The A site is usually occupied by La, Ca or rare earth element while Ni, Cu, Co, Pt or Fe fill the B site [59]. AB5 alloys are known to have a CuCa5-type hexagonal crystal structure which normally belongs to space group P6/mmm (#191) [60,61]. Singh et al. [62] evaluated the effect of milling MmNi4:6Fe0:4 alloy, where Mm refers to mischmetal, varying the milling period from 10, 20 and 30 min, at a milling speed of 200 rpm. As shown by Figure 5, mechanical milling of this alloy results in slight shift of diffraction peaks towards lower 2θ values, reflecting increase in lattice parameters, and peak width broadening. Peak width broadening is a reflection of decrease in particle size. In general, the lattice structure of MmNi5-type alloy consists of two stacked planes at ᑰ = 0 and ᑰ = <sup>1</sup> <sup>2</sup> , and the first plane contains 1a and 2c, while the latter plane contains 3g crystallographic sites. Both Mm and Ni atoms occupy the 1a and 2c sites, while the 3g sites are occupied by Ni atoms only. Therefore in the presence of Fe to form MmNi4:6Fe0:4 alloy, Fe atoms replace Ni on the 3g sites [63]. Electronic configuration of Fe and that of Ni are very close (3d64s2 for Fe and 3d84s2 for Ni) and as a result covalent character, which is stronger than ionic character, is expected in the Ni/Fe-bearing planes. Thus, during mechanical milling of an alloy material with such character, strained platelet shaped particles are expected to form [64]. Such particles allow smooth movement of hydrogen during sorption better than fine powdered particles. This statement is supported by Table 3 as it is observed from the table that materials that were milled for longer period of time are associated with lower hydrogen uptake capacity compared to their counterparts that were milled for shorter period. This is due to the mechanical energy that is being pumped into the alloy matrix during milling, which breaks the particles, while forming strain in the matrix, and thus, for longer milling periods fractured and much finer powders, which are not recommended may result [63,64].

**Figure 5.** X-ray difractograms of the unmilled; (**a**) and mechanically pulverized; (**b**) MmNi4:6Fe0:4 alloys and the inset picture focus on broadening of full width at half maximum (FWHM). reproduced from [61], with permission from Elsevier, 2004.



To be more accurate, not only particle size gives precise measurement of hydrogen uptake capacity by a material, but it also provides the nature of agglomeration, as well as homogeneity of the particles [62]. Short milling process yields non-homogeneous microstructural (wide distribution of nano-crystallite sizes and amorphous phase) and morphological alloy powders [25]. On the other hand, Parker et al. [35] recommends defects, fractures and porosity as faster absorption kinetics agents. However, it is very challenging to fully release hydrogen from a porous and fractured material. Figure 6 depicts SEM images of un-milled MmNi5 alloy and those of MmNi5 milled with different transition metals at concentration of 2.0 wt. %. for un-milled alloy, its structure lacks strain and consists of very large particles, and thus, hydrogen diffusion is difficult for such material [67].

In contrast to MmNi4:6Fe0:4 alloy, neither covalent nor ionic bonding is expected between the bulk alloy MmNi5 and the added transition metals (Co, Ni, Mn and Fe), since there was no substitution or melting during mechanical milling, thus, a physical attachment was involved [67]. As a result, the extent to which the particles of each alloy crush and become reduced in size during milling depends largely on ductility and hardness of the transition metal used. According to Mohs scale of metal hardness, Mn has Mohs hardness of 6, followed by Co with hardness of 5 and the last two are Fe and Ni both with hardness of 4.5 and 4, respectively [68]. Therefore, during ball milling, it is difficult to fracture and form finer powders with Mn particles than with Ni particles.

**Figure 6.** SEM micrographs of (**a**) bulk MmNi5 alloy; (**b**) ball-milled MmNi5 alloy; (**c**) MmNi5 alloy ball-milled with Co; (**d**) Ni; (**e**) Mn and (**f**) Fe transition metals at concentration of 2.0 wt. %, obtained from [60].

## 3.3.2. Hydrogen Absorption Behaviour of Mechanically Alloyed AB5-Type Alloys

These metal hydride alloys are typified by LaNi5 and they can be easily activated for hydrogen absorption [69]. A reaction between these alloys and hydrogen at a room temperature and pressure above equilibrium plateau pressure results in a ternary metal hydride [69]. Of 37 voids available within LaNi5 lattice structure, only 6 voids are occupied by hydrogen atoms to form LaNi5H6 ternary hydride [70]. It is notable that the hysteresis effect in AB5-type alloy is relatively small as compared to other low temperature systems with a large and distinct miscibility gap [59]. Mechanical alloying has also been employed to synthesize AB5 alloys [25,62,67,71]. For example, MmNi5 material mechanically alloyed at 200 rpm for milling time of 180 min exhibits faster hydrogenation kinetics as compared to arc-melted MmNi5 material [67]. Furthermore, Figure 7a,b present the absorption and desorption kinetic curves of MmNi5 alloy ball-milled with transition metals at concentration of 2.0 wt. %, respectively, showing reversible hydrogen absorption and desorption between 1.0 to 1.6 wt. % [67]. Singh and co-workers [62] observed that longer milling of MnNi4.6Fe0.4 alloy material improved hydriding kinetics, but this elevated kinetics behaviour comes at a cost of maximum uptake capacity as it deteriorated with increasing milling time. On the same note, an increase in rotation speed from 200 rpm to 300 rpm at constant milling time resulted in a reduction of hydrogen capacity. Similar trend was observed by Ares and co-workers [25] and they summarized that the trend is a result of extra hydrogen-trapping sites in amorphous domains, exhibited by short-milled alloys. For LaNi5 alloy, it is reported that ball-milling convert a part of LaNi5 to a non-absorbing state and as a result poor hydrogen uptake capacity is observed [71]. It was seen that there is quite a number of published work found in the literature for mechanical alloying of AB5-type alloys [62,67]. However, understanding the general extent to which mechanical alloying enhance the ability of these type of alloys to absorb hydrogen, as compared to their traditionally or arc-melted counterparts is still a gap. Firstly, to our knowledge no proper measurements of the hydrogen sticking or diffusion probabilities on either un-milled or milled AB5-type particles have been documented. The second issue is that mechanical alloying tempers with several physical properties that are vital for hydrogen sorption performance; these include expansion of the unit cell volume, strain and homogeneity of particles, particle size distribution, distance between adjacent voids and several more.

**Figure 7.** (**a**) Absorption kinetic curves of MmNi5 alloy ball-milled with transition metals at concentration of 2.0 wt. %, and (**b**) Desorption kinetic curves of MmNi5 alloy ball-milled with transition metals at concentration of 2.0 wt. %. Reprinted from [67].

## *3.4. AB2-Type Alloys*

## 3.4.1. Structural Characteristics of Mechanically Milled AB2-Type Alloys

In the AB2 Laves phase alloy, metal A is usually Ti or Zr, and metal B is another transition metal including V, Cr or Mn. These alloys crystallize in Lave phase structures [72,73]. Three types of laves phases; namely: hexagonal C14 phase (MgZn2), hexagonal C36 phase (MgNi2) and cubic C15 phase (MgCu2) represent AB2- type alloys [72]. In comparison, hydrogenation performance of hexagonal C36 phase has been reported to be poor [73], and it is not reviewed here. C15 and C14 phases (corresponding to atomic ratio of RA/RB = 1.225) have shown better performance as hydrogen absorbents [73]. Based on their structural stability, which is dependent on the atomic ratio, electronegativity and valence of the elements used [74,75], hexagonal C14 ZrCr2 Laves phases are more stable compared to cubic C15 [76].

Figure 8 represents example of powder diffraction patterns for C14 Laves phase type alloys. As seen with Mg-based and AB5-type alloys discussed earlier in this review, longer milling periods results in reduction of crystallite size and particle size. Furthermore, several other factors such as disappearance of peaks (between 2θ ≈ 48 and 100 in Figure 8), as well as peak broadening occur. To some extent, intense change of these factors may result in deterioration of hydrogen performance.

**Figure 8.** Powder diffraction patterns of alloy Ti0.5Zr0.5MnCr before and after 5, 15, 30 and 60 min of ball milling, reproduced from [65], with permission from Elsevier, 2019.

3.4.2. Hydrogen Absorption Behaviour of Mechanically Alloyed AB2-Type Alloys

The reported hydrogen storage of AB2-type alloys is nearly 2 wt. %, with low hysteresis [75]. But binary AB2-type alloys prepared by equilibrium techniques suffer from poor surface activity and poisoning intolerance [77]. As a result, efforts through synthesizing AB2 alloys utilizing non-equilibrium techniques such as mechanical alloying have been exploited [66]. For example, Ti0.5Zr0.5MnCr material prepared through mechanical alloying for a milling period of 5 min was reported to exhibit larger

hydrogen absorption capacity compared to as-cast Ti0.5Zr0.5MnCr and materials milled much longer (Table 3 and Figure 9) [65].

**Figure 9.** Hydrogenation kinetics of Ti0.5Zr0.5MnCr alloys before and after ball milling for 5, 15, 30 and 60 min, reproduced from [58], with permission from Elsevier, 2019.

However, those that are milled longer than 5 min exhibit faster hydriding kinetics and poor hydrogen absorption capacity compared to as-cast alloy [65]. Based on the literature data contained in this document, we understand that fractured fine, smaller particle sizes and amorphous nature, which are obtained through milling time above 25 min favour faster hydrogenation kinetics. Whereas combination of large to moderate particles and nano-crystalline structure, obtained during short milling time of less than 10 min favours large hydrogen absorption capacity.

## *3.5. Comparison between Hydriding Kinetics of Materials Prepared by MA with Other Techniques*

Apart from mechanical alloying, melt spinning and spark plasma sintering techniques are also prevailing synthesis methods and thus we opted to make some comparisons between these three methods. Melt-spinning technique is a useful method that is used to change structures of as-cast alloys to amorphous structure. This method improves the amorphous nature of materials by reducing their grain size [44]. The only problem related to this method is the poor cycle stability of the melt-spun alloys, due to the disappearing of the metastable structures generated by melt-spinning [78]. Another attracting technique is spark plasma sintering (SPS), due to its ability to easily synthesize metal powders, with different melting temperatures within just 5–10 min [79]. This technique involves generation of spark plasma through direct pulse current and working concurrently with a uniaxial pressure. Sintering takes place at low temperatures in a small period of time and this makes the alloys to have low porosity [80]. SPS offers numerous favourable advantages over customary sintering procedures such as atmospheric furnace. For instance, the heating rate around 1273 K per min can be accomplished [79]. Further focal points are the lower sintering temperature, short holding time, no need of pre-compaction and shorter sintering time results in the likelihood to sinter nanometric powder to near theoretical density [81]. Although this technique produces alloys with high purity surfaces, it is known that the hydriding kinetics of composites, prepared by SPS, is greatly improved only when hydrogen absorption is conducted at high temperatures (573 K) [79].

Huang et al. [82] reported that alloys with amorphous surfaces absorb hydrogen faster compared to their crystalline counterparts. Schwarz and Johnson [83] recommended that two conditions should be fulfilled for an amorphous phase to form from a mixed elemental powder blend. The first condition is that the metals involved must have a vast negative heat of blending in the liquid state. According to a study conducted by Termsuksawad et al. [84], metals that have large negative heat of mixing exhibited a complete amorphous phase, while those with smaller negative heat displayed partial amorphous nature. Second, the condition is that diffusion rates of such metal systems must differ greatly. Metal alloy materials with nano-crystalline and/or amorphous phases are easily obtainable through several preparation techniques, including mechanical milling, spark plasma sintering and melt spinning [85,86]. There are numerous published papers supporting the above statement, and we combined some of those findings in Table 4. All the as-milled, as-spun and sintered alloy materials in Table 4 are reported to exhibit amorphous phases, and they are compared with their as-cast crystalline counterparts. According to the table, mechanical alloying of as-cast materials usually has a negative impact on the maximum hydrogen absorption capacity as all materials showed drastic decrease in their respective hydrogen content, except for Ti0.5Zr0.5MnCr alloy. Similar to other authors, Khajavi et al. [65] rationalized a loss in hydrogen capacity by deducing that milling process decreases the crystallite size, resulting in peak broadening. Nonetheless, milling significantly enhanced the absorption kinetics, and this is indicated by exceptional rate constants associated with all the as-milled counterparts in Table 4. Almost all the rate constants presented in the table were estimated from kinetic curves using chemisorption model [65] as shown by Equation (4),

$$
\alpha = kt \tag{4}
$$

where α is the reacted fraction, *t* is hydrogenation time and *k* is rate constant. Metastable alloys (as amorphous, nano-crystalline, quasicrystalline, polycrystalline) can be prepared through melt spinning technique, since it allows quick quenching of melted metals. Most as-spun materials reported in this review have amorphous (examples being Mg11Y2Ni2 and Mg3LaNi0.1) and nano-crystalline phases (examples being Mg-10Ni-2Mn and Mg2Ni0.9Co0.1). On a contrary to mechanical alloying, melt spinning technique enhances both the hydrogen storage capacity and the kinetics (Table 4). The speedy hydrogenation kinetics for as-spun materials can be related to fine particle size and the crystal defects shaped as a result of rapid solidification, which facilitates swift nucleation and diffusion of hydrogen [23]. The enormous quantity of interfaces and grain boundaries created within the as-spun alloy offers easy access for hydrogen diffusion and facilitate easy absorption of hydrogen [23]. Another interesting technique featured in this review is spark plasma sintering. In the same way as MA and melt spinning, SPS-synthesized alloy materials exhibit an amorphous/nanocrystallite phase [87,88]. The absorption kinetics and capacity of sintered materials are affected by the amount (percentage), the type of material, and sintering conditions (temperature and pressure) used for SPS process. For example, hydrogen content of metallic Mg was reduced extremely from 7.26 wt. % to 2.52 wt. % upon blending with 77% V77.8Zr7.4Ti7.4Ni7.4 through SPS while hydrogen content of V35(Ti,Cr)65 sintered with ZrMn2 remained almost the same [87,88]. Interestingly, the rate constants of sintered alloy materials in Table 4 are better than those of as-cast materials. This is a clear indication that the SPS method has a positive effect on hydrogen absorption kinetics. Thus, it can be seen that all these three techniques are good candidates for improvement of hydrogen absorption kinetics, but alloys prepared using SPS technique in most cases have low hydrogen capacity. Hence, we take a further comparison between the prevailing techniques–MA and melt spinning.


**Table 4.** Comparison of hydrogenation kinetics and capacity of as-cast alloys and those prepared by MA, SPS and melt spinning.

Zhang et al [44] and Wu et al [92] made some efforts to prepare and compare structures and hydrogenation behaviour of as-milled and as-spun alloy materials. Their findings are summarized in Table 5. From the table, it can be seen that the hydrogen absorption capacity of Mg-10Ni-2Mm alloy is favoured by melt spinning technique as larger amount of hydrogen was stored by the material prepared through this technique than that prepared using mechanical alloying. However, YMg11Ni in Table 5 showed a different trend. The rate constants of materials prepared by mechanical alloying are better than those of materials synthesized by melt spinning technique. This indicates that mechanical alloying plays a significant influence on increasing the hydrogenation kinetics of the alloys as compared to melt spinning.


**Table 5.** Comparison between kinetics and hydrogen content of as-milled and as-spun techniques.

Hydrogenation of metal alloys is usually governed by three steps, namely (i) dissociation of hydrogen molecules into atomic hydrogen, (ii) the penetration of atomic hydrogen through grain boundaries, and (iii) the binding of atomic hydrogen on the metal, transforming into metal hydride [44]. The rate-determining step of the as-milled alloy is most likely ascribed to the ball milling rendering the second step, i.e., the penetration of atomic hydrogen through grain boundaries. The milling of as-cast alloys during mechanical alloying results in reduction of particle size and increment of high surface-to-volume ratio of alloy particles [93]. Due to these two changes, the distance for atomic hydrogen penetrating through the material is extremely reduced, which facilitates absorption greatly.

### *3.6. Surface Modification of Ball Milled Metal Alloys*

Better and improved hydrogenation kinetics are achieved by nano-crystalline and/or amorphous metal alloys synthesized by mechanical alloying, but such materials still encounter several limitations, including impurity interactions, which include poisoning of the metal by impure gases such as carbon monoxide, sulphur and carbon dioxide; disproportionation during absorption/desorption cycling; retardation; innocuous and difficulty of initial activation [94]. These shortcomings result in rapid loss of hydrogen capacity with cycling, caused by impurities strongly or irreversibly absorbed on the surface active sites, and loss in absorption kinetics due to surface blanketing [95]. To overcome the above-mentioned problem, nano-structuring of ball milled materials and surface modification technology can be employed before hydrogen sorption can take place [96]. Surface modification involves the formation of a protective layer on the surface of metal alloy, thus preventing oxide-containing film to form. The most common surface modification technologies include fluorination, potassium borohydride, Ni-weak acid, ion implantation, hydrochloric (HCl) acid, platinum group metals (PGMs) decomposition and Y3O2 deposition. Yeung et al. [97] reported that the surface modification techniques PGMs deposition is the most attractive technology due to its ability to simultaneously prevent deactivation by guarding the catalytically active sites on the surface of bulk material against and enhance the hydriding kinetics during hydrogen absorption [97–100]. Moreover, this technique does not only enhance the hydriding rate, but also the absorption capacity of the alloys [99]. In particular, palladium is capable of catalyzing hydrogen sorption reactions, promoting easy activation, and improving poisoning tolerance. It has high hydrogen affinity and it is impermeable to larger molecules such as CO2 and sulphur [82]. Figure 10a–c show different mechanisms of hydrogen dissociation on the surface of Pd-treated AB5-type alloys [100]. During the hydrogen spillover mechanism hydrogen molecules are dissociated into hydrogen atoms on the surface of palladium, which then spillover to the active sites of the bulk alloy [98]. Pd thin film on the surface of AB5 alloy dissociates hydrogen molecules and allows the resulting hydrogen atoms to pass through the film as they diffuse into the bulk alloy. However, this technique is costly as huge amount of Pd is utilized [99].

**Figure 10.** Different mechanisms of hydrogen dissociation on the surface of Pd-treated alloy material; (**a**) hydrogen diffusion through Pd window, (**b**) hydrogen spillover with oxide layer being reduced by spilt hydrogen from palladium, and (**c**) hydrogen spillover with hydrogen atoms diffusing through the oxide layer, reproduced from [100], with permission from Elsevier, 2009.

Moreover, PGM-treated materials also exhibit the power to absorb hydrogen at low temperature, without the activation step prior to absorption, and enhanced kinetics of hydrogen sorption processes [101]. This method is extremely active towards dissociative hydrogen chemisorption and permeable for hydrogen atoms; enabling hydrogen to rapidly absorb into the bulk material while still maintaining the hydrogenation activity even after exposure to the impurities inhibiting hydrogen sorption [102]. The properties being improved include surface catalytic activity, specific

surface area layer, activation characteristics, as well as protective nature of the complexes against poisoning materials, such as sulphur, water vapour and carbon oxides [82].

Davids et al. [103] exploited hydrogenation behaviour of ball-milled TiFe alloy surface modified using a metal-organic chemical vapour deposition technique (MOCVD), by the thermal decomposition of palladium (II) acetylacetonate (Pd[acac]2), mixed with the powder of the parent alloy. The authors reported that alloy materials containing 0.5% Pd have better hydrogenation behaviour than materials containing 1% Pd. The same trend in deterioration of hydrogenation performance, with an increase in Pd content, was observed by Pasquini and co-workers [104] who varied Pd content (corresponding to 4, 8 and 13 wt. %) on the surface of Mg metal. The observations are due to the theory that large amount of metal deposits on the surface of TiFe alloy and Mg metal results in slow penetration of hydrogen over a given period [104], as a result of fracture and defect blockage of the parent alloy material, which are usually formed during ball milling. Doyle et al. [105] recommended that the total weight of PGMs during surface modification of metal hydride-forming alloy should be in trace amounts (≤0.1wt. %) to avoid losses in hydrogenation behaviour. Another interesting study in literature was conducted by Modibane et al. [106] in which the authors carried out ball milling of AB5-type materials out using a FRITSCH Pulverisette planetary mill and subsequently deposited Pd nanoparticles using autocatalytic Pd deposition technique. Upon ball milling of the parent alloy (Figure 11a), porous agglomerates of micro-size particles (Figure 11b,c) were obtained due to blending and grinding of parent particles. Porous particles are fancied by Parker et al. [35] for enhancement of absorption kinetics. Palladium nanoparticles, which showed discontinuous nature and near-spherical shape, partially cover the AB5 materials (Figure 11d–f) and are responsible for dissociation of hydrogen molecules [106].

