Application of mechanical alloying/milling for synthesis of nanocrystalline and amorphous materials

Development of nanocrystalline structure

Grain refinement of powders to the nanometre size is governed by the plastic deformation induced during milling. In general, the grain size decreases continuously with milling time until a minimum (saturation) size is approached. ^7,42–45 A typical variation of grain size with milling time is presented in Fig. 4.

OPEN IN VIEWER

4.

Variation of Ni and Nb grain sizes as a function of milling time ⁷

Further refinement seems to be difficult to achieve for a fixed set of experimental conditions. Transmission electron microscopy observations have revealed that at the early stages of milling, a very fine microstructure is obtained which contains some shear bands and many dislocations. ^42,43 The latter are converted into a dislocation cell structure which subsequently creates low-angle grain boundaries. With prolonged processing and absorption of more dislocations into the boundaries, this structure is transformed to a fully nanocrystalline structure with completely random orientation of neighbouring grains which are separated by high-angle grain boundaries. ^43,46 The grain refinement process during ball milling is schematically illustrated in Fig. 5.

OPEN IN VIEWER

5.

Schematic representation of development of nanocrystalline structure in a single grain during ball milling: a original grain with a low dislocation density, b a high dislocation density created by severe plastic deformation of grain, c dislocation alignment and formation of low-angle grain boundaries and d transformation to high-angle (true) grain boundaries

Minimum grain size

Experimental findings have revealed that the minimum grain size for a given material is a constant value independent of milling conditions. In fact, the minimum grain size is determined by intrinsic properties of materials such as melting temperature, crystal structure and deformation behaviour. The milling conditions can only affect the rate at which the grains refine and approach the minimum size. Eckert et al. ⁴³ observed that the minimum grain size d _min in a series of FCC metals including Al, Cu, Ni, Pd, Rh and Ir scales inversely with the melting point T _m or the bulk modulus B of the respective metals. As shown in Fig. 6, the higher the T _m or B, the smaller the d _min is obtained. Eckert et al. ⁴³ have also shown that for pure FCC metals there is a linear relationship between the minimum grain size d _min and the critical equilibrium distance between two edge dislocations L _c in a pileup. L _c is given by equation (1) ⁴⁷ (1) where G is the shear modulus, b is the magnitude of the Burgers vector, ν is Poisson’s ratio (0·33), and H is the hardness of the materials.

OPEN IN VIEWER

6.

Minimum grain size obtained by ball milling for several FCC metals versus melting temperature ⁴³

However, no clear inverse correlation between the minimum grain size and melting temperature was obtained for BCC metals Cr, Fe, Nb, W as well as for HCP metals Hf, Zr, Co, Ru. As shown in Fig. 7, the minimum grain size of these series of BCC and HCP metals suggested a constant size (9 and 13 nm respectively) independent of melting temperature. Koch ¹ observed that only the lower melting temperature FCC metals (≤T _m for Pd) exhibit an inverse dependence of d _min on T _m.

OPEN IN VIEWER

7.

Minimum grain size of ball milled FCC, BCC and HCP metals versus melting temperature

The dependence of minimum grain size versus melting temperature has been discussed with respect to the competing rates of creating dislocations due to work hardening and recovery phenomena which scale with the melting point. ^43,48 Thereby, the minimum grain size achieved by ball milling is determined by the balance between creation and annihilation of dislocations during processing. This correlation explains why the minimum grain size for Al (T _m = 933 K) is about 22 nm while it can be reduced to 7 nm for Pd (T _m = 1825 K) as seen in Fig. 6.

Mohamed et al. ^49–51 analysed the minimum grain size that was obtained by ball milling by plotting the normalised minimum grain size d _min/ b (where b is the Burgers vector) as a function of melting temperature T _m. The resultant curve is shown in Fig. 8. It is observed that the data for all metals and alloys, regardless of their crystal structure, define a single curve which is divisible into two regions. First region occurs at T _m<2000 K and includes the data for the lower melting metals such as Al, Cu, Ag, Ni, and Pd (FCC), Co and Zn (HCP), and Fe (BCC). In this region, d _min/ b decreases rapidly with increasing T _m. In second region, d _min/ b decreases very slowly with increasing melting temperature. This region occurs at T _m>2000 K and contains data for the higher melting temperature metals: Ir (FCC), Hf and Zr (HCP), and Cr, Nb, and W (BCC). Similar trend was also obtained when d _min/ b is plotted versus the activation energy for self-diffusion Q and normalised hardness H/G (where H is the hardness and G is the shear modulus).

OPEN IN VIEWER

8.

Normalised minimum grain size (d _min/ b ) obtained by milling versus melting temperature T _m ⁵¹

The normalisation of the grain size by dividing it by the Burgers vector b was justified in terms of the following three considerations: (a) the Burgers vector appears in expressions and is related to dislocation reactions, deformation processes, activation volumes associated with thermally activated processes, and the velocity of migrating boundaries; (b) the normalisation of engineering parameters provides consistent trends in correlations; and (c) the normalisation of the grain size as d/ b permits a close comparison between the behaviour of various metals and alloys. The use of the activation energy of self-diffusion Q as a measure for recovery was based on the observations that thermal recovery processes are in general characterised by activation energies, and also the fact that melting temperature T _m scales with the activation energy for self-diffusion, as Q/RT _m≈λ where λ is about 17–18. This is because both Q and T _m represent a measure of the atomic bond strength. Also, the relationship between d _min/ b versus the normalised hardness H/G can be accounted for by the concept that as the grain size reaches a critical value in the nano-range, the dislocation structure cannot be sustained within the grains. This concept has been quantitatively developed by considering a balance between the repulsive force per unit length between two edge dislocations (the limit of a dislocation pile-up) and an effective force. Later, Mohamed ⁵⁰ developed a dislocation model by adopting the concept that the minimum grain size d _min obtainable by milling is governed by a balance between the hardening rate introduced by dislocation generation and the recovery rate arising from dislocation annihilation and recombination. This model is based on the similarities between milling and creep. For example, similar to minimum grain size in milling process, the steady-state creep represents a balance between the competing factors of strain hardening rate and thermal recovery rate. Moreover, the creep rate decreases with time, approaching to a steady-state region with minimum creep rate. The model expresses the normalised minimum grain size obtainable by milling, d _min/ b , in terms of several material parameters as follows (2) where A is a constant, R is the gas constant, k is Boltzmann’s constant, T is the absolute temperature, D _po is the frequency factor for pipe diffusion, γ is stacking fault energy. In addition, ν _o  = l/D _o where D _o is the frequency factor for lattice diffusion and l is the effective radius of the impurity atmosphere. β = Q _p−Q _m/Q where Q _m and Q _p are the activation energy for vacancy migration and activation energy for pipe diffusion, respectively. Based on equation (2) d _min/ b was plotted versus different materials parameters such as normalised stacking fault energy γ/Gb the activation energy for the self-diffusion Q melting temperature T _m (Fig. 9) and normalised hardness H/G. ⁵² In all cases, the data of different metals regardless of their crystal structure can be well fitted with straight lines, which additionally accords well with the prediction of dislocation model. On the basis of this model, the minimum grain size scales inversely with hardness, proportionally with the stacking fault energy and exponentially with the activation energy for recovery.

