DEVELOPMENT OF NI-MN-BASED FERROMAGNETIC SHAPE MEMORY ALLOYS. Zhigang Wu

Transcription

1 DEVELOPMENT OF NI-MN-BASED FERROMAGNETIC SHAPE MEMORY ALLOYS Zhigang Wu School of Mechanical and Chemical Engineering The University of Western Australia This thesis is presented for the degree of Doctor of Philosophy of Engineering of The University of Western Australia (2011)

2 Abstract Since the discovery of Ni 2 MnGa ferromagnetic shape memory alloys some 15 years ago, intensive research has been conducted to search and develop new and more powerful magnetically activated shape memory alloys. The effort has been severely hampered by the low magnetic driving force, intrinsically limited by the magnitude of magnetic crystallographic anisotropy, for mechanical actuation. The discovery of metamagnetic phase transformation in Ni-Mn-Z (Z=In,Sn,Sb) system in 2004, with their large magnetization difference across the transformation, made a breakthrough and brought new promise for creating magnetically activated shape memory alloys. This study is concerned with the development of Ni-Mn-Z (Z=In,Sn) ternary ferromagnetic martensitic alloys. Whereas having high promise owing to their large magnetization difference between their nonmagnetic martensite and ferromagnetic austenite, these alloys face the challenges of high mechanical resistance to deformation and brittleness. In response to these challenges, this study is focused on two main objectives: (1) to further enhance the magnetization difference of the metamagnetic reverse transformation of the alloys, and (2) to improve the toughness and ductility of the alloys, through alloying. (1) Enhance the Magnetization Difference New alloy design is accomplished in order to increase the magnetization difference between the austenitic and martensitic phases in Ni-Mn-Z (Z=In,Sn) alloys. The first step of the composition design is to maximise the use of Mn content to provide the potentially largest magnetization. Then, the proportion between Ni and In/Sn contents is adjusted to alter the chemical order for obtaining ferromagnetic structure. Lastly, Co addition is employed to modify the e/a ratio and to enhance the magnetic ordering of these alloys. In the new compositions of Mn 50 Ni 40-x In 10 Co x and Mn 50 Ni 42-x Sn 8 Co x alloys, a martensitic transformation from an Hg 2 CuTi-type austenite to body centred tetragonal martensite was observed. In both systems, the magnetization of the austenite increased significantly, i

3 whereas that of the martensite changed much less prominently with increasing the Co substitution for Ni, leading to the significantly enhanced magnetization difference across the transformation. The increased magnetization of the austenite is attributed to (i) formation of ferromagnetically coupled Mn-Mn atoms due to the new atomic configuration in off-stoichiometric composition, (ii) higher magnetic moment contribution of Co relative than Ni, and (iii) widening of temperature window for ferromagnetic austenite. The low magnetization of the martensite, relative to that of the austenite, is due to the significantly shortened distance between Mn-Mn, which leads to the disappearance of the local ferromagnetic structure in a tetragonal martensitic structure. (2) Improve the Alloy Ductility Fe is utilised to substitute for Mn in Ni-Mn-Z(Z=In, Sn) alloys to form a phase in the matrix to increase the ductility of the alloys. Whereas much attention has been given to the ductility improvement, metallurgical origins of the influences of fourth element addition on the martensitic and magnetic properties are much less understood. In Ni 50 Mn 38-x In 12 Fe x and Ni 50 Mn 40-x Sn 10 Fe x alloys, a martensitic transformation from a B2 austenite to an orthorhombic martensite was realised. Substitution of Fe for Mn at above 3 at% introduced an fcc phase in the microstructure, the amount of which increased with increasing Fe addition in both systems. The Curie temperature of the parent phase increased slightly, whereas the Curie temperature of the martensite increased rapidly with increasing Fe addition. Changes in the temperatures of the martensitic and magnetic transformations are confirmed to directly relate to the e/a ratio of the matrix caused by the formation of phase. Fe addition effectively weakens the antiferromagnetic ordering of the austenite in the matrix phase, leading to the increase of magnetization difference across the martensitic transformation. The relative shape memory effect decreased from 94 % to 37 % after 4 at% Fe addition. These findings clarify the metallurgical origins of the side effects of Fe addition on martensitic and magnetic properties and provide reference on alloy design for Ni-Mn-based alloy systems. ii

4 Publications arising from this thesis This thesis is written as a series of research publications, and my contribution to each publication is indicated as following: 1. Wu Z (70%), Liu Z, Yang H, Liu Y, Wu G, Metamagnetic phase transformation in Mn 50 Ni 37 In 10 Co 3 polycrystalline alloy, Applied Physics Letters, 2011, 98, pp (1-3). (1 st paper in Chapter 2) 2. Wu Z (80%), Liu Z, Yang H, Liu Y, Wu G, Effect of Co addition on martensitic phase transformation and magnetic properties of Mn 50 Ni 40-x In 10 Co x polycrystalline alloys, Intermetallics, 2011, 19, pp (2 nd paper in Chapter 2) 3. Wu Z (80%), Liu Z, Yang H, Liu Y, Wu G, Martensitic phase transformation and magnetic properties of Mn 50 Ni 42-x Sn 8 Co x polycrystalline alloys, Journal of Physics D: Applied Physics, 2011, 44, (1-8). (3 rd paper in Chapter 2) 4. Wu Z (80%), Liu Z, Yang H, Liu Y, Effect of Fe addition on the martensitic transformation behaviour, magnetic properties and mechanical performance of Ni 50 Mn 38-x In 12 Fe x polycrystalline alloys, submitted to Journal of Alloys and Compounds. (1 st paper in Chapter 3) 5. Wu Z (60%), Liu Z, Yang H, Liu Y, Wu G, Woodward RC, Metallurgical origin of the effect of Fe doping on the martensitic and magnetic transformation behaviours of Ni 50 Mn 40-x Sn 10 Fe x magnetic shape memory alloys, Intermetallics, 2011, 19, pp (2 nd paper in Chapter 3) Candidate signature:... Coordinating supervisor signature:... iii

5 Acknowledgements I acknowledge my outstanding supervisor Winthrop Professor Yinong Liu. I received world-class academic trainings of being a scientist, and enjoyed our many discussions about scientific issues we had through my PhD study. He has been caring, wise, friendly and supportive, and my debt to his is enormous. I acknowledge Associate Professor Hong Yang, who has been a mentor to my research. I appreciate her genuine helps to my wife and myself to set up our life in Australia when we first arrived in Perth and all the assistances afterwards through the years. Colleagues in our research group have had a huge influence on my career, which is reflected in this thesis. In particular, I acknowledge Dr Zhuhong Liu, Qinglin Meng, Mu Zhang, Jingyang Li, Xiaoxue Xu, Yimeng Yang, Bashir Samsam, Mazlina Mat Darus, and Mingliang Wang. Their wisdom, support and friendship over the years have been most important to me. I acknowledge Dr Alexandra Suvorova, Dr Martin Saunders and Dr Janet Muhling, who gave me all possible assistances in using the facilities in CMCA. Their knowledge and experience on materials characterisation are valuable contribution to my research work. I acknowledge Dr Robert Woodward in School of Physics, who assisted me to measure magnetic properties of my samples, which weigh ~50% of the total experimental work, and helped me a lot with understanding of magnetism in many discussions we had. I have always valued the contribution of my wife, Meifang Lai, who gave enormous support to my research. She has been a great listener and a true friend in my life. It is her company and encouragement that made my PhD research possible. I dedicate this work to her. I have been also receiving tremendous support from my parents in China, who gave me best education at that time. They have made huge sacrifices to allow me to pursue my dreams, and for their unconditional support and love, I will always be so grateful. iv

6 Table of contents Abstract i Publications arising from this thesis iii Acknowledgements iv Table of Contents v Chapter 1: Introduction 1 Chapter 2: Increasing magnetic driving force of Ni-Mn-based alloys 37 Chapter 3: Increasing ductility of Ni-Mn-based alloys 97 Chapter 4: Closing Remarks 142 v

7 CHAPTER 1. Introduction 1.1 Magnetomechanical effect of materials Transducing materials are becoming increasingly important in modern technologies, which combine large strain, high specific force output and fast dynamic response during an actuation event. The functionality of these materials is based on the physical mechanisms responsible for the thermal, electrical, optical, chemical or magnetic energy transformations into mechanical work, which produce actuation. For example, conventional shape memory alloys (NiTi) are widely used as thermal actuators. By heating up a typical NiTi alloy to above the martensitic transformation temperature, a strain of ~6% can be produced, or recovered, accompanying a force output of up to ~850 MPa. During such an actuating event, thermal energy is converted to mechanical work. Magnetic-field-induced mechanical actuation is another type of energy conversion, which has the advantages of high response frequency, good cycling stability, and environmentally friendliness compared to conventional shape memory alloys. The most well known magnetoactuators are the traditional magnetostrictive materials. Magnetostriction is a common phenomenon for all solid magnetic materials, but only in a few the effect is large enough for engineering exploitation. It refers to the phenomenon in which a material changes its physical dimensions in response of changes in magnetisation state. The best known magnetostrictive materials are cubic Laves-phase intermetallics, often in the form of (RE)(TM) 2 (TM=Fe, Co, Ni and Mn) [1-3]. The largest ever measured Laves-phase intermetallic is in TbMn 2 (0.6 % at 40 K) and the most successful in application is Tb x Dy 1-x Fe 2, the infamous Terfenol-D [4]. CHAPTER 1 1

8 The magnitude of magnetostrain in magnetostrictive materials (<0.6%) is considered rather small, which severely hinders their engineering applications in many aspects. Therefore, development of new types of magnetoactuation materials with large magnetostrains has become an intensive research interest to widen the application realm of magnetoactuation. 1.2 Large magnetostrain by martensite variant reorientation In 1996, Ni 2 MnGa alloy was found to generate a strain of ~0.2% under the influence of an applied magnetic field of 8 koe [5]. This material combines the properties of ferromagnetism with those of a thermoelastic martensitic transformation, thus denoted as a ferromagnetic shape memory alloy (FSMA). With this discovery, tremendous effort has been made on searching for larger magnetostrains by adjusting the compositions to offstoichiometric derivatives in Ni-Mn-Ga alloys. These alloys in their martensite state allow for a stress- or magnetic-field-induced rearrangement of twin variants, resulting in giant magnetostrains, 5-10% having been reported in the literature [6-9]. The enhanced magnetostrains, caused by twin boundary motion in Ni-Mn-Ga single crystals, led to intensively active research in the interdisciplinary field of ferromagnetic martensite in the following decade. The aforementioned record-breaking values of magnetostrain and the extreme magnetocaloric effect rekindled the interest in Ni-Mn-Ga and related multifunctional materials nowadays. FSMAs have been developed into a new class of functional materials that are capable of magnetic-field-induced actuation, mechanical sensing and magnetic refrigeration Mechanism of magnetostrain in Ni Mn Ga via martensite reorientation The magnetostrain in Ni-Mn-Ga alloys is associated with the orientation change of the martensite variants via twin boundaries movement. The change of variant orientation CHAPTER 1 2

