19.1 Introduction

The magnetocaloric effect (MCE) was discovered in iron by Warburg (1881) and explained theoretically by Debye (1926) and Giauque (1927). It consists of increasing (decreasing) the temperature of a ferro- or paramagnetic specimen under adiabatic conditions in a magnitude ΔT_ad when an external magnetic field H is suddenly applied (removed). During adiabatic magnetization of the specimen, the total entropy of the specimen ΔS_T remains constant. The increase (decrease) in magnetic order will be compensated by an increase (decrease) of disorder in atomic order given by a lattice entropy change ΔS_L. This leads to an increase (decrease) of temperature ΔT_ad and a decrease (increase) in the magnetic entropy ΔS_M of the magnetic system. Thus, the MCE of a material is characterized by both the isothermal magnetic entropy change ΔS_M(T) and the adiabatic temperature change ΔT_ad(T) on application or removal of an external magnetic field. The magnetic cooling process has been extensively used since the 1930s as technology to obtain very low temperatures by adiabatic demagnetization of a paramagnetic salt, namely Gd₂(SO₄)₃·8H₂O (Giauque and MacDougall, 1933). However, in 1997 the discovery of the giant magnetocaloric effect (GMCE) in the intermetallic ternary compound Gd₅Ge₂Si₂ along with the development of the first energetically efficient room-temperature magnetocaloric refrigerator based on metallic Gd as magnetic refrigerant have strongly stimulated both the search of new families of GMCE materials and the development of magnetic refrigeration technology with emphasis in the room-temperature range (Pecharsky and Gschneidner, 1997; Tegus et al., 2002).

The four steps of magnetic refrigerator are schematically shown in Figure 19.1 (they are similar to those of the classical Carnot cycle):

1. Adiabatic magnetization. Within this cycle, a magnetic specimen is magnetized quickly (to get adiabatic process). As a result, its temperature increases.

2. Isomagnetic cooling. Within this regimen, the magnetic specimen is cooled down to ambient temperature under applied magnetic field.

3. Adiabatic demagnetization. Within this cycle, magnetic field is quickly removed. As a result, the temperature of specimen decreases.

4. Isomagnetic heating. The specimen is heated up by cooling agent without applied magnetic field. Inversely, the temperature of cooling agent decreases.

In a ferromagnetic material, the MCE is primary characterized by the ΔS_M(T) and ΔT_ad(T) curves across its paramagnetic-to-ferromagnetic transition (i.e., its Curie point); they show an asymmetric peak, usually referred to as “caret-like” shaped, close to T_c (Chikazumi, 2010). Therefore, the best alloys for room temperature magnetic refrigeration are those having their Curie temperature close to the room temperature.

The MCE can be considerably enhanced in materials exhibiting a coupled first-order magneto-structural transition (FOMT; Krenke et al., 2005a). In this case, the total field-induced entropy change ΔS_T is given by a sum of magnetic ΔS_M and structural ΔS_str entropy variations (Gschneidner et al., 2012). In several materials, such as La(Fe,Si)₁₃, MnAs, La(Fe,Si)13H_y (La,Ca)MnO₃, MnFe(P,As), some Heusler alloys in the Ni–Mn–Ga system and MnCoGeBx (Trung et al., 2010), such type of magneto-structural coupling gives rise to the so-called GMCE. Moreover, in nonstoichiometric Heusler alloys in the Ni–Sn–X (X = Sn, In, Sb), a giant inverse MCE has been observed (Chernenko et al., 1998; Krenke et al., 2005a,b; Han et al., 2006; Kumar Pathak et al., 2007). The effect is due to the positive magnetization change ΔM associated to the reverse martensitic transformation because the high-temperature phase (austenite, AST) shows a higher spontaneous magnetization than martensite (MST). Because the magnetization of the high-temperature phase is higher than that of low temperature, the total entropy change is positive. Hence, alloys with structural transition close to the room temperature are welcome.

19.2 Heusler alloys

The Heusler alloy has been known since 1903 when Friedrich Heusler prepared the alloy Cu₂MnAl (Heusler et al., 1903). This composition was very interesting because it shows ferromagnetic behavior, even though it consists of nonferromagnetic elements. A huge variety of Heusler alloys exists now (Graf et al., 2011) showing interesting semiconducting (Wood et al., 1985), superconducting (Klimczuk et al., 2012), thermoelectric (Mikami et al., 2008), magneto-optical (Carey et al., 2000) properties, half-metallicity (de Groot et al., 1983), magnetic shape memory (Ullakko et al., 1996) or magnetocaloric (Krenke et al., 2005a,b) properties.

