The Effect of Thermal Treatment on Microstructure and Thermal-Induced Martensitic Transformations in Ni₄₄Fe₁₉Ga₂₇Co₁₀ Single Crystals

Abstract

Heat treatments of single crystals of Ni₄₄Fe₁₉Ga₂₇Co₁₀ (at.%) shape memory alloys cause various microstructures of the high-temperature phase. The nanodomain structure, consisting of regions of the L2₁- and B2-phases, and nanosized particles are the main parameters that change during heat treatments and determine the mechanism of nucleation and growth of martensite crystals, the size of thermal-induced martensite lamellae, the temperature M_s, and the temperature intervals of the martensitic transformation. In the as-grown single crystals, the high-temperature phase has only the L2₁-structure and the MT occurs at low (M_s = 125 K) temperatures due to the motion of the practically single interphase boundary in narrow temperature ranges of 3–7 K. The reduction in the volume fraction of the L2₁-phase to 40% and the formation of nanodomains (20–50 nm) of the L2₁-and B2-phases due to annealing at 1448 K for 1 h with quenching causes an increase in the MT temperatures by 80 K. The MT occurs in wide temperature ranges of 40–45 K because of multiple nucleation of individual large (300–500 µm) martensite lamellae and their growth. After aging at 773 K for 1 h, the precipitation of nanosized particles of the ω-phase in such a structure additionally increases the MT temperatures by 45 K. The MT occurs due to the multiple nucleation of packets of small (20–50 μm) martensite lamellae.

1. Introduction

NiFeGa-based alloy is one of the most widely studied alloys among the Heusler alloys undergoing thermoelastic martensitic transformations (MT). Over the past decade, NiFeGa alloys with cobalt additives have sparked more interest. The shape memory effect (SME) and superelasticity (SE) have been carefully studied on NiFeGaCo alloys with C_Co = 0–6 at.%. These alloys have a wide SE temperature range (400 K in tension and 250 K in compression), large reversible strain (up to 13.5% in tension and 6.2% in compression), high MT temperatures (273–300 K), narrow thermal and stress hysteresis (3–7 K, 3–5 MPa), etc. [1,2,3,4,5,6]. Recent studies have shown that alloying with cobalt more than 6 at.% leads to a dramatic decrease in MT temperatures, up to the complete suppression of MTs at a cobalt content of more than 10 at.% [7,8,9]. In this case, the large number of point defects causes a strain glass transition (from an unfrozen strain glass state into a frozen strain glass state) during cooling, which results in a strong distortion of the crystal lattice. Of particular interest is the study of Ni₄₄Fe₁₉Ga₂₇Co₁₀ (at.%) alloys with the cobalt content close to that of strain glass alloys and the possibility of thermal-induced MT. It is assumed that a large number of point defects will have a significant effect on the microstructure during thermal treatments (annealing and aging). It will change the parameters of the second-phase particles and will have a great impact on MT temperatures after aging in comparison with cobalt-free NiFeGa alloys. Such studies have not been carried out to date. The authors of this paper are aware of only one piece of research [7] on Ni₃₅Fe₁₈Ga₂₇Co₂₀ fibers: low-temperature aging at 523–673 K leads to the precipitation of ω-like particles and to a change in the stress level of martensite formation. Thus, systematic studies of the effect of thermal treatments on MTs and the microstructure in bulk NiFeGaCo alloys C_Co >10 at.% are necessary and topical. The current research has been carried out on Ni₄₄Fe₁₉Ga₂₇Co₁₀ (at.%) single crystals.

2. Materials and Methods

The Ni₄₄Fe₁₉Ga₂₇Co₁₀ (at.%) single crystals were obtained using the Bridgman method. The studies were conducted on the as-grown single crystals and crystals after annealing at 1448 K for 1 h with subsequent water quenching (quenched crystals). The quenched single crystals were aged (aged crystals) at 773 K for 1 h; aging was followed by quenching. The choice of the annealing and aging temperatures was based on our previous studies of NiFeGaCo alloys with cobalt content of 0–6 at.% and preliminary exploratory research [10,11,12,13,14,15,16]. The choice of the annealing temperature was determined by the formation of a single-phase microstructure; the aging temperature was chosen as aging at 773 K provides the maximum start temperature of MT, M_s [15,16]. The temperature dependence of the electrical resistance was used for determining the MT temperatures. The order–disorder transition temperature was determined by calorimetric curves obtained with a DSC 404 F1 calorimeter. For metallographic observations, a universal Keyence VHX 2000 complex with a Kammrath and Weiss console for in situ stretching was used. The electron microscopy studies were conducted with Hitachi HT-7700.