9 induced by the magnetic field is a process to allow the growth of the martensite variants with the easy magnetisation axis aligned with the applied magnetic field at the expense of others. This requires the martensite of Ni-Mn-Ga to have both structural anisotropy and magnetisation anisotropy. Structural anisotropy of the martensite The stoichiometric Ni 2 MnGa undergoes a martensitic phase transformation at 202 K. The austenite shows a superlattice cubic structure, i.e. Heusler structure or L2 1 structure at higher temperature, a=0.582 nm, while the martensite exhibits a tetragonal structure, a=b=0.590 nm, c=0.544 nm [10]. The cubic structure of the austenite contracts along the c direction by 4.45 % and elongates along the a and b directions by 1.63 % to complete the structural transformation. The crystal structures of martensite are strongly sensitive to the chemical composition. With the increase of Mn substitution of Ga, the martensite structure exhibits 5 M, 7 M and can be also non-modulated martensite, leading to the transformation from tetragonal to orthorhombic [11]. The tetragonal or orthorhombic structure of the martensite provides the structural anisotropy for the potential shape change. Magnetic anisotropy of the martensite Both the austenitic and martensitic phases are ferromagnetic, although the magnetisation of martensite is slightly bigger than that of austenite in modified composition [12]. Within the martensite structure, the easy magnetic axis lies along the tetragonal c-axis, i.e. the short axis. Figure 1 shows the magnetisation measurement of Ni 48 Mn 30 Ga 22 single crystal along the easy magnetisation direction ([001] axis) and the hard magnetisation direction ([100] axis). The magnetic crystallographic anisotropy energy (K u ), which is the enclosed area between the magnetisation curves along a and c axes, provides the magnetic driving force for field-induced deformation in the tetragonal martensite. CHAPTER 1 3

10 Figure 1 Magnetisation curves along easy ([001]) and hard ([100]) axes of Ni 48 Mn 30 Ga 22 constrained in single variant martensite. [13] High mobility of twin boundary of martensite To complete the magnetostrain in the martensite, sufficient magnitude of K u and good twin boundary mobility are essential. It is important to note that the K u is orientation dependent and limited with a saturated field. A typical K u for Ni-Mn-Ga alloys is between kj/m 3 [14-16]. Given the shape change is typically 6%, this yields a magnetically generated stress of 5-8 MPa. Therefore, to achieve a magnetic-field-induced shape change, the detwining stress level needs to be lower than the magnetostress. Figure 2 shows stress-strain curve of a single-variant sample of the Ni 48.8 Mn 29.7 Ga 21.5 alloy along the [100] direction by a compression test at 300 K. The critical stress for martensite reorientation is very small, at around 1-2 MPa. Such conditions can be satisfied by the alloys which transform from cubic austenite to 5 M and 7 M martensite mentioned above [7, 17]. The magnetostrain is restricted to the tetragonality or orthorhombility of martensite, denoted as 1-c/a. So far, giant magnetostrain of 6% for 5 M and 9.5% for 7 M were successfully obtained in single crystals [6, 7]. Very recently, Straka et al have successfully lowered the critical stress of initiating the twin boundary motion CHAPTER 1 4

11 down to 0.1 MPa [18] by modifying the twins microstructure, which greatly improves the ease of magnetostrain in Ni-Mn-Ga alloys. Figure 2 Stress-strain curve for compression of a single-variant sample of the Ni 48.8 Mn 29.7 Ga 21.5 alloy along the [100] direction at 300 K [7]. As aforementioned, the crystallographic anisotropy, magnetisation anisotropy, the extremely good mobility of the twin boundaries of the martensite determines the success of yielding large magnetostrains in Ni-Mn-Ga alloys Development of other FSMAs However, there are some serious concerns with Ni-Mn-Ga alloys for their industrial applications. One is the brittleness of the material, which is due to its intrinsic nature of an intermetallic compound. Furthermore, the high cost of pure element Ga impedes its practical production on a large scale. Last but not least, the mechanical work output of Ni- Mn-Ga, provided by the intrinsically weak magnetic anisotropy energy, is extremely small as a driving force for mechanical actuations. Even though the critical stress of twinning can be modified to as low as 0.1 MPa [19], the force generated is still restricted by the limited magnitude of K u, at a few MPa [20]. To overcome these problems and to increase the fundamental knowledge of this alloy system, other ferromagnetic shape memory alloys with Heusler structure have been investigated since last years. CHAPTER 1 5

12 Ni-Mn-Al has been developed as another candidate of FSMAs. The austenite shows a B2 or L2 1 structure, while the martensite shows non-modulated tetragonal phase with low Al and Mn content and 5M and 7M tetragonal martensite with high Al and Mn content alloys [21]. The magnetostrain of Ni-Mn-Al is rather small: at about 0.17% in single crystals and 0.01% in polycrystals. A high magnetic field (7 T) is required to yield the magnetostrain [22]. The ductility of Ni-Mn-Al alloys is improved by the precipitation of γ phase particles with the addition of other elements, such as Fe, Co or Cr [23]. Thermoelastic martensitic transformations in the ferromagnetic state were obtained in a large compositional range of Ni-Fe-Ga alloys. Austenite phases have L2 1 structures while martensite phases have 5M and 7M orthorhombic structures [24]. The critical stress for variants reorientation is very low, at 2-3 MPa [25]. The K u in single crystals of 7M martensite is kj/m 3 [26]. Nevertheless, the magnetostrain is much smaller in Ni- Fe-Ga than those in Ni-Mn-Ga alloys. The largest strain reported for Ni 54.2 Fe 19.3 Ga 26.5 single crystals in single variant state is only 0.02 % at 100 K [26]. The magnetostrain can be enhanced to 0.7 % by doping Co in Ni-Fe-Ga alloys [27]. The stress-assisted magnetostrain of 8.5 % was also achieved in Ni-Fe-Ga-Co alloys [28]. The ductility of Ni- Fe-Ga is improved by the presence of the precipitates of γ phase, and the amount and distribution of γ phase can be modified by suitable heat treatments [29-31]. Though the ductility is increased, the transformation strain is reduced due to the presence of γ phase of the material [32]. Co-Ni-Al is being investigated in the last years as another ferromagnetic shape memory alloy system. The parent phase has the B2 structure and martensitic phase has the L1 0 structure [33-35]. The K u of Co-Ni-Al is about 320 kj/m 3 at 5 K and 200 kj/m 3 at 300 K for Co 41 Ni 32 Al 27 alloys in single variants state [36]. However, the values of magnetostrain reported so far are very low: 0.06% in single crystals [37] or 0.013% in polycrystals [38]. The reason for the small strain induced by the relative large magnetic field may be due to the elevated critical stress for variant reorientation of L1 0 structure. Similar to Ni-Mn-Al and Ni-Fe-Ga, the ductility of Co-Ni-Al alloys is improved by the γ phase. CHAPTER 1 6

13 Co-Ni-Ga is another promising ferromagnetic shape memory alloy system induced by Wuttig et al in 2001 [39]. It shows the similar properties as in Co-Ni-Al, like the crystal structures of the two phases (B2 for austenite and L1 0 for martensite respectively) and the martensitic transformation happens around the room temperature [39]. The values of magnetostrain are still small: 0.011% in melt-spun ribbons and 0.003% in polycrystals [40]. Giant magnetostrain triggered by martensite reorientation have been investigated in ordered Fe-Pt and disordered Fe-Pd alloys. The parent phase of Fe-based alloys is fcc phase (γ phase) which can be retained at the room temperature by quenching and the martensite structures of these two alloys are both fct [41]. In Fe-Pt alloys, the martensite transformation temperature is always much below the room temperature (85 K for Fe 3 Pt) [42]. In Fe-Pd alloys, the transformation temperature is around the room temperature and decreases sharply with the increment of Pd concentration [43]. In Fe-Pt single crystals, the amount of magnetostrain is up to 2.3 % measured at 4.2 K [41]. However, very low martensitic phase transformation temperature restricts its application. For Fe-Pd alloys, the values of magnetostrain of 3.1 % have been measured in single crystals and 0.01 % to 0.05 % in polycrystals depending on the size and shape of the grains [44, 45]. Clearly, the magnetostrains obtained in the aforementioned FSMAs are significantly lower than those found in Ni-Mn-Ga alloys. To date, Ni-Mn-Ga shows the best performance of magnetostrain with the mechanism of martensite reorientation under the influence of a magnetic field. 1.3 Magnetoactuation via martensitic transformation Because of the success of Ni-Mn-Ga magnetic shape memory alloys, Ni-Mn- Z(Z=In,Sn,Sb) were introduced for In, Sn and Sb are the neighbor elements within the same or neighbor groups as Ga. This satisfies the requirement of Z position should be taken by sp element in Heusler structure denoting as X 2 YZ and then taken as the potentially ideal substitution for Ga. In 2004, Sutou et al discovered Ni 50 Mn 50-x Z x (Z=In,Sn,Sb; x= ) alloys system [46], and this new system has attracted much interest due to its distinctive CHAPTER 1 7

14 magnetic properties. The stoichiometric composition of Ni 2 MnZ(Z=In,Sn,Sb), exhibit T C lower than Ni-Mn-Ga, but can be elevated by increasing the Mn content. In fact, Ni 50 Mn 50- xz x (Z=In,Sn,Sb; x= ) alloys show thermoelastic martensitic transformation below the T C temperature. In this case, the magnetic actuation is possible to be carried out by magnetic-field-induced martensitic phase transformation. This group of alloys is normally regarded as Ni-Mn-based FSMAs Concurrent structural and magnetic transformation The charm of Ni-Mn-based FSMAs lies on their concurrent martensitic and magnetic transformation. Figure 3 shows the thermomagnetisation behaviour of Ni 50 Mn 34 In 16, Ni 50 Mn 37 Sn 13 and Ni 50 Mn 37 Sb 13 alloys [46]. The T C temperature is defined as the temperature at which the slope of the magnetisation versus temperature is the largest. The martensitic transformation can be identified by the abrupt dropping upon cooling and rising upon heating of the magnetisation curves with the variation of temperature, with an obvious temperature hysteresis. Similar with other conventional FSMAs, the parent phase is ferromagnetic at higher temperature, with a L2 1 structure. However, the martensite phase performs a much weaker ferromagnetism at lower temperature, with a 4M Orthorhombic structure [46]. The largest saturation magnetisation difference between the austenite and martensite (ΔM) among Ni 50 Mn 50-x Z x (Z=In,Sn,Sb) alloys is ~60 emu/g, found in Ni 50 Mn 34 In 16 alloy [47, 48]. The large ΔM is beneficial for magnetic-field-induced martensitic transformation. By slightly adjusting the composition, alloy Ni 46 Mn 41 In 13 has been found to present an enhanced ΔM of ~100 emu/g, which holds the highest ΔM record in Ni 50 Mn 50-x Z x (Z=In,Sn,Sb) alloys. The crystal structure is still L2 1 structure even the all these three element have deviated from their own proper concentration, evidenced by XRD and TEM results [49]. However, the low phase transformation temperatures at around 200 K still impede the further mechanical study and real applications. CHAPTER 1 8