OPEN IN VIEWER

9.

Normalised minimum grain size d _min / b obtained by milling versus melting temperature T _m along with the best fit plot of the data. HCP metals are represented by solid hexagons ⁵²

It is worth noting that the data for Ag, Cu, Co, and Zn–22% Al with stacking fault energies γ of <60 mJ m⁻² was fitted with a separate straight line. ⁵² Since the metals with lower stacking fault energy present larger separation of two partial dislocations, it is quite possible that, due to the severe deformation and high internal stresses developed during milling, the separation between the two partial dislocations may significantly decrease, leading to an ‘effective’ stacking fault energy, which is much higher than that obtained in the absence of external forces. ⁵²

As mentioned earlier, the dependence of minimum grain size d _min and melting temperature (Fig. 9) can be understood by the fact that the self-diffusion activation energy Q scales with melting temperature T _m.

Eckert et al. ⁴³ and Mohamed ⁵⁰ observed that the minimum grain size d _min for FCC and BCC metals scales with the bulk modulus B. In order to examine whether the behaviour of HCP metals is consistent with the above trend, Mohamed ⁵² plotted the available Q value of all metals against B b ³ on a logarithmic scale as shown in Fig. 10a and observed that the data regardless of crystal structure can be fitted by a straight line. Furthermore, the logarithm of d _min / b for all metals was plotted versus B b ³ in Fig. 10b . The data of HCP metals along with those of FCC and BCC metals fall into the same straight line. Mohamed ⁵² pointed out that the dependence of minimum grain size d _min and bulk modulus B is not related to the presence of the shear modulus G (shear modulus in equation (2) has no net effect) but most likely arises from the relation between the bulk modulus and the activation energy for self-diffusion Q.

OPEN IN VIEWER

10.

a The activation energy for the self-diffusion Q versus the bulk modulus multiplied by the cube of the Burgers vector B b ³. b The logarithm of the normalised minimum grain size d _min / b obtained by milling versus the bulk modulus multiplied by the cube of the Burgers vector B b ³. HCP metals are represented by solid hexagon ⁵²

Koch has argued that a different grain size can be obtained by altering the rate of work hardening (e.g. by changing the ball-to-powder weight ratio or the type of mill) and/or rate of dislocation recovery (e.g. by changing milling temperature). ⁵³ Hence, the grain size in final material can be controlled by changing the equilibrium rate of dislocation creation and recovery.

Experiments show that lowering milling temperature leads to a finer grain size. ⁵⁴ This behaviour is also predicted by the dislocation model of Mohamed. Furthermore, it should be noted that the magnitude of the decrease in d _min with decreasing milling temperature is predicted to be significant when the activation energy for recovery is high. In this case, the rate of recovery is very low and the minimum grain size is essentially governed by the rate of hardening. For example, the model predicts that: (a) when the milling temperature is decreased from 300 to 90 K, d _min in Al does not decrease appreciably, and (b) the amount of decrease assumes significance in the following order: Al, Cu, Ni, and Ir. These predictions are in agreement with experimental trends. First, Zhou et al. ⁵⁵ reported that d _min in Al is not very sensitive to milling temperature; the difference between d _min produced at 300 K and that produced at 90 K is within the range of experimental error. Second, when milling temperature was changed from 303 to 188 K, d _min in Cu decreased from 26 to 17 nm. ⁵⁴

The dependence of d _min on milling temperature has directed researchers to another version of ball milling process known as ‘cryomilling’. This process benefits from ball milling at cryogenic temperatures. Two different methods have been applied for cryomilling. In most cases, a cryogenic liquid (e.g. liquid N₂) flows continuously inside the mill vessel. The raw materials are ball milled with the cryogenic liquid forming a slurry. In some set-ups the cryogenic temperature during ball milling has been achieved by circulation of cryogenic liquid around the outside surfaces of milling vessel. Basically, cryomilling experiments can be carried out in any type of milling apparatus although the attrition ball mill is often preferred. This type of mill comprises of a static cylindrical vessel with a rotating impeller with which a rather large quantity of powders (several kilograms) can be processed making it potentially suitable for commercial exploitation. The cryogenic liquid can be readily flowed inside/outside of milling vessel.

The grain refinement mechanism during cryomilling was found to be similar to that observed during conventional room-temperature ball milling. ⁵⁶ With increasing interior strain, the dislocation density increases gradually. At a certain strain level, these dislocations will form sub-boundaries by aligning, annihilating, and recombining with each other. Upon further deformation, the low-angle sub-boundaries are changed to high-angle boundaries (true boundaries) by the absorption of more dislocations into the boundaries or accompanying grain rotation. Eventually, the initial large grains decompose into small grains separated by true boundaries. Finally, the grain size reaches a given constant size. Once this size is reached, further refinement ceases.

Cryomilling process includes several advantages over room temperature ball milling. A benefit of cryomilling is that the rate of diffusional process like annihilation of dislocation is extremely limited at cryogenic temperature. As discussed earlier, under this condition the grain refines with a much higher rate. Hence the milling time required to attain a certain grain size is significantly reduced. Shorter milling time in case of cryomilling implies that a large quantity of powder can be processed in large-scale low-energy ball mills within a reasonable milling time. It appears that the cryomilling does not significantly affect the minimum grain size. The experimental results show that the minimum Al grain size obtainable by ball milling falls in the range of 20–26 nm. ^43,57 These values were obtained by processing of Al powders in high-energy ball mills such as SPEX8000 as well as in low-energy ball mills. Consistently, cryomilling of Al powder resulted in a nanocrystalline structure with an average grain size of about 26 nm. ^58,59

