Magnetostructural Transformation and Magnetocaloric Properties of (Ni_37.5Co_12.5Mn₃₅Ti₁₅)_100−xB_x (x = 0.0 and 0.4) Melt-Spun Ribbons

Abstract

The effect of B-doping on the martensitic transformation (MT), microstructure, room temperature (RT) crystal structure, and magnetocaloric properties of a typical all-d-metal Ni_37.5Co_12.5Mn₃₅Ti₁₅ quaternary alloy was studied by synthesizing melt-spun ribbon samples of nominal composition (Ni_37.5Co_12.5Mn₃₅Ti₁₅)_100−xB_x with x = 0.0 and 0.4. For B-free samples, SEM images show a grain-oriented microstructure formed by the columnar in shape-elongated grains with their major axis oriented along the thermal gradient during solidification. By contrast, the B-doped samples show smaller grains whose orientation tends to be perpendicular to the contact surface with the copper wheel. For all samples, austenite (AST) and martensite (MST) phases exhibited a cubic B2-type and 5M monoclinic crystal structure, respectively. The martensitic transition temperature (T_M) and the Curie temperature of the austenite phase (T_C^A) were reduced from 295 K to 253 K and 333 K to 276 K, respectively, with the addition of B. The effect of thermal annealing for different times (from 30 min to 4 h) at 1073 K was studied. Thermal annealing increases the martensitic transformation temperature, whereas T_C^A remains unchanged. The maximum magnetic field-induced entropy changes |ΔS_T|^max for B-doped samples were around 4.5 J kg⁻¹ K⁻¹ and 4.7 J kg⁻¹ K⁻¹ for as-solidified and annealed samples (1073 K–4 h), respectively, compared to that found for the undoped samples (i.e., ΔS_T = 16 J kg⁻¹ K⁻¹). However, the entropy reduction is accompanied by an increase in the full width at half-maximum of the ΔS_T(T) curve.

1. Introduction

Motivated by the search for new, more energy-efficient, and environmentally friendly solid-state-based refrigeration technologies, magnetic refrigeration has received considerable attention over the last 25 years [1]. This technology is based on the magnetocaloric effect (MCE), a magneto-thermal phenomenon quantitatively characterized by the thermal dependence of the change in the temperature ΔT_ad(T) or magnetic field-induced entropy change ΔS_T(T) under adiabatic and isothermal conditions, respectively, linked to a second or first-order magnetic transition when the material is subject to a magnetic field change μ_oΔH [2,3,4,5,6]. In this sense, in most of the efforts to develop magnetic refrigerants that are able to be used as working substances at room temperature (RT), magnetic refrigerators have focused on magnetocaloric material systems exhibiting a giant MCE linked to first-order magnetic transitions (FOMT). The term “giant magnetocaloric effect” (GMCE) refers to the sizeable magnetic-field-driven change in magnetic entropy and adiabatic temperature changes that a magnetically ordered material experiences in the temperature range at which it undergoes a first-order magnetoelastic or magnetostructural transition. Among the material families showing GMCE, (Ni, Mn)-based Heusler-type alloys derived from ternary Ni-Mn-X systems, X = Ga, In, Sn, and Sb have been the focus of extensive research over the last 20 years since in some alloys a remarkable magnetocaloric effect has been experimentally determined [7,8,9,10,11]. However, the brittleness of these alloy systems is one of the essential drawbacks of their practical utilization in cooling devices and any other application. In this regard, the more recently discovered all-d-metal-type alloys in the Ni-Co-Mn-Ti system combine a giant magnetocaloric effect with excellent mechanical properties [12,13]. Wei et al. were the first to report the magnetostructural transformation in bulk Ni_50−xCo_xMn₃₅Ti₁₅ with x = 8.0 and 9.5 [14]. They demonstrated that martensitic transformation takes place from a high-temperature ferromagnetic (FM) cubic B2-type austenite to a low-temperature weak-magnetic monoclinic martensite (space group p2/m) [14,15]. However, the maximum field-induced entropy changes ΔS_T^max reported for bulk were moderate (10 J kg⁻¹ K⁻¹ and 18 J kg⁻¹ K⁻¹ at 2 T and 5 T, respectively) [14,15]. Nevertheless, some years later, Neves Bez et al. demonstrated the efficacy of the melt-spinning technique to optimize the magnetocaloric effect [16]: ΔS_T^max values at μ_oΔH = 2 T (ΔS_T^max = 27.2 J kg⁻¹ K⁻¹) around room temperature were found in as-solidified Ni_37.5Co_12.5Mn₃₅Ti₁₅ ribbons [16].