15 Figure 3 Thermomagnetisation curves of (a) Ni 50 Mn 34 In 16, (b) Ni 50 Mn 37 Sn 13 and (c) Ni 50 Mn 37 Sb 13 alloys Magnetic driving force for magnetostrain in Ni-Mn-Z (Z=In,Sn,Sb) alloys Different from the magnetic driving force in Ni-Mn-Ga alloys, which is the magnetic crystallographic anisotropy constant (K u ), the Zeeman energy (ZE) is responsible for triggering the actuation of Ni-Mn-based FSMAs. The comparison between the magnetic driving forces for Ni-Mn-Ga and Ni-Mn-based alloys are illustrated in Figure 4. The magnitude of K u is the enclosed area between the magnetisation responses from two differently oriented variants shown in Figure 4 (a). Once the K u is larger than the energy required for the twin boundary motion, the variants with the easy magnetisation direction CHAPTER 1 9

16 parallel to the magnetic field will grow at the expense of others, resulting in the macroscopic shape change. It is obvious that the magnitude of the maximum K u is limited. Therefore, the low force output has been proven to be a main limitation for the application of these materials for mechanical actuation. (a) (b) Figure 4 Illustration of the maximum magnetocrystalline energy (K u ) in Ni-Mn-Ga alloys responsible for magnetic-field-induced martensite variant reorientation and Zeeman energy (ZE) in Ni-Mn-based alloys responsible for magnetic-field-induced phase transformation. An intrinsic solution to this problem is to increase the magnetic driving power for the martensitic transformation. This mechanism is analogous to stress- or temperatureinduced martensitic transformations in conventional shape memory alloys. Different from the K u, the ZE plays an important role in magnetic-field-induced phase transformations, which stems from the difference in the saturation magnetisations of the phases as shown in Figure 4 (b). Unlike the K u, the ZE does not strongly depend on crystal orientation, which provides an opportunity to utilise polycrystals for actuator applications. With increasing the applied field, the ZE grows continuously with an open end until the phase transformation occurs. However, for a realistic point of view, one should always expect to achieve a magnetostrain at a reasonable magnitude of field. In this case, the ZE should be increased by enhancing the ΔM, such as when a ferromagnetic phase transforms to a paramagnetic or antiferromagnetic phase, or vice versa Effect of Co addition on increasing ΔM CHAPTER 1 10

17 The magnetostrain cannot be achieved in Ni-Mn-Z(Z=In,Sb,Sb) ternary alloys until the Co substitution for Ni was taken as an effective modification for improving the distinctive magnetic properties between the austenite and martensite. As a matter of fact, the ΔM can be significantly increased by substitution Co for Ni in Ni-Mn- Z(Z=In,Sn,Sb,Al,Ga) alloys [50-55]. The most successful compositions with respect to the production of magnetostrain are Ni-Co-Mn-In alloys, which have been found to demonstrate a magnetostress level of 140 MPa/T with 1.2% axial strain under compression [56]. The magnetostrain and magnetostress levels are both significantly higher than those from the existing magnetostrictive materials and Ni-Mn-Ga alloys. Figure 5 shows the thermomagnetisation measurements of the Ni 45 Co 5 Mn 36.6 In 13.4 alloy in several magnetic fields. In this alloy, Co was added into Ni-Mn-In alloy to increase the Curie temperature. The magnetisation of the austenite is increased and that of the martensite is decreased after Co addition, resulting in an enlarged ΔM of about 100 emu/g across the martensitic transformation. The martensitic transformation temperatures decreased with increasing magnetic field. The increase of magnetic field from 0.5 to 70 koe resulted in a decrease in the transformation temperature of about 30 K. Figure 5 Thermomagnetisation curves of the Ni 45 Co 5 Mn 36.6 In 13.4 alloy measured in several magnetic fields by the sample extraction method. CHAPTER 1 11

18 Figure 6 shows that magnetisation curves of the Ni 45 Co 5 Mn 36.6 In 13.4 alloy between 200 K and 320 K. The alloy presents non-magnetic and ferromagnetic behaviours at 200 and 320 K, respectively. The field induced reverse martensitic transformation from a nonmagnetic phase to a ferromagnetic phase is achieved at 270 and 290 K with a large hysteresis. The enlarged ΔM between the phases greatly increases the magnetic driving force for inducing a magnetic-field-induced reverse phase transformation. Figure 7 shows large magnetostrain of 2.9% in Ni 45 Co 5 Mn 36.7 In 13.3 alloy [51]. The alloy is of martensite state at the testing temperature of 298 K. A compressive pre-strain of about 3% was applied to the alloy, with the magnetic field applied in parallel to the compressive axis of the specimen and the length change parallel to the compressive axis was measured. The shape recovery is due to magnetic-field-induced reverse transformation, which is called the metamagnetic shape memory effect by the authors. Figure 6 Magnetisation versus magnetic field curves for the Ni 45 Co 5 Mn 36.6 In 13.4 alloy between 200 K and 320 K. CHAPTER 1 12

19 Figure 7 Recovery strain at 298K induced by a magnetic field for Ni 45 Co 5 Mn 36.7 In This magnetostrain is a true magnetic shape memory effect, as it involves the reverse martensitic transformation. The martensitic transformation temperatures are around room temperature and T C is 387 K. The parent phase shows L2 1 Heusler ordered structure where a= nm and martensite phases have the modulated structure of monoclinic where a= nm, b= nm, c=2.989 nm and β=93.24, respectively, confirmed by XRD. Wang et al have investigated the magnetic-field-induced martensitic transformation behaviour in Ni 45 Co 5 Mn 36.6 In 13.4 polycrystalline alloy, with or without an imposed stress, at various temperatures using a high-energy synchrotron X-ray diffraction. The reversible magnetic-field-induced martensitic transformation was observed with the application of 5 T under stress, suggesting the potential of the application in the real world [57]. The further investigation on the mechanical properties of Ni 45 Co 5 Mn 36.6 In 13.4 single crystal was systematically done by Karaca et al in The effects of temperature and bias stress on the pseudoelastic response and the shape memory effect were explored. A transformation strain of 5.4% was obtained by thermal cycling under 125 MPa. Temperature hysteresis changes from 50 to 75 K depending on the applied stress level. Pseudoelastic response was CHAPTER 1 13

20 obtained with a large stress hysteresis of 110 MPa and a Clausius-Clapeyron slope of 2.1 MPa/K [58]. Ni 43 Co 7 Mn 39 Sn 11 is another successful alloy which yields large magnetostrain by magnetic-field-induced martensitic transformation discovered by Kainuma et al later in 2006 [50]. The idea of substitution of Co for Ni is similar to that in Ni-Co-Mn-In alloys. The martensite and reverse transformations were detected in the temperature range from 300 to 350 K and T C is about 430 K. The crystal structures of parent phase and martensite phase are as same as in Ni 45 Co 5 Mn 36.7 In 13.3 alloy with slightly different lattice parameters. The recovery strain of 1 % which is 77 % of the pre-stain of 1.3 % was confirmed in a magnetic field strength of 8 T in polycrystalline samples. Moreover, a length change of 0.3 % after releasing the magnetic field was detected, known as the two-way shape memory alloy effect [50]. Similarly, the substituent of Co atoms in Ni-Co-Mn-Sb alloys help align the Mn moments in a ferromagnetic ordering, giving rise to a significantly enhanced magnetisation in the austenite and a large ΔM across the transformation [52]. Since the effect of Co doping on increasing the magnetisation difference across the phases in Ni-Mn-Z(Z=In,Sn,Sb) alloys has been well realised, similar effect is then also expected in early found Ni-Mn-Ga and Ni-Mn-Al alloys. It is found that Co substitution for Ni in Ni 50 Mn 30 Ga 20 alloy significantly lowered the martensitic transformation temperature, and elevated the Curie transition temperature. The magnetisation for the ferromagnetic austenite has been largely increased, while that of the martensite has been lowered to some extent. This results in the increase of the ΔM across the phases, leading to successful magnetic-field-induced phase transformation in Ni 33 Co 13 Mn 32 Ga 18 alloys [54] and in Ni 40 Mn 33 Co 10 Al 17 alloy [53]. Co substitution for Ni effectively increases the ΔM across the martensitic transformation, thus enhancing the magnetic driving force for magnetoactuation. The positive effects of Co can be summarised to a few aspects [52]: (i) it decreases the martensitic transformation temperatures and increases the Curie transition temperature of the austenite, thus guaranteeing the concurrent martensitic and magnetic transformation in a large temperature window, (ii) Co atoms at Ni site contribute larger magnetic moment CHAPTER 1 14

21 (~1.0 µ B ) compared to that of Ni (~0.3 µ B ) in the austenite, (iii) it strengthens the ferromagnetic ordering in the austenite by turning the magnetic moments of Mn atoms into a ferromagnetic ordering instead of the previous antiferromagnetic one [55]. 1.4 Energy evaluation of magnetostrain associated with martensitic transformation in Ni Mn based FSMAs To evaluate the current FSMAs with regard to their potential for magnetic actuation, it is essential to consider the energy conversion from magnetic energy to mechanical work associated with the magnetic-field-induced martensitic transformation. For a complete magnetomechanical actuation, the magnetic energy must overcome the mechanical resistance of the matrix to deformation, and the remainder of the magnetic energy transforms to mechanical work output. To clarify the ability of magnetic actuation of FSMAs, each energy term involved in a magnetic-field-induced shape change via martensitic transformation is analysed. These energy terms can be grouped into three components: (i) the magnetic driving force, (ii) the frictional resistance and (iii) the work output. It is also a useful tool to draw a common criterion for the present FSMAs for practical applications Thermodynamics for magnetomartensitic phase transformations Owing to its lattice distortion, a martensitic transformation is a mechanical event as well as a thermodynamic event. The Gibbs free energy change of such an event may include not only the more familiar internal energy, volume-pressure energy and temperature-entropy energy, but also all other reversible energies involved, such as the force-displacement elastic potential energy (F L), magnetic energy (B M), optical energy, etc. The relations among the multiple energy terms can be expressed as: G U P V T S F L B M (1) CHAPTER 1 15

22 For a normal thermoelastic martensitic transformation, we may consider only the thermal and the mechanical energies. It gives G U P V T S F L (2) Based on F A, L L, 1 AL V, Equation (2) can be written as G H T S (3) When a martensitic phase transformation occurs, the system is in equilibrium state, thus G 0 and hence d S (4) dt This is the famous Clausius-Clapeyron Equation. Similarly, for an isothermally magnetic-field-induced martensitic transformation, we may only consider thermal and magnetic energies. So the equation (1) is now G U P V T S B M (5) Once the magnetic energy term BΔM makes a large contribution to the Gibbs free energy change, the transformation occurs, and then G 0. It gives db dt S (6) M The transformation temperature change (dt) induced by the magnetic field change M (db) is determined by. To utilise magnetic field to induce martensitic transformation, S a combination of a large ΔM and small ΔS is required for the martensitic transformation. CHAPTER 1 16