For those cryomilling processes in which the powders are mixed with liquid N₂, the reaction of constituents with N₂ must be considered. This reaction seems to proceed to a great extent during ball milling as plastic deformation; flattening and fracturing of powder particles repeatedly produce fresh, clean and highly reactive surfaces. The powder-N₂ reaction can be more serious in case of reactive elements such as Al and Ti. However, experimental results do not support this prediction. For example, during cryomilling of 5083 Al powder the nitrogen picked up was found to be 0·16 wt-%. ^59,60 Nitrogen dissolved interstitially in Al lattice forming a supersaturated solid solution. It should be noted that because of low temperature, formation of nitrogen-containing dispersoids such as AlN are kinetically improbable, although there is a large driving force for the formation of such compounds (e.g. ΔG ^AlN = −287·1 kJ mol⁻¹). However, subsequent heating during differential scanning calorimetry (DSC) run or upon powder consolidation of cryomilled 5083 Al led to the formation of various secondary phases including AlN and Al₂O₃. The presence of these precipitates was confirmed by atom probe field-ion microscopy (FIM) and high-resolution transmission electron microscopy (HRTEM) observations but the volume fraction of precipitates were not determined. Furthermore, the XRD patterns of cryomilled powder after heat treatment showed no traces of secondary phases suggesting that the amount of the dispersoids are virtually very small <1 vol.-%. ⁵⁸ It is worth noting that the dispersion particles formed during cryomilling acquired nanoscale size ranging from 1 to 30 nm, with majority of 3–10 nm. The average spacing between the dispersoids was measured to be less than 10 nm. In another study, it was found that cryomilled pure Al contains fine particles of aluminiumoxynitride with a diameter of 2–10 nm and spacing of 50–100 nm. ⁶¹ Later, it was observed on HRTEM that in this material the dispersoids are 10–15 nm in two dimensions but their thickness is only a few atomic layers. ⁶² These dispersions were found to be incoherent in nature, highly stable at high temperature and insoluble in matrix. ⁶³ As will be discussed later in the ‘Bulk nanocrystalline/amorphous alloys prepared by consolidation of mechanical alloying/mechanical milling powders’ section, the presence of these finely dispersed particles can play a critical role in retarding grain growth in the nanocrystalline materials through the so-called Zener mechanism even though their volume fraction is low. Furthermore, they could effectively obstruct dislocation motion and therefore reduce the minimum grain size.

It has been shown that the minimum grain size is also affected by the alloy composition. For example, Eckert et al. ⁴³ demonstrated the influence of composition on the minimum grain size in Fe–Cu nanocrystalline alloys. Also, Rafiei et al. ⁶⁴ performed MA of Fe–Al–Ti ternary system to study the effect of Ti addition on alloying and formation of nanocrystalline structure in Fe–Al system. It was found that (Fe,Ti)₃Al intermetallic compound achieves a smaller crystallite size compared with Fe₃Al phase. Xun et al. ⁵⁶ studied the development of nanocrystalline structure in Zn–22 wt-% Al using cryomilling. They reported that the final structure after 16 h of ball milling consisted of two phases: Al(Zn) and Zn(Al) solid solutions. The minimum grain size d _min was measured to be 33 nm for the Al(Zn) and 41 nm for Zn(Al) phases. The d _min of Zn(Al) phase is much larger than that reported for pure Zn (17 nm). In contrast, the d _min of Al(Zn) phase is higher than that of pure Al (26 nm). The presence of alloying elements can significantly affect the rate of lattice defects generation, dislocations mobility and the rate of defects annihilation during ball milling process by changing the value of several parameters such as work hardening rate, stacking fault energy, melting temperature and diffusivity. Besides, the intrinsic properties such as melting temperature and crystal structure, the history of the as-received material and the level of contamination introduced during the deformation process (e.g. oxygen and nitrogen) can also affect the accumulation and annihilation of dislocations by, for example, blocking the dislocations slip by grain boundaries (pile-up) or pinning of dislocations by foreign atoms. Owing to these complexities and since the relative effect of each factor is not well-understood, especially for ball milling process in which an intensive plastic deformation at high strain rate is involved, the rate of grain size reduction in metals and alloys cannot be easily explained and compared with each other in terms of their physical, crystallograpical and mechanical characteristics.

Finally, the finely dispersed particles, second phase precipitates and solute segregation can lock dislocations as well as grain boundaries, reducing the minimum grain size. ⁵³ For instance, Chung et al. ⁶⁵ found that cryomilled pure Ni attained a grain size of 132 nm after milling for 8 h. Addition of 0·5 and 2·0 wt-% AlN powder with an initial particle size of 2 μm reduced the Ni grain size down to 65 and 37 nm, respectively. In this case, AlN particles drastically decreased in size during cryomilling leading to a nanoscale size of 20–300 nm. It is worth noting that if the initial microscaled reinforcements do not fracture to nanoscaled particles during milling process, they have no significant effect on the grain refinement of the matrix. ^66,67

Enhancement of lattice strain

Owing to the generation of lattice defects, especially dislocations, the grain refinement to nanoscale size during MA/MM is accompanied with an increase in internal strain. The internal strain eventually approaches a saturated value (Fig. 11). In some cases like MM of AlRu, ⁴² however, the internal strain reached first a maximum and decreased subsequently at longer milling times when dislocations were consumed by increasing the number of grain boundaries. Like the minimum grain, it has been observed that the amount of internal strain is influenced by such factors as composition, melting temperature, crystal structure and strain rate (low/high intensity milling). ^43,53

OPEN IN VIEWER

11.

Enhancement of internal strain in Ni and Nb during mechanical alloying of Ni₆₀Nb₄₀ powder ⁷

Amorphisation by MA

Amorphisation by MA has been suggested to be the same as that of solid-state amorphisation in alternate layered structures, where the amorphisation reaction is governed by interdiffusion of constituents during annealing. The application of this mechanism to MA is based on the observation that a similar layered structure, although the layers are not perfectly planar and uniform, develops during MA in which a solid-state amorphisation reaction could occur.

Thermodynamic and kinetic aspects

Johnson, ^99,100 Schultz ¹⁰¹ and Bakker et al. ¹⁰² have reviewed the thermodynamics and kinetics of different solid-state amorphisation methods.

Based on experimental evidence, Johnson ⁹⁹ proposed the following two criteria for the solid-state amorphisation reaction by annealing of an alternate layered structure:

A fine multilayered structure, providing an extensive interface area, is needed for this kind of interdiffusion reaction. Furthermore, it has been suggested that the formation of the amorphous phase requires suitable nucleation sites (e.g. grain boundaries). This possibility is supported by the experiments of Vredenberg et al. ¹⁰³ and Meng et al., ¹⁰⁴ which showed the lack of amorphisation reaction at Ni/Zr interface in the absence of high-angle Zr grain boundaries.

