Microstructural Evolution and Precipitate Control in Boron-Doped Ni-Mn-Ti Shape Memory Alloys via Thermal Processing

Abstract

Elastocaloric cooling, which leverages stress-induced phase transformation in shape memory materials, represents a sustainable and energy-efficient alternative to conventional vapor-compression cooling systems. Central to optimizing these materials is understanding how thermal processing history dictates phase formation, microstructure, and thermal properties. In this study, we investigated the (Ni₅₀Mn_31.5Ti₁₈)_99.8B_0.2 compound synthesized via vacuum induction melting and arc melting, followed by water quenching. Induction melting results in needle-like, boron-rich precipitates within the martensite lattice. In contrast, vacuum arc melting promoted precipitate growth at the grain boundaries. The vacuum arc melting sample exhibits ~82% martensite phase fraction, a near-ambient transformation temperature of ~277 K, a large transition entropy change of ~75 J·kg⁻¹·K⁻¹, and moderate thermal hysteresis of ~24 K. These results underscore the pivotal role of thermal history in tailoring phase stability and transformation thermodynamics, providing essential design guidelines for subsequent mechanical performance optimization in elastocaloric shape memory alloys for energy-efficient and sustainable thermal management applications.

1. Introduction

The rising global demand for cooling, driven by climate change and urbanization, underscores the urgent need for sustainable, energy-efficient refrigeration technologies. Conventional vapor-compression cooling systems contribute significantly to greenhouse gas emissions due to their reliance on hydrofluorocarbon (HFC) refrigerants and inherently high energy consumption. Consequently, solid-state cooling technologies have garnered increasing attention as environmentally friendly alternatives. Among these, caloric effects activated by external stimuli, such as magnetocaloric (magnetic field) [1], electrocaloric (electric field) [2], barocaloric (hydrostatic pressure) [3], and elastocaloric effects (uniaxial stress) [4,5,6], represent promising candidates for next-generation refrigeration. The elastocaloric effect is particularly compelling due to its mechanical simplicity and high potential for cost-effective scaling in compact systems [4,7,8].

Elastocaloric cooling leverages stress-induced reversible martensite phase transformations in shape memory materials. This technology has demonstrated exceptional potential for high energy efficiency and energy saving, as recognized by the U.S. Department of Energy [9]. When a mechanical stress is applied to an elastocaloric material, it induces a first-order martensite phase transition accompanied by a significant entropic change, releasing latent heat and increasing the material temperature. Conversely, the removal of stress triggers a reverse transformation into the austenitic phase, absorbing heat from the surroundings and producing a cooling effect [10]. For practical viability in sustainable thermal management, it is critical that these materials operate near or below room temperature, ensuring seamless integration into residential and commercial cooling systems while maximizing thermodynamic performance [11,12].

The development of high-performance elastocaloric materials is fundamental in advancing solid-state refrigeration. While extensive research has focused on NiTi-based [13,14,15], Cu-based [16,17,18], Fe-based [19,20], and NiMn-based [6,21] shape memory alloys (SMAs), recent attention has shifted toward microalloying with nonmetallic dopants to refine functional properties. Silicon (Si) has been shown to enhance transformation entropy [22], while nitrogen (N) promotes microstructural uniformity and cyclic stability [23]. Among these, boron (B) has emerged as one of the most effective dopants for enhancing grain boundary cohesion and refinement, thereby improving both mechanical strength and caloric cyclability [24,25,26]. Cong et al. reported a colossal elastocaloric effect in ferroelastic Ni-Mn-Ti SMAs [6], achieving a high reversible adiabatic temperature change of 31.5 K and an isothermal entropy change of 45 J·kg⁻¹·K⁻¹. By adjusting the Mn/Ti/B ratios in (Ni₅₀Mn_31.5Ti_18.5)_100−xB_x alloys, the working temperature can be finely tuned between ~250 K and ~500 K. However, the complex relationship between thermal processing history, microstructure, and the resulting caloric properties in these B-doped systems remains insufficiently explored.