23 1.4.2 Energy conversion of magnetostrain via martensitic transformation Magnetic energy input The magnetic energy input is ZE shown in Figure 4 (b), which is the area between the magnetisation curves of the austenite and martensite. It can be expressed as ZE B M, corresponding to the last energy term in equation 5. Obviously, with increasing the magnitude of the applied field B or magnetisation difference ΔM, the ZE increases. Resistances for magnetomartensitic actuation For actuation via magnetic-field-induced martensitic transformation, the ZE must overcome two energy barriers: (1) Gibbs free energy deficit for the phase transformation at the testing temperature and (2) mechanical resistance of the matrix to shape change. (1) Gibbs free energy deficit for magnetomartensitic transformation Figure 8 shows thermal- and magnetic-induced martensitic phase transformations, in which the elastic energy of the phase transformation is neglected. T o is the equilibrium transformation temperature. Because of the irreversible energy of the structural transformation, the transformation hysteresis always exists between the forward and reverse phase transformation. Therefore, the transformation temperatures can be regarded as T M (the forward transformation temperature) at below T o and T A (the reverse transformation temperature) at above T o. To induce a reverse transformation at any given temperature T below T A, the thermodynamic energy deficit may be estimated to be E ( T T) S (7) th A CHAPTER 1 17

24 Figure 8 Schematic diagrams of thermal- and magnetic-field-induced martensitic transformations without elastic energy. It is seen that the irreversible energy consumption of the transformation can be given by E ( T T ) S. This energy consumption is due to the friction stemming from ir A o the phase interface movement for the crystallographic transition. However, this irreversible energy barrier can be avoided by deliberately choosing the testing temperature close to T A, and then applying the magnetic field. Accordingly, a small ΔS and testing temperature close to T A are expected for easy magnetic actuations. In a real situation of structural transition, the elastic energy always accompanies, i.e. the elasticity of the transformation. This leads to the temperature-, stress- and magnetic field-span between the starting and finishing of the transformation. Therefore, the transformation temperatures are commonly measured as forward and reverse transformation starting and finishing temperatures (M s, M f, A s and A f ) shown in Figure 9. Within the transformation span, there is a frictional energy, as part of E ir. For a complete transformation being induced at any given temperature T, the total thermal deficit can be rewritten as CHAPTER 1 18

25 E th A f SdT (8) T Given the testing temperature is chosen at A s, the frictional energy to be overcome is shown as the blue shadow area in Figure 9, which corresponds to the minimum energy requirement for the completion of a magnetic-field-induced transformation: E min Af SdT (9) As Figure 9 Schematic diagrams of thermo- and magnetic-field-induced martensitic transformation with transformation elastic energy. (2) Mechanical resistance for magnetomartensitic transformation from pre straining For obtaining a magnetostrain via martensitic transformation, a pre-strain to the alloy at the martensite state is required before applying the magnetic field. The process of CHAPTER 1 19

26 generating a pre-strain is to convert the self-accommodated martensite variants to become reoriented variants, and thus the recovery from the deformed martensite back to the austenite requires extra energy due to the pure mechanical resistance from the crystallographic transition. This energy is denoted as E mech. The magnitude of E mech roughly equals the mechanical work for obtaining the oriented martensite from a selfaccommodated state near A s temperature, which is illustrated in Figure 10. The critical stress for inducing the martensite variant rearrangement is denoted σ o, and the maximum strain is ε max, thus E mech σ o ε max (10) Figure 10 Schematic illustration of stress-strain curve for martensite reorientation at A s temperature. In fact, E mech equals to the extra thermal energy requirement for the transformation from the oriented martensite back to austenite by heating, which is known as the martensite stabilisation behaviour in NiTi alloys [67-69]. Figure 11 shows the illustration for these extra thermal or magnetic energies for thermal- or magnetic-induced martensitic reverse transformation. In the S-T relation, it is seen that the transformation temperatures (A s and A f ) shift to higher temperature range for inducing the reverse transformation from the orientated martensite to austenite relative to those from the self-accommodated martensite to austenite. Similarly, in M-H relation, the magnetic field increase to higher magnitude to meet the requirement from the transformation between a pre-strained CHAPTER 1 20

27 martensite to austenite magnetically. The extra energy corresponding to E mech is caused by the martensite stabilisation given as the red shadow area in Figure 11. In this case, the mechanical resistance due to the martensite stabilisation can be also converted to the form of thermodynamic energy deficit (shadow area in Figure 11), which is A' f Emech SdT ( A' f Af ) S (11) Af Figure 11 Extra thermal and magnetic energy requirement for inducing reverse martensitic transformation caused by pre-strained martensite Criteria of evaluation for FSMAs Criteria I: completion of magnetic field induced martensitic transformation The resistance for a magnetic-field-induced reverse transformation comes only from the thermodynamic deficit for the reverse transformation. Therefore, the magnetic energy input for driving a martensitic transformation at temperature T is given by: CHAPTER 1 21

28 E mag E th As Emag B M, and A f T E th A f SdT, so the condition is T B M SdT (12) Apparently, the minimum magnetic energy requirement for completion of magnetic-fieldinduced martensitic transformation is Af B M SdT. Criteria II: completion of magnetostrain via martensitic transformation As In addition to the thermal deficit, to transform the deformed martensite back to austenite magnetically requires extra energy input to overcome the mechanical resistance brought by martensitic stabilisation. The magnetic driving force now needs to meet E E E As mag th mech Emag B M, E th A f SdT and E T mech A' f Af SdT, where A ' f is the martensitic reverse transformation temperature for an oriented martensite transforming to austenite. Af A' f A' f (13) B M SdT SdT SdT T Af T The minimum magnetic energy requirement for shape recovery via magnetic-field-induced martensitic transformation is A' f B M SdT. As Criteria III: completion of two way magnetostrain via martensitic transformation To obtain a two-way shape memory effect by a magnetic field, the magnetising temperature T must be at lower than M f temperature to have the austenite transforming back to the martensite after releasing the magnetic field, as seen in Figure 11. The minimum magnetic energy input can be obtained when T=M f, that equals CHAPTER 1 22

29 A' f B M SdT (14) M f Based on the analysis above, the energy barriers for a magnetic-field-induced martensitic transformation or any event related can be ascribed to a thermodynamic energy deficit. The magnetic driving force must be larger than this thermodynamic energy deficit to accomplish the energy conversion from magnetic energy to mechanical work output, i.e. magnetic actuation. 1.5 Challenges for magnetostrain via martensitic transformation in Ni Mn based FSMAs Although the magnetostrain has been successfully achieved in Ni 45 Co 5 Mn 36.7 In 13.3 single crystal and Ni 43 Co 7 Mn 39 Sn 11 polycrystalline alloys, no work output was produced. For instance, the success for inducing magnetostrain of 2.9% in Ni 45 Co 5 Mn 36.7 In 13.3 single crystal is still significantly smaller than the theoretical transformation strain (5-6% based on the compression direction shown in ref [51]). This indicates that the magnetic field (~3T) is not able to accomplish the complete martensitic transformation with full strain recovery. Based on the analysis on energy conversion in Section 1.4, the best alloy (Ni 45 Co 5 Mn 36.7 In 13.3 single crystal) discovered so far cannot fully satisfy energetic criteria II. This is because the magnetic driving force (or energy input) is completely consumed during the reverse transformation process, largely by the mechanical resistance, thus resulting in nil energy to output. Besides, the brittleness of Ni-Mn-based FSMAs still impedes their engineering applications. The main challenges for developing FSMAs as engineering magnetoactuators include three aspects as followings To increase the magnetic driving force Failures to have magnetostrain in Ni-Mn-based FSMAs can be always ascribed to the limited magnetic driving force for such an actuating event. In fact, since Ni-Mn- Z(Z=In,Sn,Sb) alloys have distinct magnetic states between the austenite and martensite, CHAPTER 1 23

30 the ZE can be infinitely increased by applying larger magnetic field B. However, the magnetostrain needs to be induced using a reasonable magnitude of magnetic field for a practical consideration. Therefore, the challenge of increasing the magnetic driving force becomes to increase the ΔM between the austenite and martensite for FSMAs, which is illustrated in Figure 12. With this consideration, solutions should be sought to increase the magnetisation of the austenite or to decrease that of the martensite, or both. s M A s M M Figure 12. Illustration of enhancement of Zeeman energy (ZE) in Ni-Mn-based alloys responsible for magnetic-field-induced phase transformation To decrease the mechanical resistance On the other side of the being a successful FSMA candidate, small resistance during a magnetic-field-induced shape change is also essential, apart from the required large magnetic driving force. Based on the discussion in Section 1.4, it is known that the resistances include thermodynamic barrier (E th ) and mechanical resistance (E mech ). The thermodynamic barrier can be realistically decreased by selecting the testing temperature close to A s, and the minimum energy requirement is E min Af As SdT. However, the pure mechanical resistance originating from transformation between the prestrained martensite and the austenite is inevitable. It is known that this part of energy corresponds to the extra CHAPTER 1 24

31 thermodynamic energy deficit as the martensite stabilisation or roughly equals to mechanical energy required for turning the self-accommodated martensite to reoriented martensite. This portion of energy must be overcome by the magnetic energy input to meet criteria II. It is known that the success of having large magnetostrain in Ni-Mn-Ga alloys is simply due to the high mobility between the twin boundaries of the martensite variants. Unlike Ni-Mn-Ga alloys, the twining movement of the martensite variant has been proven very poor in Ni-Co-Mn-In alloys, with a compressive stress of ~100 MPa to initiate detwining [51, 56, 58]. Therefore, the ZE is mainly depleted by the mechanical resistance due to the poor mobility of martensite twin boundaries in Ni-Mn-Co-In alloys. Owing to this consideration, to lower the mechanical resistance is as equally important as to increase the magnetic driving force during, as it saves the magnetic driving force in another sense, thus possibly to yield work output Brittleness of FSMAs From an engineering application point of view, reasonable mechanical properties are needed to draw attention of FSMAs, such as the ductility, the cyclic stability, the ambient stability and frequency etc. Among all of these requirements, good ductility is most crucial for the real engineering applications. It is known that Ni-Mn-Z (Z=Ga, Al, In, Sn and Sb) alloys are intermetallic compounds, which are intrinsically brittle. For a polycrystalline, the ductility is even worse, due to the volume change before and after the martensitic transformation. Consequently, this may easily induce cracking during the actuation process in polycrystalline alloys. Therefore, one of the most the realistic challenges is to increase the ductility of Ni-Mn-based FSMAs. 1.6 Solutions to the challenges of FSMAs In this thesis, solutions for increasing the magnetic driving force and ductility of Ni- Mn-based FSMAs are explored. On one hand, new composition design in Mn-Ni-In(Sn)-Co alloys has been carried out with the aim to increase the magnetic driving force for actuation. On the other hand, introduction of the phase by substitution of Fe for Mn in Ni- CHAPTER 1 25