Solid-state amorphisation by annealing has been observed in a large number of elemental pairs such as Ni–Zr, Cu–Zr, Fe–Zr, Cu–Ti, Ni–Ti, Co–Zr as reviewed by Suryanarayana and Inoue ⁷⁴ and Johnson. ⁹⁹ All investigated pairs exhibit a large negative heat of mixing and a large atomic size mismatch. The large difference in diffusivity of the two species has been experimentally confirmed in some cases including Fe–Zr, Co–Zr and Ni–Zr systems where Fe, Co and Ni diffuse several powers of ten faster than Zr (for more detail see Cahn ¹⁰⁵ and Greer et al. ¹⁰⁶ ). For the detailed experimental observations and interpretation for solid-state amorphisation in the binary metal systems, the readers are referred to the review articles. ^98–104

Mechanical alloying leads to the development of an ultrafine composite of constituents in which a solid-state amorphisation reaction can occur. It is therefore reasonable to extend the model for solid-state amorphisation by annealing to amorphisation by MA. However, MA additionally involves a high density of lattice defects such as vacancies, interstitials, dislocations and grain boundaries which can influence both the thermodynamics and the kinetics of the amorphisation reaction by MA. They raise the free energy of the crystalline system and thus destabilise it with respect to the amorphous state. At extremely small crystallite or particle size, specific surface area and hence interfacial Gibbs energy component of overall Gibbs energy of the system can increase to such an extent that amorphous state may turn out to be energetically favourable than its crystalline form. This thermodynamic aspect of solid-state amorphisation has been explained by Manna et al. ¹⁰⁷ On the other hand, the diffusion processes are assisted by the high defect density.

Enayati ⁷ studied MA amorphisation in Ni–Nb (20, 40 and 60 at-% Nb) system using a laboratory Fritsch Pulverisette 6 centrifugal ball mill. Pulverisette 6 is a low-energy type mill, capable of achieving up to eight times gravitational acceleration. Higher energy ball mill systems, as exemplified by a planetary ball mill, are capable of producing a higher acceleration (e.g. 45 times gravitational acceleration). This allowed slower processing and easier progressive observation of structure at intermediate milling times. The evolution of powder morphology and structure during MA of multi-component powders are schematised in Fig. 12. There is a tendency to form a finer microstructure at the edges of particles which is consistent with the non-uniform deformation occurring in MA at extremely high strain rates. This causes the particle edges to amorphise before the centres. However, the amorphous edges provide a strong and resilient layer, preventing further fracture. Even though the impact force is not high enough to rupture the amorphous layer at the edges, the regions underneath can be plastically deformed and then transform to the amorphous phase. The amorphous layer at the edges of particles grows progressively towards the centres with increasing milling time until the structure becomes completely amorphous. The gradual progress of the amorphisation reaction from the edges towards the centre of particles is not often observed in high-energy milling where a much higher force is imposed on particles. This leads to continuous fracturing of amorphous phase and thus a random distribution of the amorphous phase throughout the whole of the particle volumes.

OPEN IN VIEWER

12.

Schematic representation of evolution of powder morphology and structure during MA. Three stages are observed. stage I (agglomeration): Successive cold welding of Ni and Nb powder particles leads to a continuous increase in particle size. A layered structure is formed, with a progressively refined layer thickness with increasing milling time. Stage II (fragmentation): Fracturing process dominates. Powder particles and layers thickness refine. Owing to non-uniform plastic deformation a nanoscale layered structure develops at edges of particles late in this stage while centres exhibit a significant coarser layer. Stage III (steady-state): Powder particle size remains nearly constant. The strong and resilient amorphous layer formed at edges of particles prevents powder particles from further fracturing. However under impact force of colliding balls the region underneath can be plastically deformed and then transform to amorphous phase

Experimental observations suggest that MA amorphisation can occur through different paths. Most cases involve direct transformation of constituents to the amorphous structure after sufficient milling and no intermediate crystalline phase forms as a precursor to the amorphous phase. Whereas in some alloy systems a crystalline phase (i.e. solid solution, intermetallic phase) forms initially as an intermediate phase and then transforms to the amorphous structure on further milling.

Role of heat of mixing and atomic size mismatch

In contrast to solid-state amorphisation via annealing, a negative heat of mixing (ΔH _mix) appears not to be a necessary requirement for amorphisation by MA. Solid-state amorphisation by MA is found in some alloy systems such as Fe–W (0 kJ mol⁻¹), ¹⁰⁸ Cu–W (33 kJ mol⁻¹) ¹⁰⁹ and Cu–Ta (3 kJ mol⁻¹) ¹¹⁰ where the heat of mixing is almost zero or even slightly positive. In contrast, amorphous structures have not been obtained by annealing of the corresponding multilayer system. ¹¹¹

The MA amorphisation mechanism in the absence of a negative heat of mixing is not well-understood yet. Chakk et al. ¹¹¹ have proposed a theoretical model based on the atomic size ratio criterion. This model suggests that during ball milling process, under impact of colliding balls, the atoms with smaller size A can penetrate into the interstitial positions of the B lattice and thereby introduce local distortion of the lattice. When for a certain atom fraction of A in B this distortion reaches a critical value, the long-range order of the lattice is destroyed and the amorphous state is obtained. Based on the heat of mixing and the atomic size ratio, Chakk et al. constructed a diagram (Fig. 13) to analyse amorphisation in binary systems. Chakk et al. argued that pure elements and systems with low negative heat of mixing and small difference in atomic radii cannot be amorphised by MA and annealing whereas those systems which exhibit a large negative heat of mixing together with a large atomic size mismatch can be successfully amorphised by both methods. Alloy systems with a small negative heat of mixing, however, can be amorphised only by MA. Chakk et al. concluded that the negative heat of mixing can only facilitate the amorphisation reaction. In other words, a small negative or even slightly positive heat of mixing makes the amorphisation process more energy and time consuming but does not prevent it. One indication for this model is the work of Cocco et al. ¹¹² who studied the glass formation of Ni₅₀Mo₅₀. With a combination of X-ray diffraction and X-ray absorption spectroscopy, they observed that the amorphous NiMo nucleated after an insertion of Mo into the FCC Ni lattice, until a critical internal stress was reached. This was followed by the fragmentation of crystallites and loss of long-range order. However, the authors pointed out that Mo atoms occupied substitutional position in the Ni lattice which is in contrast to model proposed by Chakk et al. ¹¹¹ Furthermore it was suggested that the minimum solute concentration rule developed for amorphisation by rapid quenching can be also applied for amorphisation by MA. ¹¹¹ Egami and Waseda ¹¹³ investigated the correlation between the atomic size ratio of the constituent elements and the glass formability for both metal–metal and metal–metalloid systems to present a rule based upon the consideration of the local atomic level stresses theory in solid solutions and the stress criteria for the instability of the solid solution. They found that for a binary alloy system the minimum solute concentration necessary to obtain a stable amorphous phase by rapid quenching is inversely correlated with the atomic volume mismatch.