Alloying elements, such as Fe [17], Cu [16,18,19], V [20], Cr [21], and B [12,22], have been added to Ni-Co-Mn-Ti alloys to tune the temperature of the structural transition or to improve the mechanical and magnetocaloric properties. Due to its small atomic radio and depending on the amount added, B is an element that may occupy interstitial or substitutional sites in Heusler-type Ni-Mn-X alloys [23,24,25,26]. Recently, Zhang et al. studied the effect of B on the thermal hysteresis ΔT_hyst of the magnetostructural transition and magnetocaloric properties of Ni₃₆.₅Co_13.5Mn₃₅Ti₁₅ alloys [22]. They demonstrated that interstitial B in the (Ni_36.5Co_13.5Mn₃₅Ti₁₅)_99.6B_0.4 alloy significantly reduces ΔT_hyst (~4.4 K) and simultaneously increases ΔS_T^max from 3.8 J kg⁻¹ K⁻¹ (9.7 J kg⁻¹ K⁻¹) to ~11 J kg⁻¹ K⁻¹ (24.3 J kg⁻¹ K⁻¹) for a magnetic field change of 2 T (5 T). Likewise, Guan et al. reported improved mechanical properties with B-doping in Ni₃₇Co₁₃Mn₃₄Ti₁₆, which maintained a high value of ΔS_T^max (~40 J kg⁻¹ K⁻¹) at a magnetic field of 7 T [12]. However, the extended thermal treatment of 1123 K for two days and 1173 K for six days was used for the alloys studied by Guan et al. [12] and Zhang et al. [22], respectively.

Guided by these previous works, we addressed the synthesis of melt-spun ribbons of (Ni_37.5Co_12.5Mn₃₅Ti₁₅)_100−xB_x alloys with x = 0.0 and 0.4 to assess the effect of B on the martensitic transformation and related magnetocaloric properties. The B-containing samples were also thermally annealed for 4 h at a moderate temperature of 1073 K to obtain preliminary information on how it affects the magnetostructural martensitic transformation temperature and related ΔS_T(T) curve.

2. Materials and Methods

The initial bulk alloys to prepare rapidly solidified melt-spun ribbon samples had a nominal composition (Ni_37.5Co_12.5Mn₃₅Ti₁₅)_100−xB_x, with x = 0.0 and 0.4. The elements used to prepare them were 99.99% Ni (Alfa Aesar, Haverhill, MA, USA), 99.95% Co (Alfa Aesar), 99.9% Mn (Alfa Aesar), 99.99% Ti (Aldrich, St. Louis, MO, USA), and 99.4% B (Aldrich). Small pellets of 4 g were produced via arc melting under a highly pure argon atmosphere (99.999%) in a water-cooled copper crucible. Samples were flipped and remelted four times to ensure a good starting element mixture. An excess of Mn of 3 wt. % was added to both samples to compensate for the loss of this element by evaporation. The initial and final mass of the alloys prepared were the same.