Modern materials design increasingly view heat treatment not merely as a stress-relief step but as a precise tool for microstructural programming. Recent studies on all-d-metal Ni-Mn-Ti systems have demonstrated that solution treatment and subsequent aging can be used to control the coherency and distribution of secondary phases, which directly modulates the transformation hysteresis and latent heat [27]. Furthermore, quenching kinetics play a vital role in solute trapping, where rapid cooling prevents Boron segregation to grain boundaries, inducing significant lattice strain that shifts the transformation temperature [24]. It has also been highlighted that optimizing the annealing temperature is essential in achieving high structural integrity and cyclic stability, as it governs the trade-off between grain refinement and the dissolution of brittle intermetallic precipitates [28]. Consequently, understanding the synergy between melting techniques and subsequent thermal history is vital for the development of high-performance elastocaloric cooling materials.

While the elastocaloric effect ultimately depends on stress-induced transformations, previous research has already established the mechanical behavior of Ni-Mn-Ti-B alloys [6]. In this study, we isolate the impact of distinct thermal processing routes, comparing induction melting and arc melting followed by water quenching, specifically on a (Ni₅₀Mn_31.5Ti₁₈)_99.8B_0.2 compound. The results demonstrate that thermal processing history significantly influences the martensite-to-austenite ratio and precipitate morphology. During rapid water quenching, boron is trapped in a non-equilibrium state, interstitially positioned within the martensite lattice, where it forms boron-rich precipitates and stabilizes the austenite phase. Specifically, the water-quenched sample prepared via vacuum arc melting achieved a martensite phase fraction of ~82% and a near-ambient phase transformation temperature of ~277 K. This leads to a substantial transition entropy change of ~75 J·kg⁻¹·K⁻¹ with a moderate thermal hysteresis of ~24 K. These findings highlight the pivot role of thermal processing in tailoring phase stability and transformation thermodynamics, providing valuable insights for the optimization of sustainable cooling applications.

2. Materials and Methods

2.1. Alloy Synthesis

The designed (Ni₅₀Mn_31.5Ti₁₈)_99.8B_0.2 samples were fabricated using vacuum arc melting (VAM) (MAM-1, Edmund Buehler, GmbH, Bodelshausen, Germany) and vacuum induction melting (VIM) (MTI Corporation, Richmond, CA, USA) under an Ar atmosphere. Raw materials of Ni (foil, 99.98%, Sigma Aldrich, Darmstadt, Germany), Mn (foil, 99.95%, Sigma Aldrich, Darmstadt, Germany), Ti (wire, 99.7%, Sigma Aldrich, Darmstadt, Germany), and B (granule, 99%, Sigma Aldrich, Darmstadt, Germany) were weighed stoichiometrically with a total mass of 10 g. To ensure complete mixing and homogeneity, the ingots were re-melted 10 times for VAM with a 1 min holding time per melting cycle, and 3 times for VIM with a 3 min holding time per melting cycle. The as-cast ingots prepared by VIM were naturally cooled down inside the crucible by turning off the furnace (~10⁻²–10⁻¹ K/s cooling rate). The as-cast ingots prepared by VAM were cooled down by the circulating water under the copper hearth, with a high cooling rate (~10–100 K/s). The pictures of the as-cast ingots are shown in Figure 1. The as-cast ingots were sealed in a quartz tube under vacuum (~10⁻⁵ mbar), annealed at 900 °C for 36 h in a muffle furnace (Lindberg Blue M, Riverside, MI, USA), followed by water quenching.