Generally, the Heusler alloys are divided into two main groups: full-Heusler alloys characterized by a chemical composition X₂YZ that crystallize in the L2₁ structure consisting of four face-centered cubic sublattices and half-Heusler alloys that are given by chemical composition XYZ (being X and Y the transition metals and Z is semiconductor or nonmagnetic metal) that crystallize in the C1_b structure, which is obtained by removing one X sublattice from the L2₁ structure (Hirohata et al., 2006).

Some stoichiometric (full-Heusler) and nonstoichiometric Ni₂MnX-based (X = Ga, In, Sn) full-Heusler alloys show desired properties for high MCE – T_c close to room temperature and structural transition from martensitic to austenitic phase close to the T_c (Chernenko et al., 1998; Krenke et al., 2005a,b; Han et al., 2006; Kumar Pathak et al., 2007) that makes them ideal candidates for magnetocaloric application.

Usually, Heusler alloys are produced as bulk polycrystalline samples by arc or induction melting of the pure starting elements (Moya et al., 2007). Later on, a long-term thermal treatment at high temperatures must follow to get homogeneous samples (Oikawa et al., 2006).

A few years ago, rapid solidification using the melt spinning technique has been employed to produce highly homogeneous Heusler alloys in the Ni–Mn–X with X = Sn, In, Sb (Hernando et al., 2008a,b; Sánchez Llamazares et al., 2008). This technique offers two important advantages for the fabrication of Heusler alloys: the avoiding, or time reduction, in the thermal annealing process necessary to reach a homogeneous single phase alloy as in the case of bulk samples, and the synthesis of highly textured polycrystalline ribbons with well-oriented columnar-like polycrystalline microstructure (Figure 19.2). Moreover, preferred orientation of the crystalline grains growth results in a well-defined preferential magnetic anisotropy with the easy axis parallel to the ribbon's plane as can be inferred from the hysteresis loops measured in parallel and perpendicular direction with respect to the ribbon's plane (see Figure 19.3).

In addition, such melt-spun ribbons, similar to their bulk counterpart, are characterized by a first-order structural transition from a martensitic low-temperature phase to a cubic austenitic high-temperature phase similar to the bulk Heusler samples as shown in Figure 19.4 (Hernando et al., 2008a,b; Lyange et al., 2014; Sánchez Llamazares et al., 2008; Zheng et al., 2013).

19.3 Glass-coated magnetic microwires

One of the key factors for magnetocaloric applications is the heat exchange between the magnetocaloric material and surrounding that must be cooled. To get a high exchange rate, a high surface/volume ratio fraction is desirable. Such a condition is fulfilled in the case of glass-coated magnetic microwires.

Glass-coated magnetic microwires are composite materials (see Figure 19.5) that consist of a metallic nucleus (typically of diameter ~ 0.1–50 μm) that is covered by a glass-coating (of thickness ~ 2–20 μm) (Zhukov et al., 2004; Vazquez, 2007; Chiriac et al., 2011). They are produced by drawing and quenching (by a water jet) of molten master alloy (the so-called Taylor–Ulitovski method introduced by Taylor in 1924 [Taylor, 1924] and extended by Ulitovski in 1950 [Ulitovski, 1950]). As a result, strong stresses are introduced into metallic nucleus that arises from drawing and quenching as well as from the different thermal expansion coefficients of metallic nucleus and glass coating (Chiriac et al., 1995). From the point of view of applications, the biggest advantage of glass-coated microwires is their nonexpensive and easy production process, small dimensions that do not affect mechanical properties of composite, in which they are included and glass-coating that prevents the metallic nucleus from oxidation, short-circuit and improved mechanical properties of potentially brittle crystalline nucleus.

Although known for almost 90 years, glass-coated microwires came to the attention of scientists in last two decades because of peculiar magnetic properties of amorphous microwires. Either it is magnetic bistability characterized by single Barkhausen jump at a switching field for microwires with positive magnetostriction or giant magnetoimpedance (GMI) effect for microwires with negative and almost vanishing magnetostriction. Both of these phenomena can be applied in many sensing devices to sense magnetic field, position, mechanical stresses, temperature, and so on (Vazquez et al., 2007; Hudak et al., 2013; Praslicka et al., 2013). Naturally, the idea to employ glass-coated microwires for magnetocaloric applications appeared shortly after introduction of rapid quenching for production of Heusler alloys.