Table 6 reveals some of reported literature on surface modified metal alloys synthesized through mechanical alloying. Surface modification techniques herein are based on PGMs deposition and co-deposition. All surface modified alloys exhibit better hydrogenation kinetics as compared to their as-milled counterparts. Variations arise when comparing the hydrogen capacities. For instance, Pt deposition on as-milled La0.9Pr0.05Nd0.05Al0.3Mn0.4Co0.65Ni3.5 alloy results in increase of capacity, while Ru deposition on the same alloy lowers the hydrogen capacity. This is attributed to poor or little hydrogen affinity associated with Ru.


**Table 6.** Hydrogenation properties of surface modified metal alloys prepared by mechanical alloying.

**Figure 11.** Low-((**a,b**), Hitachi X-650 EM) and high-resolution ((**c**–**f**), Zeiss Ultra) SEM images of raw AB5 alloy (**a**), the alloy ball milled without Pd black (**b**), and the alloy ball milled with Pd black after autocatalytic Pd deposition at different magnifications (**c**–**f**), reproduced from [106], with permission from Elsevier, 2018.

#### **4. Conclusions**

Hydrogenation kinetics and storage in solid state materials is a major concern in developing an up-and-coming hydrogen technology and economy. The present work reviews some challenges regarding the hydride-forming alloys, particularly their hydriding kinetics and capacity. In this regard, developments on hydrogenation performances of metal alloys, prepared by the mechanical alloying technique, were discussed. Moreover, this technique was compared with other preparation techniques, such as spark plasma sintering and melt spinning. The work is carried out in an attempt to facilitate suitable alloy preparation technique that can extensively improve the hydrogenation kinetics of different alloy materials. Although, much attention has been given to the techniques that are discussed in this work, with the aim of minimizing dehydrogenation temperature, enhancing the kinetics and poisoning-tolerance of different alloys, further research is needed. The collected literature data reveal encouraging improvements in terms of hydrogen storage kinetics through mechanical alloying. In the light of the achievements, through the combination of mechanical alloying and melt spinning techniques there is high potential in developing better metal hydride materials with high hydriding kinetics. Mechanically alloyed materials still suffer from cycling stability; they are prone to oxide layers formation, which hinders rapid absorption of hydrogen; and they also exhibit poisonous-intolerance. As a result, we further reviewed different surface modification techniques that can overcome these drawbacks. Surface modification of mechanically alloyed materials was overlooked at and it was seen that it promotes hydrogen storage kinetics by forming a protective layer on the surface of bulk alloy, thereby preventing the formation of oxide layer. Moreover, surface modification through deposition of PGMs on the surface of alloy materials stimulates poisonous tolerance of parent alloys.

**Author Contributions:** Conceptualization, K.D.M., and M.V.L.; funding acquisition, K.D.M. and M.J.H.; project administration, K.D.M.; Supervision, K.D.M., and M.V.L.; writing—original draft, T.R.S., T.C.M., M.W.D., K.D.M., M.J.H. and M.V.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by the National Research Foundation under Thuthuka program (UIDs. 117727 and 118113), University of Limpopo (Research Development Grants R202 and R232) and Sasol Inzalo Foundation, South Africa.

**Conflicts of Interest:** The authors declare no conflict of interest.

## **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article*

## **A Novel Non-Equiatomic (W35Ta35Mo15Nb15)95Ni5 Refractory High Entropy Alloy with High Density Fabricated by Powder Metallurgical Process**

## **Bohua Duan, Yingrui Yu, Xinli Liu \*, Dezhi Wang and Zhuangzhi Wu**

School of Materials Science and Engineering, Central South University, Changsha 410083, China; dbh72@163.com (B.D.); yyr021189@163.com (Y.Y.); dzwang@csu.edu.cn (D.W.); zwu2012@csu.edu.cn (Z.W.) **\*** Correspondence: liuxinli@csu.edu.cn

Received: 10 October 2020; Accepted: 25 October 2020; Published: 29 October 2020

**Abstract:** A non-equiatomic refractory high entropy alloy (RHEA), (W35Ta35Mo15Nb15)95Ni5 with high density of 14.55 g/cm<sup>3</sup> was fabricated by powder metallurgical process of mechanical alloying (MA) and spark plasma sintering (SPS). The mechanical alloying behavior of the metallic powders was studied systematically, and the microstructure and phase composition for both the powders and alloys were analyzed. Results show that the crystal consists of the primary solid solution and marginal oxide inclusion (Nb5.7Ni4Ta2.3O2). In addition, the maximum strength, yield strength and fracture strain are, 2562 MPa, 2128 MPa, 8.16%, respectively.

**Keywords:** refractory; high entropy alloy; mechanical alloying; phase transformation; mechanical properties

## **1. Introduction**

With the rapid development of advanced science and technology in the fields of armor piercing projectile, counter weight balance, radiation shields, shaped charge liner in ordnance industry and high-density gradient materials in dynamic high-pressure physics, the demand for materials with high density has boosted wide study on heavy metals and alloys [1]. Recently, a novel type of advanced refractory metallic alloys, refractory high entropy alloys, have been regarded as a potential candidate. High entropy alloy is a kind of alloy with high entropy of more than 1.61 R (when the alloys have 5 components). The different atoms with different bonding energy and crystal structure make the crystal lattice complex, and solute phases lead to severe lattice distortion in high entropy alloy (HEAs). These distortions are proposed to be more severe than in conventional alloys. The diffusion of atoms in HEAs mainly occurs via the vacancy diffusion mechanism. Due to the different melting points of the components, the element with high migration rate will preferentially diffuse to the vacancies. However, different atoms in HEAs have different bonding energies. Once atoms fill in vacancy, the energy of the system will decrease or increase. As a result, compared to the conventional alloys, the atomic diffusion and phase transformation processes in HEAs are relatively slow. This is called the sluggish diffusion effect in HEAs [2]. Refractory high entropy alloys (RHEAs) possess many excellent properties, such as high strength and hardness, and excellent creep resistance, due to its lattice distortion effect and sluggish diffusion effect [2]. Unlike conventional alloys with one or two major elements, RHEAs are generally made up of multiple principle refractory elements (≥5), such as Cr, W, Ta, Mo, Nb, Ti, V, Zr, Hf, and other elements of Al, Ni, Co, Si, C and N [3] in equimolar or near-equimolar proportions (5–35 at%). For the moment, the HEAs was recognized consisting of at least four major metallic elements in non-equimolar ratios (5–35 at%). The most common RHEAs include WTaMoNb, HfNbTaTiZr, NbTiVZr, AlNbTiV and CrMoVW, which have received extensive attention due to the easy way to tailor the phase stability and phase microstructure to enhance mechanical properties [4].

Up to now, various fabrication methods have been adopted to manufacture the HEAs, such as vacuum arc-melting and casting, sintering, magnetron sputtering and laser cladding. Among these methods, vacuum arc-melting and casting is generally applied to product bulk HEAs. However, the defects such as segregation and inhomogeneous dendritic microstructures generally exist in HEAs which are prepared by casting, and may be harmful to mechanical properties [5]. A few HEA systems were manufactured by mechanical alloying (MA) process [6]. HEA powders produced by MA process exhibit many advantages, such as more homogeneous morphologies and precise composition control. However, the milling time which achieves alloying is overwhelmingly long [6,7]. Some researches described appropriate addition of Ni can reduce reaction energy barrier and significantly accelerate the reaction among brittle phase during MA process [8]. Spark plasma sintering (SPS) technique is a new technology for preparing materials, which has the distinctive characteristics of fast heating speed, short sintering time, controllable structure, energy saving and environmental protection, and can be used to prepare metals materials, ceramic materials and composite materials, et al. The MA and SPS processes can exhibit some significant advantages, such as lower processing temperature, finer grain and more homogeneous microstructure, and excellent mechanical properties of the bulk alloys [7,9].

Our motivation is to develop a HEA material with high strength, closely related to the well-known refractory WTaMoNb HEA [10]. We design (W35Ta35Mo15Nb15)95Ni5 RHEA with high theoretical density of 15.12 g/cm<sup>3</sup> via regulating constituent molar ratio and addition of fcc-Ni to accelerate the alloying process. Given the high melting temperature of refractory elements and the traits of fabrication methods, we will adopt powder metallurgical processes to fabricate bulk RHEA.

In this study, a novel (W35Ta35Mo15Nb15)95Ni5 RHEA is fabricated by mechanical alloying and spark plasma sintering process. The alloying behavior, phase compositions were analyzed, and the mechanical property was also studied.

### **2. Materials and Methods**

#### *2.1. Synthesis of RHEA*

The schematically preparation process was presented in Figure 1. The first process is the synthesis of (W35Ta35Mo15Nb15)95Ni5 refractory high entropy alloy (RHEA) powders by mechanical alloying (MA). The RHEA powders were prepared by high-energy ball milling (YXQM-4L, MITR Co., Ltd., Changsha, China) of elemental W, Ta, Mo, Nb and Ni powders, and all the powders were purchased from Beijing Xingrongyuan Technology Co., LTD, China with purity of more than 99.9% and diameter of less than 5 μm. The W, Ta, Mo, Nb and Ni powders with atomic ratio of 33.25:33.25:14.25:14.25:5 were placed in a stainless-steel tank filled with tungsten carbide balls. The MA process was carried out in a high-purity argon gas atmosphere to prevent oxidation with a high-energy planetary ball mill at 300 rpm, and a ball to powder weight ratio of 10:1. To analyze the MA behavior of the HEA powders, powders were sampled after 0.25, 2, 4, 10 and 12 h of milling. All of these procedures were carried out in a glove box filled with high-purity argon gas to prevent oxidation of the HEA powders. In order to ensure the metallic powders fully alloying, the powders ball milled for 12 h were selected to sinter via spark plasma sintering (SPS, HP D25, FCT Systeme GmbH, Rauenstein, Germany) at pressure of 30 MPa under vacuum. The metallic powders were placed in the carbon mold, and then the upper and lower carbon die punches are put into the carbon mold. The SPS process are carried out applying a specific sintering power resource and pressing pressure to the sintered powder. The powders were heated to 420 ◦C within 2 min, and from 420 ◦C to final temperature of 1700 ◦C at a heating rate of 100 ◦C min<sup>−</sup>1, followed by 10 min of holding time at the sintering temperature.

**Figure 1.** Schematic demonstration of the processing route adopted for the fabrication of (W35Ta35Mo15Nb15)95Ni5 high entropy alloy (HEA).

## *2.2. Characterization*

The particle size distribution of the HEA powder was determined by a laser particle size analyzer (MALVERN, MICRO-PLUS, UK) using water as dispersant. The crystal structure of the alloyed powders and sintered bulk specimen were identified by X-ray diffraction (XRD, Rigaku, D/max 2550VB, Japan, scan rate of 8 deg min<sup>−</sup>1, scan-step size of 0.02 deg, scan angle range of 20–90 deg) using CuK<sup>α</sup> radiation at 40 KV and 450 mA. The microstructure and alloying component distribution of the mechanically alloyed HEA powder were analyzed by transmission electron microscopy (TEM, JEM-2100F, JEOL, Tokyo, Japan). The microstructure of the (W35Ta35Mo15Nb15)95Ni5 HEA was characterized by field emission scanning electron microscopy (SEM, NOVATM NanoSEM230, FEI, Eindhoven, the Neitherland) equipped with the energy dispersive (EDS, GENESIS60S). The secondary electron (SE) imaging mode was used to observe the morphology of the samples. The back-scattered electron (BSE) imaging mode was used to distinguish the different phases of the microstructure. The alloying element distribution was measured by electron probe microanalysis (EPMA, JXA-8230, JEOR, Tokyo, Japan) with back-scattered mode. Metallographic specimens were prepared by grinding, polishing and chemical etching in an aqueous solution (mKOH:mk3FeCN6:mH2O = 1:1:18). The density of the (W35Ta35Mo15Nb15)95Ni5 HEA was determined using Archimedes' method with distilled water. Compressive stress–strain testing was carried out in universal testing machine (3369, INSTRON, Bosto, MA, USA) with cylindrical specimens (<sup>Φ</sup> 6 mm <sup>×</sup> 8 mm) at an engineering strain rate of 2 mm min−<sup>1</sup> at room temperature. The representative data was obtained by averaging three values of the test results.

#### **3. Results and Discussion**

## *3.1. Alloying Process*

Figure 2 depicts the XRD patterns of (W35Ta35Mo15Nb15)95Ni5 HEA powders after different milling time. The mixed powders after 0.25 h of milling contains diffraction patterns of all constituent elements, while the least atomic weight leaded to that the characteristic peak of intensities of Ni were relatively lower than those of the other elements, because the peak intensities of each element are related to the atomic number of the constituents [11]. The intensities of the diffraction peaks declined sharply and the characteristic peak of Ni disappeared after 2 h of milling, as shown in Figure 2a, which recommends sufficient dissolution of Ni. With the milling time prolonged to 4 h, the characteristic peaks of Mo and Nb clearly merged into the peaks of Ta and W. After 10 h of milling, the constituent elements peaks are no longer observed, and it is obvious that four new diffraction peaks appearing at 2θ ≈ 39.94◦, 57.78◦, 72.56◦ and 86.19◦ are quite well with (110), (200), (211) and (220) planes of a simple solid solution structure. The declined and broadened diffraction peaks are largely ascribed to the refinement of the grain size and the increment of the lattice strain. All diffraction peaks are consistent with a single bcc phase. The minimum alloying time for (W35Ta35Mo15Nb15)95Ni5 RHEA powders is shorter than that of some common RHEAs [6,7] powders may due to the addition of Ni. In fact, the solid solution sequence of the alloying elements is closely related to their melting points [12]. Generally, elements with lower melting points would have higher alloying rate than that with higher melting points. From what has been discussed above, we can draw a conclusion that the mechanical alloying procedure is that the decrease in the peak intensity, broadening and merging of the diffraction peaks and the subsequent disappearance, which is due to the factors: crystal size refinement, lattice strain increasement and reduced crystallinity [13].

**Figure 2.** (**a**) XRD patterns of (W35Ta35Mo15Nb15)95Ni5 HEA powders after different milling time, and (**b**) magnified XRD patterns of the 2, 4, 10 and 12 h-milled powders at 2θ range of 38–45◦.

The breadth of the diffraction peak is a combination of both instrument and sample dependent effects. To eliminate instrumental contributions, it is necessary to obtain a diffraction pattern from the line broadening of a standard material such as silicon powders that are annealed at 1100 ◦C for 30 min to determine the instrumental broadening. The instrumental-corrected broadening β*<sup>D</sup>* corresponding to the diffraction peak of (W35Ta35Mo15Nb15)95Ni5 was evaluated using the relation as follows [14]:

$$
\beta\_D^2 = \beta^2\_{\text{measured}} - \beta^2\_{\text{instrumental}} \tag{1}
$$

The grain size was determined according to the Scherrer equation:

$$D = \frac{k\lambda}{\beta\_D \cos \theta} \tag{2}$$

where *D* is the grain size in nanometers, λ is the wavelength of the radiation (0.154056 nm for CuKα radiation). *k* is a constant equal to 0.94, β*<sup>D</sup>* is the peak width at half-maximum intensity and θ is the peak position. The lattice strain of HEAs were calculated according to the Williamson–Hall equation [15]:

$$
\beta \cos \theta \,\, = \frac{k\lambda}{D} + \varepsilon \sin \theta \,\,\tag{3}
$$

where β is assumed to be β*D*, ε is the lattice strain.

The dislocation density ρ was calculated using the following equation [16]:

$$
\rho = \frac{2\sqrt{3}\varepsilon}{Db} \tag{4}
$$

where ε, *D* and *b* are lattice strain, grain size and Burgers vector, respectively. *b* is equal to √ 3 <sup>2</sup> *a* (*a* is the lattice parameter determined in the means of the Nelson-Riley approach with the weighting function 1/2(cot <sup>2</sup> θ + cos θ cot θ) [17]) for a bcc structure, ε and *D* are calculated from the XRD patterns based on Equation (2).

Figure 3a shows grain size and lattice strain under different milling duration. By and large, with increasing milling time, the grain size become lower, while the lattice strain become larger. As shown, the grain size refined with a large margin to 18.1 nm after 10 h milling time and then slightly decreases to 16.3 nm after 12 h MA, along with the lattice strain increases to 5.87% from 4.15%. Finally, the grain sizes and lattice strain of the powders reaches a dynamic balance after milling for 10 h. The dislocation density gradually become larger with prolonged milling time according to the Equation (4). The morphology and particle size distribution of the HEA powders obtained after 12 h ball milling treatment is shown in Figure 3b. The mechanically alloyed powders with mean size of 5.29 μm are made up of a large amount of refined powders and relatively much larger powders. The approximate spherical morphologies of the HEA powders demonstrate great fluidity, promote the relative high density of as-sintered bulk materials. We choose the powders with milling time of 12 h to sinter, which has been completely alloyed.

**Figure 3.** (**a**) Grain size and lattice strain for the powders at different milling time, (**b**) the morphology and the powder particle size distribution curve of the 12 h-milled powders.

Figure 4 is the high-resolution TEM (HRTEM) image of HEA powders after 12 h milling, and fast Fourier transformation images from the region is shown in the inset. The interplanar distance is 0.23 nm determined by the HRTEM image. Scanning TEM-energy dispersive X-ray analysis (STEM-EDX) was also conducted to qualitatively verify the elemental distribution in HEA powder. Figure 5 clearly illustrates that the element of the (W35Ta35Mo15Nb15)95Ni5 RHEA are uniformly dispersed.

**Figure 4.** HRTEM image of the HEA powder after 12 h milling and fast Fourier transformation image.

**Figure 5.** TEM image and corresponding EDX map for each component of HEA powder after 12 h milling, (**a**)TEM image, (**b**–**f**) EDS mappings of W, Ta, Mo, Nb and Ni, respectively.

## *3.2. Phase and Microstructure of the Bulk RHEA*

The crystal structure and phase analysis were investigated by XRD. Figure 6 shows the XRD pattern of the (W35Ta35Mo15Nb15)95Ni5 bulk alloy consolidated by SPS. The pattern of the powders milled for 12 h is also presented for comparison. The theoretical crystal lattice parameter *amix* of the bcc solid solution in non-equiatomic (W35Ta35Mo15Nb15)95Ni5 RHEA is calculated to be *amix* = 0.324 nm using the rule of mixtures, while the lattice parameter of as-sintered bulk RHEA determined by the XRD is 0.320 nm based on Nelson-Riley approach [17]. It can be concluded that the calculated value for the (W35Ta35Mo15Nb15)95Ni5 RHEA is very close to the experimental result. Compared with the XRD pattern of 12 h-milled powder, peaks shift can be clearly observed for the as-sintered bulk. The peaks position transfer toward lower Bragg angle (2θ), indicating that the lattice parameter of as-sintered bulk RHEA is larger than that of 12 h-milled RHEA powder (*a* = 0.319 nm). The XRD patterns indicate that the bulk specimens exhibit a single bcc structure. Besides, the minor diffraction peaks in the XRD pattern closely matched with cubic Nb5.7Ni4Ta2.3O2 (PDF#82-0947). A few diffraction peaks intensity is inconsistent with that of the PDF card may because the intensities of the XRD diffraction peaks decreases due to the appeared extremely refined oxide inclusion. As shown in Figure 6, the predominant crystal structure for both (W35Ta35Mo15Nb15)95Ni5 bulk sample and powders is a single-phase solid solution with bcc structure, which demonstrates there was no phase decomposition during the SPS process.

**Figure 6.** XRD patterns of HEA powders after milling for 12 h and bulk specimen sintered at 1700 ◦C.

Generally, the formation of metallic materials decides their physical and chemical properties. High entropy alloys are deemed as a kind of novel metallic materials, thus the phase structure and stability prediction have great significance for the design and application of these types of material. Since the concept of HEA was introduced by Yeh et al. [18], many literatures have reported physical parameters to predict the phase structure and stability of HEAs. Especially, the parameters δ (the atomic size difference) and Ω (a thermodynamic parameter) that were proposed [19] can predict the structure and stability of HEA. The parameter δ is defined as:

$$\delta \ll = \sqrt{\sum\_{i=1}^{n} c\_i \left(1 - \frac{ri}{\bar{r}}\right)^2} \tag{5}$$

where *ci* is the atomic fraction of the *<sup>i</sup>*th element, *ri* is the atomic radius of the *<sup>i</sup>*th element and <sup>−</sup> *r* is the average atomic radius, which is calculated by <sup>−</sup> *<sup>r</sup>* <sup>=</sup> *<sup>n</sup> i*=1 *ciri*.