2.2. Structural and Thermal Characterizations

The crystal structure of all the samples was confirmed using X-ray diffraction (XRD, PANalytical X-ray Diffractometer, Malvern Panalytical Ltd, Westborough, MA, USA) with Cu Kα radiation and a Bragg–Brentano configuration over a 2θ range from 10° to 60°. Instrumental broadening was corrected using a silicon standard reference material. The microstructures of the alloys were characterized by field emission scanning electron microscopy (FESEM) (FEI Verios G4, Thermo Fisher Scientific, Waltham, WA, USA) and energy-dispersive spectroscopy (EDS) (Oxford Aztec, Oxford, UK). Thermal analysis was performed using a differential scanning calorimeter (DSC 214) (NETZSCH Group, Selb, Germany) with the heating and cooling rate of 10 K/min. The phase fraction analysis was performed using ilastik software (version 1.4.0) developed by the ilastik team [29] and Fiji [30].

3. Results and Discussion

3.1. Subsection Austenite and Martensite Phase

The development of novel materials increasingly relies on the synergy between high-throughput computational exploration and rigorous experimental validation [31,32,33]. While simulation studies on binary or ternary systems have elucidated how composition-dependent bond energies dictate phase stability and transition temperatures [27,34], the present work extends these fundamental insights to complex quaternary Ni-Mn-Ti-B alloys. X-ray diffraction (XRD) was employed to characterize the phase evolution of (Ni₅₀Mn_31.5Ti₁₈)_99.8B_0.2 samples synthesized via vacuum induction melting (VIM) and vacuum arc melting (VAM), both in the as-cast state and after water quenching (denoted as VIM-cast, VIM-Q, VAM-cast, and VAM-Q, Figure 1a). In the as-cast state, XRD patterns reveal a dual-phase contribution of austenite and martensite across all the samples. The VAM-cast sample exhibited a slightly higher martensite fraction (~60%) compared to the VIM-cast counterpart (~49.7%). This divergence is attributed to the distinct solidification kinetics of each process, where VIM involves more uniform heating but slower cooling, leading to the formation of larger precipitates that stabilize the austenite phase. Conversely, the high localized heating and rapid cooling inherent to VAM promote a more uniform phase distribution and result in a higher initial martensite content.

Upon thermal annealing and subsequent water quenching (denoted as VIM-Q and VAM-Q), a substantial increase in martensite fraction was observed. This shift stems from the transition from the face-centered cubic austenite (γ-FCC, space group $Fm\overline{3}m$) to the orthorhombic martensite (α, space group Pmma). Thermal annealing above the transformation temperature facilitates a more uniform phase distribution while water quenching with rapid cooling rate locks this state, preventing a reversion to the austenite phase. Notably, the VIM-Q sample showed a dramatic increase in martensite content from ~50% to ~98.5% (Figure 1b–d), whereas the VAM-Q sample increased from ~60% to ~82.1% (Figure 1e–g). The pronounced responsiveness in the induction-melted sample is likely due to the dissolution or reorganization of the larger precipitates formed during the initial VIM process, which clears the path for a nearly complete martensite transformation. In contrast, the transformation in the VAM-Q sample is slightly constrained by the significant elemental segregation, dendritic structure, and finer grains resulting from the intense localized temperature gradients of the VAM process. These XRD findings dictate the critical role of thermal processing history in governing phase constitution, highlighting the VIM processed alloy’s superior sensitivity to microstructural tuning via water quenching.

3.2. Microstructure

The microstructural evolution of the (Ni₅₀Mn_31.5Ti₁₈)_99.8B_0.2 alloys was investigated, revealing notable differences between the VIM and VAM processing routes (Figure 2 and Figure 3). The VIM-cast sample exhibits relatively coarse precipitates (~2–10 μm). Energy-dispersive X-ray spectroscopy (EDS) mapping identifies these as B-rich (yellow regions in Figure 3a) and NiTi-rich phases (Figure 2a). These large precipitates act as nucleation sites for the austenite phase, effectively stabilizing it and hindering a complete transformation to martensite. In contrast, the VAM-cast sample displays a more refined, homogeneous microstructure with smaller dendritic features and a uniform distribution of precipitates (~1–2 μm with ~0.89 vol%, Figure 2c and Figure 3c). This finer distribution offers less resistance to martensitic transformation, resulting in a higher initial martensite fraction. However, localized elemental segregation within these dendritic regions may still stabilize the austenite phase internally, influencing the material response to subsequent thermal treatment.