19.4 Heusler glass-coated microwires

With a small diameter (which gives high surface/volume ration), glass-coated microwires are the ideal material for magnetocaloric applications. Moreover, glass coating improves the mechanical properties of a metallic nucleus that is usually very brittle in the case of Heusler alloys and improves the corrosive resistance of metallic nucleus.

One of the first studied compositions was the model Ni₂MnGa Heusler alloy (Varga et al., 2011). Heusler-type Ni_50.95Mn_25.45Ga_23.6 ingot has been previously arc-melted from pure elements. Two magnetic Ni_50.95Mn_25.45Ga_23.6 glass-coated microwires, one with diameter of metallic core, d ≈ 44 μm, surrounded by glass coating (of the thickness of 18 μm) and the second one with diameter of metallic core, d ≈ 8 μm, surrounded by glass coating (of the thickness of 7 μm) have been later fabricated by the Taylor–Ulitovski method. A single wire having a length of 1.5 cm has been used for structural and magnetic characterization as well as for magnetocaloric studies.

As-prepared microwires did not show ferromagnetic ordering. However, annealing at 823 K (annealing time 5 min) results in a drastic change of magnetic properties: Annealed samples show magnetization versus temperature dependence typical for ferromagnetic behavior with Curie temperature about 315 K (see Figure 19.6). However, in contrast to the bulk or rapidly quenched ribbon (see also Figure 19.4), no structural transition appears. Most probably, this is the effect of strong stresses applied on metallic nucleus by glass coating, as it has been shown that mechanical stress affects strongly the transition temperature (Mañosa et al., 2010). Neither of microwires (8 or 44 μm in diameter) exhibits the structural transition.

On the other hand, the crystalline structure of microwires depends strongly on the applied stress. X-ray diffraction shows a single crystalline structure of tetragonal martensitic structure with a lattice parameters a = 3.75 Å and c = 6.78 Å for microwire with 44 μm in diameter (Figure 19.7). However, the structure is elongated along the c axis (compared to bulk materials – Cong et al., 2007; Wang et al., 2006). It is assumed that the high levels of stress originating from the difference between the thermal expansion coefficients of the metallic core and the glass coating may be responsible for these deviations in lattice parameters. In contrary, Ni₂MnGa microwire with 8 μm in diameter shows a single crystalline structure of cubic austenitic structure with a lattice parameter a = 5.810 Å. Hence, it confirms that both phases are possible to produce in glass-coated microwires; just the transition between them is hindered.

Hysteresis loops measured in parallel and perpendicular direction with respect to the wire's axis confirms strong anisotropy introduced during the production process. Hysteresis loops measured in parallel direction show easier saturation with saturation field 3 kA m^− 1, whereas that one measured in perpendicular direction shows almost unhysteretic behavior and much elevated saturation field (over the maximum field applied in experiment 10 kA m^− 1 – see Figure 19.8). These results point to the easy axis parallel to the wire's axis.

To estimate refrigerant capacity (RC) of microwires, the change of entropy has been obtained from magnetization measurements measured for 44 μm microwire within the temperature range from 250 to 350 K (see Figure 19.9). The magnetic entropy change for different applied magnetic field ΔS_M(T,H), as a function of temperature, has been calculated from the Maxwell relationship (Pecharsky and Gschneidner, 1999a,b):

$\mathrm{\Delta }{S}_{\mathrm{M}}\left(T,H\right)={\displaystyle {\int }_0^H{\left(\frac{\partial M}{\partial T}\right)}_H\mathrm{d}H}$

The entropy change ΔS (Figure 19.10) shows a wide peak with a maximum at about 315 K that corresponds to the Curie temperature estimated from magnetization measurement (Figure 19.9). The maximum of ΔS is slightly above 0.7 J kg^− 1 K^− 1 and is comparable to that found in rapidly quenched ribbons (Hernando et al., 2008a).

Another group of studied microwires was NiMnIn-based (Vega et al., 2012). The authors showed that Ni₅₉Mn_23.5In₁₅ microwire having 19.3 μm of metallic nucleus are characterized by AST structure. The Curie temperature has been measured around 250 K. Similar to the Ni₂MnGa microwire, no structural transition is observed neither in as-cast nor in annealed state. However, the entropy change was higher than in Ni₂MnGa microwire (up to 2 J K^− 1 kg^− 1 in magnetic field 30 kOe) giving the RC up to 150 J kg^− 1.