3. Results

Figure 1 represents the bright-field and dark-field images of the as-grown single crystals (Figure 1a,b) and single crystals after thermal treatments (Figure 1c–f) [10]. The γ-phase particles (with face-centered cubic lattice) are observed in the as-grown crystals. The size of the particles is 200–500 nm and volume fraction is ~1–2%. The superstructure <111>_L21 reflexes in the selected area electron diffraction pattern (SAEDP) is evidence of the L2₁-structure. In these reflexes, the thermal antiphase boundaries can be observed in the dark-field images.

The quenched crystals do not contain secondary phases (Figure 1c,d) [10]. The SAEDP with the [011]_L21 zone axis also demonstrated <111>_L21 reflexes. The domains of 20–50 nm in size were observed in the dark-field images from the <111>_L21 reflexes. The bright domains (L2₁-phase) are separated by the dark regions (B2-phase) with a similar size. Hence, the high-temperature phase of the quenched crystals is a mixture of B2 and L2₁-phases.

In the B2-lattice, Ni atoms are located on one sublattice, and Fe and Ga atoms are located on the other one. Fe and Ga atoms are randomly located on the sites of the sublattice (not ordered) [17]. Co atoms can occupy the positions of Ni and Fe, since the radius of Co (0.1252 nm) is close to the radii of Ni (0.1246 nm) and Fe (0.1274 nm) and less than the radius of Ga (0.1353 nm) [18,19,20,21]. Cobalt additives change the distances and exchange interactions between pairs of Ni–Fe atoms to Co–Ni and Co–Fe. As a result, the elementary cell is distorted. The difference between the L2₁-lattice and B2-lattice lies in the ordered arrangement of Fe and Ga atoms at the sites of the sublattice. The volume fraction of L2₁-domains derived from the dark-field images is estimated to be 40% in the quenched crystals. The same mixture has also been observed in NiFeGa alloys [12,13,22] and in NiCoMnIn [23,24] alloys after annealing at high temperature above the order–disorder transition temperature.

Aging at 773 K for 1 h was carried out on the crystals after annealing and quenching. First, after aging, the domains of the B2- and L2₁-phases increased up to 100–200 nm (Figure 1f) compared to the quenched state (Figure 1d). The estimate of the volume fraction of the L2₁-phase is carried out on the basis of the dark-field images. The volume fraction makes up 60–65%. Secondly, during aging, the dispersed particles of 70–100 nm with a volume fraction of 4–6% precipitated (Figure 1e). In this case, the reflexes from γ or γ’-phases, typical of NiFeGa(Co) alloys, were not detected. Instead, the SAEDPs obtained from the matrix and particles show the reflexes at the 1/3 and 2/3 {224}_L21 positions (marked by the circles in Figure 1e). The dark fields obtained in these reflexes represent particles. These reflexes from particles were also observed by other authors [7], where low-temperature aging of Ni₃₅Co₂₀Fe₁₈Ga₂₇ fibers was carried out at 523–673 K. It was shown in [7] that these particles have a hexagonal ω-like structure.

The recorded changes in the microstructure after all heat treatments have a strong effect on the MT temperatures and the morphology of the thermal-induced martensite. The MT temperatures are determined from the temperature dependences of the electrical resistance (Figure 2) and are presented in

Table 1. Characteristic MT temperatures for Ni₄₄Fe₁₉Ga₂₇Co₁₀ as-grown single crystals, after quenching and aging.

Thermal Treatment	Ms, K	Mf, K	As, K	Af, K	Δ1 = Ms − Mf, K	Δ2 = As − Af, K	ΔT = As − Ms, K
As-grown single crystals	125	122	139	146	3	7	21
Annealing at 1448 K for 1 h with quenching	203	163	173	218	40	45	15
Annealing + aging at 773 K for 1 h with quenching	251	217	228	262	34	34	21

. The martensite morphology during in situ cooling is shown in Figure 3. In the as-grown crystals, the MT has an explosive nature, with narrow intervals Δ₁ = M_s − M_f = 3 K and Δ₂ = A_s − A_f = 7 K. Cooling a few degrees below M_s causes an instantaneous growth of a large volume fraction of martensite (1–2 lamellae). The MT occurs due to the rapid interfacial boundary motion.

Annealing and quenching leads to an increase in the temperature M_s by 80 K and a significant broadening of the MT temperature intervals up to 40–45 K; it is necessary to significantly supercool below M_s to obtain a large volume fraction of martensite. In the quenched crystals, several martensite lamellae of 300–500 μm in size appear in a wide temperature range in contrast with the initial as-grown crystals.

