Crystallization of Mechanically Alloyed Ni₅₀Ti₅₀ and Ti₅₀Ni₂₅Cu₂₅ Shape Memory Alloys

Parameters of manufacturing process, applied for alloys production, highly impact their structure. This effect is clearly visible when mechanical alloying is used, in which the material can be obtained in amorphous and/or nanocrystalline form. Shape memory effects do not occur in the amorphous state. However, the use of thermal treatment gives the possibility to shape the appropriate grain size. Thus, the shape memory effects can be controlled. Crystallization of amorphous materials is complex phenomenon due to nucleation and/or growth processes of residual precursor nuclei introduced into the alloy during either amorphization or due to impurities introduced through manufacturing method (Ref 1). Hence, understanding the crystallization process in amorphous alloys is crucial to controlling and/or obtaining a nanocrystalline or micrograins.

NiTi-based alloys can be obtained in amorphous state by various methods—repeated cold rolling (Ref 2), severe plastic deformation (Ref 3), high-pressure torsion (Ref 4), melt spinning (Ref 5) and mechanical alloying (Ref 6‐8). Adjustment of parameters such as grinding time, grinding speed or ball-to-material ratio allows acquiring either amorphous alloy and/or alloy in nanocrystalline form. Specially, the nanocrystalline form gives the possibility to influence the course of the reversible martensitic transformation via controlling the grain size.

Therefore, in the presented work, we focused on a detailed analysis of the crystallization process taking place in amorphous Ni₅₀Ti₅₀ and Ti₅₀Ni₂₅Cu₂₅ alloys produced by high-energy ball milling, using non-isothermal approach. In the crystalline state, both alloys show the presence of the one-step reversible martensitic transformation; however, it differs in its final product: in the Ni₅₀Ti₅₀ alloy, the low-temperature phase is the B19’ monoclinic martensite and in the case of Ti₅₀Ni₂₅Cu₂₅—the B19 orthorhombic one.

Figure 1 (top) shows x-ray diffraction patterns measured for freely mixed powders used as a starting mixture for the high-energy ball milling. The patterns consisted from diffraction lines characteristic for individual elemental powders. After 100 h of grinding in the high-energy ball mill (HEBM), the powders were alloyed. This was confirmed from the measured diffractograms (Fig. 1—middle). Diffraction lines from individual alloying elements disappeared, and instead of them two diffused maxima appeared. The extended value of the full width at half maximum (FWHM) indicated the presence of an amorphous phase in both alloys. However, detailed observations carried out using the HRTEM technique revealed that, in addition to the amorphous phase, nanocrystalline regions were formed (Ref 11).

Figure 2 shows an example of HRTEM image observed for the Ni₅₀Ti₅₀ alloy milled for 100 h. Similar images were observed for the Ni₂₅Ti₅₀Cu₂₅ alloy. It can be seen that large part of the volume is amorphous, in which the nanocrystalline particles of the bcc phase were formed. After annealing, done at temperature above 777 K, in both alloys the crystalline state was confirmed from XRD studies in both alloys (Fig. 1—bottom). In the case of the crystallized Ni₅₀Ti₅₀, the presence of the B2 parent phase was identified. In addition, a small amount of the B19’ monoclinic martensite was found. Due to the fact that the Ni₂₅Ti₅₀Cu₂₅ alloy shows the presence of B2-B19 martensitic transformation, at range of temperatures between 310 and 350 K, the B19 orthorhombic martensite was identified. Also, an amount of the Ti₂Ni equilibrium phase was present in the alloy.

In order to characterize crystallization process, the DSC heating curves were measured with different heating rates. Results are shown in Fig. 3. From the measured thermograms, characteristic temperatures of the crystallization peak were determined. In general, both alloys revealed similar thermal behavior. Regardless of the heating rate, on the thermograms, there was a single crystallization peak in the temperature range from 777 to 821 K and from 758 to 827 K for the Ni₅₀Ti₅₀ and Ti₅₀Ni₂₅Cu₂₅ alloys, respectively. As the heating rate increased, the characteristic temperatures for crystallizations peak increased their values. The introduction of 25 at% of copper, instead of the nickel into the alloy, lowered the crystallization temperature of about 18 degrees with respect to the binary alloy. In contrast, the temperature of the end of crystallization for both alloys showed similar values. Moreover, the temperature range of the crystallization process was wider in the case of the Ti₅₀Ni₂₅Cu₂₅ alloy in comparison with the Ni₅₀Ti₅₀ of about 20 degrees. In both alloys, its value increased depending on the increase in the heating rate. Also significant parameter of the transformations is an enthalpy, which in the case of a ternary alloy (depending on the heating rate ranged from 71 to 78 J/g) was almost twice as high as for a binary alloy (from 33 to 38 J/g). However, the most important parameter, from the point of view of determining the activation energy, was the peak temperature. The crystallization peak temperature T_p ranged from 788 K at slowest heating rate to 813 K for the fastest one in the Ni₅₀Ti₅₀ alloy. Similar behavior has been observed for the Ti₅₀Ni₂₅Cu₂₅ alloy. Value of the peak temperature increased from 777 K at heating rate of 5 K/min to 802 K at 40 K/min.