Melt-spun ribbons were fabricated under a highly pure Ar atmosphere in an Edmund Bühler model SC melt-spinner system. Once molten, the alloys were ejected by applying an argon overpressure of 200 mbar onto the polished surface of a 200 mm diameter copper wheel rotating at a linear speed of 20 ms⁻¹. The as-solidified ribbon samples typically had the following physical dimensions: 1.0–1.5 mm width and 20 mm length. B-containing ribbon samples were sealed in vacuumed quartz ampoules to be thermally annealed at 1073 K for 4 h; thermal annealing (TA) ended with water quenching at 373 K for 20 min. Hereafter, we focus on characterizing these three ribbon samples, referred to as x = 0.0 AS, x = 0.4 AS, and x = 0.4 TA, respectively.

X-ray diffraction (XRD) patterns were collected at room temperature in a high-resolution Rigaku Smartlab diffractometer (CuKα, λ = 1.54059 Å; 30° ≤ 2θ ≤ 90°, Rigaku Corporation, Tokyo, Japan). The main goal of XRD analyses was the determination of the phase constitution at room temperature for the ribbon samples studied.

Differential scanning calorimetry (DSC) curves were measured on a TA-DSC Q200 system (TA Instruments, New Castle, DE, USA) in the temperature interval 220 K–320 K at a heating and cooling temperature sweep rate of 10 K min⁻¹. The initial and final temperatures of the direct and reverse martensitic transition (M_S and M_f, and A_S and A_f, respectively) were determined by a simple extrapolation from the DSC and M(T)^5mT curves. The thermal hysteresis ΔT_hyst of the magnetostructural transition was estimated as ΔT_hyst = A_f − M_S.

Microstructural and semi-quantitative elemental chemical composition analyses were performed in an FEI QUANTA 250 SEM (FEI Company, Hillsboro, OR, USA) equipped with an energy dispersive spectroscopy (EDS) system from EDAX.

Magnetization measurements were conducted in a 9 T Quantum Design PPMS^® Dynacool^® system (Quantum Design, San Diego, CA, USA) using the vibrating sample magnetometry (VSM) option. The magnetic field was applied along the ribbon length to reduce the internal magnetic demagnetizing field. The thermal dependencies of the magnetization, the M(T) curves, were measured under static magnetic fields of 5 mT, 2 T, and 5 T at a temperature range of 100–400 K with a temperature sweep rate of 1.5 K min⁻¹ following zero-field-cooling (ZFC) and field-cooling (FC) regimens.

3. Results

Figure 1 compares the typical granular microstructure exhibited by as-solidified (AS) ribbons on the contact (CS, Figure 1a,b) and non-contact (NCS, Figure 1c,d) surfaces, and for the cross-section (Figure 1e,f). Columnar-shaped grains grow along the parallel direction to the thermal gradient during solidification, i.e., the longer axis of grains tends to orient perpendicular to both ribbon surfaces (see Figure 1e,f). The cross-sectional views denote this reduction in the average thickness of B-doped ribbons (~20 μm) compared to B-free samples (~38 μm). Also, a slight but perceptible increase in the average grain size was observed, suggesting that B promotes grain growth. As the pictures show, the B-free sample shows grains with and without martensitic domains, denoting the coexistence of austenite and martensite. By contrast, in the B-doped sample, no martensitic domains appeared. The latter is consistent with the temperature shift of the martensitic transformation with B addition, as the DSC and low-field M(T) curves show (see below). EDS semi-quantitative analyses were performed in several selected zones on the CS and NCS (the typical spectra recorded are shown in Figure S1 of the Supplementary Material file). From them, and not considering the B amount, the average elemental chemical compositions determined were as follows: Ni_35.6±1.2Co_12.1±1.0Mn_37.0±1.1Ti_15.3±0.6 for x = 0.0 AS, Ni_35.9±1.3Co_10.7±1.0Mn_37.8±1.1Ti_15.6±0.5 for x = 0.4 AS, and Ni_35.9±1.2Co_10.4±1.0Mn_37.5±1.2Ti_16.2±0.6 for x = 0.4 TA. Within the error in the determination, the semi-quantitatively determined composition is close to that of Ni_37.5Co_12.5Mn₃₅Ti₁₅.