It has been shown that a small addition of Co can successfully enhance the MCE (Wu et al., 2011; Maziarz, 2012; Sahoo et al., 2013). Hence, MCE has also been investigated in glass-coated Ni₅₀Mn₃₅In₁₅, Ni₄₅Mn_36.5In_13.5Co₅, and Ni_42.5Mn_37.5In_12.5Co_7.5 microwires with the metallic core diameter 18, 32.2, and 24.3 μm, respectively, covered by the Pyrex glass with total diameter of 40, 41.1, and 33.3 μm, respectively. Similar to the Ni₂MnGa microwire, the microwires were annealed at 823 K for 10 min in a protective helium atmosphere to release the stresses introduced during the production process. A single wire having a length of 1.5 cm has been used for structural and magnetic characterization as well as for magnetocaloric studies. The bunch of 100 microwires has been used for X-ray analysis.

Figure 19.11 shows the SEM images taken from a section of the microwires. Highly ordered polycrystalline structure is observed, in which the individual crystals grow perpendicular to the wire's axis starting from the point where the water jet touches the wire surface for the first time (see also Chapters 9 and 10). When the glass coating is too thick (Ni₅₀Mn₃₅In₁₅ – Figure 19.11, left), the cooling rate is lower, and no such effect is visible. However, when the glass coating is thin (Ni₄₅Mn_36.5In_13.5Co₅ and Ni_42.5Mn_37.5In_12.5Co_7.5), the starting point for crystallization is clearly recognized (marked by arrows in Figure 19.11). The size of the crystals reflects the cooling rate being high for low cooling rate in the center of the wire and low below the surface where the cooling rate is higher.

X-ray difractograms confirm AST single phase with cubic bcc L2₁ crystal structure with lattice parameter a = 5.935 Å for Ni₅₀Mn₃₅In₁₅, a = 5.920 Å for Ni₄₅Mn_36.5In_13.5Co₅, and a = 5.910 Å for Ni_42.5Mn_37.5In_12.5Co_7.5 (Figure 19.12). The lattice parameter slightly decreases with the decrease of In content (increase of Co).

Preferred orientation of the crystal growth leads to the well-defined magnetic anisotropy. It can be recognized from the shape of the hysteresis loop that shows high susceptibility at low fields and fast approach to saturation for magnetic field orientation parallel to the wire's axis (Figure 19.13), having 115 Oe for Ni₅₀Mn₃₅In₁₅, 87 Oe for Ni₄₅Mn_36.5In_13.5Co₅, and 64 Oe for Ni_42.5Mn_37.5In_12.5Co_7.5. On the other hand, magnetization growth gradually with the field oriented perpendicularly to the wire's axis, having coercivity 122 Oe for Ni₅₀Mn₃₅In₁₅, 80 Oe for Ni₄₅Mn_36.5In_13.5Co₅, and 157 Oe for Ni_42.5Mn_37.5In_12.5Co_7.5. These results point to the easy axis parallel to the wire's axis and perpendicular to the major axis of large grains. Similarly, it was found in rapidly quenched ribbons (Hernando et al., 2008a,b).

Strong stresses introduced on the metallic nucleus hinder structural transformation, and the temperature dependence of saturation magnetization confirms single phase in a wide range of temperatures from 100 to 400 K. On the other hand, the strong and complex stress distribution results in the appearance of the so-called Hopkinson maximum (Cullity and Graham, 2009) – sharp increase of magnetization just below the Curie temperature when it is measured at low fields (see Figure 19.14, left). As a result, the temperature dependence of magnetization curves shows peculiar dependence showing the range where magnetization increases with temperature at low fields, whereas high-field magnetization curves show continuous decrease with the temperature (Figure 19.14, right) similar to the case of Ni₂MnGa microwires. The Hopkinson maximum is finished by a sharp decrease of magnetization in a very narrow temperature range. The stronger is the anisotropy, the sharper is the Hopkinson maximum. Hence, in Ni₅₀Mn₃₅In₁₅, which has thickest glass-coating (see Figure 19.11) and slowest cooling rate, the Hopkinson maximum almost vanishes. On the other hand, a high quenching rate (thinner glass-coating) in case of Ni₄₅Mn_36.5In_13.5Co₅ and Ni_42.5Mn_37.5In_12.5Co_7.5 results in well-pronounced stress distribution (see Figure 19.11) and sharp Hopkinson maximum.

The appearance of Hopkinson maximum results in a peculiar temperature dependence of magnetization curves measured at low fields (Ryba et al., 2013). Figure 19.15, left, shows the magnetization curves measured at fields bellow 500 Oe in the temperature range close to the Curie temperature (240–360 K). As a result of Hopkinson effect, the magnetization increases with temperature in the field range 0–350 Oe. Above 280 K, the magnetization starts to decrease steeply with temperature in the whole field range.