The other parameter (Ω) is defined as:

$$
\Omega = \frac{T\_m \Delta S\_{\text{mix}}}{|\Delta H\_{\text{mix}}|} \tag{6}
$$

where *Tm* is melting point calculated as weighted average of the constituting elements, Δ*Smix* is the entropy of mixing for HEAs, which is calculated by <sup>Δ</sup>*Smix* <sup>=</sup> <sup>−</sup>*<sup>R</sup> <sup>n</sup> i*=1 *ci* ln *ci* and Δ*Hmix* is the enthalpy of mixing of HEAs, which is calculated by <sup>Δ</sup>*Hmix* <sup>=</sup> <sup>n</sup> *i*=1,*ij* 4Δ*HmixABcicj*. According to Yang [19], as δ ≤ 6.6% and Ω ≥ 1.1, a stable, solid solution phase is constructed in a HEA system.

The literature has proposed another important parameter [20], the valence electron concentration (*VEC*). *VEC* is calculated by:

$$VEC = \sum\_{i=1}^{n} c\_i (VEC)\_i \tag{7}$$

where (*VEC*)*<sup>i</sup>* is the *VEC* for the *i*th alloying element. Guo et al. [20] criticized that a low *VEC* (<6.87) is partial to sole bcc structure's solid solution phase, whereas a high *VEC* (≥8) favors sole solid solution phase with fcc structure. When 6.87 ≤ *VEC* < 8, the mixture of bcc and fcc solid solution phase is displayed.

According to the Equations (5)–(7), the calculated values of δ, Ω and *VEC*for the (W35Ta35Mo15Nb15)95 Ni5 HEA are 3.27%, 4.16 and 5.73 respectively. Hence, the formation of the current HEA is supposed to be a formation of single-phase solid solution with bcc structure. Our experimental results coincided exactly with the prediction.

Figure 7a shows the SEM image of the as-sintered (W35Ta35Mo15Nb15)95Ni5 HEA. Likewise, the back-scatter electron imaging mode was utilized to distinguish the different phases of the bulk HEA as shown in Figure 7b. It suggests that the (W35Ta35Mo15Nb15)95Ni5 HEA is composed of a light gray matrix and dark gray inclusion. We adopt SEM-EDS to determine chemical components in region A and region B. The EDS results are presented in Figure 7c,d. The EDS analysis result presents the region B were enriched in Ta, Nb and O, and had a much higher Ni content than region A, while the W, Mo elements in region B are deficient.

**Figure 7.** SEM image of the bulk (W35Ta35Mo15Nb15)95Ni5 RHEA (**a**,**b**) with secondary electron and back-scattering electron mode, respectively, and EDS analysis of (**c**) region A and (**d**) region B marked in (**b**).

In order to further analyze the composition of the second phase in the bulk alloy, the back-scatter electron imaging mode and corresponding EDS-mapping were used to measure the distribution of the alloying elements in inclusion and matrix. The EDS-mapping analysis results are shown in Figure 8. The distribution of Ta, Nb elements was homogeneous in the HEA, whereas, the distribution of W, Mo, Ni elements were inhomogeneous. The light matrix is enriched in W, Mo and the distribution of W, Mo elements are homogeneous, but in the dark gray region are deficient. The dark gray region is enriched in Ni element, but in the light grey matrix were deficient. These results are consistent with the XRD analysis that the dark gray oxide inclusion should be Nb5.7Ni4Ta2.3O2 phase. The microstructure and element distribution of the alloy was further confirmed by EPMA. Figure 9 presents the EPMA mapping of alloying elements of the (W35Ta35Mo15Nb15)95Ni5 HEA. As can be seen, alloying components Ni, Nb and Ta content is rich, while W and Mo are extremely minor in the region demonstrated by yellow circle. Thus, the EPMA analysis agrees with the XRD and the SEM results.

**Figure 8.** SEM-BSE image and corresponding EDS map for each component of (W35Ta35Mo15Nb15)95Ni5 HEA bulk specimen, (**a**) SEM image, (**b**–**g**) EDS mappings of W, Ta, Mo, Nb, Ni and O respectively.

**Figure 9.** The EPMA-BSE image and the related mapping of alloying elements of the (W35Ta35Mo15Nb15)95Ni5 HEA.

Figure 10a shows the microstructure of the (W35Ta35Mo15Nb15)95Ni5 HEA bulk after chemical etching. The bulk materials via powder metallurgical process have a lot of pores as marked by red arrows. From the image, we can see that a large amount of inclusion marked by yellow arrows, are uniformly dispersed in the grain boundaries of the matrix combing with Figure 7b. We can clearly observe the grain marked by red dotted line, and the grain size of the bulk HEA was also measured from Figure 10a by nano-measurer software. Figure 10b depicts the statistical distribution of grain size and its Gaussian fitting. From Figure 10b, we can see that the grain size mainly ranges from 2.25 to 4.25 μm, and the average grain size is around 3.25 μm.

**Figure 10.** (**a**) SEM image of the bulk (W35Ta35Mo15Nb15)95Ni5 RHEA after etching, (**b**) grain size distribution of HEA matrix.

## *3.3. Mechanical Properties*

Compression test was performed at room temperature. Figure 11 shows the engineering compression stress–strain curve for the (W35Ta35Mo15Nb15)95Ni5 RHEA. The compressive mechanical property of the bulk (W35Ta35Mo15Nb15)95Ni5 RHEA and typical refractory HEAs reported were summarized in Table 1. In contrast with other typical RHEAs, the (W35Ta35Mo15Nb15)95Ni5 RHEA possesses the advantages of both high strength and high fracture strain. It is necessary to take the microstructure and crystal structure of each phase into account to analyze the factors of the outstanding mechanical properties for the (W35Ta35Mo15Nb15)95Ni5 RHEA. Firstly, solid solution strengthening is an intrinsic concept of HEAs on account of a severe lattice distortion [2]. Furthermore, unlike the typical as-casted RHEAs that have larger coarse-grains (grain size: 50–100 μm) [21], the as-sintered (W35Ta35Mo15Nb15)95Ni5 RHEA displays a much finer and more homogeneous microstructure (grain size was approximately 3.25 μm), implying that grain boundary strengthening play a dominant part in the strength improvement. Grain boundary strengthening mechanism is described by the Hall–Petch formula (Δσ*gb* = *ky*/ √ *d*, where *ky* is the Hall–Petch coefficient and d is the average grain size). In addition, it should be noted that the (W35Ta35Mo15Nb15)95Ni5 RHEA in this study has marginal volume fraction of oxide inclusion phases Nb5.7Ni4Ta2.3O2 which contribute to the precipitation strengthening. Generally, precipitation strengthening mechanism complies with Orowan dislocation bypass mechanism or dislocation shearing mechanism. Given the two mechanisms traits, the high shear modulus and hardness of the oxide inclusion, Orowan dislocation bypass mechanism is more appropriate to assess the contribution of precipitate-reinforce mechanism. Finally, it is inevitable that interstitial elements such as O, N, C, etc, are introduced during the MA and SPS process, which contributes to interstitial solid solution strengthening [22]. Figure 11b,c depict compressive fracture morphologies of the (W35Ta35Mo15Nb15)95Ni5 RHEA at room temperature. We can deduce that the compressive fracture mode is cleavage fracture from the fracture morphologies present a mixture of river-like pattern and steps.

**Figure 11.** (**a**) Engineering stress–strain curve of the bulk (W35Ta35Mo15Nb15)95Ni5 RHEA at room temperature, (**b**) fracture surface morphology and inserted compression sample image and (**c**) magnified fracture morphology.



To sum up, the as-sintered (W35Ta35Mo15Nb15)95Ni5 RHEA may be regarded as a potential candidate for high temperature structural material with high density and outstanding compressive strength owing to a combination of the four strengthening mechanisms.

### **4. Conclusions**

A novel high strength (W35Ta35Mo15Nb15)95Ni5 RHEA with high density was successfully fabricated by MA and SPS process in this study. The 12 h-milled powders with an average crystalline size of about 16.3 nm exhibit a single bcc solution phase. After the SPS process at 1700 ◦C, the bulk RHEA has a relative density of exceeding 96% and the average grain size is 3.25 μm. Meanwhile, minor oxide inclusion (Nb5.7Ni4Ta2.3O2) are detected in the bulk sample. The compressive yield stress, maximum stress and fracture strain of the (W35Ta35Mo15Nb15)95Ni5 RHEA are 2128 MPa, 2562 MPa and 8.16%, respectively, which are superior to some typical RHEAs. The extraordinary high mechanical properties are attributed to the following factors: solid solution strengthening due to a severe lattice distortion, which is an intrinsic concept of HEAs, grain boundary strengthening, precipitation strengthening mechanism by second phases.

**Author Contributions:** Investigation, B.D. and Y.Y.; methodology, D.W.; project administration, X.L.; writing original draft, Y.Y.; writing—review & editing, Z.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by the Natural Science Foundation of Hunan Province, China (No. 2018JJ3677).

**Conflicts of Interest:** The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

## **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **E**ff**ects of the Addition of Fe, Co on the Azo Dye Degradation Ability of Mn-Al Mechanically Alloyed Powders**

## **Wael Ben Mbarek 1, Joan Saurina 1, Lluisa Escoda 1, Eloi Pineda 2, Mohamed Khitouni <sup>3</sup> and Joan-Josep Suñol 1,\***


Received: 15 October 2020; Accepted: 22 November 2020; Published: 25 November 2020

**Abstract:** Azo compounds are used in the textile and leather industry. A significant step during the azo dyes treatment of water is the degradation by breaking the N=N bonds. This break produces the decolorization of water. In this research work, 10% atomic of Fe or Co was added to produce ternary Mn-Al-rich, nanostructured, mechanically alloyed powders in order to improve the decolorization of Reactive Black 5 solutions and to check Fe and Co addition's influence. The microstructure was followed by X-ray diffraction, the morphology and composition by electronic microscopy and energy-dispersive X-ray spectroscopy (EDS) microanalysis. The dye degradation was monitored with ultraviolet/visible absorption spectrophotometry. After degradation, the remaining organic compound was checked by high-performance liquid chromatography (HPLC) and the functional groups of the powdered alloys by infrared spectroscopy. Fe addition to Mn-Al displayed faster kinetics and a higher efficiency than the Co addition. The Mn-Al-Fe solution (0.25 g/100 mL) was fully decolorized in 5 min. On the other side, Mn-Al-Co powders were able to successfully decolorize the dyed solution in 10 min under the same conditions. Thus, nanocrystalline Fe-doped Mn-Al alloys are good candidates for use in the decolorization process, in comparison with Co-doped and other intermetallic particles.

**Keywords:** mechanical alloying; reactive black 5; decolorization; UV-visible spectrophotometry; LC-MS analysis

## **1. Introduction**

The industrial water effluent is one important source of pollution in the environment [1]. The dye effluents from textile industries are significant sources of pollution (high toxicity, low biodegradation rate) [2]. Various processes and materials have been proposed in the scientific literature for wastewater treatment. Some examples are: adsorption onto granular activated carbon [3], hollow fiber ultrafiltration [4], coagulation-flocculation with Detarium microcarpum [5], biological oxidation [6], or some photocatalyst methods and materials [7,8] including a magnetic photocatalyst [9]. One of the methods is reduction with zero valent metals or alloys [10]. They have been applied for removing azo dyes [7] as well as metals and chlorinated organics [11–14].

The metallic particles (MPs) have been applied as elements or as alloys to facilitate the reduction process. The microstructure of the particles can be amorphous or crystalline. Usually, the best results are found in amorphous because the corrosion resistance and chemical interaction are usually improved if the comparison is performed with crystalline alloys with the same composition [15]. Thus, as microstructure affects the functional behavior, in this work, we analyze some nanocrystalline Mn-rich alloys. Nanocrystalline materials have a crystalline structure, and between 10 and 50% of atoms are in the grain boundaries between crystalline grains. In the scientific literature, there are interesting works about the degradation induced by MPs in dyes (by chemical reaction, breaking bonds). One of the main topics is the analysis of the degradation reaction by taking into account the particle's surface activity (with the organic contaminants) of the metallic particles [16,17]. The reaction of degradation of the dye molecules is based on a redox process. On the surface of the metallic alloys, the metallic atoms lose electrons. These electrons are effective to cleave the active bonds (as the azo dye -N=N- bonds) of organic molecules [18–22].

There are some methods that help to develop metallic particles: precipitation (magnetite, iron) [23,24], gas atomization (high-entropy alloys) [25] or mechanical alloying (Mn-Al-based alloys) [19,22]. In the mechanical alloying (or ball milling) process, particles yield severe plastic deformation, producing stress and strains. Thus, the mechanical deformation favors the development of metastable microstructures (amorphous, nanocrystalline) and, subsequently, the mechanical and functional properties are different from crystalline alloys with the same composition [26].

In this work, we analyze the azo dye degradation ability by improving the rate of the degradation reaction of ternary Mn-Al- (Fe, Co) nanocrystalline powders produced by mechanical alloying. A high degradation rate will favor the integration of this material as a preliminary step procedure for the full degradation of azo dye wastewaters.

In previous works, we reported the high efficiency of Manganese-Aluminum (Mn-Al) particles in the degradation of azo-dyes: reactive black 5 (RB5) [19] and orange II [21]. It should be remarked that Mn and Al (as biocompatible elements) are used in environmental and biomedical applications [27].

#### **2. Materials and Methods**

The Mn60Al30Fe10 and Mn60Al20Co10 (at.%) alloys were produced from elemental high-purity (<99.9 at.%) Mn, Al, Fe or Co by mechanical alloying (MA) under an Ar atmosphere for 15 h at 500 rpm. Cycles (10 min milling/5 min off) are applied to prevent the excessive heating, with the inversion of the rotation in each cycle (to prevent powders sticking to milling media).

The morphology, composition, surface area, and microstructure of the powders were explored with scanning electron microscopy (SEM) at 15 kV. Being integrated in the SEM equipment, microanalysis was performed by energy-dispersive X-ray spectroscopy (EDS, Vega©Tescan, Brno, Czech Republic). The reactivity is linked to the specific surface area of the Mn-Al- (Fe, Co) powders. This specific area is determined by the gas multilayer adsorption method in a BET device (Micromeritics ASAP 2010 M, Norcross, GA, USA) under nitrogen, after degassing the powder at 300 ◦C for 24 h. The microstructure was determined at room temperature by X-ray diffraction (XRD, Siemens/Bruker D500, Billerica, MA, USA) using Cu-Kα radiation.

In order to assess the colorant (RB5) degradation reaction, an aqueous solution (dye concentration: 40 mg L−1). The procedure was described and portrayed in previous works [19,20]. The produced supernatants were separated, and the color intensity was measured at the maximum absorption wavelength by ultraviolet-visible absorption spectrophotometry (UV-Vis, Shimadzu 2600, Kyoto, Japan), wavelength scan in the range between 200 and 800 nm).

To check whether these materials can be used in the integrated treatment of wastewater, it is necessary to analyze the metallic powders after degradation as well as the remaining wastewater liquid (the procedure described in references 19 and 20). The powder collected after degradation was analyzed using Fourier transform infrared spectroscopy (FTIR) (Bruker, Billerica, MA, USA) and the concentrations of the metal ions checked by absorption spectrophotometer according to standard methods [19]. The organic compounds from the RB5 degradation were analyzed by high-performance liquid chromatography (HPLC) (Agilent, Santa Clara, CA, USA) coupled with

tandem mass spectrometry (LC-MS/MS, Bruker, Billerica, MA, USA) in a Beckman Gold chromatograph fitted with a Proshell 120 Pheny Hexyl column at room temperature.

#### **3. Results**

First, we check powders morphology and composition. Figure 1a,b displays the surface morphology of the MA Mn-Al-Fe powder. As exhibited in the figure, many corrugations appear on the surface of the nanostructured powders. The EDS microanalysis finding reported in Figure 1c does not reveal significant contamination by the production process (milling tools) and the MA powder is mainly composed (in the volume) of the precursors. The C is due to the sputtering process used to prepare samples for SEM observation. Oxygen contamination is typical of powdered metals with a high surface/volume ratio due to contact with the air before and after the MA process in the Ar atmosphere. Furthermore, oxides were undetected in the powders. The atomic concentration of the elements (Mn, Al and Fe) is 57:31:12, These values are similar to those of the nominal composition: 60:30:10. It should be remarked that EDS is considered as a semi-quantitative method. The size distribution of the powders (by neglecting big powders associated with conglomerates) is provided in Figure 1d (mean particle size is ≈10 μm, the accuracy of the size is 0.5 μm and the standard deviation 3.6 μm). The micrographs of the Mn-Al-Co as-produced powder are illustrated in Figure 2a,b. As expected in ball-milled alloys, smooth surfaces, particle aggregation and corrugations were found. The EDS microanalysis (Figure 2c) proves that the Mn:Al:Co ratio was approximately equal to 55:34:11, similar to the 60:30:10 nominal composition. Likewise, the distribution of the particles size is shown in Figure 2d. For statistical comparison, the particle size distribution was also analyzed. The average particle size found to be slightly lower than those of Mn-Al-Fe particles, at ≈9 μm (this slight difference can be influenced by the particle selection manual procedure for SEM analysis), and the standard deviation is 1.9 μm.

The decolorization was checked by analyzing the solutions described in the materials and methods section. Figure 3a presents the image of RB5 solutions before and after the addition of MA powders (25 ◦C). It is clear that the discoloration of the RB5 aqueous solution RB5 is very successful. The supernatants of the aqueous solution of RB5 treated with the two ternary powders Mn-Al-Fe and Mn-Al-Co were separated, respectively, at 0.5, 1, 2, 3, 4 and 5 min as well as 0, 1, 5, 10, 15 and 20 min. Figure 3b,c shows the evolution of absorbance of different aqueous supernatant portions which were measured by UV-Vis absorption spectrophotometry. For the ternary compound Mn-Al-Fe, the solution is fully discolored in 5 min (Figure 3b), whereas the solution treated with the Mn-Al-Co powder is fully discolored in 20 min (Figure 3c). The peak at 597 nm of the absorbance is associated to the azo dye concentration in the solution [19,20,28]. This peak in the visible region is associated to the "-N=N-" bonds and to the dye molecule concentration in the solution. Thus, the decrease in the intensity (as the reaction time in the solution increases) at 597 nm is an indicator of the degradation and RB5 chromophores' evolution. Likewise, the bands (at 230 and 310 nm) in the ultraviolet region are assigned, respectively, to the benzene and naphthalene rings of the molecule dye [19,20,29,30]. These bands, at λmax, become weaker with degradation time. This effect is linked to: (a) the cleavage of the azo bands (also confirmed by the increase of 246 nm in the absorbance peak), (b) the formation of (-NH2) groups, (c) the decomposition of RB5 in the solution. This result was also found by Zhang and coworkers [31], who confirmed the reductive degradation by amorphous Fe of azo acid orange II solution and Ben Mbarek et al. [19,20], who highlighted the rapid degradation of azo-dye with mechanically alloyed nanocrystalline Mn-Al alloys, as well as the high-efficiency decolorization of azo dyes by Ca-Al particles.

**Figure 1.** (**a**,**b**) Micrographs of the mechanical alloying (MA) Mn-Al-Fe. (**c**) Energy-dispersive X-ray spectroscopy (EDS) microanalysis and (**d**) distribution of particle sizes of the mechanically alloyed powder.

The mechanism of the redox reaction is viewed to be practically the same as for the case of the binary Mn-Al. However, the rates of RB5 bleaching reaction under the same dosage and temperature conditions using the Mn-Al-Fe and Mn-Al-Co powders appear to be different. The different potentials of the reduction in Fe and Co could account for these discrepancies, as well as the different solubilities of their hydroxides. It is known that the standard oxidation potential Co/Co2<sup>+</sup> is less negative than the oxidation potential of H+/H2, whereas the standard oxidation potential of Fe/Fe2<sup>+</sup> is more negative (−0.29 and <sup>−</sup>0.44 V, respectively), which makes the transfer of electrons between Co and H<sup>+</sup> slower than that occurring between the Fe and H<sup>+</sup> [32]. For the improvement in the azo bond-breaking, Co/Fe are active actors favoring the production of hydrogen from water [31,33–35].

Another factor to consider is the different adsorption capacity of both elements. The valence electrons (metal atoms) are shared by all atoms to form a cloud of electrons in a highly shared state. The valence electron configurations are 3d64s2 (Fe) and 3d74s2 (Co). Magnetic data showed that there was an average of 2.2 and 1.7 holes in the d band for Fe and Co, respectively [36]. The more holes in the bands, the more unpaired electrons there are. Therefore, the interaction between unpaired electrons and adsorbate molecules favors the formation of a localized adsorption bond. Thus, higher adsorption capacity is linked to the unpaired electrons. Simultaneously to the adsorption of the RB5 molecule, a reduction in the azo group might take place on the surface of the metallic powder. Thus, the selected value to follow the kinetics reaction was the maximum absorption wavelength. The compound Mn-Al-Fe exhibits the highest decolorization process reaction rate in the first five minutes, since it has the highest concentration of d-band holes per unit atom [36]. However, the compound Mn-Al-Co displays a lower adsorption rate (Figure 3d).