Aging of the quenched crystals additionally increased the MT temperatures by 45 K. In the aged single crystals containing fine particles, the martensite lamellae are significantly reduced in size to 20–50 µm. A similar change in the morphology of martensite during the precipitation of nanosized particles was also observed in NiFeGa alloys during aging at 773 K for 1 h [15,16]. In the aged single crystals, the MT occurs in a wide temperature range of 34 K, as in the quenched crystals. However, in the quenched crystals, several martensite lamellae appear whose size increases upon cooling, whereas in the aged crystals martensite appears as packets of small lamellae, whose size does not change upon cooling.

A possible martensitic phase that is realized in Ni₄₄Fe₁₉Ga₂₇Co₁₀ crystals is L1₀, which follows from our previous work [10]. Layered structures (10 M, 14 M) were not observed during stress-free cooling in opposite to NiFeGa(Co) alloys with C_Co = 4–6% [1], due to low MT temperatures and difficulties with stabilizing martensite (tetragonal martensite L1₀ is always the final phase).

4. Discussion

4.1. Martensitic Transformations in As-Grown Single Crystals

In the initial as-grown Ni₄₄Fe₁₉Ga₂₇Co₁₀ single crystals, the MTs are characterized by very low temperatures determined by the cobalt content. In the as-grown single crystals, the MT temperature is 150 K lower than in the cobalt-free single crystals [1,2,3,4,5,6]. This is in agreement with the works [9,19,25], which state that an increase in the cobalt content leads to a strong decrease in the temperature M_s.

The γ-phase particles contained in the initial crystals have little effect on the nucleation and growth of martensite crystals due to their low volume fraction. The highly ordered homogeneous structure of single crystals makes it possible to observe large thermal-induced martensite lamellae (Figure 2), which grow instantly with a large volume fraction in narrow temperature ranges of 3–7 K. In accordance with the Tong–Weiman classification, MTs differ in the ratio of the contributions of the reversible and dissipated energies to the balance of the driving forces of the transformation. Based on [26], the reversible and dissipated energies can be written as:

$$\Delta {\mathrm{G}}_{\mathrm{rev}}\left(1\right)=\left({\mathrm{M}}_{\mathrm{s}}-{\mathrm{M}}_{\mathrm{f}}\right)\frac{\Delta {\mathrm{S}}_{\mathrm{ch}}}{2}+\left({\mathrm{A}}_{\mathrm{s}}-{\mathrm{A}}_{\mathrm{f}}\right)\frac{\Delta {\mathrm{S}}_{\mathrm{ch}}}{2}, \Delta {\mathrm{G}}_{\mathrm{irr}}=\left({\mathrm{A}}_{\mathrm{f}}-{\mathrm{M}}_{\mathrm{s}}\right)\frac{\Delta {\mathrm{S}}_{\mathrm{ch}}}{2}.$$

(1)

Therefore, their ratio is:

$$\Delta {\mathrm{G}}_{\mathrm{rev}}\left(1\right)/\Delta {\mathrm{G}}_{\mathrm{irr}}=\frac{{\Delta }_1+{\Delta }_2}{2\Delta \mathrm{T}}.$$

(2)

In the as-grown single crystals, the energy ratio $\Delta {\mathrm{G}}_{\mathrm{rev}}\left(1\right)/\Delta {\mathrm{G}}_{\mathrm{irr}}$ = 0.23 < 1, which corresponds to the first type according to the Tong–Weiman classification. This means that practically no elastic energy is stored during the MT, and the reverse MT begins at temperature A_s, which is higher than the temperature M_s. Such an energy balance leads to narrow intervals and an “explosive nature” of the MT, which makes it possible to consider the MT in the as-grown single crystals as a transformation with almost a single interfacial boundary, which is confirmed by the optical metalography (Figure 3a). The hysteresis in this case is due to friction forces during the interface motion.

4.2. Martensitic Transformations in Quenched Single Crystals

High-temperature annealing at 1448 K for 1 h and subsequent quenching change the microstructure of the single crystals (Figure 1). As has been shown above, after annealing and quenching, the γ-phase particles dissolve. This has little effect on the MT temperatures and the mechanism of martensite nucleation and growth because the volume fraction of the γ-phase dissolved during annealing is very small, only ~1–2%. The most significant change in the microstructure, which affects the temperature M_s, morphology, and the mechanism of nucleation and growth of martensite, is the formation of the nanodomains of B2- and L2₁-phases.