32 Mn-In(Sn) alloys effectively improves the ductility of FSMAs, however, the research focuses on the metallurgical origin of the changes in transformation behaviour and magnetic properties brought by the formation of the second phase in these alloys Increasing the magnetic driving force new compositions design of Ni Mn based FSMAs Step I-maximise the use of Mn The magnetic moments of the constituents in Ni-Mn-based FSMAs can be estimated from those in the stoichiometric compositions. The magnetic moments are µ B /Mn, µ B /Ni and ~0 µ B /Z, and the net magnetic moments are µ B /f.u. in Ni 2 MnZ(Z=Ga,In,Sn,Sb) alloys [70]. It is seen that the net magnetic moment mainly comes from the contribution of Mn atoms in the unit cell. For this reason, it is reasonable to consider that the magnetisation of the austenite may be maximised by increasing the Mn content up to 50 at% with the assumption of ferromagnetic alignment between Mn atoms. The first step of alloy design is to employ as much Mn content as possible in the new composition. Therefore, Mn 50 Ni 25 Z 25 (Z=In,Sn) was chosen as the base alloy compositions. Step II-stacking order adjustment The type of magnetic interaction between the Mn atoms is very sensitive to the distance between them. Early studies found that the type of magnetic interaction between Mn atoms changes from antiferromagnetic to ferromagnetic when the Mn-Mn distance is increased to above a critical value of approximately 0.30 nm and exhibits a maximum around 0.37 nm [71-73]. In case of Mn 2 NiIn and Mn 2 NiSn alloys, which have an Hg 2 CuTi superlattice structure, Mn atoms occupy A and B sites, which form the nearest neighbour. The distance between Mn(A) and Mn(B) is 3/4a, which is ~0.26 nm, in case of a=0.6 nm. The short distance between Mn(A)-Mn(B) leads to the antiferromagnetic interaction. One solution is to substitute Z element by Ni in the nominal composition. In this case, the composition becomes Mn 50 Ni 25+x Z 25-x. It is seen that some portion of Mn atoms at A site CHAPTER 1 26

33 have been replaced by Ni atoms, and these new Mn atoms share D site with In atoms. This hypothesis is based on the rule of preferential site occupation in Mn 2 YZ (Y: 3d elements; Z: III-V A group elements) alloys reported by Liu et al [63]. They observed that Y elements on the right hand side of Mn in the Periodic Table of Elements prefer to occupy (A,C) sites, whereas Y elements to the left of Mn have strong preference for B site occupancy. In Mn 2 YZ (Y = V, Cr, Mn, Fe, Co and Ni; Z =Al, Ga, In, Si, Ge, Sn and Sb) Heusler alloys, this rule of atomic occupancy has been shown to be well obeyed [61, 63, 66]. According to this principle, Ni substitution for In will have the priority to take A site in preference to Mn. The distance between the new Mn atoms at D site and Mn atoms at B site is ~0.3 nm, which may favor the ferromagnetic exchange interaction between the Mn atoms. Therefore, the magnitude of antiferromagnetic alignment between Mn(A) and Mn(B) is reduced, and the new Mn atoms at D site form ferromagnetic interaction with the Mn atoms at B site. The second step of alloy design is to substitute Ni for Z element to separate Mn(A)- Mn(B) to become Mn(B)-Mn(D), thus the new composition becomes Mn 50 Ni 25+x Z 25- x(z=in,sn). Step III-Co doping One main problem with continuous substitution of Ni for Z is that the martensitic transformation temperature increases rapidly and exceeds the Curie transition temperature of the austenite at the Ni content of 40 at%, leading to the transformation being from paramagnetic austenite to paramagnetic martensite. The solution is to utilise Co to substitute for Ni in the nominal composition, which increases the Curie transition temperature of the austenite and decreases of the martensitic transformation temperatures in the meantime time, thus giving rise to more temperature room for a concurrent martensitic and magnetic transformation. The third step of alloy design is to substitute Co for Ni element to separate the Curie transition and martensitic transformation temperature, hence the final composition becomes Mn 50 Ni 25+x-y Z 25-x Co y (Z=In,Sn). CHAPTER 1 27

34 1.6.2 Ductility improvement and metallurgical origins of changes in martensitic and magnetic properties caused by Fe addition of Ni Mn based alloys The only solution to increase the ductility is to introduce a ductile second phase ( phase) into the matrix of Ni-Mn-based alloys. The phase was first found in Co-Ni-Al [23, 38, 74], Co-Ni-Ga [75] and Ni-Fe-Ga [76] alloy. Fe and Co as dopants were also found effectively to form phase in Ni-Mn-based alloys [77-82]. Whereas the purpose is to improve ductility, addition of a fourth element to the ternary Ni-Mn-Z alloys inevitably alters the matrix composition, hence changing the structure, thermal and magnetic properties. Whereas much attention has been given to the influences of fourth element addition on ductility improvement and transformation properties of these alloys in the literature, given the level of complexity associated with the quaternary systems, much less is understood of the metallurgical origins of these influences. In this thesis, this fundamental issue was examined by investigating the effects of Fe substitution for Mn in Ni-Mn-In and Ni-Mn-Sn alloys. Fe bears much resemblance to Mn in these alloy systems, which provides an opportunity to examine the metallurgical influence of Fe addition to the properties of the alloys, in addition to being a selected element for ductility improvement for some common ferromagnetic shape memory alloys. 1.7 Thesis organisation This thesis is arranged as a series of 5 papers, including 4 published and 1 submitted papers. Below is an overview of the structure of the thesis Chapter 1 (Introduction) Chapter 1 has provided a concise literature review on ferromagnetic shape memory alloys, including the development of FSMAs, survey of magnetostrain in various compositions of FSMAs, and current knowledge in the mechanisms of magnetostrain. It CHAPTER 1 28

35 also includes a detailed analysis on the energy conversion of magnetic actuation associated with the martensitic transformation. Three energetic criteria are established for evaluating the feasibility of magnetostrain of FSMAs. Followed by the energy analysis, the problems are identified of FSMAs, leading to the objectives of this thesis. The objectives of the thesis can be summarised as following: 1. To increase the magnetic driving force by optimising the composition of Ni-Mnbased alloys. 2. To investigate the metallurgical origins of Fe addition on martensitic and magnetic properties of Ni-Mn-based alloys Chapter 2 (Paper 1, Paper2 and Paper3) The martensitic transformation and magnetic behaviour of the newly designed compositions of Mn-rich Ni-Mn-based alloys are illustrated in Paper 1, Paper 2 and Paper 3 in details with the concern of increasing ΔM across the transformation. Paper 1: Metamagnetic phase transformation in Mn 50 Ni 37 In 10 Co 3 polycrystalline alloy, Zhigang Wu, Zhuhong Liu, Hong Yang, Yinong Liu, Guangheng Wu, Applied Physics Letters, 2011, 98, pp (1-3). This work reports on a new composition design of Mn 50 Ni 37 In 10 Co 3, in which a large magnetisation difference of 89 emu/g was obtained. The complete magnetic-fieldinduced martensitic transformation was achieved. It is well demonstrated that the magnetic driving force in Mn-rich Ni-Mn-based alloys was successfully increased. Paper 2: Effect of Co addition on martensitic phase transformation and magnetic properties of Mn 50 Ni 40-x In 10 Co x polycrystalline alloys, Zhigang Wu, Zhuhong Liu, Hong Yang, Yinong Liu, Guangheng Wu, Intermetallics, 2011, 19, pp This work reports a complete alloy series design of Mn 50 Ni 40-x In 10 Co x alloys. The effects of Co addition on the martensitic and magnetic properties were investigated. The origin of the increase of ΔM across the transformation was well interpreted and the CHAPTER 1 29

36 magnetic moment interactions between the constituents were demonstrated in the proposed model. Paper 3: Martensitic phase transformation and magnetic properties of Mn 50 Ni 42- xsn 8 Co x polycrystalline alloys, Zhigang Wu, Zhuhong Liu, Hong Yang, Yinong Liu, Guangheng Wu, Journal of Physics D: Applied Physics, 2011, 44, (1-8). A complete alloy series design of Mn 50 Ni 42-x Sn 8 Co x alloy was carried out aiming to increase the ΔM for magnetic actuation in this work. Co substitution for Ni was found effective on increasing the magnetisation of the austenite, while that of the martensite remained unchanged at a very low level, leading to the continuously gained ΔM across the martensite transformation Chapter 3 (Paper4 and Paper5) Changes on the martensitic transformation behaviour, magnetic properties, mechanical properties and shape memory effect caused by the formation of phase with the original purpose of increasing the ductility are clarified in Ni-Mn-based FSMAs. The studies are carried out in Ni-Mn-In and Ni-Mn-Sn alloys with Fe addition illustrated in Paper 4 and Paper 5 respectively. Paper 4: Effect of Fe addition on the martensitic transformation behaviour, magnetic properties and mechanical performance of Ni 50 Mn 38-x In 12 Fe x polycrystalline alloys, Zhigang Wu, Zhuhong Liu, Hong Yang, Yinong Liu, submitted to Journal of Alloys and Compounds, September This work investigates on the martensitic transformation, magnetic properties and mechanical behaviour of Ni 50 Mn 38-x In 12 Fe x alloys. Fe substitution for Mn at above 3 at% was found to create a phase, which greatly alters the composition of the matrix phase. The martensitic transformation and magnetic transition temperatures were affected by the change of composition of the matrix phase. Mechanical behaviour and shape memory effect were also investigated. CHAPTER 1 30

37 Paper 5: Metallurgical origin of the effect of Fe doping on the martensitic and magnetic transformation behaviours of Ni 50 Mn 40-x Sn 10 Fe x magnetic shape memory alloys, Zhigang Wu, Zhuhong Liu, Hong Yang, Yinong Liu, Guangheng Wu, Robert Woodward, Intermetallics, 2011, 19, pp The metallurgical origin of the effect of Fe doping on the martensitic transformation and magnetic properties was investigated in Ni 50 Mn 40-x Sn 10 Fe x alloys in this work. The findings clarify the origin of the effect of Fe addition and provide useful reference on alloy design in Ni-Mn-Sn alloy system Chapter 4 (Closing remarks) The final chapter presents a short summary of the main findings and the significance of the work in this thesis. The immediate future works as continuation of the current work are proposed. CHAPTER 1 31

38 Reference [1] Clark AE, Abbundi R, Savage HT, McMasters OD. Physica B+C 1977;86-88:73. [2] Levitin RZ, Markosyan AS. Journal of Magnetism and Magnetic Materials 1990;84:247. [3] Ibarra MR, del Moral A, Abell JS. Journal of Magnetism and Magnetic Materials 1984;46:157. [4] Clark AE, Wohlfarth EP. Magnetostrictive rare earth-fe2 compounds. Handbook of Ferromagnetic Materials, vol. Volume 1. Elsevier, p.531. [5] Ullakko K, Huang JK, Kantner C, Handley RCO, Kokorin VV. Applied Physics Letters 1996;69:1966. [6] Murray SJ, Marioni M, Allen SM, O'Handley RC, Lograsso TA. Applied Physics Letters 2000;77:886. [7] Sozinov A, Likhachev AA, Lanska N, Ullakko K. Applied Physics Letters 2002;80:1746. [8] Karaca HE, Karaman I, Basaran B, Chumlyakov YI, Maier HJ. Acta Materialia 2006;54:233. [9] Jiang C, Wang J, Xu H. Applied Physics Letters 2005;86: [10] Ullakko K, Huang JK, Kantner C, Handley RC, Kokorin VV. Applied Physics Letters 1996;69:1966. [11] Jiang C, Muhammad Y, Deng L, Wu W, Xu H. Acta Materialia 2004;52:2779. [12] Wu GH, Yu CH, Meng LQ, Chen JL, Yang FM, Qi SR, Zhan WS, Wang Z, Zheng YF, Zhao LC. Applied Physics Letters 1999;75:2990. [13] Likhachev AA, Ullakko K. Physics Letters A 2000;275:142. [14] Kakeshita T, Fukuda T. International Journal of Applied Electromagnetics and Mechanics 2006;23:45. [15] Kakeshita T, Fukuda T, Takeuchi T. Materials Science and Engineering: A 2006; :12. [16] Kakeshita T, Kim J-H, Fukuda T. Materials Science and Engineering: A 2008; :40. [17] Liu G, Chen J, Cui Y, Liu Z, Zhang M, Wu G, Bruck E, de Boer FR, Meng F, Li Y, Qu J. Solid State Communications 2004;130:687. CHAPTER 1 32