**Figure 2.** (**a**,**b**) Particle morphologies of the MA Mn-Al-Co powder. (**c**) EDS microanalysis and (**d**) distribution of particle sizes of the ball milled powder.

As shown in Figure 4, a straight line of ln(C/Co) versus reaction time was achieve, with coefficients of determination R<sup>2</sup> (0.9322 and 0.9813, values > 0.9). Therefore, the decolorization of R5 solution is consistent with first-order kinetics. The slope of the alloy with Fe addition is 0.9725 s<sup>−</sup>1, higher than 0.1717 s−<sup>1</sup> (Mn-Al-Co). The Mn-Al-Co value is also higher than those of Mn-Al (0.2154 s−1). Thus, Co addition does not provoke a significant change in the decolorization efficiency, whereas Fe addition favors the process.

**Figure 3.** (**a**) Image of black 5 solutions processed by MA powder before and after degradation. (**b**) UV absorption spectra at different times for Mn-Al-Fe alloy. (**c**) UV absorption spectra at different times for Mn-Al-Co alloy. (**d**) The decolorization % from UV absorption intensity at 597 nm versus reaction time for both alloys.

**Figure 4.** Ln(C/Co) versus reaction time of the MA Mn-Al-Fe and Mn-Al-Co alloys.

The appearance of bubbles was detected inside the solution. This effect is associated with the formation of hydrogen gas. Resting upon this fact, the proposed mechanism of the dye degradation operates as a result of the release of: (a) H2 from the reduction in water, (b) the cleavage of -N=Nbonds and, (c), the formation of -NH2 groups. Under these conditions, the release of hydrogen gas is accompanied by the formation of insoluble metal hydroxides [19,37]. The mechanism for Mn-Al-Fe and Mn-Al-Co, in acid conditions, are based on the reactions of these references. Here, we introduce Fe and Co reactions. The reactions involving Fe are:

$$\text{Fe} \rightarrow \text{Fe}^{3+} + \text{3e}^- \tag{1}$$

$$\text{Fe}^{3+} + \text{3(OH)}^{-} \rightarrow \text{Fe(OH)}\_{3} \tag{2}$$

The main difference with previous works in Mn-Al alloys is the role of Fe. The mechanism for Mn-Al-Co is similar, taking into account the Co reactions (pH < 7) in substitution of Fe reactions [38]

$$\text{Co} \rightarrow \text{Co}^{2+} + 2\text{e}^- \tag{3}$$

$$\text{Co}^{2+} + 2\text{(OH)}^{-} \rightarrow \text{Co(OH)}\_{2} \tag{4}$$

The aluminum oxide can act as an acid or a basic agent relying on the environment by the Al (OH)3 and [Al (OH)4] − formation. The pH of the solution increases (from 6.3 to 10.8 or 11.2 in Mn-Al-Fe/Mn-Al-Co, respectively) due to OH− ion enrichment. The mechanism proposed by the Mn-Al-Fe and Mn-Al-Co reaction, in basic conditions, (pH > 7), is given in references [19,38]. The reactions involving Fe and Co are the same.

To assess the product of the discoloration reaction by the Mn-Al-Fe (or Mn-Al-Co) powder, we observed the surface of the particles by scanning electron microscopy. The micrographs obtained are plotted in Figures 5 and 6 and unveil that some products of the reaction are spread (uniformly distributed) over the entire particle's surface. The same effect was found in other metallic compounds [39]. These crystalline precipitates are mainly composed of the element Al and oxygen. In addition, a high amount of corrosion holes was detected on the surface of the alloys, indicating that corrosion of the alloy particles during the degradation process of RB5 occurs by pitting [19]. It is well known that the corrosion of the Al-based alloy is dependent on the damage of the Al2O3-passive layer. In addition, it is believed that pitting corrosion is one of the main degradation mechanisms of RB5. This result is very expressive, the addition of an acid to the initially alkaline textile industrial wastewater unnecessary. Likewise, the cationic metals concentrations of Mn, Al, Fe and Co are obtained from absorption spectrometry: 2.12, <0.59, <0.16 and <0.20 mg L−<sup>1</sup> (detection limit) for Mn, Al, Fe and Co, respectively. The dissolution of Al (OH)3 results from the excess of OH−, and therefore more surface is exposed to water favoring the progression of the reactions.

**Figure 5.** SEM micrographs of MA Mn-Al-Fe after RB5 degradation at different magnifications. Scale bars: (**a**) 100 μm, (**b**) 500 μm, (**c**) 10 μm and (**d**) 1 μm.

**Figure 6.** SEM micrographs of MA Mn-Al-Co after RB5 degradation at different magnifications. Scale bars: (**a**) 100 μm, (**b**) 500 μm, (**c**) 10 μm and (**d**) 1 μm.

After the decoloring reaction, the size distributions of the Mn-Al-Fe and Mn-Al-Co alloys are illustrated in Figure 7a,c, respectively, showing mean sizes of the order of 20 and 17 μm. This difference in size is associated to the precipitates on the surface of the nanostructured powders. The corresponding EDS analyzes depicted in Figure 7b,d indicate the approximately compositions of the elements, Mn 50.0/Al: 15.0/Fe: 8.0/O: 27.0 for the Mn-Al-Fe alloy, and Mn: 47.0/Al: 17.0/Co: 10.0/O: 26.0 for the Mn-Al-Co alloy. Finally, the microstructural study of Mn-Al-Fe and Mn-Al-Co powders before and after RB5 dye degradation (pH < 7) is performed by XRD. The results are shown (see Figure 8a,b). The diffraction patterns obtained before the degradation of the two Mn-Al-Fe/Mn-Al-Co powders reveal the coexistence of a solid solution of MnAlFe2 (JCPD 00-054-0388) and MnAlCo2 (JCPD 03-065-5185) phases, respectively, (B2, space group Fm3m) with traces of very small proportions of MnAl phase (η) rich in Al and tetragonal face-centered (FCT) structure [40]. The average crystallite size of the solid solution B2 is of 60 ± 3 nm. This nanocrystalline structure may stand for another factor increasing the reactivity of these metal powders within the aqueous dyed solutions by increasing atoms in the boundary regions. As two phases are detected after MA, the samples are not chemically homogeneous. Nevertheless, the fast discoloration kinetics allow us to state that this chemical in-homogeneity is probably not determinant. The inspection of the X-ray diffraction patterns of both powdered alloys obtained after the discoloration reaction demonstrates the appearance of new phases identified as: Al(OH)3 (JCPD 00-003 -0915), Fe(OH)3 (JCPD 00-038-0032), and Co(OH)2 (JCPD 00-051-173) beside the B2-MnAl (Fe, Co)2 and η-MnAl phases (Figure 8a,b). The products resemble micro-precipitates of Al(OH)3, Fe(OH)3 and Co(OH)2 hydroxides, covering the surface of the particles. These hydroxides are minor phases (>10 at.%). It should be remarked that it is likely that in all phases, including hydroxides, all elements (Mn, Al, Fe or Co) are present: one with a high amount and the other two elements in solid solution. Likewise, EDS analysis gives information about the volume of the powder near the surface, not the overall volume of the powders.

**Figure 7.** Distribution of particle sizes of MA powders after decoloring reaction: (**a**) Mn-Al-Fe, (**b**) Mn-Al-Co. EDS microanalysis: (**c**) Mn-Al-Fe and (**d**) Mn-Al-Co.

**Figure 8.** XRD diffraction patterns: (**a**,**b**) MA Mn-Al-Fe powder before and after degradation and (**c**,**d**) MA Mn-Al-Co powder before and after degradation.

It has been reported that MA can favor a higher specific surface area [22]. This parameter, determined by BET, provides best information about the active surface than the statistical analysis of powders' size distribution from SEM micrographs. The higher the specific surface, the higher the interaction between particles and the dye molecule. The specific surface area was 0.50 <sup>±</sup> 0.02 m2 g−<sup>1</sup> for the Mn-Al-Fe alloy and 0.45 <sup>±</sup> 0.02 m<sup>2</sup> g−<sup>1</sup> for the Mn-Al-Co alloy. There is apparently a slight difference in the specific surface. Nevertheless, fast discoloration was found in the sample with Fe

addition (the slopes of Figure 4 are 0.9725 and 0.1727 s−<sup>1</sup> in alloys with Fe and Co, respectively). This high difference is due to the high surface area of the sample with Fe, but can also be partially attributed, as discussed above, to the different chemical compositions of the alloys.

Thus, the results make these two alloys promising candidates for the fast degradation of different azo dyes. In Mn-Al MA powders, the specific surface values were found to range between 0.55 and 0.48 m<sup>2</sup> g<sup>−</sup>1. Values between 0.50 and 0.55 m2 g−<sup>1</sup> were found by other authors [41].

The FTIR spectra of RB5 before and after degradation were shown in Figure 9. The main peaks and bands identification are given in Table 1.

**Figure 9.** FTIR spectra: (**a**) RB5 powder before degradation selection. (**b**,**c**) Mn-Al-Fe and Mn-Al-Co powders after degradation, respectively.



From FTIR spectra, it is clear that some characteristic RB5 peaks decrease during the reaction, while some new peaks appear. The decreasing peaks are the azo bond (-N=N-, at 1635 cm−1) and

the naphthalene ring (C=C aromatic skeletal vibration, at 1400–1600 cm−1). A minor peak remains in all FTIR spectra, the 2963 cm−<sup>1</sup> peak, assigned to the skeletal vibration of the benzene ring. Some new peaks appear after degradation, located at 1742, 1528 and 1260 cm−<sup>1</sup> (stretching vibration of C=C, N-H and C-N bonds, respectively). This demonstrates that the reductive cleavage of the -N=Nbond [19,44] favors the presence of amines. It should be remarked that amines are also water-pollutant. Mn-Al based alloys' application in industrial processes will be combined with an ulterior amine adsorption process. Thus, metallic nanostructured particles are candidates for the first step of a combined multi-step procedure.

In addition, these results are in good accordance with the previously stated UV-Vis (Figure 3b,c) regarding the cleavage of the azo bands [42,44–46]. The FTIR confirms that the results indicated that the azo bond and the naphthalene ring are broken by mechanical alloying of Mn-Al-(Fe, Co) and result in the formation of amino and alkyl compounds.

The HPLC chromatograms are shown in Figure 10. Before degradation, the major peak has a molecular weight of 991. Similar results were elaborated by Shilpa and coworkers [46], Patel and coworkers [47] and Ben Mbarek [19,20]. The RB5 chromatography should be compared with the HPLC chromatography and LC-MS mass spectrums of the solutions after degradation with the metallic alloys.

**Figure 10.** (**a**) HPLC profile of RB5 dye solution. Mass spectrums of peaks from 10.5 min at (**b**) λ = 311 nm and (**c**) λ = 220 nm.

Figure 11 displays the LC-MS analyses of the extract of the two solutions obtained following the degradation reactions carried out by the two ternary powders Mn-Al-Fe and Mn-Al-Co. These analyses reveal several retention time peaks. In addition, analyses of the major peaks eluted between 10 and 15 min demonstrated the existence of two peaks at 349 and 280.1 m/z in relation to the final amino molecules, related to 1-2-7-triamino-8-hydroxy,3-6-naphthalenedisulfonate and 1-sulfonic acid, 2-(4-aminobenzenesulfonyl) ethanol, as depicted in Figure 10a,b. All these results confirm the appearance of aromatic amino compounds emerging from the decomposition reaction of the large RB5 organic molecule. This is inferred through the appearance of signals at 248.9 m/z, related to the aromatic amines obtained after bond cleavage. The mechanism of discoloration of RB5 with these metal powders can only be identical to that identified for the case of the Mn-Al binary following the cleavage of the azo bonds [19].

**Figure 11.** HPLC evolution profiles after degradation of the RB5 dye and mass spectrum of peaks from 10.5 to 15.2 min of the MA powders: (**a**) Mn-Al-Fe and (**b**) Mn-Al-Co.

## **4. Conclusions**

In sum, Mn-Al-Fe and Mn-Al-Co powders were successfully manufactured by mechanical alloying. The efficiency and kinetics of the discoloration reaction of aqueous solutions of RB5 are systematically evaluated to determine the capacity of these two alloys in the bleaching reaction.


consistent with first-order kinetics. Hence, the high efficiency of Mn-Al based alloys recorded in decolorization treatments of dyed wastewaters, already emphasized in previous works, can be modified and increased by minor alloying with other transition metals.

Thus, these promising materials have interesting characteristics to be applied, probably in a multi-step procedure in the wastewater treatment in the textile industry.

**Author Contributions:** Methodology, L.E., J.S.; formal analysis, W.B.M.; investigation, W.B.M.; resources, E.P., J.-J.S.; data curation, L.E., M.K.; writing—original draft preparation, W.B.M., J.-J.S.; supervision, L.E.; funding acquisition, E.P., J.-J.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** Work funded by Spanish MINECO Grant No. FIS2017-82625-P, Catalan GenCat Grant No. 2017SGR0042 and University of Girona UNIGE-2-2019 project is also acknowledged.

**Acknowledgments:** The authors acknowledge the technical facilities of the STR support to research units of the University of Girona.

**Conflicts of Interest:** The authors declare no conflict of interest.

## **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Structure-Phase Transformations in the Course of Solid-State Mechanical Alloying of High-Nitrogen Chromium-Manganese Steels**

**Kirill Lyashkov 1, Valery Shabashov 2,\*, Andrey Zamatovskii 2, Kirill Kozlov 2, Natalya Kataeva 2, Evgenii Novikov <sup>2</sup> and Yurii Ustyugov <sup>2</sup>**


**Abstract:** The solid-state mechanical alloying (MA) of high-nitrogen chromium-manganese austenite steel—MA in a planetary ball mill, —was studied by methods of Mössbauer spectroscopy and transmission electron microscopy (TEM). In the capacity of a material for the alloying we used mixtures of the binary Fe–Mn and Fe–Cr alloys with the nitrides CrN (Cr2N) and Mn2N. It is shown that ball milling of the mixtures has led to the occurrence of the α → γ transitions being accompanied by the (i) formation of the solid solutions supersaturated with nitrogen and by (ii) their decomposition with the formation of secondary nitrides. The austenite formed by the ball milling and subsequent annealing at 700–800 ◦C, was a submicrocrystalline one that contained secondary nano-sized crystalline CrN (Cr2N) nitrides. It has been established that using the nitride Mn2N as nitrogen-containing addition is more preferable for the formation and stabilization of austenite—in the course of the MA and subsequent annealing—because of the formation of the concentrationinhomogeneous regions of γ phase enriched with austenite-forming low-mobile manganese.

**Keywords:** austenitic alloys; mechanical alloying; high-nitrogen steels; atomic redistribution; point defects; Mössbauer spectroscopy

### **1. Introduction**

In the modern technology of production of the materials with specific (improved) functional properties a great emphasis is placed on the design of the high-nitrogen steels, alloys, and composites [1–3]. This is due to the (i) need for economically alloyed steels and (ii) feasibility of substitution of nitrogen for the high-cost nickel and manganese. The current technology for melting high-nitrogen steels under nitrogen pressure requires expensive equipment and high energy expenditure. In the capacity of an alternative and cheaper method for producing high-nitrogen steels, today powder metallurgy methods using mechanical alloying are proposed. In most studies on the mechanical alloying (MA) of high-nitrogen steels, the metal powder was saturated with nitrogen from the gas phase in a flowing atmosphere of nitrogen or ammonia [4–13]. In these works, the authors usually employed vibrationally assisted ball milling with long-term processing times (100 h or more). In the works [14–20], for the purpose of nitriding Fe-based steels, the authors were the first to have employed the method of solid-phase MA (i.e., MA of materials in a solid state). This method does not require additional gas equipment to create a flowing nitrogen-containing atmosphere in the mill vessel. In addition, the method does not require such a long time of mechanical processing and allows for one to use planetary-type ball mills. To the basis of the approach in [15–18] their authors have placed the cyclically

**Citation:** Lyashkov, K.; Shabashov, V.; Zamatovskii, A.; Kozlov, K.; Kataeva, N.; Novikov, E.; Ustyugov, Y. Structure-Phase Transformations in the Course of Solid-State Mechanical Alloying of High-Nitrogen Chromium-Manganese Steels. *Metals* **2021**, *11*, 301. https://doi.org/ 10.3390/met11020301

Academic Editors: Joan-Josep Suñol and Carlos Garcia-Mateo Received: 16 December 2020 Accepted: 4 February 2021 Published: 9 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

evaluating deformation-induced phase transitions «dissolution—precipitation» of disperse particles of the prime low-stable nitrides (of the types CrN and Fe4N) located in metallic matrices. In result of the deformation-induced dissolution of prime (initially present) nitrides, in the metallic matrix of alloys one can observe the formation of both the nanostructured solid solutions supersaturated with nitrogen and highly disperse secondary nitrides. Within the framework of the proposed approach with the employment of the (i) method of severe plastic deformation (namely, by high pressure torsion in the rotating Bridgman anvils) [15,16], (ii) friction-induced external action [16], and (iii) ball milling (BM) [17–20], it became feasible to have performed alloying of the Fe–Ni–Cr–N and the Fe–Mn–Cr–N austenite. For the development of high-nitrogen steels as competitors to austenite Fe–Ni–Cr–N alloys, the alloys based on the system Fe–Cr–Mn–N were proposed in [21,22] for consideration. First experiments on the solid-state MA of Fe–Cr–Mn–N steels using BM were performed on the mixtures of the binary Fe–Mn alloy and CrN (Cr2N) nitrides [17], as well as on the mixtures of the pure metals Fe, Mn, and Cr—on the one hand, and the nitrides CrN and Mn2N, on the other [19].

The purpose of this work is the investigation of the possibility to obtaining nitrogencontaining austenite by means of employment of the solid-phase MA where in the capacity of the starting material a researcher uses initial mixtures of the binary alloys and nitrides together with the alloying elements that traditionally are taken as constituents of the stainless steels from the austenite class of Fe–Cr–Mn–N type. The subject of the study was the analysis of the mechanism, kinetics, and properties of the products of the MA—in the course of BM—and subsequent thermal anneals, in dependence of initial compositions of the initial matrices and nitrides. In particular, an optimum composition of the austenite to obtain was of interest from the point of view of reducing the volume of alloying manganese and chromium in the composition of MA-produced austenite.

#### **2. Experimental**

#### *2.1. Sample Compositions and Their Treatment*

For obtaining MA Fe–Cr–Mn–N alloys, in the capacity of their metallic matrix we used powders of the binary alloys with BCC crystal lattice of compositions: Fe–*X*Mn (*X*, wt% = 0, 4.0, 6.7, 8.9) and Fe–*Y*Cr (*Y*, wt% = 0, 4.7, 8.6, 14.2), and in the capacity of "donor" of the alloying element «nitrogen» there were chosen the nitrides CrN (Cr2N) and Mn2N, which are low-stable in conditions of deformation. We analyzed the mixtures in the series *A* and *B* (see Tables 1 and 2), each of which had several compositions of constituents with the varying contents of the metallic matrix and 10 or 20 wt% proportion of the nitrogen-containing addition:

> *A*: Fe-*X*Mn + 20CrN (Cr2N); *B*: Fe-*Y*Cr + (10 or 20Mn2N).


*Metals* **2021** , *11*, 301

 =

 -

Element formulas of compositions were selected based on the data from phase diagrams for the states of equilibrium of stainless austenitic steels, including those doped with nitrogen [21,22]. The binary alloys were smelted, homogenized, and then filed. The nitrides were synthesized using the technology of self-propagating high-temperature synthesis [23]. According to the data of X-ray diffraction (XRD) analysis, the chromium nitrides have presented by themselves the mixture of composition 80% CrN + 20% Cr2N [18,19], and the manganese nitrides have been represented by 80% Mn2N in mixture with the nitrides that have had an enhanced content of magnesium. According to the results of chemical analysis, the synthesized magnesium nitrides have contained 9.1 wt% (28.3 at.%) nitrogen [20]. Mechanical alloying was performed with the employment of a planetary ball mill «Pulverisette-7». The speed of rotation of the ball mill platform was 800 rpm. The vessel and balls were made of high-strength ball-bearing steel containing 1.5 wt% Cr and 1.0 wt% C, iron in balance. This steel is the most stable compared to materials such as tungsten carbide and stainless steel [24]. The weight ratio for the balls (the total quantity of 15 pieces, 10 mm in diameter each) and a powder sample was 6:1. After loading the balls and powder mixtures into the vessel, air was pumped out to 10−<sup>3</sup> mm Hg and the vessel was filled with an inert gas-argon. The duration of milling time was varied from 5 to 20 h. The temperature at the external sides of the vessels did not exceed 70 ◦C. The average size of the particulates of the initial powders of the alloys was about 200 μm. After termination of ball milling, part of the powders were annealed in vacuum at temperatures of 700 ◦C (*a*) and 800 ◦C (*b*) for 1 h, to specify—in a vacuum of 10−<sup>5</sup> mm Hg. Possible contamination of the test samples with wear products was controlled by weighing the mass of vessels, balls, and powder before and after mechanical alloying. The difference in the masses of powders did not exceed 0.3–0.5 wt%. The method of estimating powder contamination by weighing the mass of the vessels, balls, and powder is not absolute, i.e., it is not free of «errors». However, in this work, (i) taking into account the same conditions of exposure of different mixtures to mechanical processing and (ii) assuming the same systematic errors (in terms of the degree of contamination), the differences in the results of mechanical alloying of mixtures can be considered reliable. According to the results of chemical analysis, in all the cases of ball milling, the oxygen concentration in the MA samples did not exceed 0.8 wt%.