In contrast to the low field magnetization curves, the measurement at higher fields (up to 30 kOe) shows continuous and monotonous decrease of magnetization in the whole measured temperature range (Figure 19.15, right).

Finally, the temperature dependence of the entropy change shows a maximum at the Curie temperature (T_max ~ 290 °C, see Figure 19.16). The maximum is narrower for low-field and wider for high-field measurement reflecting the temperature dependence of magnetization that shows a sharp decrease of M at low field due to Hopkinson maximum, whereas it decreases monotonously at high fields. Moreover, the amplitude of maximum is not proportional to the applied magnetic field. At low field it steeply increases, having a maximum at 500 Oe (Figure 19.16), whereas it decreases with the applied field at the fields from the range of 10–30 kOe.

From the temperature dependence of magnetic entropy change, the RC for each corresponding field can be calculated (see Figure 19.17). The RC increases with increasing applied magnetic field having a maximum at 20 kOe. However, to estimate the efficiency of MCE, we have calculated the ratio between the RC and applied magnetic field, which better expresses the energy obtained by cooling with respect to the energy supplied to get cooling. Such parameters show maximum at the applied field of 300 Oe and gradually decreases with the applied field increase. As a result, the ration is more than five times higher at 300 Oe than that obtained for 30 kOe that points to the five times higher efficiency at low fields. This is a result of Hopkinson maximum and the sharp variation of magnetization connected to this effect.

These results show the advantages of microwires to employ them into the magnetocaloric applications that use second-order transition. However, no first-order transition appeared in all studied samples as it was observed in other Heusler alloys of the same compositions. As given earlier, such FOMT should increase the efficiency of MCE in microwires and enhance their application possibilities into the region of shape memory alloys.

Various phases have been shown earlier to appear in the different microwires from MST to AST. This points to the fact that it should be possible to get both phases in the microwire and therefore also to observe the first-order transition. However, the key parameter that hinders the structural transition is the high stress applied by glass coating on metallic nucleus (Zhukov et al., 2013; Zhukova et al., 2014). The first promising result appeared recently (Aronin et al., 2013) showing that structural phase transition could appear in the microwire after glass removal. Although the positive effect of glass coating is removed, too, it points to the fact that shape memory alloys based on glass-coating microwires could be possible when the glass thickness is selected properly.

19.5 Applications perspectives of glass-coated microwires

The trend towards miniaturization in all technical fields demands new concepts for the manufacture of cooling devices as well. To meet these technical challenges, the construction of such devices requires optimized geometries on a micrometer scale. MCE brings advantages in microcooling devices. There are some approaches already confirmed to be successful in application of MCE (Tsukamoto et al., 2012). Small dimensions, insulating glass coating and specific magnetic properties, make glass-coated microwires also ideal material for miniaturized applications. Small dimensions of microwires allow their applications at higher frequency because of reduction of eddy current during magnetization (in comparison with the bulk materials). Moreover, higher refrigerant efficiency at low fields due to Hopkinson effect enhances their applicability for devices where high fields are not allowed or hardly to be produced.

In addition, cost-effective mass-production is a prerequisite for a competitive market position. Microwires offer a quite simple and cheap production process (bringing high added value of the microwires), which allows the application of glass-coated microwires into the microcooling devices with low energetic requirements.

By solving the problem with the stress applied by glass coating, the structural transition could be observed that enhances the application range into the field of magnetic shape memory microdevices (Ullakko et al., 2012).

19.6 Conclusions

In this chapter, we have shown that it is possible to produce glass-coated microwires based on Heusler alloys. They are characterized by small dimensions (that brings high surface-to-volume ratio and allows application into a wide range of materials without variation of their mechanical properties) and insulating glass coating (that prevents microwire from chemical and electrical interaction with surroundings and enhances mechanical strength of Heusler-based metallic nucleus). Rapid quenching employed during microwire's production leads to the single-phase polycrystalline material. NiMnGa- and NiMnInCo-based microwires are characterized by the Curie temperature close to room temperatures. High stresses introduced during the production process results in the complex anisotropy distribution. The complex anisotropy distribution leads to the Hopkinson effect – huge and sharp variation of magnetization just below the Curie temperature. Finally, such wires are characterized by enhanced refrigerant efficiency (reflected in the elevated RC/applied magnetic field ratio) at low fields, which present maximum at the field of 300 Oe.