39 [18] Straka L, Lanska N, Ullakko K, Sozinov A. Applied Physics Letters 2010;96:3. [19] Straka L, Hanninen H, Soroka A, Sozinov A. Journal of Physics: Conference Series 2011;303: [20] Gans E, Henry C, Carman GP. Reduction in required magnetic field to induce twinboundary motion in ferromagnetic shape memory alloys. vol. 95: AIP, p [21] Morito S, Otsuka K. Materials Science and Engineering A 1996;208:47. [22] Fujita A, Fukamichi K, Gejima F, Kainuma R, Ishida K. Applied Physics Letters 2000;77:3054. [23] Ishida K, Kainuma R, Ueno N, Nishizawa T. Metallurgical and Materials Transaction A 1991;22:441. [24] Oikawa K, Ota T, Ohmori T, Tanaka Y, Morito H, Fujita A, Kainuma R, Fukamichi K, Ishida K. Applied Physics Letters 2002;81:5201. [25] Sutou Y, Kamiya N, Omori T, Kainuma R, Ishida K, Oikawa K. Applied Physics Letters 2004;84:1275. [26] Morito H, Fujita A, Fukamichi K, Kainuma R, Ishida K, Oikawa K. Applied Physics Letters 2003;83:4993. [27] Morito H, Oikawa K, Fujita A, Fukamichi K, Kainuma R, Ishida K. Scripta Materialia 2005;53:1237. [28] Morito H, Fujita A, Oikawa K, Ishida K, Fukamichi K, Kainuma R. Applied Physics Letters 2007;90: [29] Zheng HX, Liu J, Xia MX, Li JG. Journal of Alloys and Compounds 2005;387:265. [30] Omori T, Kamiya N, Sutou Y, Oikawa K, Kainuma R, Ishida K. Materials Science and Engineering A 2004;378:403. [31] Santamarta R, Font J, Muntasell J, Masdeu F, Pons J, Cesari E, Dutkiewicz J. Scripta Materialia 2006;54:1105. [32] Masdeu F, Pons J, Santamarta R, Cesari E, Dutkiewicz J. Materials Science and Engineering: A;In Press, Corrected Proof. [33] Liu ZH, Wang HY, Yu SY, Dai XF, Chen JL, Wu GH, Liu Y. Scripta Materialia 2006;54:1299. [34] Karaca HE, Karaman I, Chumlyakov YI, Lagoudas DC, Zhang X. Scripta Materialia 2004;51:261. [35] Murakami Y. Acta Materialia 50 (2002) CHAPTER 1 33

40 [36] Fujita A, Morito H, Kudo T, Fukamichi K, Kainuma R, Ishida K, Oikawa K. Metallurgical and Materials Transaction 2003;44:2180. [37] Morito H, Fujita A, Fukamichi K, Kainuma R, Ishida K, Oikawa K. Applied Physics Letters 2002;81:1657. [38] Liu J, Zheng H, Huang Y, Xia M, Li J. Scripta Materialia 2005;53:29. [39] Wuttig M, Li J, Craciunescu C. Scripta Materialia 2001;44:2393. [40] Sato M, Okazaki T, Furuya Y, Wuttig M. Materials Transactions 2003;44:372. [41] Kakeshita T, Takeuchi T, Fukuda T, Tsujiguchi M, Saburi T, Oshima R, Muto S. Applied Physics Letters 2000;77:1502. [42] Sakamoto T, Fukuda T, Kakeshita T. Materials Science and Engineering: A 2006; :1036. [43] Sugiyama M, Oshima R, Fujita FE. Trans. JIM 1984;25:585. [44] Fukuda T, Sakamoto T, Kakeshita T, Takeuchi T, Kishio K. Materials Transactions 2004;45:188. [45] Wada T, Liang Y, Kato H, Tagawa T, Taya M, Mori T. Materials Science and Engineering A 2003;361:75. [46] Sutou Y, Imano Y, Koeda N, Omori T, Kainuma R, Ishida K, Oikawa K. Applied Physics Letters 2004;85:4358. [47] Krenke T, Acet M, F. Wassermann E, Moya X, Manosa L, Planes A. Physical Review B 2006;73: [48] Ito W, Imano Y, Kainuma R, Sutou Y, Oikawa K, Ishida K. Metallurgical and Materials Transactions A: Physical Metallurgy and Materials Science 2007;38:759. [49] Oikawa K, Ito W, Imano Y, Sutou Y, Kainuma R, Ishida K, Okamoto S, Kitakami O, Kanomata T. Applied Physics Letters 2006;88: [50] Kainuma R, Imano Y, Ito W, Morito H, Sutou Y, Oikawa K, Fujita A, Ishida K, Okamoto S, Kitakami O, Kanomata T. Applied Physics Letters 2006;88: [51] Kainuma R, Imano Y, Ito W, Sutou Y, Morito H, Okamoto S, Kitakami O, Oikawa K, Fujita A, Kanomata T, Ishida K. Nature 2006;439:957. [52] Yu SY, Ma L, Liu GD, Liu ZH, Chen JL, Cao ZX, Wu GH, Zhang B, Zhang XX. Applied Physics Letters 2007;90: [53] Kainuma R, Ito W, Umetsu RY, Oikawa K, Ishida K. Applied Physics Letters 2008;93: CHAPTER 1 34

41 [54] Yu SY, Cao ZX, Ma L, Liu GD, Chen JL, Wu GH, Zhang B, Zhang XX. Applied Physics Letters 2007;91: [55] Umetsu RY, Sheikh A, Ito W, Ouladdiaf B, Ziebeck KRA, Kanomata T, Kainuma R. Applied Physics Letters 2011;98:3. [56] Karaca HE, Karaman I, Basaran B, Ren Y, Chumlyakov YI, Maier HJ. Advanced Functional Materials 2009;19:983. [57] Wang YD, Huang EW, Ren Y, Nie ZH, Wang G, Liu YD, Deng JN, Choo H, Liaw PK, Brown DE, Zuo L. Acta Materialia 2008;56:913. [58] Karaca HE, Karaman I, Brewer A, Basaran B, Chumlyakov YI, Maier HJ. Scripta Materialia 2008;58:815. [59] Ito W, Nagasako M, Umetsu RY, Kainuma R, Kanomata T, Ishida K. Applied Physics Letters 2008;93: [60] Liu GD, Dai XF, Yu SY, Zhu ZY, Chen JL, Wu GH, Zhu H, Xiao JQ. Physical Review B: Condensed Matter 2006;74: [61] Luo H, Liu G, Feng Z, Li Y, Ma L, Wu G, Zhu X, Jiang C, Xu H. Journal of Magnetism and Magnetic Materials 2009;321:4063. [62] Chakrabarti A, Barman SR. Applied Physics Letters 2009;94: [63] Liu GD, Dai XF, Liu HY, Chen JL, Li YX, Xiao G, Wu GH. Physical Review B 2008;77: [64] Krenke T, Acet M, F. Wassermann E, Moya X, Manosa L, Planes A. Physical Review B 2005;72: [65] Liu GD, Chen JL, Liu ZH, Dai XF, Wu GH. Applied Physics Letters 2005;87: [66] Luo HZ, Meng FB, Feng ZQ, Li YX, Zhu W, Wu GH, Zhu XX, Jiang CB, Xu HB. Journal of Applied Physics 2009;105:4. [67] Liu YN. Materials Science and Engineering a-structural Materials Properties Microstructure and Processing 1999;273:668. [68] Liu YN, Favier D. Acta Materialia 2000;48:3489. [69] Liu YN, Tan GS. Intermetallics 2000;8:67. [70] Sasioglu E, Sandratskii LM, Bruno P. Physical Review B 2004;70: [71] Yamada T, Kunitomi N, Nakai Y, Cox DE, Shirane G. Journal of the Physical Society of Japan 1970;28:615. CHAPTER 1 35

42 [72] Khovaylo VV, Kanomata T, Tanaka T, Nakashima M, Amako Y, Kainuma R, Umetsu RY, Morito H, Miki H. Physical Review B 2009;80: [73] Chieda Y, Kanomata T, Fukushima K, Matsubayashi K, Uwatoko Y, Kainuma R, Oikawa K, Ishida K, Obara K, Shishido T. Journal of Alloys and Compounds 2009;486:51. [74] Tanaka Y, Oikawa K, Sutou Y, Omori T, Kainuma R, Ishida K. Materials Science and Engineering: A 2006; :1054. [75] Liu J, Xie H, Huo Y, Zheng H, Li J. Journal of Alloys and Compounds 2006;420:145. [76] Masdeu F. Journal of Magnetism and Magnetic Materials (2005) [77] Ma Y, Yang S, Liu Y, Liu X. Acta Materialia 2009;57:3232. [78] Yang S, Ma Y, Jiang H, Liu X. Intermetallics 2011;19:225. [79] Wang HB, Sui JH, Liu C, Cai W. Materials Science and Engineering: A 2008;480:472. [80] Feng Y, Jiehe S, Zhiyong G, Wei C. International Journal of Modern Physics B 2010;23:1803. [81] Feng Y, Sui JH, Gao ZY, Dong GF, Cai W. Journal of Alloys and Compounds 2009;476:935. [82] Feng Y, Sui JH, Gao ZY, Zhang J, Cai W. Materials Science and Engineering A 2009;507:174. CHAPTER 1 36

43 CHAPTER 2. Increasing magnetic driving force of Ni-Mn-based alloys Paper 1 Metamagnetic phase transformation in Mn 50 Ni 37 In 10 Co 3 polycrystalline alloy Zhigang Wu 1, Zhuhong Liu 2, Hong Yang 1, Yinong Liu 1, Guangheng Wu 3 1 School of Mechanical and Chemical Engineering, The University of Western Australia, Crawley, WA 6009, Australia 2 Department of Physics, University of Science and Technology Beijing, Beijing , China 3 Beijing National Laboratory for Condense Matter Physics, Institute of Physics, Chinese Academy of Science, Beijing , China This paper reports on an alloy design of Mn 50 Ni 37 In 10 Co 3 based on the principle of Mn-Mn ferromagnetic coupling via Co doping. The alloy is shown to exhibit a metamagnetic martensitic transformation and a high saturation magnetization of 118 emu/g in its austenitic state. The transformation generates a large magnetization difference of 89 emu/g, more than 200% of what reported in the literature for similar alloys. A complete magnetic field induced martensitic transformation was achieved at 170 K. Such high magnetization difference provides a strong driving force for magnetic-field-induced transformation, making this material a promising candidate for magnetic actuation applications. CHAPTER 2 37