## *2.2. Mössbauer and TEM Analysis of MA Samples*

The Mössbauer measurements of the product results of the MA and subsequent annealing of the powders were carried out at room temperature on the spectrometer unit MS-1101 in regime of constant accelerations with the source 57Co(Rh). The spectra were calibrated using the absorbent material of α-Fe at a room temperature. Both the structure and the phase composition of an MA powder mixture were studied with the help of a transmission electron microscope JEM-200CX, with analyzing XRD patterns and dark-field images.

The Mössbauer spectra (transmission of gamma-quanta *A*, % as function of Doppler velocity *V*, mm/s) of MA samples have revealed in themselves a multicomponent structure typical of the solid solutions in both α-ferrite and γ-austenite phase states. Taking into account the complicated character of Mössbauer spectra, we used for their calculation the application-oriented soft-ware package MS Tools (borrowed from) [25]. Figures 1–5 show the experimental Mössbauer spectra and the results of their calculation. In particular, the program DISTRI was employed, which is usually used for—in the case of—locally inhomogeneous systems with a multicomponent structure and poor spectrum resolution. With its help one can improve the quality of a resolution of such spectra via restoring the distributions *p*(*V*) and *p*(*H*), which present by themselves the probabilities of resonance absorption represented on the scale of the Doppler velocities *V* and effective field *H*. Further, based on the (i) type of distributions *p*(*V*) and *p*(*H*), (ii) analysis of a priori data, and (iii) selection of the model for representing Mössbauer spectra, we made use of the soft-ware program SPECTR. For the calculating the parameters of the hyper-fine structure and partial contribution of the spectra components, we employed a standard procedure of

approximating integral spectra by a superposition of the components with a Lorentzian shape of the lines.

**Figure 1.** Mössbauer spectra and the distribution *p*(*V*) for MA samples of Series *B*. Treatment, composition: (**a**) ball milling (BM), composition 2 (Fe–14.2Cr + 20% Mn2N); (**b**) BM, composition 1 (Fe + 20% Mn2N); (**c**) BM, annealing at 800 ◦C, 1 h, composition 1 (Fe + 20% Mn2N). In the distributions *p*(*V*) a "deciphering" of the model of nitrogen-containing austenite is given.

**Figure 2.** Mössbauer data of the MA samples after BM of Series *A*: (**a**–**c**) the Mössbauer spectra and distributions of the *p*(*H*) of the sextet for the α phase; (**a'**) the Mössbauer spectra and distribution of the *p*(*V*) of the center of the spectrum; (**b'**,**c'**) distributions of the *p*(*V*) of the center of the spectra. Treatment, composition: (**a**,**a'**) BM, composition 1 (Fe + 20% CrN); (**b**,**b'**) BM, composition 3 (Fe–6.7Mn + 20% CrN); (**c**,**c'**) BM, annealing at 800 ◦C, 1 h, composition 4 (Fe–8.9Mn + 20% CrN).

**Figure 3.** Mössbauer data of the MA samples after BM of Series *B*: (**a**,**b**) the Mössbauer spectra and distributions of the *p*(*H*) of the sextet for the α phase; (**a'**) the Mössbauer spectra and distribution of the *p*(*V*) of the center of the spectrum; (**b'**) distribution of the *p*(*V*) of the center of the spectrum. Treatment, composition: (**a**,**a'**) BM, composition 4 (Fe–4.7Cr + 10% Mn2N); (**b**,**b'**) BM, composition 5 (Fe–8.6Cr + 10% Mn2N).

For verification of the results of the mechanical alloying and modeling of the spectra of the austenite of MA alloys, special experiments were performed on the samples of the compositions 1 and 2 of the *B* series: Fe–14.2Cr + 20Mn2N and Fe + 20Mn2N, see Table 2. Element formulas of the special compositions in the case of mechanical alloying permitted us to compare the obtained spectra with those well-known of the alloys produced by means of traditional metallurgy [26–36]. In the composition 1 from the series *B*, in accordance with the phase diagram borrowed from [21,22], in the case of MA one can expect the realization of the spectrum for paramagnetic nitrogen austenite, whereas in the composition 2 from the series *B*—one has the right to expect the realization of the spectrum of/from γ phase with a hyper-fine magnetic structure, the formation of which being the result of an anti-ferromagnetic (AFM) ordering in the Fe–Mn alloy with large (≥20%) content of manganese [29,32,33]. The results of calculation of the *p*(*V*) and spectra for the MA alloys of compositions 1 and 2 of (a) *B* series are shown in Figure 1a,b. The distribution *p*(*V*) reveals by itself a hyper-fine structure of the spectra of/from nitrided austenite, with that the spectra restored with the help of the data over *p*(*V*) describe the experiment well. Indeed, the spectra of/from the MA alloy of composition 1 of *B* series, judging by their shape and hyper-fine parameters agree well with the well-known data on the spectrum of/from nitrided paramagnetic austenite [16,30,31] (see Figure 1a and Table 3). In the case of samples of composition 2 from the series *B*, the measurements at room temperature show for the spectrum of/from the alloy the realization of broadening of the central singlet, which is characteristic of the AFM ordering in the FCC Fe–Mn austenite that contains manganese in quantity equal or greater (≥) than 20 wt% [29,32,34], Figure 1b and Table 1. To confirm the AFM origin of the broadening in the spectra of/from the MA alloy of composition 2 of series *B* there serves the emergence of the narrowing of the central singlet, which is due to the (i) partial decomposition of FCC solid solution and the (ii) exit of the manganese and nitrogen atoms from the matrix into nitrides, Figure 1c. When approximating the distribution *p*(*V*) by Gaussian forms, the model of the spectrum can be represented by the superposition of the components—*D*(0) + *D*(1) + *D*(2) + *S*AFM with the parameters

given in Table 3. The selection of the Gaussian forms for the approximation of the lines and the very modeling itself are based on a priori info about the spectrum of the stainless steel [34,35] and nitrogen austenite [16,30,31]. The doublet *D*(0) corresponds to the atoms of resonance-exhibiting iron without admixture of nitrogen, which is located in the octahedral interstitial positions—in the crystal lattice—of the first coordination shell (CS), and—by the magnitudes of the quadrupole shift and width of Gaussian forms—is close to the values of distribution (of probabilities of observation) of the electric field gradient and isomer shift for stainless steel [34,35]. The doublets *D*(1) and *D*(2) are of hyperfine parameters (namely, isomer shift *Is* and quadrupole shift *Q*s) in value close to the parameters of the doublets from intrusion of one and two (so-called dumbbell configurations) atoms of nitrogen into the octahedral interstitial positions in the crystal lattice of FCC iron [30,31]. The sextet *S*AFM with an hyperfine magnetic structure corresponds to a part of the formed γ phase with a large (≥20%) content of manganese [32,33]. The spectra and distributions *p*(*V*) are presented in Figure 1, and the parameters, in Table 3. Thus, special experiments with the samples of compositions 1 and 2 in the series *B* testify to the attainment of a desired mechanical alloying and to the possibility of modeling of spectra of MA alloys on the basis of models of Mössbauer spectra of nitrided austenite produced by traditional metallurgical methods [29–34].

**Figure 4.** Mössbauer data of the MA samples after BM and annealing of Series *A*: (**a**–**d**) distributions of the *p*(*H*) of the sextet for the α phase; (**b'**–**d'**) distributions of the *p*(*V*) of the center of the spectra. Treatment, composition: (**a**) BM, annealing at 700 ◦C, 1 h, composition 3 (Fe–6.7Mn + 20% CrN); (**b**,**b'**) BM, annealing at 800 ◦C, 1 h, composition 3 (Fe–6.7Mn + 20% CrN); (**c**,**c'**) BM, annealing at 800 ◦C, 1 h, composition 4 (Fe–8.9Mn + 20% CrN); (**d**,**d'**) BM, annealing at 800 ◦C, 1 h, composition 4 (Fe–8.9Mn + 20% CrN).

**Figure 5.** Mössbauer data of the MA samples after BM and annealing of Series *B*: (**a**–**d**) distributions of the *p*(*H*) of the sextet for the α phase; (**a'**–**d'**) distributions of the *p*(*V*) of the center of the spectra. Treatment, composition: (**a**,**a'**) BM, annealing at 700 ◦C, 1 h, composition 4 (Fe–4.7Cr + 10% Mn2N); (**b**,**b'**) BM, annealing at 800 ◦C, 1 h, composition 4 (Fe–4.7Cr + 10% Mn2N); (**c**,**c'**) BM, annealing at 700 ◦C, 1 h, composition 5 (Fe–8.6Cr + 10% Mn2N); (**d**,**d'**) BM, annealing at 800 ◦C, 1 h, composition 5 (Fe–8.6Cr + 10% Mn2N).

**Table 3.** Hyper-fine parameters of simulated spectra of the MA nitrogen-containing austenite, stainless austenite Fe–22Mn– 18Cr–0.8N steel, and nitrogen-containing γ iron.


Spectra of/from powders of the initial ferromagnetic alloys Fe–*X*Mn and Fe–*Y*Cr, as well as of the α phase in the body of MA samples, present by themselves the sextets with broadened lines which are a superposition of the subspectra *S*(*n*1, *n*2) correspondent to the non-equivalent surroundings of atoms of resonance-exhibiting iron by the impurities of chromium and manganese in the positions (lattice sites) of substitution in the solid solutions with BCC crystal lattice [26–28]. In result of the mechanical alloying with the formation of solid solutions, along with the sextet from the α phase, in the center of the spectra there appears a broadened asymmetrical singlet from austenite. In a first approximation (without taking into account the difference in the values of Debye–Waller factor of phase components), the numerical ratio of the areas under the integral sextet and central singlet corresponds—and is equal in volume percentages—to the quantitative ratio of the volumes of the α and γ phases in the MA alloy.

Quantitative assessment of the nitrogen content *CN* (in at.%) in austenite was carried out, similarly to that performed in [15,16,30] under the assumption of repulsive distribution (i.e., mutual repulsion) of nitrogen atoms in a solid solution, by the contribution of the atomic configuration represented by the *D*(1) doublet (Figure 1) from iron atoms with one nitrogen atom in the nearest octahedral interstitial positions. The value of the integral intensity from the doublet *D*(1) is related to *CN* by the relationship [30]:

$$S\_{D(1)} = 6p(1 - p)\_\prime$$

where *p* = *CN*(1 − *CN*) is the fraction of the octahedral interstitial sites in austenite occupied by nitrogen atoms. In order to account for the self-absorption effect, the intensity of *SD*(1) was extrapolated to the zero thickness of the absorber. The change in the integral intensity of the doublet *D*(1) can be explained by the self-transfer of nitrogen from nitrides to the solid solution of γ phase in the course of deformation-induced dissolution and vice versa during the decomposition of solid solution under conditions of nitrogen binding with chromium or manganese into next-emerging secondary nitrides [15,16].

Calculations of the spectra of/from the ferrite α matrix in the MA alloys and, in particular, of the quantity of impurity of substitution in this matrix were performed in the assumptions that (i) the contributions of chromium and manganese to the change in the isomer shift *Is* and effective field *H* at Fe atoms was additive in nature and (ii) the (binomial) distribution of impurities (namely, Cr and Mn) in the solid solution was of chaotic character [26–28,36]. Estimation of the total content of the chromium and manganese in the solid solution of ferrite matrix of the MA alloy, α*C*Cr + Mn, was performed via establishment of a partial contribution to the spectrum and distribution *p*(*H*) from the impurity-free atomic surroundings of resonance-exhibiting iron—i.e., from the sextet *S*(*n*1, *n*2) with *n*1 = *n*2 = 0, where *n*1 and *n*2 are the quantities (number) of atoms of chromium and manganese in the first and second CS, respectively. The sextet *S*(0,0) in the diluted Fe–Mn and Fe–Cr solid solutions is easily separated in the form of the peak in the distribution *p*(*H*) with magnitudes of fields in the vicinity of 330 kOe, see Figure 2a. Nitrogen in the ferromagnetic component of the MA alloy weakly affects (i) the hyper-fine parameters of the spectrum of α solid solution, as well as (ii) the value of the parameter of the mean effective field <*H*> [17]. Therefore, we present no quantitative estimations of the nitrogen content in the α solid solution of the MA alloy.

#### **3. Mechanical Alloying in the Course of the Ball Milling and Thermal Annealing**

*3.1. Mössbauer Analysis of Mechanical Alloying in the Cases of A and B Series*

In the case of samples of the series *A*: Fe–*X*Mn + 20CrN (Cr2N), the results for MA mixtures in the compositions with *X* = 0, 6.7, 8.9 wt% Mn are presented in Figure 2 and in Table 1. It is seen that an increase of the concentration of manganese in the initial matrix from 0 to 8.9 wt% leads to the growth of intensity of the central singlet of paramagnetic austenite from 6 to 36 vol.%. In its turn, in the singlet of/from the austenite, the doublets *D*(1) and *D*(2) are well pronounced, which testifies that (i) nitrogen «hits» into the octahedral interstitial positions in the crystal lattice of FCC phase and (ii) nitrogen-containing austenite is formed. In the spectrum for the α phase—being a ferrite component of the MA alloy, changes also are observed, namely, there takes place a pronounced lowering in the magnitude of the mean effective magnetic field <*H*> of the integral sextet at the expense of lowering of a partial contribution to the spectrum and *p*(*H*) of the superpositional sextet *S* (*n*1,*n*2) with *n*1 = *n*2 = 0, correspondent to the impurity-free surroundings of iron by the atoms of chromium and manganese [17,18,26,36]. The estimation of the content of impurity of substitution in the MA α solid solution testifies to the increase in the concentration of the

chemical elements of substitution in the matrix of α phase, see the results of BM in Table 1. Thus, based on the spectra results of the change—in result of ball milling (BM)—in the ferrite and austenite components in the samples of series *A*, the chemical elements from the initial nitrides CrN partially pass to the body of the α and γ solid solutions.

In the case of samples of the series *B*: Fe–*Y*Cr + 10Mn2N, the results for the MA mixtures are presented in Figure 3 and in Table 2. The spectra for the MA results have qualitatively the same "appearance" as in the case of series *A*. In the sextet that describes a ferromagnetic α solid solution, one can observe the decrease in the magnitude of <*H*> as a consequence of the enrichment of metallic matrix by manganese, i.e., the dissolution of nitrides has taken place and passing of manganese atoms from the nitrides to the α matrix of the MA alloy has occurred in parallel. However, a noticeable quantitative distinction of the experimental results exists relative to those for the series *A*.

This, first of all, concerns the increase (i) in the volume of austenite, as well as (ii) of the degree of alloying of the α phase by manganese. In the case of the samples of the series *B* with the composition formulas 3 and 4 (see Table 2), the ball milling preceding sample compaction brings about the quantity of austenite from 10 to 24 vol.%, whereas in the case of the samples of the series *A* with the composition formulas from 1 to 3 (see Table 1) the processing mentioned entails the increase in the quantity of austenite component from 4 to 7 vol.%. At the same time, the content of alloying elements of substitution in the ferrite matrix, α*C*Mn+Cr, increases in a more pronounced extent in comparison with that one which is characteristic of the case of series *A*. Besides, in the central singlet of/from austenite the component related with the AFM ordering is distinguished to stand obviously apart, see Figure 3a,c. This testifies to the formation in the γ-phase of the regions enriched in manganese, i.e., having the manganese content greater than that is characteristic of the average for the composition. The values of the parameters for the sextets "typical" of the AFM ordering are presented in Table 3. The content of nitrogen in austenite for the series *A* and *B* amounts from 0.7 wt (2.6 at.) to 3.7 wt (13.1 at.) %.

#### *3.2. Mössbauer Analysis of MA Samples after Their Annealing*

In the case of samples of the series *A*, the result of an annealing of the MA mixtures at 700 ◦C consists in the reverse evolution of the spectra with considerable lowering of the intensity of the central singlet of/from austenite, Figure 4a,c. The content of substitution impurity in the α solid solution approaches the initial—or a lower one than that is characteristic of the initial binary Fe–*X*Mn alloy, Table 2 *a*. In particular, in the composition Fe–8.9Mn + 20CrN, after the MA and annealing at 700 ◦C, for 1 h, the total content of the manganese and chromium in the α phase does not exceed (i) 1.7 wt% at 8.9 wt% Mn in the initial binary alloy and (ii) 10.0 wt% in the MA alloy after BM. The quantity 36 vol.% of austenite remains virtually the same (i.e., unaltered), but the content of nitrogen exhibits a lowering from 1.1 wt (4.2 at.) % to 0.5 wt (1.8 at.) %. An annealing at a higher temperature, at 800 ◦C, for 1 h, facilitates the occurrence of the α → γ transition in the samples of compositions 3 and 4 with *X* = 6.7 and 8.9 and makes lower the degree of decomposition of the solid solutions formed in the course of MA. In samples of all the compositions of series *A*, an annealing at a higher temperature preserves the values of <*H*> and of the intensity of *S*(0,0) component of the sextet of/from the α phase from alteration, i.e., the annealing stabilizes the chemical elements of the substitution in the α matrix. The intensity of *D*(1) decreases in a lesser extent, i.e., nitrogen is partly preserved in the solid solution of paramagnetic austenite, see Figure 4b,d and Table 1 *b*.

In the case of series *B*, the annealing of the MA samples differs, first of all, by increasing the quantity of austenite in the cases of all compositions, see Figure 5 and Table 2. After annealing at 800 ◦C of the samples of composition 5 with *Y* = 8.6 wt%, up to 100% of austenite is formed. In the austenite(-stipulated) singlet of/from the MA alloy one can observe the preservation of the component related to the AFM ordering, which is described by the sextet with the combined magnetic and quadrupole splitting, see Table 3. The same but in a better way as it takes place in the series *A*, annealing at 800 ◦C stabilizes the elements of mechanical alloying in the ferrite matrix; the comparison of the results of annealing of samples of the compositions in the series *A* and *B* is shown in Table 2.

#### *3.3. TEM Data on the Results of the MA and Subsequent Annealing*

According to the TEM data on the MA alloyed powder mixtures *A* and *B*, the same as in the case of mixtures based on the Fe–Ni matrix [18], one observes there the formation of a submicrocrystalline structure (SMC). In Figure 6 the dark-field image and SAED pattern of the powder mixture Fe–8.6Cr + 15Mn2N are presented. The SAED pattern (Figure 6b) contains ring (circular-wise disposed) reflections from greatly misoriented grains of austenite. According to [37], reflections from the nitrides CrN and Cr2N are located near the reflections from austenite. In the dark-field image taken in the reflection «austenite + nitrides» one can observe the images of the nitrides and fragments of austenite. The size of the fragments of matrix (grains and subgrains) amounts to 50–80 nm. Also visible are more disperse (finer) precipitates of size about 2 nm, which correspond to the secondary nitrides of the types CrN and Cr2N.

**Figure 6.** TEM structure of the MA powder mixture Fe–8.6Cr + 15Mn2N: (**a**) dark-field image taken in the combined reflection ((111)<sup>γ</sup> + CrN, Cr2N); (**b**) SAED pattern.

After termination of the annealing at 800 ◦C for 1 h, an SMC structure of the powder mixture remained unaltered, see Figure 7. The phase composition of the powder mixture Fe–8.6Cr + 15Mn2N presents by itself—after the milling and annealing—the austenite and chromium nitrides (CrN and Cr2N). Annealing of the MA samples leads both to an increase in the size of chromium nitrides, which have been formed in the course of mechanical alloying, and to their precipitation at the dissolution of supersaturated γ solid solution. In the SAED pattern shown in Figure 7a, one can see two first reflection rings that correspond to nitrides of chromium (CrN and Cr2N). In the dark-field image taken in the reflection «CrN + Cr2N», (Figure 7b), it is these nitrides that shine.