Firstly, after annealing at 1448 K and quenching, the volume fraction of the L2₁-phase decreased from 100 to 40% and regions of the B2-phase appeared. The decrease in the volume fraction of the L2₁-phase is due to the fact that annealing was carried out at a temperature higher than the order–disorder transition temperature, which was determined by differential scanning calorimetry (Figure 4). Thus, the temperature is 930 K. The obtained value is close to the values of the order–disorder transition temperature in NiFeGa(Co) alloys [20,22,23].

A number of studies have shown that the transition from the second-neighbor ordered phase (L2₁) to first-neighbor ordered phase (B2) leads to an increase in MT temperatures [12,13,22,27,28,29]. In accordance with these works, a decrease in the order degree of the high-temperature phase reduces the stability of austenite and leads to an increase in the equilibrium temperature of the phases and, consequently, to an increase in the temperature M_s. Consequently, the appearance of the B2-phase with a volume fraction of ~60% and a reduction in the volume fraction of the L2₁-phase to ~40% after quenching cause an increase in the MT temperatures compared to the as-grown single crystals after growth.

Secondly, the formed domain structure consists of the nanosized objects of 20–50 nm in size. Such a structure can hinder the growth of martensite crystals, which affects the morphology of martensite crystals, reduces the size of lamellae, and changes the mechanism of nucleation and growth of martensite. Namely, the MT does not occur due to the motion of single interfacial boundary, but due to the multiple nucleation of various martensite lamellae and subsequent increase in their size (Figure 3b). To increase the volume fraction of martensite, significant supercooling is necessary, i.e., the nucleation of martensite crystals occurs at different temperatures, since the nanodomain structure significantly increases the friction force and prevents the single interface motion.

Therefore, according to (1), an increase in the MT intervals indicates an increase in the stored elastic energy during the forward MT. In this case, according to the Tong–Weiman classification, quenched single crystals see an MT of the second type develop. At such a type, the value of the accumulated elastic energy is large $\Delta {\mathrm{G}}_{\mathrm{rev}}\left(1\right)/\Delta {\mathrm{G}}_{\mathrm{irr}}$ > 1 and counteracts the forward MT (large MT intervals) and contributes to the reverse MT, which starts at A_s < M_s.

Thus, the modification of the microstructure, including the appearance of a domain nanostructure, leads to an increase in the MT temperatures and a change in the morphology of the thermal-induced martensite crystals.

4.3. Martensitic Transformations in Aged Single Crystals

Aging at 773 K for 1 h, which was carried out on the quenched samples, leads to the precipitation of small particles of the second phase and an increase in the domain size and the volume fraction of the L2₁-phase. The separation of particles of the second ω-phase in this case plays the most significant role.

Typically, in NiFeGaCo alloys, γ-phase is precipitated. This phase is rich in cobalt and causes a decrease in the cobalt content in the matrix and an increase in MT temperatures [28,30]. It was shown in [18,25] that a change in the cobalt content by 1 at.% can shift MT temperatures by 50–80 K. The precipitation of the ω-phase was found only in alloys with a high content of cobalt [7]. It is assumed that the ω-phase also has a similar effect on the chemical composition in the matrix as the γ-phase. ω-phase particles are quite large, up to 100 nm, so they can partially lose their complete coherence with the matrix and be the sources of martensite nucleation, which also increases the temperature M_s and, most importantly, causes a strong decrease in thermal-induced martensite crystals.

Aging was carried out at 773 K below the order–disorder transition temperature of 930 K (see Figure 3). Therefore, this caused a reduction in the ratio of the volume fractions of the B2- and L2₁-phases to 35–40% and 60–65%, respectively. Moreover, it also reduces the size of L2₁-domains from 20–30 nm to 100–200 nm. Therefore, changes in the fraction volume and size of the domains are expected to decrease the MT temperatures. However, this factor has a weaker influence compared to the change in the chemical composition.

According to the Tong–Weiman classification, after aging, the second type of MT is preserved in both aged and quenched crystals—A_s < M_s and $\Delta {\mathrm{G}}_{\mathrm{rev}}\left(1\right)/\Delta {\mathrm{G}}_{\mathrm{irr}}$ > 1. The MT intervals are wide, and significant supercooling is necessary to form a large volume fraction of martensite. However, the morphology of martensite of the aged single crystals differs from the one of the quenched crystals, where the MT occurs due to the multiple nucleation of various martensite lamellae and subsequent increase in their size. After aging, the martensite lamellae are reduced by an order compared to the quenched single crystals. The martensite lamellae appear in groups upon cooling. At the same time, their size practically does not change during cooling, i.e., the motion of the interfacial boundary is not optically fixed. The reason for this is the presence of small particles of the second phase. The particles are too large to be included in martensite crystals and serve as sites of nucleation of lamellae and limit the continuous motion of interfacial boundaries. As can be seen from Table 1, the intervals of forward and reverse MT in the aged crystals are smaller (by 6–10 K) than in the quenched ones. Consequently, the value of the elastic energy stored during the forward MT in the aged single crystals is lower than in the quenched crystals.