OPEN IN VIEWER

13.

Heat of mixing (ΔH _mix) versus atomic size ratio (r _A/r _B) map for prediction of the amorphisation tendency in binary alloy systems by MA. ¹¹¹ Presence of either a large heat of mixing (regions III and IV) or a large atomic size mismatch (regions II and IV) is necessary for amorphisation by MA. Therefore the cases which fall into region I cannot be amorphised by MA

Amorphisation by MM (defects-induced amorphisation)

In MM, the starting material all has a single composition. Although no thermodynamic driving force for diffusion is available, still an amorphisation reaction can occur. Further in some MA cases, a crystalline intermetallic compound forms as a precursor to the amorphous phase during initial stages of MA of a multi-component system. In second stage this chemically homogeneous crystalline phase transforms to the amorphous structure on further milling. The subsequent transition from chemically homogeneous intermediate phase to the amorphous phase in second stage cannot be easily explained by the diffusion-induced destabilisation model as proposed by Johnson ¹⁰⁰ and Schultz. ¹⁰¹

Johnson, ⁹⁹ Schwarz and Koch ¹¹⁴ and Bakker et al. ¹⁰² proposed that the lattice defects produced by deformation during milling are responsible for MM amorphisation in a similar way as irradiation-induced amorphisation. The accumulation of lattice defects raises the free energy of the system from the equilibrium crystalline state G _x to amorphous state G _a. The thermodynamic condition for amorphisation can be defined as G _x+ΔG _d>G _a where ΔG _d is the free energy increase due to defects.

Energy can be stored in a deformed material in the form of defects such as vacancies, interstitials, dislocations, grain boundaries and atomic disorder. There is still much uncertainty regarding which defects play a major role in raising the free energy of the crystalline state. Dislocations have not been observed to be a significant contribution. Atomic chemical disordering and/or nanocrystalline grain boundaries are more likely sources of energy storage to induce amorphisation as suggested by Seki and Johnson ¹¹⁵ for MM of CuTi₂, and by Cho and Koch ^5,116 for MM of CoZr and Nb₃Sn.

Considering only pure defects accumulation mechanism, it should be possible to amorphise pure elements. However, the amorphisation of pure Si, as a metalloid, is the only available case. ⁴⁵ Partially amorphised Si powder has been obtained by ball milling of polycrystalline elemental Si with a purity of 99·9%. The volume fraction of amorphous phase was about 15%. The authors have suggested a pressure-induced amorphisation mechanism to describe the amorphisation of Si by MM. In case of Si, cubic diamond crystal structure (Si-I) transforms to β-Sn metallic phase (Si-II) at pressure of 11·3 GPa under hydrostatic compression at room temperature ¹¹⁷ and at stresses between 10 and 12 GPa in uniaxial shock-wave experiments. ¹¹⁸ Owing to the ball-powder-ball compression force, this polymorphic transformation can occur during MM process. Owing to the rapid unloading rate, it is possible that the high-pressure crystalline form of Si cannot transform back fast enough to the equilibrium structure so that the amorphous phase is formed. This mechanism is supported by recent finding of Gamero-Castano et al. ¹¹⁹ who reported amorphisation of single-crystal Si bombarded by nanodroplets.

Polymorphic (irreversible) or allotropic (reversible) phase transition below a critical grain size at room temperature has been also reported for several early transition metals (Ti, Nb, Zr). This phenomenon was attributed to the structural instability caused by negative hydrostatic pressure (from crystal-core to surface) due to formation of nanocrystalline structure and/or high strain rate deformation in MA/MM. ^31–34

The crystal-to-amorphous transition in Si can be also explained by the defects-induced destabilisation. Partial amorphisation may be related to the presence of a non-uniform nanocrystalline grain size distribution. Only those regions that reach the critical grain size satisfy the thermodynamics condition and can transform. The critical grain size for amorphisation has been estimated to be about 3 nm for Si which is consistent with high resolution electron microscopy observations. ⁴⁵

Solid-state amorphisation in both MM and MA processes is proposed to be aided by pumping in free volume (vacancies) and/or creating new atomic coordination (or short-range order) that renders long-range periodicity untenable. These two aspects have been extensively studied and explained by positron annihilation spectroscopy and nuclear magnetic resonance in case of solid-state amorphisation of Al–Ti–Si system. ^120–122

To conclude, although the MM and MA amorphisation are not understood completely yet, it is clear that the high density of lattice defects such as vacancies, interstitials, dislocations, grain boundaries and specific surface area play an important role for the thermodynamics and the kinetics of the amorphisation reaction. They raise the free energy of crystalline system and destabilise it with respect to the amorphous state. On the other hand, the high defects density produced during milling aids the diffusion processes and therefore the amorphisation reaction. It has been suggested that for both MM and MA the rate-limiting step is the production of lattice defects by milling. ¹¹⁴

Amorphisation criteria

Mechanical alloying and rapid solidification are two different approaches for producing amorphous structures.

In rapid solidification process, if the cooling rate of the melt exceeds a critical value R _c the crystallisation is precluded and amorphous structure is obtained. This implies that the alloys with lower critical cooling rate are better glass former and can be obtained in fully amorphous structure with a higher thickness. In other words, slower crystallisation allows a decreased critical cooling rate and enables stable BMG formation and fabrication by conventional casting techniques.