**Figure 7.** TEM structure of the MA powder mixture Fe–8.6Cr + 15Mn2N after annealing at 800 ◦C for 1 h: (**a**) SAED pattern; (**b**) dark-field image taken in the combined reflection (CrN, Cr2N).

#### **4. Discussion**

Using as metallic matrix precisely the Fe–Mn and Fe–Cr alloys instead of the pure metals Fe, Mn, and Cr with admixture of nitrides, has its own peculiar features that affect the (i) mechanism of structure-phase transitions, (ii) kinetics, and (iii) properties of the products of MA. This influence is stipulated by the physicochemical properties of the components under alloying and, in particular, by the phase diagram of the alloys and nitrides, as well as (by) the difference in the chemical activity and diffusion mobility of the elements of alloying (Cr, Mn, and N), in the course of the MA and subsequent annealing.

As it follows from the results of Mössbauer investigations, in the samples of series *A* and *B* one can observe—under BM—the smelting of the initial binary Fe–Mn and Fe–Cr alloys, on the one hand, and the nitrides CrN and Mn2N, on the other. Experiments on the mechanical alloying in ball mills—used in the capacity of metallic matrix the alloys, —are of common regularities, with SPD-induced (by severe plastic deformation) structurephase transformations upon the friction- and HPT-providing treatments executed on the nitrided surface of the iron alloys and stainless steels [15–17]. In the course of severe deformation, in MA samples we encounter the formation of solid solutions with the element and phase composition similar to that typical of Fe–Cr–Mn–N alloys produced in a traditional metallurgical way, see data of Tables 1 and 2. The ferromagnetic component having BCC crystal lattice also exhibits a change in its composition by revealing the altering of the quantity of the manganese and chromium in the α solution. In result of an annealing at 700 and 800 ◦C of the MA samples, both in the series *A* and in the series *B*, there occurs the α → γ phase transition accompanied by (i) the decomposition of the formed supersaturated solid solutions and (ii) by the exit of nitrogen and of the alloying elements manganese and chromium from the metallic matrices into secondary nitrides, see Tables 1 and 2 *a*, *b*. Moreover, at BM and annealing, in the processes of decomposition of MA solid solutions there participate the chemical elements of an alloying (by chromium and manganese) of the initial binary alloys. After annealing, the content of the chromium and manganese in the α matrix of MA alloy is noticeably decreased in comparison with the content of these elements in the alloys of initial composition. For instance, in the case of samples of composition 4 (Fe–8.9Mn + 20CrN) of *A* series, the total content of substitution elements in the α phase, α*C*Mn+Cr, decreases to 1.7 wt%, whereas the concentration of nitrogen in austenite does not exceed 0.5 wt (2 at.)%. In the case of series *B*, at 700 ◦C a qualitatively similar pattern (picture) is observed, see data for the samples of compositions 4 and 5 in Table 2 and in Figure 5a,c. We can assume that active exit of substitution elements from the α matrix is connected not only with the decomposition of supersaturated solid solutions and the formation of nitrides, but also with the diffusion exchange of the chromium and manganese atoms between the α and γ phases in accordance with the phase diagrams of these alloys. Moreover, the nanostructurization of metallic matrix in the course of BM contributes to the formation of segregations of the chromium, manganese, and nitrides on along the developed grain boundaries [38,39]. The structure of the nitrogen-containing MA austenite after the thermal annealing and cooling to room temperature experiences a better stabilization precisely after annealing at a higher temperature, namely, of 800 ◦C, see Tables 1 and 2 *a*, *b*. It can be seen from the Tables that the stabilization of nitrogen-containing austenite at the annealing of higher temperature is explained by the presence of chemical elements of mechanical alloying that have finally remained in the body of a volume of the solid solution.

The data of TEM have confirmed the validity of results of Mössbauer measurements concerning the formation of nitrogen-containing austenite together with the secondary fine nitrides (CrN and Cr2N) in the course of MA of the powder mixtures Fe–Mn–*X*CrN and Fe–Cr–*Y*Mn2N. At high-temperature anneals there occurs both the increase in the size of chromium nitrides that have been formed at mechanical alloying of powder mixtures and the precipitation of new fine ones during the decomposition of supersaturated γ solid solution. The preservation at high-temperature annealing of the submicro- and nano-sized austenite grains in MA alloys can be in part explained by the formation of the secondary nitrides CrN and Cr2N along the developed grain boundaries, which evolve in parallel of the α → γ phase transition, slows down the process of recrystallization-controlled growth of grains, see Figure 7a,b.

The alloys obtained by the MA and subsequent annealing at 700 and 800 ◦C, are metastable and at cooling to room temperature they can pass by the mechanism of shear transformation just to the α phase. The small size of austenite grains [40], short-range-order atomic concentrational separation [41], and precipitation of disperse nitrides coherently matched with the matrix [42], all this play, in the case under consideration, a stabilizing role in relation to the occurrence of the γ → α transition. An additional stabilization of austenite can be connected with the presence of the concentrational inhomogeneity over a material. The stabilizing role of concentrational inhomogeneities was exemplified on the austenitic Fe–Ni alloys [43]. The formation of the concentrational inhomogeneities over manganese in the austenitic MA samples of series *B*, and more specifically, of the regions of the γ phase with an increased content of magnesium, appears both after ball milling and after annealing right in the form of the component related with the AFM ordering, see Figures 1c and 5b. In the spectrum for the alloy Fe–4.7Cr + 10Mn2N annealed at 800 ◦C, for 1 h, these regions of a structure are described by the sextet with a combined magnetic and quadrupole splitting, see Figure 5b,d and Table 1.

From the results of the BM and anneals of samples from the series *A* and *B*, it follows that the series *B* turns out to be more efficient from the point of view of the volume of a stable austenite being under formation at the lesser (i) quantity *of* alloying elements (*of* chromium, manganese, and nitrogen) and (ii) amount of the primary nitrides in the samples of these compositions, see Table 2. A more efficient formation of the nitrogencontaining austenite and its stabilization in the samples of series *B* can be explained by several reasons.

First, (it can be explained) (i) by a more efficient dissolution of Mn2N nitrides in a metallic matrix and (ii) by the formation of the Mn-rich austenite-forming regions in a metallic matrix of MA alloy with the concentration of manganese ~20 at. %. Manganeserich regions of austenite can be formed by the actual smelting of nitrides both with austenite and with a paternal α phase. A more active alloying of the initial matrix in the samples of series *B* is well illustrated by data from Tables 1 and 2. The manganese-rich regions of the α phase in this very case possess an enhanced stimulus to/for the α → γ transition in accordance with the phase diagram (in) [21,22]. Accommodation stresses between (i) the γ phase enriched in manganese during its formation in the course of MA treatment and (ii) the rest matrix, stabilize austenite in relation to the γ → α transition [42,43].

Second, the high efficiency of the formation of austenite in the samples of series *B* can be explained by a low diffusion mobility of the atoms of manganese (having a larger atomic radius in comparison with that of chromium), which (i) makes weaker the process of dynamic aging (at BM) concurrent with the process of dissolution of nitrides and (ii) facilitates the preservation of manganese to remain in the solid solution. The (in-case) presence of the dynamic aging concurrent with the dissolution is obvious from the slight chromium doping of the initial Fe–*X*Mn matrix, see the result of BM in Table 1 that presents data for the case of series *A*. The accelerated concurrent process in the samples of series *A* presumably is explained by the enhanced chemical and diffusion activity of chromium. It is noticeable that it is in the case of series *B* there more actively occur both dissolution of Mn2N nitrides and alloying of the α phase by manganese, see Table 2. The formation of the regions enriched in manganese is most distinctly revealed in conditions of the BM at relatively moderate temperatures. Synergetic nature of deformation-induced processes, which manifests itself in the decomposition of supersaturated solid solutions in the course of concurrent mechanical alloying, was demonstrated in the works on the short-range order investigations into the Fe–Ni and Fe–Cr alloys [44,45]. The influence of the diffusion mobility of the alloying elements of substitution on this process was shown in the works on the dynamic aging in the Fe–Ni–*M*e alloys where *M*e = Ti, Al, Si, Zr [44,46,47].

The third reason for the high efficiency of the formation of austenite in the case of series *B* consists in the lower thermal stability of Mn2N nitrides in comparison with that of the nitrides CrN (Cr2N) in conditions of the local heating-up of the mixture at the BM and high-temperature anneals. According to the data from [48], the nitride Mn2N undergoes decomposition at 900 ◦C, while the nitride CrN decomposes at sufficiently higher temperatures (>1000 ◦C). The effect of the thermal stability and chemical activity of the constituents of would-be MA products on the kinetics of the dynamic aging and formation of secondary phases was demonstrated in the example of deformation-induced dissolution of nitrides, as well as intermetallics in [44,46,47].

#### **5. Conclusions**

Using the methods of Mössbauer spectroscopy and electron microscopy, we have investigated the structure-phase transitions in the course of solid-phase mechanical alloying in a ball mill and at subsequent annealing of the mixtures of the Fe–Cr and Fe–Mn alloys, on the one hand, and the nitrides CrN (Cr2N) and Mn2N, on the other. The samples from the series *A*, of compositions Fe–*X*Mn + 20CrN (Cr2N), and the samples from the series *B*, of compositions Fe–*Y*Cr + 10, 20Mn2N, —were the objects of investigation. It is shown that the ball milling of the mixtures from the series *A* and *B* leads to the smelting of the components of mixtures and to the formation of the Fe–Cr–Mn–N solid solutions with FCC and BCC crystal lattices. Increasing the concentration of the elements Cr and Mn in the initial BCC alloys, as well as the proportion amount of nitrides in the studied compositions in conditions of the ball milling and subsequent annealing, is accompanied by the increase in the volume of forming austenite with nitrogen concentration of 3.7 wt (13.1 at.)%. A result of the mechanical alloying of samples from the series *B* is the formation of the FCC phase, part of which is of AFM ordering development with the content of manganese equal or greater (≥) than 20 wt%.

The process of the ball milling and subsequent isothermal annealing of the MA mixtures correspondent to the series *A* and *B* is accompanied by the α → γ phase transitions, formation of nitrogen-supersaturated solid solutions, and their decomposition with the formation of secondary nitrides. Austenite formed in the course of ball milling and subsequent annealing at 700 and 800 ◦C, for 1, has submicro- and nano-sized grains, which along with the coherent disperse (fine) secondary nitrides and alloying elements are preserved in the matrix, which is a reason for the stabilization of austenite. It has been established experimentally that annealing of MA samples at 700 ◦C leads to a sharp decrease in the amount of the alloying elements of manganese and chromium in the α phase, which along with the decomposition and formation of secondary nitrides can be a consequence of diffusion of chromium and manganese from the volume of the α into γ phase. It is shown that the *B* series—in its samples with the primary nitrides Mn2N—is both at ball milling and at subsequent annealing more efficient from the viewpoint of forming the greater volume of MA austenite and its saturation with nitrogen at a lesser—than is characteristic of the case of the series *A*—content of the alloying chromium and manganese. The higher stability of MA austenite in the case of the series *B* is explained by the (i) formation of concentration-inhomogeneous regions of the matrix enriched in austeniteforming manganese, (ii) thermal stability of Mn2N nitrides being lower in comparison with that of CrN (Cr2N), and (iii) low diffusion mobility of manganese in the course of the ball milling and subsequent annealing.

**Author Contributions:** Conceptualization, V.S.; methodology, V.S., E.N., K.L., A.Z., N.K. and K.K.; validation, V.S., K.L., N.K. and K.K.; formal Analysis, V.S., A.Z., N.K., K.L. and K.K.; investigation, K.L., K.K. and E.N.; writing—original draft preparation, V.S. and K.L.; writing—review and editing, Y.U., K.L. and K.K.; project administration, V.S.; funding acquisition, K.L. and V.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** The reported study was funded by the Russian Foundation for Basic Research (project no. 19-33-60006). The research was carried out within the state assignment of Ministry of Science and Higher Education of the Russian Federation (theme "structure" no. AAAA-A18-118020190116-6.

**Acknowledgments:** The authors are grateful to Gennady Dorofeev for the providing of nitride powders and their attestation.

**Conflicts of Interest:** The authors declare no conflict of interest.

## **References**


## *Article* **Fe-X-B-Cu (X = Nb, NiZr) Alloys Produced by Mechanical Alloying: Influence of Milling Device**

**Albert Carrillo, Joan Saurina, Lluisa Escoda and Joan-Josep Suñol \***

Department of Physics, University of Girona, C/Universitat de Girona 3, 17003 Girona, Spain; albert.carrillo.berlanga@gmail.com (A.C.); joan.saurina@udg.edu (J.S.); lluisa.escoda@udg.edu (L.E.) **\*** Correspondence: joanjosep.sunyol@udg.edu; Tel.: +34-872418757

**Abstract:** In this work, we analyze the influence of the milling device in the microstructural evolution of two Fe-X-B-Cu (X = Nb, NiZr) alloys produced by mechanical alloying (MA). The two milling devices are a planetary mil (P7) and a shaker mill (SPEX 8000). Microstructural analysis by X-ray diffraction detects the formation of a Fe rich solid solution. In the Fe-Nb-B-Cu alloy produced in the shaker mill also appears a Nb(B) minor phase, whereas in the Fe-NiZr-B-Cu alloy produced in the planetary mill, a minor disordered phase is formed. The comparative study regarding the energy transferred per unit of time in both devices determines that the shaker mill is more energetic. This fact explains that in the Fe-Nb-B-Cu alloy, Nb has not been introduced in the main Fe rich phase, whereas in the Fe-NiZr-B-Cu alloy milled in the shaker mill was formed the highly disordered phase. With regard to thermal analysis, the values of the apparent activation energies of the main crystallization process (above 200 kJ/mol) correspond to the crystalline growth of the nanocrystalline Fe rich phase.

**Keywords:** mechanical alloying; microstructure; Fe based alloys

**Citation:** Carrillo, A.; Saurina, J.; Escoda, L.; Suñol, J.-J. Fe-X-B-Cu (X = Nb, NiZr) Alloys Produced by Mechanical Alloying: Influence of Milling Device. *Metals* **2021**, *11*, 379. https://doi.org/10.3390/met11030379

Academic Editor: Tomasz Czujko

Received: 21 January 2021 Accepted: 21 February 2021 Published: 25 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

### **1. Introduction**

Nanocrystalline soft magnetic alloys have been obtained by rapid solidification techniques as melt-spinning, MS, (ribbon shape) or by mechanical alloying, MA (powder shape) [1,2]. Melt-spun ribbons are usually amorphous and further annealing is needed to develop nanocrystalline structure by primary crystallization [3]. Mechanical alloying favors the direct formation of powdered nanocrystalline alloys, that can be compacted/sintered to produce bulk pieces [4].

The nanocrystalline alloys can be classified based on the crystallographic structure of the nanocrystalline phase. The bcc Fe based alloys developed by Suzuki and coworkers are known as Nanoperm [5]. Nanoperm alloys, particularly nanocrystalline Fe-(Zr, Nb)- B-Cu alloys, have good magnetic properties: High effective permeability (~20,000) and low coercivity (<3 A/m) and core loss (<100 W/kg) at room temperature. There are soft magnetic offering reduced core losses over a wide range of applied frequencies. These alloys are an alternative for industrial (automotive, green energy) applications: High precision energy meters, absorber cores for suppression of motor bearing currents, high frequency current transformers, push-pull transformers for power transistors, and current sensors [3,4,6]. Due to magnetocaloric effect of the ferromagnetic to paramagnetic transition, these materials have also been of interest for magnetic refrigeration [7]. The composition is a key factor. It is known that an excessive content of B favors the formation of an amorphous phase, as it was found in Fe75Nb10B15 [8], the increase of the coercivity and a decrease of the magnetization of saturation [9].

In the mechanical alloying process, the processing parameters influence the final microstructure and the functional properties. Milling parameters include the milling intensity, time and atmosphere, the ball-to-power weight ratio (BPR), the addition of a process control agent (PCA), and the filling factor of the vials. Some parameters, such as PCA, are linked to grain refinement [10], but the majority has a high influence in the energy transferred to the powders [11].

In this work, two Nanoperm alloys (Fe80Nb7B12Cu1 and Fe80(NiZr)7B12Cu1) have been produced in two milling devices (shaker mill, planetary ball mill) to detect their influence on the nanocrystalline bcc Fe rich phase formation. In the SPEX 8000 shaker mill, the jar containing powders is agitated at a high frequency in a complex cycle that involves motion in three orthogonal directions (combining back-and-forth swings with short lateral movements, each end of the vial describing a figure like an eight). In the P7 planetary mill, two jars are arranged eccentrically on a platform (opposite direction of rotation between platform and jars). Furthermore, crystallographic microstructure (after 80 h MA) was discussed by energy transfer models.

#### **2. Materials and Methods**

Mechanical alloying has been performed in a Spex 800 shaker mill (Spex, Metuchen, NJ, USA) and in a Fritsch Pulverisette 7 planetary mill (Fritsch, Idar-Oberstein, Germany). The experiments were performed in jars sealed inside an Ar-filled glove box to prevent oxidation. The milling (10 g of powders) was performed in both devices with Cr-Ni steel jars (45 mL capacity) and balls (7 balls, 12 mm diameter) for 5, 20, and 80 h. The initial filling factor was close to one half. In SPEX, the information by the provider is that the number of cycles (rotational speed or frequency of rotation) is 875 rpm, whereas in the P7, the selected cycles of the platform, Ω, and jars, ω, are the same, 600 rpm. The R radius of the platform (measured from the geometric center of the jar position to the center of the platform) of the P7 is rp = 7 cm and the internal radius of the jars rj = 1.7 cm. The internal radius of the SPEX jar is 2.1 cm. The ball radius rb is 0.6 cm in the Spex and 0.5 in the P7. The ball to powder weight ratio, BPR, used in both devices has been the same: 5.

The nominal composition of the two produced alloys were Fe80Nb7B12Cu1 and Fe80(NiZr)7B12Cu1 (at.%), labelled as A and B, respectively. Thus, depending on the mill device, the samples are labelled as: P7—A, SPEX—A, P7—B and SPEX—B. Elemental Fe, Nb, B, Cu powders, and NiZr compound (Sigma-Aldrich, Saint Louis, MO, USA) with high purity (<99.5%) and low particle size (Fe and Cu ~ 10 μm, B, B and Nb ~ 10 μm, NiZr < 100 μm) are mixed to obtain the desired composition. Likewise, NiZr compound was selected to avoid undesired oxidation of Zr.

The morphology, microstructure, and thermal behavior evolution as a function of milling time has been analyzed by scanning electron microscopy (SEM) coupled with compositional EDS spectroscopy microanalysis, X-Ray diffraction (XRD), and differential scanning calorimetry (DSC). SEM observations and EDS microanalysis were performed in a D500 device (Siemens/Bruker, Billerica, MA, USA); at least five micrographs/microanalyses for samples milled for 80 h. Thermal analysis was performed in a DSC822e calorimeter (Mettler-Toledo, Columbus, OH, USA); one experiment at 20 K/min for samples milled for 5 and 20 h, and four experiments for samples milled for 80 h. XRD experiments (before milling and after 5, 20, and 80 h of milling) were performed in a D8 Advance diffractometer (Siemens/Bruker, Billerica, MA, USA) using Cu-Kα radiation. XRD patterns analysis was performed by applying Rietveld refinement and the free software MAUD (Maud, Trento, Italy) [12].

## **3. Results and Discussion**

#### *3.1. Microstructure and Thermal Analysis*

The XRD diffraction patterns of alloys A and B are given in Figures 1 and 2, respectively. The main component is elemental Fe in both alloys. Thus, as expected, the main phase (higher reflections intensity) in all the diffraction patterns is the bcc Fe phase. The reflection peaks of this phase are (110), (200), (211), and (220). The minor peaks before milling correspond to Cu, Nb, and NiZr compound. Initial elemental B was amorphous. These reflections disappear after milling. The minor elements were introduced in solid solution in the bcc Fe rich phase, or remains in the grain boundaries [8,9]. Regarding the shape of the

peaks, its width has increased considerably with the milling time, phenomena associated to a decrease in the crystalline size. The parameter of the goodness of fit, GOF, obtained from Rietveld refinement is lower than 1.25 for all XRD patterns. Regarding the shape of the peaks, its width has increased considerably with the milling time that seems to indicate a decrease in the crystalline size.

**Figure 1.** XRD (X-Ray diffraction) patterns of alloy A (Fe80Nb7B12Cu1). In the P7: (**a**) Before milling (0 h) and after milling ((**b**) 5 h, (**c**) 20 h, (**d**) 80 h); or in the Spex mill: (**e**) Before milling (0 h) and after milling ((**f**) 5 h, (**g**) 20 h, (**h**) 80 h). Bcc Fe rich solid solution (+), Nb phase (X), Nb(B) phase (#).