As a result, the changes in the microstructure after aging at 773 K for 1 h (with quenching), including precipitation of particles, a change in the chemical composition of the matrix, and an increase in the volume fraction and domains of the L2₁-phase, lead to an increase in the temperature M_s and a change in the martensite morphology compared to the quenched state.

4.4. The Influence of Cobalt on Temperatures of Martensitic Transformations in Aged Single Crystals (Comparison with Literature Data)

The obtained experimental data on the effect of aging on MT temperatures can be compared with the similar data for Ni₅₄Fe₁₉Ga₂₇ single crystals [15,16].

In Ni₅₄Fe₁₉Ga₂₇ single crystals without cobalt, aging at the same regime, 773 K for 1 h (with quenching), also leads to the precipitation of particles (the volume fraction of the γ-phase is 2–3%) and the formation of a nanodomain structure, as in the single crystals under study. However, in Ni₅₄Fe₁₉Ga₂₇ the particle size is 2–3 times smaller (20–35 nm) and the size of nanodomains is more than 20 times smaller (5–10 nm) compared to Ni₄₄Fe₁₉Ga₂₇Co₁₀. Therefore, the addition of 10 at.% cobalt causes an increase in the size of the precipitated phase, which may be due to the presence of a large amount of the alloying element, and changes the structure of the precipitated phase from γ to ω.

Both in the Ni₅₄Fe₁₉Ga₂₇ single crystals and the Ni₄₄Fe₁₉Ga₂₇Co₁₀ single crystals, the MT temperatures similarly depend on the precipitation of the secondary phase and the change in second-neighbor ordering. After aging at 773 K for 1 h, the MT temperatures in the aged crystals increase relative to the initial crystals. However, the temperature ΔM_s change in the Ni₅₄Fe₁₉Ga₂₇ single crystals does not exceed 15–20 K [15,16]. This is observed even when the particle size increases up to 200–500 nm [5]. On the contrary, in the studied Ni₄₄Fe₁₉Ga₂₇Co₁₀ alloys, the value of ΔM_s reaches 50–120 K. It is assumed that this difference is associated with a strong deviation from the stoichiometric composition and the presence of a large amount of the alloying element. These differences lead to significant local strain (similar to glass strain materials) in the Ni₄₄Fe₁₉Ga₂₇Co₁₀ single crystals. An increase in the cobalt content by more than 10 at.% results in the suppression of thermal-induced MT, as shown in [7,8,9].

5. Conclusions

Heat treatment caused three different microstructures with different MT temperatures and martensite morphology in the Ni₄₄Fe₁₉Ga₂₇Co₁₀ (at.%) single crystals. It has been found that in the as-grown single crystals, the high-temperature phase has an L2₁-structure (100% volume fraction with anti-phase boundaries) with an insignificant content of γ-phase particles (1–2%). The MT in the as-grown single crystals occurs at low temperatures (M_s = 125 K), in narrow MT intervals (3–7 K), and with a morphology that represents the motion of practically single interfacial boundary.

High-temperature annealing at 1448 K for 1 h followed by quenching leads to the dissolution of γ-phase particles. The high-temperature phase is a mixture of L2₁-nanodomains and B2-phase of 20–50 nm in size. The volume fraction of the L2₁-phase is reduced to 40%, which causes an increase in the MT temperatures by 80 K and broadening of the MT temperature ranges from 3–7 K to 40–45 K compared to the as-grown crystals. The thermal-induced martensite lamellae are reduced to 300–500 microns. The MT occurs due to the multiple nucleation of various martensite lamellae and subsequent increase in their size.

If aging at 773 K for 1 h is carried out on the quenched crystals, the domain structure changes: the L2₁-domains increase to 100–200 nm and the volume fraction of the L2₁-phase increases to 60–65%. During aging, small particles of the secondary phase (ω-phase) with a size of 70–100 nm with a volume fraction of 4–6% are released, which causes a change in the chemical composition and increases the MT temperature by 45 K, compared with the quenched state (annealing 1448 K, 1 h). Small particles lead to a significant decrease in the thermal-induced martensite lamellae to 20–50 µm. Martensite lamellae appear in groups; their size practically does not change upon cooling.