Following thermal annealing and water quenching, both samples underwent significant precipitate refinement. For the VIM-Q sample, rapid cooling broke down the coarse as-cast precipitates into finer, evenly distributed precipitates (~1.47 vol%, Figure 2b and Figure 3b). B-rich zones primarily formed at grain boundaries, while Ni-rich zones appeared as light-gray acicular structures (Figure 2b). These finer precipitates provide significantly less mechanical hindrance to martensitic transformation, explaining the dramatic increase in martensite content observed in XRD analysis. While the VAM-Q sample also showed refinement, the transition was less pronounced because the microstructure of the VAM-cast sample was already conducive to martensitic transformation.

A defining feature of the water-quenched samples, particularly the VIM-Q variant (~1.47 vol% vs. ~0.66 vol%, Figure 3), is the presence of pronounced needle-like regions (Figure 3b,d). This acicular morphology is a hallmark of the martensitic phase, which forms through a rapid, diffusionless transformation involving the coordinated shear motion of atoms [35]. The growth of these elongated structures is governed by specific habit planes, such as the transition from (110) γ to (020) α, for nucleation and growth (Figure 1b–f) [36,37].

It is well established that carbon diffusion plays a crucial role in stabilizing the austenite phase and influencing martensite formation in steel and other alloys [38]. The role of boron in this system appears analogous to that of carbon in steel. During water quenching, the high cooling rate creates a diffusion-limited non-equilibrium state where boron atoms are trapped within the lattice. Given that the atomic size of boron (~0.85 Å) is substantially smaller than those of the matrix elements (Ni ~1.25 Å, Mn ~1.27 Å, Ti ~1.45 Å), boron preferentially occupies larger octahedral interstitial sites rather than substituting for larger metal atoms [39]. Additionally, boron exhibits higher solubility in the austenite matrix. The rapid quenching kinetics of the VAM-Q sample suppresses the long-range diffusion required for boron to segregate into TiB₂ precipitates. Consequently, these interstitial boron atoms induce local lattice strains that stabilize the dendritic regions and modulate the austenite-to-martensite transformation.

Notably, the VIM-cast sample exhibits significantly lower homogeneity compared to its VAM-cast counterpart. EDS analysis reveals the presence of cubit Ni₂(Ti, Mn) (space group Fd3m) and hexagonal TiB₂ (space group P6/mmm) flakes, with dimensions of ~10 μm (Figure 4). Thermodynamically, the formation of TiB₂ is highly favored due to its exceptionally low negative enthalpy of formation (ΔH_f ~ −324 kJ/mol) [40], rendering it substantially more stable than potential Ni-B (ΔH_f ~ −35 kJ/mol) or Mn-B (ΔH_f ~ −63 kJ/mol) alternatives [40,41]. Following the quenching process, these precipitates evolve into acicular (needle-like) morphologies (Figure 2b). In both VIM and VAM samples, boron resides primarily along grain boundaries as precipitates rather than being fully assimilated into the NiTi matrix. Previous studies indicate that Ni-Ti and TiB₂ precipitates can enhance the mechanical strength of SMAs by improving grain boundary cohesion while preserving superelasticity [42,43]. These B-rich precipitates and needle-like structures likely serve as additional nucleation sites for stress-induced martensite. By reducing the energy dissipation associated with nucleation and promoting a more reversible phase transition, these features contribute to the reduction in stress hysteresis.