While rapid solidification starts with a homogeneous alloy melt and relies on preventing the nucleation and growth of crystalline phases, MA starts with elemental crystalline powders mixture and involves diffusion processes in the solid state. For ball milling process, the times required for amorphisation can be considered as a measure to compare the GFA of alloys. ¹²³ Suryanarayana et al. studied the start time of MA amorphisation of Fe₄₂X₂₈Zr₁₀B₂₀ (X = Al, Co, Ge, Mn, Ni, Sn) powder mixtures based on XRD results. ¹²⁴ They observed that the amorphisation are easier (shorter time) in alloys in which the number of intermetallics in the constituent Zr–X binary phase diagrams is higher. It was also concluded that the MA amorphisation is possible for those systems which include a large number of intermetallic phases in the corresponding equilibrium phase diagrams. In contrast, if there is a room temperature solid solution phase over a wide composition range in the phase diagrams, then the amorphisation is not easy. Comparison of MA amorphisation kinetics in different alloy systems basically relies on two parameters; the onset time (t _o) and end time (t _e the required MA time to form a fully amorphous structure) of amorphisation reaction. While t _o shows the ease of nucleation of an amorphous phase across the layers interfaces in a particular system, the difference between the onset time and end time of amorphisation Δt = t _e−t _o is an indication of the growth rate of amorphous phase and implies how fast the amorphisation reaction proceeds. It is important to remember that the determination of start time of amorphisation, nucleation of amorphous phase, based on only XRD method limits the accuracy of data to the XRD resolution. To overcome this limitation Enayati ⁷ applied DSC in order to determine the onset time of MA amorphisation in Ni₆₀Zr₄₀ and Ni₆₀Nb₄₀ alloys. Differential scanning calorimetry is beneficial for determination of the onset time of the amorphisation reaction, since the presence of a small fraction of amorphous phase results in an exothermic peak corresponding to the subsequent crystallisation of the amorphous phase. Both Ni–Zr and Ni–Nb systems are known as easy glass formers and are therefore suitable systems for studying the amorphisation reaction during ball milling. The kinetics of the amorphisation reaction for Ni₆₀Zr₄₀ alloy were observed to be faster than that for Ni–Nb alloys. ⁷ The start time of the MA amorphisation reaction (t _o) for the Ni₆₀Zr₄₀ alloy was after 10 h which is half the 20 h for Ni–Nb alloys. This observation is consistent with the criterion suggested by Suryanarayana et al. ¹²⁴ as there exists much more intermetallic phases in Ni–Zr system compared to Ni–Nb system. Ni–Nb and Ni–Zr exhibit different thermodynamic properties, atomic size mismatch and plasticity behaviour. All of these parameters can influence the amorphisation reactions during MA. The start time of the amorphisation reaction is mainly determined by the rate at which the smaller atom Ni can diffuse in the lattice of Zr (or Nb). Since the atomic radius of Zr (1·60 Å ) is bigger than that of Nb (1·47 Å), it is predicted that Ni has a higher diffusivity in Zr than Nb. Furthermore the heat of mixing for the Ni–Zr system is −51 kJ mol⁻¹ which is more negative than −32 kJ mol⁻¹ for Ni–Nb system. This implies that a higher thermodynamic driving force is available for intermixing and thus amorphisation in the Ni–Zr system which favours the reaction kinetics by increasing the chemical diffusivity. Moreover, Zr attains a grain size of 21 nm during MA of Ni₆₀Zr₄₀, while Nb exhibits a larger grain size of 35 nm during MA of Ni₆₀Nb₄₀ after the same milling time of 10 h. ⁷ The higher volume fraction of grain boundaries in Zr aids the kinetics of the amorphisation reaction by accelerating the diffusion of Ni in Zr. The lower diffusion rate of Ni in Nb compared with Zr means that the amorphisation reaction in Ni–Nb alloys is postponed until an adequate density of lattice defects (i.e dislocations, grain boundaries) is created by continued milling. It is worth noting that the time for completion of amorphisation (t _e) was 50 h for Ni₆₀Zr₄₀ and 85 h for Ni₆₀Nb₄₀ compositions. ⁷

Effect of milling conditions on amorphisation

Several variables of the milling process can be critical in controlling amorphisation, including milling energy, weight of powder, milling temperature and time.

There are several examples of systems in which the fraction of amorphous phase for a given composition is a function of milling energy and/or temperature. For example, Eckert et al. ^13,43,125 demonstrated that if the milling energy was too high, Ni–Zr alloys which were amorphised at a low milling energy would exhibit crystalline structure. The effect of milling temperature and energy on amorphisation of some Ni–Zr compounds was also studied by Chen et al. ¹²⁶ They concluded that the ultimate fraction of the amorphous phase decreases as the milling temperature increases. Moreover, the final fraction of amorphous phase increases with increasing milling energy. Those compounds with a higher hardness exhibited a greater fraction of the amorphous phase for the same milling conditions. Similar results have also been reported by Padella et al. ¹²⁷ for Pd–Si, Suryanarayana et al. ¹²⁸ for Ti–Al, Yamada and Koch ⁸ for TiNi, Takeuchi et al. ²⁰ for Ni–Zr and Murty et al. ¹⁷ for Ti–Ni and Ti–Cu alloy systems.

These observations have been explained in terms of two competing processes occurring during the cyclic deformation introduced by ball milling, ^8,129 namely: (i) the creation of lattice defects which increases the free energy of system and (ii) recovery which reduces the density of defects and therefore, the free energy of the system. The milling time for attaining a certain fraction of amorphous phase is determined by the balance between the relative rates of these two processes. Increasing milling energy and/or ball-to-powder weight ratio causes a higher rate of creation of defects, whereas higher milling temperature increases the rate of recovery and removal of defects. Consequently, different equilibrium rates and different end products can be obtained. The greater fraction of the amorphous phase for harder materials is related to the higher rate of work hardening in these materials which enhances the rate of creation of defects. ¹²⁶ If the milling energy and/or ball-to-powder weight ratio is chosen to be too high, either a partially amorphised or a crystalline structure is obtained. Eckert et al., ¹³ Suryanarayana et al. ¹²⁸ and Takeuchi et al. ²⁰ argued that this is probably due to the generation of more heat during milling, which promotes the recovery processes or causes subsequent crystallisation of the amorphous phase.

Various attempts have been made to understand the crystalline-to-amorphous transition during MA in terms of milling energy. Abdellaoui and Gaffet ¹³⁰ and Magini et al. ¹³¹ showed that neither the impact energy nor the frequency of impacts govern the amorphisation process in a planetary ball mill if taken separately, but the shock power (the product of impact energy and the frequency of impacts) determines the outcome of milling process. In several researches, it has been shown, for different materials, that the total energy imparted during milling is responsible for the amorphisation and formation of intermetallics during MA rather than the impact energy of the individual balls. ^17,132–134

It should be noted that the total energy of milling proposed by Murty et al. ¹⁷ additionally includes further parameters such as the total number of balls, time of milling and weight of powder, in addition to the parameters contributing to shock power.