Figure 2 and Figure 3a present the RT XRD patterns and DSC scans for as-solidified (AS) B-free and B-doped samples and the B-doped thermally annealed one. XRD patterns corresponding to the x = 0.0 AS sample reveal the coexistence of a cubic B2-type austenite and a 5-modulated (5M) monoclinic martensite phase. This is consistent with what was reported in the refs. [14,15], and agrees with the calorimetry curves shown in Figure 3a since a martensitic transition took place around RT, and, therefore, a mixture of both phases coexisted in the sample. However, the patterns for the B-doped samples correspond to a pure B2-type austenite since, as Figure 3a shows, the martensitic transformation appeared below room temperature. In earlier reports, austenite showed a B2-type crystal structure [12,17,18,27]. The lattice parameters were determined using the equation $\frac{1}{d^2}=\frac{h^2+k^2+l^2}{a^2}$ and $\frac{1}{d^2}=\frac{1}{{sin}^2\beta }·\left(\frac{h^2}{a^2}+\frac{k^2{sin}^2\beta }{b^2}+\frac{l^2}{c^2}-\frac{2hl·cos\beta }{ac}\right)$ for cubic and monoclinic systems, respectively. They are summarized in

Table 1. Unit cell parameters and cell volume obtained for the (Ni_37.5Co_12.5Mn₃₅Ti₁₅)_100−xB_x ribbon samples studied (x = 0.0 AS, x = 0.4 AS, and x = 0.4 TA).

Sample	Structure	Cell Parameters (Å)			Angles (°)	Cell Volume (Å)3
		a	b	c
x = 0.0 AS	Monoclinic	4.428	5.508	21.30	α = γ = 90 β = 94.1386	518.251
	Cubic	5.901	5.901	5.901	α = β = γ = 90°	205.542
x = 0.4 AS	Cubic	5.915	5.914	5.914	α = β = γ = 90°	206.935
x = 0.4 TA	Cubic	5.699	5.699	5.699	α = β = γ = 90°	185.112

. As previously stated, if the B amount does not exceed a critical value, B atoms occupy interstitial sites in the crystal structure, increasing the cell volume [22]. This explains the increased cell volume of the B-doped ribbon samples compared to the B-free ones. However, as the samples were prepared using a non-equilibrium technique such as melt spinning, they always showed some local crystalline disorder, which tended to enlarge the cell volume. Notice that the cell volume of austenite in B-doped as-solidified ribbon samples is higher than that of the B-free ones. However, partial local disorder decreases in interstitially doped samples after the thermal annealing with a consequent cell volume reduction.

Figure 3a compares the DSC curves measured on heating and cooling for x = 0.0 AS (blue line), x = 0.4 AS (black line), and x = 0.4 TA (red line) samples. All samples show exothermic and endothermic peaks corresponding to the direct and reverse martensitic transformation.

Table 2. A_S, A_f, M_S, M_f, ΔT_hyst, and T_C^A parameters obtained from DSC and M(T)^5mT curves for the (Ni_37.5Co_12.5Mn₃₅Ti₁₅)_100−xB_x ribbon samples studied (x = 0.0 AS, x = 0.4 AS, and x = 0.4 TA).

DSC
Sample	AS (K)	Af (K)	MS (K)	Mf (K)	ΔThyst (K)	TCA (K)
x = 0.0 AS	300	309	295	281	14	✕
x = 0.4 AS	253	270	253	233	17	✕
x = 0.4 TA	270	282	265	251	17	✕
M(T)5 mT
x = 0.0 AS	303	309	292	286	17	333
x = 0.4 AS	253	268	254	235	14	276
x = 0.4 TA	272	283	272	260	11	290

lists the characteristic temperatures (A_S, A_f, M_S, and M_f) determined via simple extrapolation from the DSC and M(T) curves measured at 5 mT and the thermal hysteresis of the transformation (ΔT_hyst). The DSC curves show that the martensitic transition temperature (T_M), defined as T_M = (A_S + A_f + M_S + M_f)/2, decreased by around 42 K with the addition of B. This behavior contrasts with previous works on bulk samples that show a temperature increase in the martensitic transition temperature with the increase in the B content [12,22]. In addition, a T_M increase appears after the thermal annealing at 1073 K for 4 h. This shift to higher temperatures of the magnetostructural transition temperature is consistent with previous results [28].