The TiB₂ and Ni₂(Ti,Mn) precipitates show a dual role in modulating martensitic strain propagation. Kinetically, large and coarse precipitates (Figure 4) act as thermal and mechanical obstacles. Thermodynamically, however, they alter the local matrix composition, thereby shifting phase stability. To quantify this thermodynamic driving force, the valence electron concentration (e/a ratio) [44] was determined based on the matrix composition of the quenched samples (Figure 2b,d), representing the chemical state of the austenite/martensite phases during the DSC measurements (Figure 5a,b). The e/a ratio is defined as

$$e/a=\sum (c_i·V_i)$$

(1)

where c_i is the atomic fraction of each element in the matrix and V_i is the number of valence electrons. A shift in e/a from 7.957 (VAM-Q sample) to 8.129 (VIM-Q sample) was observed. This increase in Δ(e/a) of ~1.7 corresponds to an increase in phase transformation temperature of ~100 K (

Table 1. Austenite and martensite start, peak and finish temperatures (A_s, A_p, A_f, M_s, M_p, M_f), and thermal hysteresis ΔT_hys of VIM-Q and VAM-Q (Ni₅₀Mn_31.5Ti₁₈)_99.8B_0.2 samples.

Sample	As (K)	Ap (K)	Af (K)	Ms (K)	Mp (K)	Mf (K)	ΔThys (K)
VIM-Q	394	435	452	429	408	374	21.5
VAM-Q	277	286	292	270	259	250	24.5

), as revealed in our DSC data (Figure 5a,b). This confirms that the precipitates exert a primary thermodynamic influence by modulating matrix chemistry, rather than acting solely as passive mechanical barriers.

3.3. Thermal Analysis

The DSC profiles characterize the phase transitions of the alloys; endothermic peaks during heating and exothermic peaks during cooling correspond to the austenitic and martensitic transformations, respectively. The critical transition temperatures, including A_s (austenite start), A_p (austenite peak), A_f (austenite finish), M_s (martensite start), M_p (martensite peak), and M_f (martensite finish), along with the thermal hysteresis (ΔT_hys) and transformation entropy changes (ΔS_tr), are summarized in Table 1. The ΔT_hys is defined as

$${∆T}_{hys}={\displaystyle \frac{A_s+A_f}{2}}-{\displaystyle \frac{M_s+M_f}{2}}$$

(2)

For a reversible first-order transformation at equilibrium, the Gibbs free energy change ΔG = 0. Consequently, ΔS_tr can be derived as

$${∆S}_{tr}=∆H/T_0$$

(3)

where $∆H$ is the transformation enthalpy obtained by integrating the heat flow peaks, and T₀ represents the thermodynamic equilibrium temperature at which martensite and austenite phase coexist, often approximated as $T_0={\displaystyle \frac{M_s+A_s}{2}}$ [45]. Both ΔT_hys and ΔS_tr are crucial parameters in evaluating and optimizing elastocaloric SMA materials. A larger ΔS_tr leads to improved cooling power, while lower thermal hysteresis allows for enhanced cycling stability and mechanical durability, improved system efficiency, and reduced energy loss [21,46]. As shown in Figure 5a,b, the VIM-cast and VAM-cast samples exhibit relative flat and broad DSC curves. This is attributed to the incomplete phase transition and microstructural inhomogeneities, such as segregation of elements, dendritic structures, and unevenly distributed precipitates, which suppress distinct transformation peaks. In contrast, the quenched samples (VIM-Q and VAM-Q) display sharp, well-defined peaks, indicating a more complete and homogeneous transformation facilitated by the redistribution of precipitates during thermal treatment.

Figure 5. DSC analysis. (a) VIM-Q sample. (b) VAM-Q sample. (c) Comparison of thermal hysteresis (ΔT_hys) between VIM-Q and VAM-Q samples and the literature [6,21,47,48,49].

Figure 5. DSC analysis. (a) VIM-Q sample. (b) VAM-Q sample. (c) Comparison of thermal hysteresis (ΔT_hys) between VIM-Q and VAM-Q samples and the literature [6,21,47,48,49].