Various attempts have been also made to construct 3D or 2D milling maps for prediction of MA amorphisation in terms of milling parameters and thus identify the conditions required for amorphisation during MA. These maps involve different parameters such as impact energy and cumulative kinetic energy, ^135,136 power absorption and degree of filling, ¹³¹ ball-to-powder weight ratio and milling time, ¹²⁸ and the impact energy of the ball and the total energy. ^17,133

Bhatt and Murty ¹³⁷ recently reported an attempt to synthesise amorphous phase by MA in a number of BMG forming compositions in Zr, Ti, Fe, Cu and Ni based systems to develop milling maps/energy maps for the formation of amorphous phase. Milling parameters such as milling speed, ball-to-powder weight ratio have been varied over wide ranges in order to have a wide range of milling energies. The milling maps in Fig. 14 show the energy domains for amorphisation from the elemental blends. The results indicate that total energy of milling has a more decisive role than the impact energy of the balls in the formation of the amorphous phase by planetary ball mill. Milling energies in excessive of that required (shown as second dashed line on the right) for the formation of amorphous phase will lead to in situ crystallisation of the amorphous phase. This is caused by either the raise in temperature during milling, or due to large number of defects generated, which can enhance the diffusivity. Thus, there is definitely an energy window in which amorphous phase forms during MA. The energy required for amorphisation falls in a range of 500–600 J g⁻¹ for all the compositions studied. Further amorphisation in all the compositions has been observed to occur below a critical crystallite size of about 20 nm as observed in Fig. 15. ¹³⁷

OPEN IN VIEWER

14.

Milling maps indicating the energy required for amorphisation in a Zr₆₅Al_7·5Cu_17·5Ni₁₀ and b Fe₅₆Co₇Ni₇Zr₁₀B₂₀ ¹³⁷

OPEN IN VIEWER

15.

Crystallite size of various systems showing nano-formation as prerequisite of amorphisation ¹³⁷

Amorphous alloy-based composite materials

In addition to the use of amorphous alloys as a single-phase material, there exists a broad range of composite materials that can be obtained by introducing crystalline phase(s) in the amorphous structures. The amorphous alloys exhibit useful properties such as high strength, high hardness, large elastic limit, good corrosion resistance, and superplasticity at high temperature. However, deformation of amorphous alloys by shear bands and their sudden failure similar to a brittle material is a barrier for using these materials in load-bearing applications. ¹³⁸ Earlier studies demonstrated that preparation of amorphous nanocomposite, a microstructure consisting of nanocrystalline phases in amorphous matrix, is promising for improving the ductility of amorphous alloys. ¹³⁹ Basically, such a structure can be readily achieved by (i) partial crystallisation either directly by quenching from the melt or through annealing of amorphous precursor structures, (ii) partial amorphisation during ball milling, and (iii) blending a metallic glass with insoluble metallic (W particles ^140,141 ) or ceramic particles (oxide ^140,142 carbide ¹⁴³ or nitride ¹⁴⁴ ). The crystallisation process can be executed by thermal treatment or alternatively by plastic deformation. Both options can be applied during a separate subsequent process or during consolidation of amorphous powder to produce bulk materials.

The mechanical properties of amorphous-based nanocomposite have been found to be very promising as they possess both outstanding tensile strength and good ductility at room temperature. Hence, this class of materials seems to be superior to single-phase nanocrystalline materials which typically exhibit little ductility in tension for grain sizes smaller than about 25 nm.

For example, Kim et al. ¹⁴⁵ and Inoue et al. ¹⁴⁶ developed amorphous Al₈₈Ni₉Ce₂Fe₁ based nanocomposite by partial crystallisation of the melt-spun amorphous ribbon. The microstructure of partially crystallised alloy consists of 25 per cent volume fraction of fcc-Al particles with a size of 3–5 nm and a mean distance of 10 nm dispersed within an amorphous matrix. While the tensile fracture strength of Al–Ni–Ln (Ln = Y, La or Ce) amorphous has been reported to exhibit levels up to 1000 MPa, by partially crystallising A1₈₈Ni₉Ce₂Fe₁ amorphous alloys the tensile strength reaches as high as 1560 MPa which is about three times more than that of conventional high-strength Al alloys. Nano-sized fcc-Al particles are yielded by controlled primary crystallisation of amorphous aluminium alloys. Similar microstructure are possible when using mechanically alloyed amorphous powders as starting materials. ¹⁴⁷ Inoue and Kimura ¹⁴⁸ showed that five types of non-equilibrium structures can be obtained by controlled crystallisation of Al-based amorphous alloys including; (1) nanostructure consisting of crystalline Al and compounds, (2) partially crystallised structure of nanoscale fcc-Al particles embedded in an amorphous matrix, (3) nano-quasi-crystalline structure consisting of nanoscale quasi-crystalline particles surrounded by Al phase without grain boundary, (4) coexistent nanogranular amorphous and Al phases, and (5) nanogranular Al phase surrounded by an amorphous network. It is notable that the density of nanocrystal in amorphous matrix can attain high levels of 10²¹–10²⁵ m⁻³ to yield ultra-high strength. These non-equilibrium Al-based alloys exhibit much better mechanical properties as compared to conventional crystalline alloys developed up to date. Figure 16 schematically illustrates the microstructure and the mechanical strengths of these structures. ¹⁴⁸

OPEN IN VIEWER

16.

Schematic illustration of microstructure and mechanical strengths of five types of non-equilibrium structures obtainable by crystallisation of Al-based amorphous alloys. ¹⁴⁸

Similar microstructures have been developed in other amorphous alloys. For example, in partially crystallised Mg-rich alloys with more than 80 at-% Mg, nanoscale hcp particles (5–10 nm) are dispersed homogeneously in the amorphous matrix with a typical interparticle spacing of about 3–10 nm. ^149,217 This structure provides higher tensile strength, hardness, Young’s modulus and better room temperature ductility compared to the fully amorphous alloy. For Mg–Zn–La alloys the ultimate room temperature tensile strength increases from about 600 MPa for the amorphous phase to about 700 MPa for nanocomposite structure containing nanoscale hcp Mg particles. ^149,217 Higher room temperature ductility is attributed to the homogeneous dispersion of ultrafine (5–10 nm) hcp particles embedded in the amorphous matrix which act as effective barriers for local shear deformation and assist homogeneous plastic deformation as long as the amorphous matrix phase exhibits some ductility. Similar results have been reported in case of Zr-based amorphous nanocomposite prepared by partial crystallisation of amorphous structure. ^150–157