Using data obtained from the measured DSC heating curves, the Kissinger and Ozawa plots were prepared. From slope of the ln(β/T_p²) or ln(β) versus 1/T plot (Fig. 4), the activation energy (E_a) was determined (β—heating rate; T_p temperature at maximum of a thermal peak). Activation energy calculated for the Ni₅₀Ti₅₀ alloys was 410 kJ/mol (Kissinger) and 424 kJ/mol (Ozawa), for the Ti₅₀Ni₂₅Cu₂₅ alloy: 415 and 429 kJ/mol, respectively. Regardless of the applied calculation model, the determined values of the activation energy values are comparable. Slightly lower energies were obtained for the Ni₅₀Ti₅₀ alloy. It results from the lower crystallization temperatures (determined for the same heating rates) of the Ti₅₀Ni₂₅Cu₂₅ alloy in relation to the Ni₅₀Ti₅₀. This fact indicates a greater influence of the alloying technology on activation energy than the difference in a chemical composition of alloys. The key role plays the state of internal stresses and accumulated energy resulted in the increased number of structural defects such as dislocations and/or grain boundaries modified during manufacturing process. This dependence can be traced from the literature data listed in Table 1. For the binary NiTi alloys (with chemical composition close to equiatomic one) processed by heavy plastic deformation methods (HPD), such as cold rolling or high-pressure torsion, values of the activation energy range from 259 to 285 kJ/mol. It should be remembered that in the alloys processed by HPD, not all of the volume goes into an amorphous state. Depending on the HPD technique, some of the material remains in the nanocrystalline or ultrafine grains (Ref 2, 4, 12). In addition, the crystalline regions are strongly defected, which contributes to the increase in the internal state of stress and hence, the state of the stored energy. Completely different behaviors can be found in the alloys produced by magnetron sputtering (Ref 13-15). They are completely amorphous. Values of the activation energy are much higher and range from 403 to 416 kJ/mol. This means that their ability to crystallization is less than alloys processed by HPD methods. Gasperini et al., (Ref 16) basing on his results obtained from studies done on similar NiTi alloy manufactured by mechanical alloying (different proportions of ball to powder—5:1 and during shorter milling time), reported that E_c = 277 kJ/mol. This lower value of activation energy can be attributed to slightly different methodology of obtaining amorphous Ni46Ti54 alloy (wg%), introducing more internal stresses contributing to receiving less stable amorphous phase. In the case of NiTiCu alloys, which were produced by melt-spinning technique, the authors (Ref 12, 17, 18) obtained completely amorphous ribbons. Thus, the activation energy ranged from 350 to 406 kJ/mol. These discrepancies can be a result of the cooling rate realized in the melt spinning, which may affect in inhomogeneity of the chemical composition. The values of activation energy obtained in our research are comparable to alloys that exhibit the presence of the amorphous phase and require higher activation energy for crystallization. This may mean that extending of the grinding time up to 100 h leads to amorphisation and a decrease in internal stresses. It has been known from the literature that, in high-energy ball mills, during the collision of the balls with the material, the temperature is increased locally—even to receiving melting effect at the particle interface (Ref 19, 20). This may facilitate diffusion transport of alloying elements and contributes to the dynamic recovery of deformed particles.

The use of high-energy ball milling, with an extended grinding time to 100 h, enabled to obtain both the Ti₅₀Ni₅₀ and Ti₅₀Ni₂₅Cu₂₅ alloys in an amorphous form with few crystallized nano-areas. Despite the differences in the chemical composition of the alloys, they revealed similar values of activation energy. The ongoing process of crystallization requires activation energy of approximately 410 kJ/mol. Such a relatively high value of energy proves the high stability of the amorphous phase. In general, initiation of the crystallization process requires raising the temperature above 777 K.