The VAM-Q sample illustrates a significantly lower transformation temperature compared to the VIM-Q sample, characterized by a sharp transition near ambient temperature (A_p = 286 K and M_p = 259 K). Conversely, the VIM-Q sample shows broader transformation peaks at much higher temperatures, with A_s being ~120 K higher than that of the VAM-Q sample (Figure 5a,b and Table 1). This shift is driven by the higher martensite fraction (~98.5% in Figure 1d) in the VIM-Q sample, which generates significant internal stresses and a highly strained needle-like microstructure (Figure 2b). The high transformation temperature in the VIM-Q sample is supported by a thermodynamic balance where internal quenching stresses and precipitate-induced strain fields act as a mechanical driving force, shifting the equilibrium temperature upward as per the Clausius–Clapeyron relationship [50]. The total Gibbs free energy change for the austenite-to-martensite transformation (ΔG_A→M) is expressed as

$$∆G_{A\to M}=∆G_{chem}+∆G_{el}+∆G_{diss}$$

(4)

where $∆G_{chem}$ is the chemical free energy difference (ΔH-TΔS), $∆G_{el}$ is the elastic strain energy from lattice mismatch, and $∆G_{diss}$ is the energy dissipated by internal friction of hysteresis [51]. In the VIM-Q sample, the higher fraction of needle-like precipitates (~1.47%) contributes to a substantial internal stress (σ_int) that acts as a mechanical driving force. According to the Clausius–Clapeyron relation, this internal stress lifts the transformation temperature (ΔT_tr) as follows [50]:

$$∆T\approx {\displaystyle \frac{{\sigma }_{int}·{ϵ}_{tr}}{∆S_{tr}}}$$

(5)

where ${ϵ}_{tr}$ is the transformation strain. The local stress fields around these precipitates favor the nucleation of martensite at significantly higher temperatures (A_s ~394 K), while their relatively high fraction (~98.5%) provides a lower barrier to global phase propagation.

Boron, as a p-block element, has a very low e/a ratio contribution compared to transition metals. It does not significantly change the global e/a value but alters the γ (Fm$\overline{3}$m)–α (Pmma) phase equilibrium (Figure 1). By occupying interstitial sites within the lattice, boron alters the elastic strain energy ($∆G_{el}$) and the chemical potential of the phases. The quenching process is kinetically essential to trap boron in this interstitial state, preventing the formation of TiB₂ precipitates (Figure 2, Figure 3 and Figure 4). This solute effect fundamentally redefines the energy landscape, as evidenced by the substantial increase in the equilibrium temperature of T₀ ~100 K (Figure 5a,b) and the substantial changes in transformation enthalpy (ΔH).

From an application standpoint, the near-ambient transformation temperature of the VAM-Q sample (~273 K) is ideal for space cooling and refrigeration. Operating within this range allows for efficient heat extraction from building environments without excessive energy input. Furthermore, this proximity to room temperature minimizes mechanical stress accumulation during cycling, enhancing the coefficient of performance (COP) and ensuring the long-term durability of elastocaloric materials [4,11,12,52,53].

Figure 5c summarizes the thermal hysteresis (ΔT_hys) of the represented elastocaloric SMAs. The ΔT_hys of the VAM-Q sample is slightly higher than that of the VIM-Q sample. The difference is primarily governed by three factors: chemical compositions, microstructure homogeneity, and transition entropy change (ΔS_tr). ΔT_hys varies significantly with stoichiometry. For example, variations in Ti content (Ti₁₆ and Ti_18.5) in NiMnTi-based SMAs [6], in Ni concentration (Ni₄₇ to Ni_53.9) in NiMnGa-based SMAs [47], and in V concentration in TiNiCuV-based SMAs [21] can shift ΔT_hys by 10.4 K to 23 K, respectively. The observed ΔT_hys of ~23 K is considered moderate compared to the 30–40 K typical of binary NiTi and many all-d-metal Heusler alloys [54,55]. Microstructurally, the VAM-Q sample’s dendritic architecture and boron-rich precipitates introduce spatial variations in transformation temperatures, contributing to its slightly higher thermal hysteresis relative to the VIM-Q sample.