**Figure 2.** XRD diffraction patterns of alloy B (Fe80(NiZr)7B12Cu1). In the P7: (**a**) Before milling (0 h) and after milling ((**b**) 5 h, (**c**) 20 h, (**d**) 80 h); or in the Spex mill: (**e**) Before milling (0 h) and after milling ((**f**) 5 h, (**g**) 20 h, (**h**) 80 h). Bcc Fe rich solid solution (+), disordered phase (O).

If one focuses in the XRD patterns after 5 h of milling in alloy A (Figure 1a), the Nb peaks are still present. With the peak (110) of the Nb, it was verified that, at its lower angular reflection, we began to notice an asymmetry. The result of this asymmetry was clearer after 20 h of milling (in sample A), in which pattern appears a double peak in the place where before there was only one. Whereas in sample B, there only appears the new low angular phase. The proximity of the two peaks seems to indicate a mutual relationship. This new phase forcibly has a cell parameter slightly higher than that of the elemental Nb since it is shifted to lower angular positions. In short, asymmetry that began to be present at 5 h was an incipient peak of a new phase of Nb(B), which becomes more evident at 20 h. The creation of this phase is favored by the negative mixing enthalpy between Nb and B (−39 kJ/mol), indicating an exothermic process, which has also been detected in the scientific literature [13].

The formation of borides (undesired phases due to the associated loss of improved soft magnetic behavior) was not detected. After 80 h of milling (A alloy), in SPEX only the Fe rich bcc solid solution was detected, whereas a minor amount (2.1%) of the mechanically induced Nb(B) phase remains in alloy A milled in the P7 device.

The analysis of the XRD patterns of alloy B shows significative differences. The main phase is the bcc Fe rich phase and the reflections of minor phases disappears during milling. It should be remarked that the NiZr phase has not been detected after milling (probably the integration of Zr in the NiZr compound favors the integration of both elements in the main solid solution. From 80 h MA XRD diffraction patterns analysis, it was found that in

one case (milling in P7) there is only the majority solid solution; while in the other case (milling in SPEX), a highly disordered phase appears (about 3.2%). This phase has been previously detected [10] in alloys with Zr and is probably due to an incipient amorphous Zr rich phase.

The main parameters derived from the Rietveld refinement (Maud software); the lattice parameter, a, the crystalline size, L, and the microstrain, ε, were given in Tables 1–3, respectively (as a function of the milling time). The values are calculated with the isotropic model. Thus, they are calculated from refinement weighted analysis of the reflections of the bcc Fe rich phase, and the error is associated to both the isotropic model and the mixed refinement of all parameters.


**Table 1.** Lattice parameter, a (Å), of the cubic bcc Fe main phase.

**Table 2.** Crystalline size, L (nm), of the cubic bcc Fe main phase.


**Table 3.** Microstrain, ε (%), of the cubic bcc Fe main phase.


The milling process favors the increase of the lattice parameter. This effect is due to the introduction of mechanically induced crystallographic defects (vacancies, dislocations) and to the solid solution of minor elements in the bcc phase. The highest value, 2.8832 ± 0.0003 Å, corresponds to alloy P7—A milled for 80 h. The lower values are found in samples without an additional second minor phase. Probably, these low parameters are due to the no solid solution of Nb and Zr (a high amount of Nb and Zr atoms remains in the Nb(B) and disordered phases, respectively).

The milling process favors the decrease of the crystalline size and the increase of the microstrain. It should be remarked that low crystalline size is calculated in those alloys (P7—A and SPEX—B milled for 80 h) with a minor secondary phase. One explication is that the Nb(B) and the disordered Zr rich phases formation favors the intra-division of the highly deformed crystals. The increase on the microstrain is due to the mechanically induced crystallographic defects.

The density of dislocation (of the bcc phase) was calculated (see Table 4) from crystallographic parameters (from Tables 2 and 3) by applying the following equation [14]:

$$
\mathfrak{p} = \mathfrak{Z} \times \sqrt{\mathfrak{Z}\left(\mathfrak{e}/\mathrm{Lb}\right)},
\tag{1}
$$


**Table 4.** Dislocations density, ρ (1015 m<sup>−</sup>2), of the cubic bcc Fe main phase.

In this equation, b is the Burgers vector. For the bcc crystallographic phase, the direction of easy dislocations concentration is (1 1 1). In this case, the Burgers vector is:

$$\mathbf{b} = \mathbf{a} \sqrt{3}/2,\tag{2}$$

The general trend, as expected, it is the increase of the dislocations density as increasing the milling time. Higher values are calculated in those samples milled (80 h) in the planetary mill. The results are close to the expected limit of the dislocation density (plastic deformation) in bcc Fe (10<sup>16</sup> m−2) [15]. It should be remarked than XRD methods sometimes can provide higher values of the dislocations density than transmission electron microscopy (TEM) analysis [16]. TEM values are considered as more accurate due to direct local observation. Nevertheless, the milling process favors inhomogeneity in the crystallographic defects.

The morphology of the powders after 80 h of milling was checked with SEM micrographs (as shown in Figure 3). There are rounded particles with smooth shape.

**Figure 3.** SEM micrographs after 80 h milling of samples: (**a**) P7—A, (**b**) SPEX—A, (**c**) P7—B, and (**d**) SPEX—B.

In order to analyze the particle size and its distribution, the particle size of several micrographs of each sample have been measured. The results are shown in Figure 4. To facilitate the comparison, the results were fitted with a log-normal distribution with the mean, μ, and the shape parameter (square root of variance), σ, values are given in Table 5.

**Figure 4.** Particle size frequency distribution after 80 h milling of samples: (**a**) P7—A, (**b**) SPEX—A, (**c**) P7—B, and (**d**) SPEX—B.

**Table 5.** Parameters (mean, μ, and shape parameter, σ) of the log-normal distribution.


The higher mean value was found in sample SPEX—A, whereas the high shape parameter values are found in B samples. No correlation has been detected between these parameters (micrometric scale) and the XRD results (nanometric scale).

Likewise, compositional analysis was performed (80 h of milling) with EDS microanalysis. The measurements indicates that the contamination from the milling tools is lower than 1.5 ± 0.5 at.%; typical results coherent with those previously found [10,16]. Oxygen is also found, probably favored by the high surface area of the particles in contact with air after the milling process. Values are lower than 2.5 ± 0.5 at.%.

Additional thermal analysis has been performed from DSC scans. The analysis of the evolution of the thermal behavior has been carried out after 5, 20, and 80 h of milling at 10 K/min as the heating rate. The DSC scans (5 to 80 h of milling) of A and B samples are given in Figures 5 and 6, respectively. DSC after 80 h of milling were reported on ref. [17]. Here we discuss the process evolution and the kinetic analysis.

**Figure 5.** DSC (differential scanning calorimetry) scans of alloy A after milling (5, 20, 80 h) in the P7 (**a**) or in the Spex (**b**) mill.

**Figure 6.** DSC scans of alloy B after milling (5, 20, 80 h) in the P7 (**a**) or in the Spex (**b**) mill.

In all scans, multiple exothermic processes are observed in wide intervals of temperature and with a certain degree of overlap. Exothermic processes at lowest temperature (below 400 ◦C) were associated with relaxation phenomena (caused by a decrease of the free volume after relaxation of the mechanically induced tensions at the micro and nanoscale). Furthermore, thermal treatment also favors the reduction of the crystallographic defects and local inhomogeneity by atomic diffusion. The exothermic processes at higher temperatures were associated with crystallization (nucleation and/or crystalline growth). This complex behavior is typical of nanocrystalline alloys produced by mechanical alloying [9,18]. In all DSC at 80 h, the main peak is remarked with an arrow. The peak temperature is higher (Tp > 500 ◦C) in samples milled with a shaker mill if compared with the temperature of the samples milled in the planetary mill (Tp < 500 ◦C). Probably, higher energetic milling favors higher internal local temperature and atomic diffusion [18].

On the other hand, the apparent activation of the crystallization processes (those remarked with an arrow) has also been calculated by applying Kissinger's linear method at heating rates of 10, 20, 30, and 40 K/min. Kissinger's method is based on the relationship between the peak temperature, Tp, and the scan rate, β. The linear fitting is shown in Figure 7, and the apparent activation energies are given in Table 6. The activation energy is determined from the slope of the linear fit, by applying the following relation (R is the universal gas constant):

$$\mathbf{E} = -\mathbf{R} / \text{(1000 T\_p)},\tag{3}$$

**Figure 7.** Kissinger linear fitting to calculate the activation energy of the crystallization process (experimental point for each sample are linked to peak temperatures of samples determined at 10, 20, 30, and 40 K/min from right to left respectively). Coefficients of correlation R-squared > 0.9.



The activation energies are higher in samples milled in the shaker mill, and a possible hypothesis is that it was favored by different milling energy transfer (higher energy can favor diffusion and the homogeneity of the alloys, stabilizing against crystalline growth). Local inhomogeneities can act as preferential sites for crystalline growth, reducing the activation energy. An alternative hypothesis is the content of oxygen, higher (~0.5–0.8 at.%) in powders milled in the shaker mill. More work will be provided to reach a definitive explanation. Likewise, the energy of activation values were characteristic for each type of reaction [19]. According to the bibliography, the activation energy for the crystal growth of Fe based nanocrystalline samples are around 140 kJ/mol [20] for pure iron. In samples containing additional elements, these energies can have higher/lower values. In Fe-Co-Nb-B alloys obtained by MA, energies between 238 and 261 kJ/mol have been calculated [21]. For amorphous samples in which a new phase must be nucleated, energies have values and are around 300 kJ/mol [22] or 365 kJ/mol [23], or even 425 and 550 kJ/mol in Fe-Nb-B alloys [24].

Thus, all of the analyzed crystallization process can be associated to the crystalline growth of the pre-existing Fe rich solid solution.

#### *3.2. Mills Comparison*

If we compare the phases detected by XRD after 80 h of milling, it is found that:


The results indicate that probably the milling is more energetic and efficient in SPEX. In one case, at 80 h in the P7, it was not possible to obtain a single phase (it would surely be necessary to increase the milling time to promote the Nb(B) phase disappearance and the integration of the Nb, via solid solution or in the grain boundaries). In the other, in the SPEX, the more energetic milling favors the formation of a disordered phase.

Likewise, from DSC scans we can state that the higher thermal stability (of the analyzed crystallization process) was found in the samples milled in the shaker mill. The criteria for a higher thermal stability of the nanocrystalline phase were high temperature of transformation and high activation energy.

A comparative study was performed to analyze the transfer of energy in both milling devices. This factor must influence the milling time needed to develop specific microstructural changes, such as nanocrystallinity. In the bibliography, there are multiple comparative studies between ball milling devices at the laboratory level. These are based on the comparison via experimental data and/or simulations. Overall, these studies usually conclude that the planetary mills were less energetic than the shaker mills as SPEX. Mathematic ab initio models have been applied to study MA process [25]. Moreover, experimental data to determine the speed of balls or rotation of milling vials [26]. These works are generally based on energy transferred during milling [27]. To make a comparison a useful parameter was the ratio between the power transmitted and the powder mass (P/m).

One of the problems associated with the energetic comparison of mills is the differences in the motion. Another question is that the factors involved are multiple and there are complex relationships between them: Rotation frequencies, number and characteristics of the balls, geometry vials, degree of container filling, powder mass, energy transfer, temperature, milling atmosphere, frequency of interactions, etc. Furthermore, real processes have a distribution of energies involved in collisions due to a distribution of ball velocities in the shocks [28]. Some of the studies carried out take into account the direct comparison of shaker Spex and planetary mills [19,29].

First, for the comparison between both devices, we apply the method described in ref. [29]. In order to evaluate the total energy transferred to the powder during milling, it is necessary to know the ball velocity. In the Spex device, the ball velocity given in ref. [29] is v = 2.5 m/s. To have an equivalent value in the P7, the following formula has been chosen to determine the speed, v, of ball impact [30].

$$\mathbf{v}^2 = (\mathbf{r}\_\mathrm{P} \,\, \Omega)^2 + (\mathbf{r}\_\mathrm{j} - \mathbf{r}\_\mathrm{b})^2 \,\, \omega^2 \,\, (1 + \langle 2\omega \rangle \,\Omega) \tag{4}$$

As introduced in the Section 2, in SPEX the cycles are 875 rpm, whereas in the P7 the selected number of cycles (jar, platform) is ω = Ω = 600 rpm. The P7 radius of the platform rp = 7 cm, the ball radius rb is 0.5 cm for P7; and the internal radius of the jar rj = 1.7 cm. The calculated velocity is v = 0.79 m/s. Obviously, these values should be considered for devices comparison, and real energy transfers have a distribution of speeds. Other models provide different velocities, as an example, the expression of reference [31] provides a value of v = 2.04 m/s for P7.

Regarding the kinetic energy, Ec, transferred by a ball in the planetary mill, it can be calculated directly from the velocity. The calculated value is 0.002 J. The equivalent calculation in the shaker was calculated in ref. [29], and the kinetic energy is 0.026 J.

The normalized (by unit mass of powder) total energy (E/m) transferred at a given milling time, t, was calculated by taking into account the following expression [32]:

$$\mathbf{E}/\mathbf{m} = (\mathbf{n}\mathbf{E}\_{\mathbf{c}}\mathbf{t}/\mathbf{m})\tag{5}$$

where n is the balls number. As shown in Table 7, the calculated values for 80 h of milling are (E/m)SPEX = 5193 J/g and (E/m)P7 = 3752 J/g. Thus, milling is more energetic in the Spex device.


**Table 7.** Ball velocity, kinetic energy, and total energy of Spex and P7.

The applied model considers that the frequency of impacts (events frequency) is equivalent in both devices. This factor is inversely proportional to the time between shock events on milling. The sliding factor influences the time it takes for the balls to separate from the container wall, the takeoff time, tt. The increase in this time reduces the frequency of collisions. It is also necessary to determine a flight time of the ball before impact. Obviously, there is a distribution of speeds as well as energy and time. However, it should be taken into account that we need to stablish an easy way to compare both devices with characteristic mean parameters.

The ball detaches from the container wall when the value of the contact force is zero. The takeoff time can be estimated by applying the following expression [33]:

$$\mathbf{t}\_t \ll (1/(\Omega + \omega)) \arccos\left(-(\omega^2 \,\mathrm{r}/\Omega^2 \,\mathrm{R})\right),\tag{6}$$

In the P7 device this time is tt = 0.076 s. Likewise, the flight time tf (between detachment and collision) is also linked to the period (inverse of the frequency) in planetary mills. Thus, the period is higher in P7. The P7 flight time is determined by using the relationship given in Figure 8 [19]. The calculated time is tf = 0.013 s.

**Figure 8.** Flight time, tf, as a function of the period T, (+, this work; red line, [19]).

Consequently, the total time [19], its sum, is tt + tf (P7) = 0.089 s. In the SPEX device, the bibliography data were used: tt = 0.0073 s and tf = 0.0269 s that drive to tt + tf (SPEX) = 0.0342 s. As the total time is higher in the planetary mill, the events frequency (for unit time) is higher in the shaker mill.

Thus, from the time between collisions, the shaker mill is also more effective. Likewise, the results obtained in this study cannot be directly based in this total time because the final microstructure varies (sometimes appears a second minor phase).

Therefore, SPEX mill is more energetically efficient than the P7 planetary mill (in the processing conditions of our study).

On the other hand, it is possible to calculate an equivalent milling time between different mills [34,35]. The calculation of this equivalent time is based on the assumption that the power transferred per unit of mass (P/m) is directly proportional to the time so that in two different devices the powder achieves the same final microstructure. It can be determined by applying the following equation:

$$\mathbf{P}\left(\mathbf{P}/\mathfrak{m}\right)\_{\mathsf{SPEX}} = \mathfrak{q}\left(\mathbf{P}\left(\Omega\_{\mathsf{o}}\right)/\mathfrak{m}\right)\_{\mathsf{P}\heartsuit}\left(\Omega/\Omega\_{\mathsf{o}}\right)^{3},\tag{7}$$

This approach has been previously applied in FeNb and FeZr alloys [35]. In our study, the cycles ratio in the P7 is 1 (Ω = Ωo= 600 rpm). From the previous calculations, we can calculate ϕ that is a proportionality factor, ϕ = 1.384. Thus, equation 6 is reduced to:

$$\text{(P/m)}\_{\text{SPE}\&} = \text{\textquotesingle} \text{ (P(}\Omega\_{\text{o}}\text{)/m)}\_{\text{P}\uparrow} \tag{8}$$

The equivalent milling time, teq, can be calculated by applying [19]:

$$\mathbf{t}\_{\rm sp} = \left( (\mathbf{P}/\mathbf{m})\_{\rm SPEX} / (\mathbf{P}(\Omega\_0)/\mathbf{m})\_{\rm P7} \right) \mathbf{t}\_\prime \tag{9}$$

It has been estimated (from Table 7 data) that the time required in the P7 for the same energy transfer is 1.384 higher than for SPEX. Obviously, there are many factors modifying the consequences of this equivalent time. Sometimes the MA process is not energetic enough to obtain a concrete microstructure. For example, by modifying the milling conditions, the achieved microstructure can be nanocrystalline or amorphous [36]. Other possibilities are that the energy transfer facilitates the extension of the solubility limits or the formation of intermetallic compounds [24,37]. In alloy A, the energy/power transfer is not enough to obtain only the bcc Fe rich solid solution in the planetary mill (Nb(B) phase remains after 80 h of milling), and in alloy B the high energy/power transfer in shaker mill favors the formation of an incipient amorphous phase. Additional experiments, with alloys of similar composition and/or by repeating production process will confirm the general trend found in this work.

#### **4. Conclusions**

Two alloys, Fe80Nb7B12Cu1 (A) and Fe80(NiZr)7B12Cu1 (B), have been produced in two milling devices, a shaker mill (SPEX) and a planetary mill (P7). In the sample A milled in the shaker device and in sample B milled in the planetary mill only an Fe rich solid solution was detected by XRD after 80 h of MA. Instead, in the sample A milled in the planetary mill a low percentage of a Nb(B) phase was also detected; whereas in the sample B shaker milled the Fe rich solid solution coexist with a minor disordered phase. Additional thermal analysis permits us to establish that the analyzed crystallization process is due to crystalline growth, and the thermal stability is higher in samples milled in the shaker mill (higher temperature of crystallization and activation energy).

The results indicate that probably the milling is more energetic and efficient in the shaker mill. The equivalent time required in the planetary mill for the same energy transfer is 1.384 higher. In alloy A, the energy/power transfer is not enough to obtain only the bcc Fe rich solid solution in the planetary mill (a minor Nb(B) phase remains), and in sample B the high energy/power transfer in shaker mill favors the formation of a disordered Zr rich phase.

**Author Contributions:** Methodology, L.E., J.S.; formal analysis, A.C.; investigation, A.C.; writing original draft preparation, J.-J.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Spanish MINECO project MAT2016-75967-P and UdG project PONTOS-2020/01.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** The authors agree the Technical Research Facilities of the University of Girona.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


## *Review* **Mechanical Alloying of Elemental Powders into Nanocrystalline (NC) Fe-Cr Alloys: Remarkable Oxidation Resistance of NC Alloys**

**R. K. Singh Raman 1,2**


**Abstract:** Mechanical alloying is among the few cost effective techniques for synthesizing nanocrystalline alloy powders. This article reviews mechanical alloying or ball-milling of (NC) powders of Fe-Cr alloys of different compositions, and the remarkable oxidation resistance of the NC alloy. The article also reviews challenges in thermal processing of the mechanically alloyed powders (such as compaction into monolithic mass) and means to overcome the challenges.

**Keywords:** nanocrystalline (NC) alloy; microcrystalline (MC) alloy; mechanical alloying; ball-milling; oxidation resistance

## **1. Introduction**

Nanocrystalline (NC) structure of alloys (instead of their traditional microcrystalline (MC) counterparts) is reported to remarkably improve some of the properties, such as strength and corrosion resistance [1–4]. Mechanical alloying of powders of the constituent elements is among the most commonly employed approaches for producing NC alloys [2,4–6]. This article reviews challenges in consolidation of mechanically alloyed Fe-Cr powders into monolithic solids and their sintering for achieving close-to-theoretical density, and circumvention of the challenges. The article also presents examples of improvements in oxidation/corrosion resistance properties of a few alloy systems due to the NC structure, and their mechanistic explanations.