To obtain a high density of ultrafine crystalline particles, the crystallisation process should occur with a high nucleation rate while the growth rate being kept at a low level. Generally, it is also possible to produce a nanocrystalline structure directly by proper cooling of a melt in which the solidification proceeds by homogeneous nucleation. However, the released latent heat during solidification reduces the extent of supercooling and therefore the nucleation rate. The key strategy in achieving the nanocrystalline structure by crystallisation route is to provide a high nucleation rate, while the latent heat of crystallisation releases with low rate. These conditions can be readily accomplished during crystallisation of a solid amorphous structure where the crystallisation occurs at low temperature around T _g at which the growth rate is quite small. ¹⁵⁸

The amorphous-based nanocomposite by ball milling route can be achieved via several approaches as summarised in Fig. 17. Incomplete ball milling is the most straightforward approach. This leaves ultrafine crystalline phase(s) of constituents in the amorphous matrix. Figure 18 shows a typical microstructure obtained by incomplete amorphisation reaction in case of MA of Ni₆₀Nb₄₀ alloy. In this approach, the extent of amorphisation is either thermodynamically or kinetically limited by proper choosing the composition or milling parameters. Crystalline phase(s) can be also introduced in amorphous matrix by ball milling of amorphous phase constitutes with insoluble metallic or ceramic particles as reinforcement phase. The insoluble particles are repeatedly broken to smaller size and dispersed in the matrix simultaneously with development of amorphous structure. Yet another way for making amorphous-based nanocomposites by ball milling is to utilise the reaction of constituents with milling atmosphere or the reaction between constituents themselves (MX+R→RX+M) to obtain the reinforcing phase (RX). For example, MA of Mg–Y–Cu powders led to in situ development of amorphous-Y₂O₃ nanocomposites during processing. ¹⁴³ XRD and TEM investigations revealed that the structure of powders contains nanoscale unreacted elemental material and traces of 10–20 nm Y₂O₃ crystals homogeneously embedded in the amorphous matrix. Oxide phases are introduced mainly because of a reaction between constituents and milling atmosphere; also, they might have resulted from fracturing and dispersion of pre-existing oxide layer on surface of powder particles. ¹⁴³ It was found that the thermal stability of the amorphous matrix alloy is not markedly affected by the presence of the nanoscale Y₂O₃ dispersoids. The powders exhibit a supercooled liquid region of about 40–50 K, similar to that of melt-spun ribbons, ^143,159,160 allowing an easy consolidation of these powders by hot pressing or extrusion in the viscous state above T _g. The Vickers hardness and fracture strength of consolidated samples were reported to be about 345–385 HV and 1150–1280 MPa depending on composition. ^141,159,161 These values are significantly higher than those for as-quenched amorphous or partially crystallised ribbons or cast bulk specimens. ^162,163 Mg-based amorphous nanocomposite has been also synthesised by ex situ addition of MgO, CeO₂, Cr₂O₃, or Y₂O₃ oxide particles. ¹⁶⁴ In this way, the addition of oxide particles is not limited to a small volume fraction as is the case with in situ processing. Zr-based amorphous nanocomposites were also prepared directly via MA, by blending the metallic constituents of the amorphous phase with insoluble oxide, ^140,142 carbide ¹⁴³ or nitride ¹⁴⁴ particles. The high density of nanoscale metallic or oxide particles, i.e., a small interparticle spacing, provides superior mechanical strength after MA and also after consolidation. Dastanpoor et al. utilised mechanochemical reaction between CuO and Al powders to obtain Cu–Zr–Al/Al₂O₃ amorphous nanocomposite. ¹⁶⁵ Cu₆₀Zr₄₀ intermetallic compound was first synthesised by melting of Cu and Zr. The Cu₆₀Zr₄₀ intermetallic compound was then ball milled with Al and CuO powders. The MA of Cu₆₀Zr₄₀, Al and CuO powder mixture for 10 h led to the reaction of CuO with Al, forming Al₂O₃ particulate, and concurrent formation of Cu₆₂Zr₃₂Al₄ amorphous matrix. Nanoindentation tests showed that Cu₆₂Zr₃₂Al₄–10 vol.-% Al₂O₃ amorphous nanocomposite had a hardness value of 837 HV and an elastic modulus of 90 GPa which are higher than those for Cu₆₀Zr₄₀ and Cu₅₀Zr₄₃Al₇ amorphous alloys. In another work, nanocomposite structure containing amorphous matrix surrounding nanocrystals was synthesised by MA of (Ti_69·7Nb_23·7Zr_4·9Ta_1·7)_100−x Fe _x for x = 2 and 6 whereas for x = 0 and 10 full nanocrystalline and full amorphous structures were achieved, respectively. ¹⁶⁶ Kumaran et al. ¹⁶⁷ developed TiAl–Nb₂Al nanocomposite powders by high-energy ball milling from a mixture of prealloyed TiAl, niobium, aluminium and SiC powders. The final microstructure included nanocrystalline TiAl, amorphous phase and TiAl–Nb₂Al intermetallic nanocomposite powders with varying Nb and Al concentrations. Amorphous-based nanocomposite containing Mg₂Ni phase, some residual Ni and an amorphous phase was prepared by MA of a mixture of Mg and Ni with an atomic ratio of 2 : 1 at a certain milling intensities. ¹⁶⁸ This structure had improved hydrogen storage characteristics. Rodríguez and Botta ¹⁶⁹ performed high-energy milling on Al₉₀Fe₅Nb₅, Al₉₀Fe₇Nb₃, Al₉₀Fe₅Cr₅ and Al₉₀Fe₇Zr₃ alloys to produce Al-based amorphous alloys. Milling of all composition resulted in partially amorphised structure containing intermetallic compound. These results clearly demonstrate that nanocomposites with an amorphous matrix and metallic/non-metallic nanocrystals can be directly obtained during ball milling process by careful adjusting of powder compositions as well as milling parameters (i.e. intensity, time, temperature and atmosphere of milling). On the other hand, amorphous powders prepared by ball milling can be used as a precursor for producing nanocomposites through controlled devitrification.

OPEN IN VIEWER

17.

Different possible routes for synthesis of amorphous-based nanocomposite by ball milling process

OPEN IN VIEWER

18.

Transmission electron microscopy (TEM) images and corresponding selected are diffraction patterns of Ni₆₀Nb₄₀ powder particles after 50 h of milling time ⁷