Furthermore, SMAs exhibiting larger ΔS_tr during phase transitions typically demonstrate wider thermal hysteresis [56]. As illustrated in Figure 6, both VIM-Q and VAM-Q samples process high ΔS_tr values (>60 J·kg⁻¹·K⁻¹), with the VAM-Q sample reaching ~75 J·kg⁻¹·K⁻¹. These results are significant when compared with other state-of-the-art SMAs (Figure 6) [14,15,57,58,59,60,61,62,63]. While moderate hysteresis can lead to cumulative heating and a potential reduction in the COP, this is offset by the advantages of a colossal ΔS_tr.

A high ΔS_tr enhances elastocaloric cooling performance in several critical ways. Firstly, it increases latent heat ($Q$), according to the relation $Q=T_0·{∆S}_{tr}$, leading to higher cooling power per cycle. Secondly, it drives a larger adiabatic temperature change (ΔT_ad), defined as ${∆T}_{ad}=-T_0·{∆S}_{tr}/C_p$ [5], where C_p is the specific heat capacity. This facilitates a broader operating temperature span. Thirdly, a high ΔS_tr enhances recoverable work output (W), given by $W=∆H-T{∆S}_{tr}$, and promotes more rapid thermal cycling. In this study, we deliberately focus on the synthesis and thermal treatment history, to isolate the impact of them on phase fraction, precipitate morphology, and transformation entropy changes. Our results show similar ΔS_tr to other high-transition-entropy changes in the literature in Mn-Ti-B systems [6,54,64,65]. A colossal adiabatic temperature change for Mn-Ti-B alloys of up to −31.5 K has been reported, which surpasses the state-of-the-art NiTi (~25 K) [6,54]. Furthermore, the advantage of boron microalloying provides high compressive strength (>1500 MPa) and significant ductility (>17%) by strengthening grain boundaries [65].

Figure 6. Transformation entropy changes ΔS_tr in VIM-Q and VAM-Q samples and comparison with other SMAs [6,66,67,68,69,70,71,72,73,74,75]. The A-M and M-A indicate the austenite-to-martensite and martensite-to-austenite phase transformation. The highest values of ΔS_tr are collected under the various conditions such as stress mode and cycle numbers. The range of ΔS_tr for each system corresponds to the different elemental ratio reported from either the same or different literature.

Figure 6. Transformation entropy changes ΔS_tr in VIM-Q and VAM-Q samples and comparison with other SMAs [6,66,67,68,69,70,71,72,73,74,75]. The A-M and M-A indicate the austenite-to-martensite and martensite-to-austenite phase transformation. The highest values of ΔS_tr are collected under the various conditions such as stress mode and cycle numbers. The range of ΔS_tr for each system corresponds to the different elemental ratio reported from either the same or different literature.

4. Conclusions

This study demonstrates that thermal processing history is a decisive factor in governing the martensite–austenite phase ratio, microstructure evolution, and phase transformation kinetics in (Ni₅₀Mn_31.5Ti₁₈)_99.8B_0.2 alloys. In the induction-melted sample, water quenching (VIM-Q) induces the formation of needle-like, boron-rich precipitates within the martensite lattice. Conversely, the water-quenched vacuum arc-melted (VAM-Q) sample exhibits boron-rich precipitates localized along grain boundaries within a characteristic dendritic architecture. These precipitates facilitate austenite phase retention, effectively shifting the transformation temperature toward the ambient regime. Consequently, the VAM-Q sample achieved a significant martensite phase fraction of ~82% and a near-ambient phase transformation temperature of ~277 K. This microstructural tuning yielded a colossal transformation entropy change of ~ 75 J·kg⁻¹·K⁻¹ and a moderate thermal hysteresis of ~23 K. These findings highlight the fundamental role of thermal processing history in tailoring phase stability and transition thermodynamics, offering a robust framework for the design and fabrication of high-performance elastocaloric materials for sustainable thermal management.