44 Since the discovery of the magnetic-field-assisted shape memory effect in Mn 2 NiGa single crystal in 2005, 1 much effort has been made to develop better Mn-rich ferromagnetic shape memory alloys (FSMAs). In Mn 2 NiX (Ga,In,Sn,Sb) system, the alloys hold the promise for higher saturation magnetisation owning to its higher Mn content. Mn-rich Mn- Ni-In alloys have been found to exhibit concurrent magnetic and martensitic transformations, i.e., metamagnetic transformations. 2,3 The magnetic driving force for such metamagnetic phase transformations is provided by the Zeeman energy E Zeeman =µ 0 MH, where M is the saturation magnetization difference between the austenite and martensite and H corresponds to the strength of the applied field. This energy is dependent on the M, which is typically ~40 emu/g for Mn 50 Ni 40 In 10. 2,3 Largely due to the low ΔM, a complete reversible metamagnetic transformation is yet to be achieved in Mn-rich FSMAs. At the meantime, Co doping has been reported to have prominent effect on increasing M in Mn 48 Co x Ni 32-x Ga 20 alloys 4, due to its effect on promoting ferromagnetic alignment of the moments of the nearest neighboring Mn atoms. This paper reports on a Mn 50 Ni 37 In 10 Co 3 alloy, which has a significantly increased ΔM for its metamagnetic transformation. A polycrystalline Mn 50 Ni 37 In 10 Co 3 button ingot was prepared using an arc melting furnace in argon atmosphere from high purity (99.99 %) elemental metals. The ingot (~4 g) was heat treated at 1073 K for 24 hours in vacuum followed by quenching in water to ensure composition homogeneity. Phase identification and crystal structures were determined by means of X-ray powder diffraction (XRD) using Cu-Kα radiation, phase transformation behavior was measured by means of differential scanning calorimetry (DSC) with a cooling/heating rate of 10 K/min, and magnetic properties were studied using a superconducting quantum interference device magnetometer (SQUID). Figure 1 shows an XRD spectrum of powder sample of Mn 50 Ni 37 In 10 Co 3 alloy at room temperature. The alloy shows a single phase structure with bcc fundamental lattice reflections of (220), (400), (422) and (440) and superlattice reflections of (111), (200), (311) and (222). The superlattice structure can be determined by comparing the relative intensities of (111) and (200). 5 It is evident that I 111 /I 200 >1, implying that the superlattice is of the Hg 2 CuTi-type, consistent with other Mn 2 NiX (Ga,Sn,Sb) alloys. 5-7 CHAPTER 2 38

45 X-ray intensity Heat flow Temperature (K) ( o ) FIG. 1. X-ray diffraction spectrum of powder sample of Mn 50 Ni 37 In 10 Co 3 alloy; inset: DSC curve of the martensitic transformation behavior of the alloy. In this structure, Mn atoms occupy A (0,0,0) site and B (1/4,1/4,1/4) site, leaving C (1/2,1/2,1/2) site to Ni atoms and D (3/4,3/4,3/4) site to the third element atoms. Such structure can be expressed in a stacking order of MnMnNiX ( F43m space group) along the diagonal [111] direction of the cubic unit cell. The lattice constant is determined to be a= nm. The inset in the figure shows DSC measurement of the transformation. The peak transformation temperatures are determined to be 169 K and 195 K and the latent heat of the transformation is 2.9 J/g. Figure 2 shows the zero-field cooled (ZFC) and field cooled (FC) M(T) curves with a cooling/heating rate of 10 K/min of the alloy in magnetic fields of different strengths. At a low field of 50 Oe, the martensitic and austenitic transformation starting and finishing temperatures are determined to be respectively. M 186 K, M 153 K, A 179 K and A 212 K, s f s f CHAPTER 2 39

46 Magnetization (emu/g) M=89 emu/g 5x10-2 T 7T 5T 2T FC 5x10-3 T ZFH Temperature (K) FIG. 2. Zero-filed cooled (ZFC) and field cooled (FC) M(T) curves of Mn 50 Ni 37 In 10 Co 3 alloy under various fields measured by SQUID. The transformation hysteresis is determined to be A f M 26 K, which is s consistent with the DSC result. It is also seen that at low field strengths the magnetization behavior exhibited complete reversible phase transformation, shown as the closed ZFC and FC pathways. At high field strengths ( 2T) the FC pathway did not overlap with ZFC pathway at 10 K after a cycle of M A M transformation. It is obviously due to the kinetic arrest of the martensitic transformation under the influence of high magnetic field, which has been observed and discussed in the literature for several Ni-Mn-In alloys in the past few years The same phenomenon has also been observed in a similar Mn 49.5 Ni 40.4 In 10.1 ribbon alloy, as reported by Sanchez Llamazares recently. 11 The increased magnetization at the finishing point on the FC curves indicates that the amount of the arrested austenite increased with higher magnetic field applied. It is worth noting that upon heating (ZFC curves) ΔM between the austenite and the martensite increased progressively with increasing the magnetic field. The maximum of 7 T. M A M achieved is 89 emu/g in a field CHAPTER 2 40

47 Recently, Charkrabarti and Barman conducted theoretical calculation on the ferrimagnetism of Mn 50 Ni 25 In 25 alloy, and showed that the net magnetic moment is 0.47 µ B per formula unit (f.u.) for the austenite, 12 corresponding to 9.27 emu/g. The crystal structure of the austenite is considered as Hg 2 CuTi-type superlattice bcc structure, which is the same structure as the present Mn 50 Ni 37 Co 3 In 10 alloy. This calculation is based on the condition that Mn-Mn atoms within the lattice form antiferromagnetic coupling. The austenite of our alloy showed a much higher magnetization of ~118 emu/g at 7 T. The drastically increased magnetization is a strong indication that the Mn-Mn interaction in the lattice of the Mn 50 Ni 37 In 10 Co 3 alloy has changed to ferromagnetic coupling. This is attributed to two reasons: (i) change of magnetic exchange status due to the composition change of the lattice, and (ii) doping effect of Co. The exchange interaction between Mn atoms is known to depend strongly on Mn- Mn distance in the lattice. Early studies found that the type of magnetic interaction between Mn atoms changes from antiferromagnetic to ferromagnetic when the Mn-Mn distance is increased to above a critical value of approximately 0.30 nm. 13 For Mn-Mn-Ni-X stacking order, the distance between A site and B site is AB 3 a/4= nm and the distance between B site and D site is BD a / nm for the current alloy with a= nm. This implies that Mn(A)-Mn(B) form antiferromagnetic interaction and Mn(B) and Mn(D) form ferromagnetic interaction. In Mn 50 Ni 25 In 25 alloy, the calculated spin magnetic moments for the austenite are -3.08, 3.42, 0.13 and 0 µ B /f.u. for Mn (A site), Mn (B site), Ni (C site), and In (D site), respectively. 12 However, in Mn 50 Ni 40 In 10 alloy, the saturation magnetization of the austenite is ~75 emu/g, as reported by Sanchez Llamazares etc. 2,3 This suggests the change of antiferromagnetic coupling between Mn-Mn atoms in the stoichiometric alloy to ferromagnetic coupling in the non-stoichiometric alloy. One scenario is that in MnMnNiX-type lattice, excess Ni above 25 at% has priority to take A site, displacing Mn from A site to D site. In this case the magnitude of antiferromagnetic alignment between Mn(A) and Mn(B) is reduced, and the new Mn at D site ferromagnetically aligns with the Mn at B site. In Mn 50 Ni 40 In 10 alloy, the extra Ni (15 at%) may displace 15 at% Mn from A site to D site, thus forming domains of NiMnNiMn stacking structure. In this case, the matrix of the alloy may be considered to contain two CHAPTER 2 41

48 mixed domains: Mn-Mn-Ni-In domains (40 % in volume) and Ni-Mn-Ni-Mn domains (60 % in volume). In the MnMnNiIn domain, the Mn(A)-Mn(B) interaction is antiferromagnetic, as in the case of stoichiometric Mn 50 Ni 25 In 25. In the NiMnNiMn domain, the Mn(B)-Mn(D) interaction is ferromagnetic. Liu et al reported that the magnetic moment of Mn(A) and Mn(C) in MnNiMnGa is 2.99 µ B. 5 Using this value, and assuming the magnetic moment of Ni is unchanged at 0.13 µ B, the net moment of NiMnNiMn can be estimated to be 6.24 µ B. Combining with the net moment of 0.47 µ B for the MnMnNiIn domain, the total magnetic moment of the alloy can be calculated to be =3.94 µ B, corresponding to 76 emu/g, as summarized in Table I. This is in excellent agreement with the experimental evidences reported in the literature. 2,3 TABLE I. Magnetization calculation of Ni 50 Mn 25 In 25, Mn 50 Ni 40 In 10, and Mn 50 Ni 37 In 10 Co 3 alloys at room temperature. alloy stacking coupl e Mn( A) (µ B ) Mn( B) (µ B ) Mn( D) (µ B ) Ni (µ B ) In (µ B ) f.u. (µ B ) total (emu/g) Mn 50 Ni 25 In 25 Mn-Mn-Ni-In anti Mn 50 Ni 40 In 10 Mn-Mn-Ni-In (40%) anti Mn 50 Ni 37 In 10 C o 3 Ni-Mn-Ni-Mn (60%) ferro Mn-Mn-(Ni,Co)-In (40%) ferro Ni-Mn-Ni-Mn (60%) ferro However, the magnetization of the present alloy is still significantly higher than thus predicted. This is believed to be related to the effect of Co doping in the crystal structure of the austenite. There have been evidences that Co doping enhances ferromagnetic ordering of the austentie. 14,15 In Mn 48 Ni 32-x Co x Ga 20 alloys Co doping converts the coupling between the nearest neighboring Mn-Mn atoms from antiferromagnetic into ferromagnetic. 4 Assuming the same effect in our alloy, neglecting the net moment of the 3 at% Co, and converting the moment for Mn(A) from µ B to 3.08 µ B, the lattice magnetization of the Mn-Mn-(Ni,Co)-In (40%) domain may be estimated to be 6.63 µ B. In this case, the total magnetic moment for the alloy is expected to be =6.39 µ B, which corresponds to 123 emu/g. This is a good agreement CHAPTER 2 42

49 with the value determined experimentally (118 emu/g). It has been suggested that the presence of Co in the matrix alters the states of Mn 3d electrons, which leads to the enhanced ferromagnetic exchange in the austenite. 16 This argument is also expected to apply to the Mn 50 Ni 37 In 10 Co 3 alloy. Figure 3 shows isothermal magnetization loops of the alloy at different temperatures. The sample showed typical soft magnetic behavior in martensitic state at 5 K with a saturation magnetization of 29 emu/g under 7 T, which is consistent with previous findings. 2,3 At 200 K, the austenite magnetized in a similar way to saturation at 113 emu/g. At 170 K, which is 42 K below A f, the sample quickly magnetized at below 0.5 T, corresponding to the magnetization of the martensite, and then magnetized again at above 2 T, corresponding to the reverse M A transformation. The magnetization reached saturation level (~118 emu/g) of the austenite at 6 T, indicating the completion of the metamagnetic transformation. Similar behavior is also observed at 165 K (47 K below A f ), but the transformation occurred at moderately increased field strength. Consequently, the transformation was not complete at the maximum field applied (7 T) and the maximum magnetization reached (~108 emu/g) was slightly below the saturation magnetization of the austenite. CHAPTER 2 43