Figure 3 shows the ZFC and FC M(T) curves for x = 0.0 AS (blue line), x = 0.4 AS (black line), and x = 0.4 TA (red line) samples measured under static magnetic fields (µ_oH) of 5 mT (Figure 3b), 2 T (Figure 3c) and 5 T (Figure 3d). Notice that, with B-doping, the Curie temperature of austenite phase T_C^A (determined from the minimum of the dM/dT(T) curve at 5 mT) decreases at around 57 K. Then, the temperature difference between the magnetostructural (T_M) and magnetic transition of austenite (T_C^A) significantly reduces, making it difficult to estimate T_C^A for x = 0.4 AS and TA samples. In addition, the magnetization change ΔM across the first-order transition becomes lower. It can be noticed that at 2 T and 5 T, ΔM decreased from 64 Am² kg⁻¹ (x = 0.0 AS) to 53 Am ²kg⁻¹ (x = 0.4 AS) (see Figure 3c) and from 70 Am² kg⁻¹ (x = 0.0 AS) to 63 Am² kg⁻¹ (x = 0.4 AS) (see Figure 3d), respectively, (values that were roughly estimated from the respective M(T) curves for the magnetostructural transition on heating). For the x = 0.4 TA samples, ΔM was further reduced to 45 Am² kg⁻¹ and 59 Am² kg⁻¹ at 2 T and 5 T, respectively, due to the shift to higher temperatures of the magnetostructural transformation (which further reduced the temperature difference with the magnetic transition of austenite).

To investigate the effect of B-doping on the magnetocaloric properties of these samples, sets of isofield M(T) curves were measured on heating through the reverse martensitic transformation. Owing to the first-order nature of the magnetostructural phase transition, after one isofield M(T) curve on heating was measured, the magnetic field was set to zero, and the sample was then cooled down at the zero fields well below M_f to complete the martensitic transformation, and then heated again to the starting measuring temperature to record the next M(T) curve. The sets of isofield curves are shown in Figure S2 of the Supplementary Material file; each graph shows the starting and finishing temperature at which the M(T) curves were measured. From these measurements, the magnetic field-induced entropy changed as a function of temperature; the ΔS_T(T) curves were estimated from the numerical integration of the Maxwell relation [29]. Figure 4a compares the ΔS_T(T) curves at a magnetic field change of 2 T for x = 0.0 AS, x = 0.4 AS, and x = 0.4 T samples. The peak, or maximum, entropy change value |ΔS_T|^max decreased with B-doping from 13.6 J kg⁻¹ K⁻¹ to 4.1 J kg⁻¹ K⁻¹ for a µ_oΔH = 2 T and from 24.7 J kg⁻¹ K⁻¹ to 12.3 J kg⁻¹ K⁻¹ for a µ_oΔH = 5 T; however, it can be noticed that the ΔST(T) curves become broader. This behavior agrees with the report in Ref. [30] for B-doped samples with ≥0.2 at.%. However, it contrasts with the reported by Guan et al. [12] since these authors reported that |ΔS_T|^max remains almost unchanged after doping. Our results also differ from those of Zhang et al. [22] because they demonstrate that through B-doping, |ΔS_T|^max increased from 3.8 J kg⁻¹ K⁻¹ to around 11 J kg⁻¹ K⁻¹ at µ_oΔH = 2 T and from 9.7 J kg⁻¹ K⁻¹ to 24.3 J kg⁻¹ K⁻¹ at a µ_oΔH = 5 T. As