## **2. Mechanical Alloying for Nanocrystallinity**

A few processing techniques have been employed for synthesis of nanocrystalline (NC) alloys, such as electrodeposition [7], mechanical alloying [2,4–6], severe plastic deformation [8], sputtering, electron beam evaporation, pulse laser ablation [9], gas condensation [9,10], and sol-gel [11]. Some of these techniques (viz., sputtering, electron beam evaporation, pulse laser ablation, gas condensation and sol-gel) have limitations of producing only thin films or tiny samples of NC materials. Several critical applications, such as mechanical or corrosion testing, require bulk monolithic samples. For synthesis of bulk NC materials, mechanical alloying [2,4–6] and electrodeposition [7] have been widely employed.

A few metals [7], and Ni-Fe and Ni-Co alloys [5,6,12,13] were among the early NC metallic materials synthesized by pulsed electrodeposition. However, processing of NC metallic materials by pulsed electrodeposition requires promotion of nucleation and suppression of growth of the depositing grains, and, in order to bias nucleation, organic additives are added to the electrolyte used for electrodeposition [7]. A post-synthesis thermal decomposition is employed for removal of these additives. However, often, the organic additives still remain in the electrodeposited NC material and deteriorate its properties (such as causing embrittlement). In contrast, NC metal and alloy powders synthesized by

**Citation:** Singh Raman, R.K. Mechanical Alloying of Elemental Powders into Nanocrystalline (NC) Fe-Cr Alloys: Remarkable Oxidation Resistance of NC Alloys. *Metals* **2021**, *11*, 695. https://doi.org/10.3390/ met11050695

Academic Editors: Tadeusz Kulik and Javier S. Blázquez Gámez

Received: 12 March 2021 Accepted: 22 April 2021 Published: 23 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

mechanical alloying technique are generally free from such artefacts [2,4–6]. For critical applications such as mechanical or corrosion testing, the mechanically alloyed powders need to be compacted into monolithic mass. Depending on the thermal and mechanical characteristics of NC materials, various techniques have been employed for their compaction [14], such as hot compaction [15], high pressure/lower temperature compaction, in-situ consolidation [16], hot isostatic pressing [17], and explosive compaction [18]. Often, the consolidated material is subjected to sintering in order to achieve the theoretical density. The thermal stability of NC material governs the choice of sintering temperature and time. Examples of mechanical alloying of elemental powders into NC alloy powders of a few systems, their consolidation and sintering, and the associated challenges and circumvention are discussed in the succeeding sections. The role of the NC structure in remarkably influencing oxidation resistance is also discussed.

#### **3. Mechanical Alloying of Fe-Cr Alloy Powders and Their Consolidation**

Powders of Fe and Cr (purity: 99.9%, particle size: <10 μm) were mechanical alloyed into Fe-Cr alloy powders with different Cr contents (2, 5, 10, and 20 wt%) for the purpose of investigation of the role of nanocrystalline (NC) structure of Fe-Cr alloys in their oxidation/corrosion resistance [19]. NC Fe-Cr alloy powders were synthesized under an inert environment of toluene, using an air-cooled planetary ball-mill (300 rpm), tungsten carbide balls, and a powder-to-ball weight ratio of 1:10. The ball milling was continued for 20 h, and the crystallite grain size was determined intermittently, from the broadening of X-ray diffraction (XRD) peaks and the Scherrer equation [20]. Figure 1 shows the decrease in the grain size with increasing milling duration and confirms the grain size to have stabilized well and truly in the nanocrystalline regime (~20 nm) for all the powders, after 15–20 h of mechanical alloying [21]. It should be noted that the indirect determination of nanocrystalline grain size by XRD has been confirmed by direct measurements by transmission electron microscopy [21].

**Figure 1.** Decrease in the crystallite grain size with increasing duration of milling of elemental powders into Fe-Cr alloy powders with different Cr contents [21].

As mentioned earlier, for critical applications such as generating specimens for mechanical or corrosion testing, the mechanically alloyed powders need to be compacted into monolithic mass. However, it may be a non-trivial task to accomplish compaction of powders of NC Fe-Cr alloys, because the crystal structure of Fe and Fe-Cr alloys is body centred cubic (BCC), which resist plastic deformation and have high hardness. Therefore, the plastic deformation and material flow that are necessary for an effective compaction required for consolidation of Fe and Fe-Cr alloys necessitate application of high pressures and high temperatures. Seigel [22] reported that for plastic deformation of NC (10 nm) Fe (that has a hardness of 10 GPa), the pressure needs to be greater than the yield strength,

i.e., 3.5 GPa (~1/3rd of hardness), that can be prohibitively high. Therefore, to accomplish compaction at moderate loads, it becomes necessary to elevate the compaction temperature in order to enhance plastic flow. However, the compaction temperature cannot be too high because that will cause excessive grain growth and loss of nanocrystallinity. Therefore, it becomes essential to identify the temperature-time window for accomplishing compaction without allowing excessive grain growth and loss of NC structure.

Malow et al. [23] successfully compacted NC Fe to ~100% of the theoretical density, by employing a pressure of 2.7 GPa at 475 ◦C. However, when the NC Fe-10Cr alloy powder that was prepared as described earlier was compacted at 2 GPa and 515 ◦C, only 92% of the theoretical density could be achieved, indicating the need for increasing the compaction temperature. To identify an appropriate temperature-time window for softening of hardened NC Fe-Cr alloy powders but avoiding the excessive grain growth, the mechanically alloyed powders produced as shown in Figure 1 [21] were subjected to annealing at 500, 600, and 700 ◦C in a reducing environment for different durations (3–120 min). The average grain size was determined after intermittent durations of annealing (again, from the broadening of XRD peaks and the Scherrer equation [20]). Figure 2 shows a typical set of XRD spectra for Fe-10Cr alloy mechanically alloyed for 3–120 min at 500 ◦C. The peaks are seen to broaden with increasing time of annealing because of the increasing grain growth.

**Figure 2.** XRD spectra for mechanically alloyed Fe-10Cr alloy annealed for different durations (0–120 min) at 500 ◦C ([24], reproduced with copyright clearance).

Figure 3 shows the grain growth of Fe-10Cr alloy powder during annealing at 500, 600, and 700 ◦C for durations up to 120 min. Though the grains grow rapidly during the initial stage of annealing at each temperature, the intensity of growth is faster at higher temperatures. Grain growth plateaued after 15 min at 500 ◦C, whereas grains continued to grow rapidly at 600 and 700 ◦C. In fact, at 700 ◦C, the grain growth was so rapid that after 60 min the grains tended to grow beyond the nanocrystalline regime. The grain growth took off after 90 min also at 600 ◦C. Figure 4 compares the grain growth data for the mechanically alloyed Fe-10Cr NC alloy powder extracted from Figure 3, which are compared with those reported in the literature for ball-milled powders of NC Fe and an Fe-Al alloy [9,23–28]. There are considerable variabilities, particularly at temperatures above 400 ◦C. Such variabilities are attributed largely to the impurities in the powders. Some impurities such as oxide particles pin the grain boundaries and impede grain growth, and therefore powders with such impurities resist grain growth over higher temperature regimes. So, the purer the NC powder, the greater is its susceptibility to grain growth, and the lower is the temperature at which the rapid grain growth starts. The grain growth impeding due to impurities is more readily offered by alloying elements with greater

thermodynamic susceptibility to form oxides, for example the Fe-Al alloy in Figure 4 shows resistance to grain growth over a higher temperature regime [28]. In fact, this effect has been successfully exploited for increasing the temperature regime for stability of NC alloys, as will be discussed subsequently.

**Figure 3.** Grain growth of mechanically alloyed Fe-10Cr nanocrystalline alloy powder during annealing at 500 ◦C (**a**), 600 ◦C (**b**) and 700 ◦C (**c**) for different durations up to 120 min ([24], reproduced with copyright clearance].

**Figure 4.** Grain growth data in literature for ball-milled nanocrystalline Fe and two Fe-alloy powders during annealing at different temperatures for different durations: [1] *NanoStructured Materials, 6 (1995) 421 (Ref* [25]), [2] *NanoStructured Materials, 12 (1999) 685 (Ref* [26]), [3] *Acta Mater, 45 (1997) 2177 (Ref* [23]), [4] *Progress in Materials Science, 33 (1989) 223 (Ref* [9]), [5] *Mater. Sci. Forum, 343 (2000) 683 (Ref* [27]), [6] *NanoStructured Materials 9 (1997) 71 (Ref* [28]), and [7] *Mater Sci Engineering-A, 494 (2008) 253 (Ref* [24]).

On the basis of the grain growth data in Figure 3, a time-temperature window of 600 ◦C and 30 min was selected for improving ductility and flow of mechanically alloyed NC powder of Fe-10Cr alloy for successful compaction of the powder into monolithic discs (under a uniaxial pressure of 2.7 GPa). The consolidated discs are seen in Figure 5. The density of the discs was determined to be 98% of the theoretical density. The compacted discs were sintered at 600 ◦C for 1 h to improve the density to ~99%. Again, grain growth data in Figure 3 were the basis for the choice of sintering time and temperature. Grain size of the sintered discs (52 nm) was very much in the nanocrystalline regime.

### **4. Oxidation Resistance of Mechanically Alloyed NC Fe-Cr Alloys**

Fe-alloys with sufficient contents of specific alloying elements (namely Cr, Al and Si) develop a continuous surface layer of oxides of such alloying elements (i.e., Cr2O3, Al2O3 or SiO2) that remarkably retards oxidation rate of such alloys. Defect contents of these oxides are remarkably low [29]. As a result, the counter diffusivity of elements involved in oxidation is insignificantly low through the surface layer of such oxides, hence the remarkable oxidation resistance of such alloys. Enabling the development of a continuous surface layer of these protective oxides is the basis for corrosion resistance of alloys. Most common examples are stainless steels (i.e., Fe-Cr alloys) and Al-alloys that readily develop robust surface barrier layers of Cr2O3 and Al2O3 respectively, and hence, such alloys possess excellent oxidation/corrosion resistance [29].

The development of a full-fledged continuous external layer of chromia, alumina or silica requires availability of a sufficient amount of the specific alloying elements (viz., Cr, Al or Si) at the surface [29]. The critical concentration of such alloying content for a given alloy can be determined by Wagner's calculations [30] (e.g., stainless steels possess 18–20 wt% Cr [3]). The critical concentration is governed directly by the diffusivity of the element (and the factors that influence diffusivity, e.g., temperature). It is well-known that at moderately elevated temperatures, grain boundary diffusivity far exceeds the lattice/bulk diffusivity [31]. The temperature range for the Fe-Cr system is 300–400 ◦C (as shown in Table 1). Therefore, it is obvious that diffusivity will be higher in the case of a nanocrystalline (NC) alloy (that possess remarkably high grain boundary volume fraction [9]) than its microcrystalline counterpart. Indeed, Wang et al. [31] reported the Cr diffusion in a NC iron to be four orders of magnitude greater than that in its MC counterpart. On this basis, it was hypothesized that an Fe-Cr alloy with sufficient Cr content will possess much superior oxidation resistance when the structure of the alloy is nanocrystalline (NC) vis-à-vis the same alloy with microcrystalline (MC) structure. On the same basis, it was further hypothesized that with NC structure, considerably lower chromium contents may be sufficient to achieve oxidation resistance similar to common high chromium microcrystalline alloys. These hypotheses were nicely validated by Singh Raman et al. [19,32]. For example, as seen in Figure 5, a NC Fe-10Cr alloy (that was synthesized by mechanical alloying, as described in the preceding section) showed nearly an order of magnitude superior resistance to oxidation than the MC alloy of same composition, during air-oxidation at 300 ◦C for 3120 min [19,32]. The secondary ion mass spectroscopy (SIMS) depth profiles of Cr through the oxide scale thickness (Figure 6) demonstrate the remarkably higher Cr contents of the protective oxide layer of the NC alloy than that in the MC alloy, thus accounting for the remarkably superior oxidation resistance of the NC alloy [19]. In fact, it took some time (120 min) for sufficient Cr to enrich in the oxide layer to develop a full-fledged protective layer of Cr2O3. At this stage, the oxidation rate of NC alloy became insignificant and similar to that for a stainless steel (that possess ~20 wt% Cr), whereas the MC alloy continued to oxidize at a high rate because of its inability to form a full-fledged protective layer of Cr2O3. It was possible to explain the stark difference in colour of the NC and MC alloy oxidized for 3120 min (interested readers may refer to [32] for the explanation).


**Table 1.** Diffusion coefficients of Cr in NC and MC iron [31].

**Figure 5.** Oxidation kinetics (weight gain versus time plots) of Fe–10Cr alloy with nanocrystalline (nc) and microcrystalline (mc) structures oxidized at 300 ◦C for 3120 min (insets, show colour of the oxidized samples at different stages of oxidation) ([32], reproduced with copyright clearance).

**Figure 6.** SIMS depth profiles for Cr in the oxide scale of Fe-10%Cr alloy with nanocrystalline (nc or NC) and microcrystalline (mc or MC) structures, oxidized for 120 h in air at 300 ◦C alloys. ([19], reproduced with copyright clearance).

#### **5. Mechanical Alloying of Fe-Cr-Ni-Zr Alloy Powders and Their Consolidation**

As shown in Figure 3, mechanically alloyed Fe-Cr nanocrystalline (NC) powders undergo rapid grain growth when annealed at temperatures above 600–700 ◦C. Therefore, such powders can lose nanocrystallinity upon exposures to higher temperatures, which

means temperatures must be <700 ◦C for: (a) further thermal processing of such NC powders (such as consolidation/sintering, as described in the preceding section) and (b) applications where the NC structure of alloy is required (such as high temperature oxidation resistant NC alloys). Therefore, it is highly attractive to explore means for retaining nanocrystallinity of mechanically alloyed Fe-Cr alloyed powders at temperatures >700 ◦C. As described in the preceding section, impurities such as oxide particles pin the grain boundaries and impede grain growth, and as a result, some NC powders of Fe and Fe-Cr alloy resisted grain growth until higher temperature, as shown in Figure 4. Zr readily forms oxides that are thermodynamically stable up to very high temperatures. Following the ball-milling procedure described earlier, Fe, Cr, and Ni powders with and without 2 wt% Zr were mechanically alloyed to synthesize NC Fe-10Cr-5Ni and Fe-10Cr-5Ni-2Zr alloy powders, with an objective of increasing the high temperature processibility of the powders, such as compaction at higher temperature, due to Zr addition [33]. The alloy was added with 5 wt% Ni, for improving ductility of the compacted alloy, since NC alloys are known to suffer from poor ductility [1,2,33]. It should be noted that in the relatively recent times, researchers have successfully employed spark plasma sintering for in-situ consolidation [34].

Grain size of the Fe-10Cr-5Ni alloy powders with and without Zr addition was measured by XRD technique (as described earlier) after annealing for 1 h at various temperatures. As shown in Figure 7, NC powder of Zr-containing alloy resists grain growth during annealing at much higher temperatures. In fact, grain size of Zr-containing NC alloy powder remains well within NC regime upon annealing even at 1000 ◦C, whereas the Zr-free alloy undergoes rapid grain growth during 1 h annealing at 700 ◦C and the grains grow into the regime of >100 nm. Mechanically alloyed NC powder of Fe-10Cr-5Ni-2Zr was hot-compacted at 400 ◦C into discs that could be sintered at a very high temperature (1000 ◦C), on the basis of the grain growth data shown in Figure 7 [33]. For improving ductility, the NC Fe-10Cr-5Ni-2Zr alloy powder was mixed with an equal amount of microcrystalline (MC) Fe-10Cr-5Ni-2Zr alloy powder and compacted/sintered into discs of Fe-10Cr-5Ni-2Zr alloy with bimodal grain size distribution (50% NC + 50% MC). The alloy with bimodal grain size distribution showed improved ductility [33]. In simple terms, the bimodal alloy has considerable (50%) MC component where dislocations can travel greater distance before being pinned, and hence better ductility.

**Figure 7.** Grain growth of Fe-10Cr-5Ni alloys with and without Zr, after annealing for 1 h at different temperatures [33], reproduced with copyright clearance.

#### **6. Oxidation Resistance of Mechanically Alloyed NC Fe-Cr-Ni-Zr Alloys**

Oxidation resistance of NC Fe-Cr alloys has been shown to be considerably superior to that of their MC counterparts [3], as also seen in Figure 5, and the reasons for the superior oxidation resistance of NC alloy is discussed in the literature [32] and summarized in the preceding section. Consistent with this trend, nanocrystalline Fe-10Cr-5Ni-2Zr alloy oxidized at much slower rate than its microcrystalline counterpart, as shown in Figure 8. The role of nanocrystalline structure in improving oxidation resistance is further supported from the observation that the Fe-10Cr-5Ni-2Zr alloy with bimodal grain size distribution (i.e., 50% nanocrystalline and 50% microcrystalline content) oxidized at somewhat higher rate (Figure 8), because of their smaller nanocrystalline content. The most striking observation was that the oxidation resistance of nanocrystalline Fe-10Cr-5Ni-2Zr alloy Cr) was in the same regime as that of a microcrystalline Fe-20Cr-5Ni alloy (whose 20% Cr content is similar to a common stainless steel.) The SIMS depth profiles of Cr through the oxide scale thickness (Figure 9) show considerably lower Cr contents of the protective oxide layer of the microcrystalline Fe-10Cr-5Ni-2Zr alloy, whereas the Cr content was considerably higher in protective oxide of all the alloys with high oxidation resistance [33]. In fact, the relative Cr contents of the protective oxide scales of various alloys commensurate well with their corresponding oxidation resistance.

**Figure 8.** Oxidation kinetics of the nanocrystalline (nc), bimodal (bi), and microcrystalline (mc) alloys at 550 ◦C. The inset shows the difference in the oxidation kinetics trend during the initial stages of oxidation. ([33], reproduced with copyright clearance).

**Figure 9.** Cr depth profile for oxidised alloys described in Figure 8 (above) measured using secondary ion mass spectroscopy (SIMS). ([33], reproduced with copyright clearance).

#### **7. Opportunities**

It is well known [29] that the defect content of Al2O3 is considerably less than that of Cr2O3, and hence a surface layer of Al2O3 is more protective. However, at the Al contents required for developing a continuous protective layer of Al2O3 on a steel, the alloy will suffer from unacceptable loss in mechanical property (such as loss of toughness). However, it has been possible to develop Al2O3 layer on Fe-Cr-Al and Ni-Cr-Al alloys (that have acceptably low Al contents), with the help of the phenomenon called, "third element effect" [29]. In this phenomenon, the alloy with sufficient Cr first forms a continuous layer of Cr2O3, which stops outward diffusion of Al, thereby facilitating Al accumulation (just underneath the Cr2O3 layer) to the level that is sufficient for the development of a full-fledged layer of Al2O3. Thus, with the "third element effect" of Cr, the alloy develops a protective layer of Al2O3 in spite of the low Al content of the alloy. As discussed in the preceding sections, NC structure profoundly accelerates diffusion; hence, it is expected to facilitate the "third element effect". Therefore, it will be interesting to investigate the role of NC structure of mechanically alloyed systems in promoting the "third element effect".

#### **8. Conclusions**

Nanocrystalline (NC) powders of Fe-Cr alloys can be synthesized by mechanical alloying. However, consolidation of these powders into monolithic mass is a challenge because of their high hardness due to body-centered cubic crystalline structure of Fe-Cr systems. In order to promote plastic flow, compaction is required to be carried out at high temperatures; however, too high a temperature can cause grain growth and loss of NC structure. A temperature-time window is optimized for hot-compaction of NC powders of Fe-Cr alloys without losing NC structure. In order to enable hot processing of the powders at much higher temperatures without losing NC structure, Zr was added to the alloy during mechanical alloying that pins grain boundaries and suppresses grain growth. As a result, compaction temperature of powders of Fe-10Cr-5Ni-2Zr alloy can be as high as 1000 ◦C.

Oxidation resistance of compacted nanocrystalline (NC) Fe-10Cr alloy is found to be remarkably superior to its microcrystalline (MC) counterpart. The superior oxidation resistance of NC alloy is attributed to the much greater grain boundary volume of NC structure that causes sufficient diffusion of Cr in oxide layer and development of a protective layer of Cr-oxide, whereas the MC alloy fails to develop this protective layer because of insufficient Cr diffusion. In fact, Cr content of the protective oxide layer developed during oxidation of NC Fe-10Cr-5Ni-2Zr alloy at 550 ◦C is similar to that of a MC Fe-20Cr alloy (i.e., common stainless steel); hence, the two alloys show similar oxidation resistance. Regarding further opportunities, it will be interesting to investigate the role of NC structure of mechanically alloyed systems in promoting the "third element effect".

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is contained within the article.

**Conflicts of Interest:** The author declares no conflict of interest.

## **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Metals* Editorial Office E-mail: metals@mdpi.com www.mdpi.com/journal/metals

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com