Table 3. |ΔS_T|^max, RC-1, RC-2, RC-3, δT_FWHM, and δT^RC−3, through the AFM-FM martensitic transformation under magnetic field changes of 2 and 5 T, were linked to the MST-to-AST phase transition for the (Ni_37.5Co_12.5Mn₃₅Ti₁₅)_100−xB_x ribbon samples studied. <HL> represents the average value of the magnetic hysteresis loss at 2 T over δT_FWHM.

	x = 0.0 AS		x = 0.4 AS		x = 0.4 TA
	μoΔH = 2 T	μoΔH = 5 T	μoΔH = 2 T	μoΔH = 5 T	μoΔH = 2 T	μoΔH = 5 T
|ΔST|max (J kg−1 K−1)	13.6	24.7	4.1	12.2	4.2	14.0
<HL> (J kg−1)	−12	✕	−4.1	✕	−3.0	✕
RC-1 (J kg−1)	58	195	59	183	49	176
RC-2 (J kg−1)	47	163	48	150	43	144
Thot (K)	307	306	268	266	290	289
Tcold (K)	303	298	254	252	278	276
δTFWHM (K)	4	8	14	15	12	13
RC-3 (J kg−1)	29	105	31	97	29	92
δTRC−3 (K)	4.5	6	13	12	8	10
ThotRC−3 (K)	307	305	267	265	288	288
TcoldRC−3 (K)	303	299	255	253	280	278

shows, the thermal annealing increased the structural transition temperature and |ΔS_T|^max in the B-doped ribbons. Figure 4b depicts the temperature dependence of the magnetic hysteresis loss for a magnetic field change of 2 T (i.e., the HL(T) graph). The HL values for each temperature were determined from the area between the isothermal magnetization curves on increasing the magnetic field to 2 T and decreasing it to zero. Figure S3 depicts the magnetization curves in the martensite-to-austenite phase transition temperature interval measured for each sample. It is highlighted that, due to the first-order nature of the transition, for each temperature, the M(µ_oH) curves were measured following the proper thermomagnetic protocol. The average hysteresis loos value, <HL> value, was estimated from the HL(T) curve in the temperature interval given by the full-width at half maximum, δT_FWHM = T_hot − T_cold, of the ΔS_T(T) curve. As can be seen, the addition of B led to a decrease in <HL>. Table 3 summarizes the significant magnetocaloric properties through the AST-MST phase transition, including the refrigerant capacity (RC) values. RC is an important magnetocaloric parameter that measures the heat the material can transfer between hot and cold reservoirs if an ideal refrigeration cycle is considered. RC was estimated using the following criteria: (i) RC-1 = |ΔS_T|^max × δT_FWHM, where δT_FWHM = T_hot − T_cold; (ii) by calculating the area under the ΔS_T(T) curve between T_hot and T_cold (RC-2); and (iii) by maximizing the product ΔS_T × δT below the ΔS_T(T) curve (RC-3; Wood and Potter criterion) [31]. The effective magnetic refrigeration capacity (RC_eff) at 2 T can be calculated by subtracting <HL> from the respective RC value.

4. Conclusions

This work demonstrates how single-phase ribbon samples doped with interstitial B can be obtained in the Ni-Co-Mn-Ti system through rapid solidification using the melt-spinning technique, as confirmed using X-ray diffraction and thermomagnetic analysis curves. As in bulk samples, austenite exhibits a B2-type crystal structure. B-doping decreases the martensitic transformation temperature and the Curie temperature of austenite by around 42 K and 57 K, respectively; this radically differs from the reported for bulk alloys, demanding further study. Also, the maximum field-induced entropy change and thermal hysteresis reduce with the small amount of B added (0.4 at.%). Thermal annealing at a moderate temperature of 1073 K increases the magnetostructual transition temperature close to room temperature and does not reduce ΔS_T^max and the Curie temperature of austenite compared to the obtained as-solidified B-doped samples.