50 Magnetization (emu/g) K 170 K 165 K 5 K Magnetic field (T) FIG. 3. Isothermal magnetization loops of Mn 50 Ni 37 In 10 Co 3 alloy at different temperatures. In Summary, the Mn 50 Ni 37 In 10 Co 3 alloy shows a martensitic transformation at M 186 K. The crystal structure of the austenite is Hg 2 CuTi-type superlattice bcc s structure with lattice constant of a= nm. The saturation magnetization of the austenite is 118 emu/g and that of the martensite is 29 emu/g at 7 T, resulting in a large ΔM of 89 emu/g across the martensitic transformation. The largely improved magnetization for the austenite is attributed to (i) change of magnetic exchange status due to the composition change of the lattice, and (ii) doping effect of Co. The calculations for the magnetization of the austenite show excellent agreement with the experimental measurements. Co doping of 3 at% has increased the magnetization of the austenite by 42 emu/g. A complete metamagnetic transformation is induced isothermally at 170 K in a magnetic field up to 7 T, indicating this alloy a promising candidate for magnetic actuation applications. The authors wish to acknowledge the financial supports by the Department of Innovation Industry, Science and Research of the Australian Government in ISL Grant CHAPTER 2 44

51 CH070136, by the National Natural Science Foundation of China in Grant No and by the Fundamental Research Funds for the central universities. Reference 1 G. D. Liu, J. L. Chen, Z. H. Liu, X. F. Dai, and G. H. Wu, Appl. Phys. Lett. 87, (2005). 2 J. L. Sanchez Llamazares, T. Sanchez, J. D. Santos, M. J. Perez, M. L. Sanchez, B. Hernando, L. Escoda, J. J. Sunol, and R. Varga, Appl. Phys. Lett. 92, (2008). 3 J. L. Sanchez Llamazares, B. Hernando, V. M. Prida, C. Garcia, J. Gonzalez, R. Varga, and C. A. Ross, J. Appl. Phys. 105, 07A945 (2009). 4 L. Ma, H. W. Zhang, S. Y. Yu, Z. Y. Zhu, J. L. Chen, G. H. Wu, H. Y. Liu, J. P. Qu, and Y. X. Li, Appl. Phys. Lett. 92, (2008). 5 G. D. Liu, X. F. Dai, S. Y. Yu, Z. Y. Zhu, J. L. Chen, G. H. Wu, H. Zhu, and J. Q. Xiao, Phys. Rev. B: Condens. Matter 74, (2006). 6 R. B. Helmholdt and K. H. J. Buschow, Journal of the Less-Common Metals 128, 167 (1987). 7 H. Luo, G. Liu, Z. Feng, Y. Li, L. Ma, G. Wu, X. Zhu, C. Jiang, and H. Xu, J. Magn. Magn. Mater. 321, 4063 (2009). 8 W. Ito, K. Ito, R. Y. Umetsu, R. Kainuma, K. Koyama, K. Watanabe, A. Fujita, K. Oikawa, K. Ishida, and T. Kanomata, Appl. Phys. Lett. 92, (2008). 9 R. Y. Umetsu, W. Ito, K. Ito, K. Koyama, A. Fujita, K. Oikawa, T. Kanomata, R. Kainuma, and K. Ishida, Scripta Mater. 60, 25 (2009). 10 V. K. Sharma, M. K. Chattopadhyay, and S. B. Roy, Phys. Rev. B 76, (2007). 11 J. L. Sanchez Llamazares, B. Hernando, J. J. Sunol, C. Garcia, and C. A. Ross, J. Appl. Phys. 107, 09A956 (2010). 12 A. Chakrabarti and S. R. Barman, Appl. Phys. Lett. 94, (2009). 13 T. Yamada, N. Kunitomi, Y. Nakai, D. E. Cox, and G. Shirane, J. Phys. Soc. Jpn. 28, 615 (1970). 14 S. Y. Yu, Z. X. Cao, L. Ma, G. D. Liu, J. L. Chen, G. H. Wu, B. Zhang, and X. X. Zhang, Appl. Phys. Lett. 91, (2007). CHAPTER 2 45

52 15 S. Y. Yu, L. Ma, G. D. Liu, Z. H. Liu, J. L. Chen, Z. X. Cao, G. H. Wu, B. Zhang, and X. X. Zhang, Appl. Phys. Lett. 90, (2007). 16 B. Gao, F. X. Hu, J. Shen, J. Wang, J. R. Sun, and B. G. Shen, J. Magn. Magn. Mater. 321, 2571 (2009). CHAPTER 2 46

53 Paper 2 Effect of Co addition on martensitic phase transformation and magnetic properties of Mn 50 Ni 40-x In 10 Co x polycrystalline alloys Zhigang Wu a, Zhuhong Liu b, Hong Yang a, Yinong Liu a, *, Guangheng Wu c a School of Mechanical and Chemical Engineering, The University of Western Australia, Crawley, WA 6009, Australia b Department of Physics, University of Science and Technology Beijing, Beijing , China c Beijing National Laboratory for Condense Matter Physics, Institute of Physics, Chinese Academy of Science, Beijing , China Keywords: A: magnetic intermetallics; B: alloy design; B: shape memory effects; B: martensitic transformations; B: magnetic properties. Abstract This study investigated the use of Co to enhance the magnetic driving force for inducing the martensitic transformation of Mn 50 Ni 40-x In 10 Co x alloys. These alloys present a martensitic transformation from a Hg 2 CuTi-type austenite to a body centered tetragonal martensite, with a large lattice distortion of 15.7% elongation along the c direction and 8.2% contraction along a and b directions. The martensitic transformation temperatures, transformation enthalpy and entropy changes decreased with increasing the Co content in these alloys. The maximum magnetization of the austenite increased significantly, whereas that of the martensite changed much less prominently with increasing the Co substitution CHAPTER 2 47

54 for Ni, leading to increase of the magnetic driving force for the transformation. The magnetization increase of the austenite is found to be due to (i) formation of ferromagnetically coupled Mn-Mn due to new atomic configuration in off-stoichiometric composition, (ii) magnetic moment contribution of Co and (iii) widening of the temperature window for magnetization of the austenite. These findings clarify the effect of Co addition on martensitic transformation and magnetic properties in Mn-rich ferromagnetic shape memory alloys, and provide useful understanding for alloy design for magnetoactuation applications. 1. Introduction Ternary Ni-Mn-Z(Z=In,Sn,Sb) alloys have attracted much attention in the past few years as a new type of ferromagnetic shape memory alloys (FSMAs) since their discovery in 2004 [1]. Unlike Ni 2 MnGa alloy [2], which relies on magnetic crystallographic anisotropy of the martensite, these Ni-Mn-Z(Z=In,Sn,Sb) alloys exhibit a martensitic transformation between a ferromagnetic austenite and a paramagnetic martensite. The different magnetic states between the two phases provide a much greater magnetic driving force, thus the possibility for a magnetic-field-induced martensitic transformation. Such transformations are referred to as metamagnetic transformations in recognition of their concurrent metallurgical and magnetic changes. The magnetic driving force for a metamagnetic transformation is provided by the Zeeman Energy E Zeeman =µ 0 M H, where µ 0 is the permeability of a vacuum, M is the saturation magnetization difference between the austenite and martensite and H corresponds to the strength of the applied field. The M between the ferromagnetic austenite and the paramagnetic martensite, as in the case of Ni- Mn-Z(Z=In,Sn,Sb), is much greater than the M between the easy and hard magnetization directions of the same crystal structure, as in the case of Ni-Mn-Ga alloys, thus giving possibility for much more powerful magnetic-field-induced martensitic phase transformation and mechanical actuation. In Ni 2 MnZ(Z=In,Sn,Sb) alloys, the net magnetic moment mainly comes from the contribution of Mn [3]. By substituting Mn for X, ΔM has been found to increase in Ni 2 Mn 1+x In 1-x alloys but to decrease in Ni 2 Mn 1+x Sn 1-x alloys [4, 5]. At the meantime, CHAPTER 2 48

55 increasing Mn content also causes rapid increase of the martensitic transformation temperatures, to above the Curie temperature of the austenite [6, 7]. This results in the transformation being between a paramagnetic austenite to a paramagnetic martensite, thus losing the advantage of large magnetic driving force for transformation and jeopardizing the possibility for magnetic actuation. This limits the range of Mn content feasible in Ni 2 MnX(In,Sn,Sb) alloys. A new approach is to develop Mn 2 NiZ(Z=Ga,In,Sn,Sb) alloys. These alloys have the obvious advantage by having more Mn in the matrix, which has the highest magnetization contribution among the three constituents [3]. A magnetic-field-assisted shape change of ~4 % has been achieved in single crystalline Mn 2 NiGa [8]. However, the magnetic driving force in this alloy is small due to the limited magnetization difference (~9 emu/g) between the austenite and martensite [8, 9]. A progress has been made recently with off-stoichiometric Mn 50 Ni 40 In 10 [10, 11] and Mn 48 Co x Ni 32-x Ga 20 [12] alloys, which showed a relative large ΔM of about 40 emu/g and 30 emu/g respectively, making these alloys valid candidates for ferromagnetic shape memory actuation. To further improve ΔM, we have recently reported our study on a Mn 50 Ni 37 In 10 Co 3 polycrystalline alloy. This alloy exhibited a large ΔM of ~89 emu/g and a complete reversible metamagnetic transformation [13]. These limited early findings indicate a possible solution to challenge of enhancing magnetic driving force for inducing metamagnetic transformation, a prerequisite for magnetically actuated shape memory alloys. The saturation magnetization of the alloys depends greatly on the magnetic moment distribution from Mn atoms. However, the study on the magnetic moment distribution of Mn in the off-stoichiometric alloys is much lacking. Very recently, Lazpita et al. proposed a model of magnetic interaction between Mn atoms in the off-stoichiometric Ni-Mn-Ga alloys [14]. In their model, the excess of Mn atoms at Ga sites couple antiferromagnetically with the Mn at Mn sites when Ni atoms are at their proper sties, while the Mn at Ga sites couple ferromagnetically with the Mn at Mn sites when Mn excess occupies Ni sites. However, the systematic analysis on atomic configuration in Mn-rich off-stoichiometric Mn-Ni-Z(Z=In,Sn,Sb) alloys is still missing. Moreover, the magnetic interactions between the constituents may rise up to another level of complexity after Co doping in these ternary CHAPTER 2 49

56 alloys, since Co doping in Ni-Mn-Z(Ga,Al,In,Sn,Sb) has been found to be effective in inducing its metamagnetic transformation [15-19]. These findings all indicate that Co doping in Ni-Mn-Z alloys greatly enhances the ferromagnetic interaction of the austenite, resulting in the significantly increased the magnetic driving for metamagnetic transformation. A popular argument is that when Co enters the Ni-Mn-Z Heusler lattice, it has the effect of turning the antiferromagnetically coupled Mn-Mn atoms into ferromagnetically couples ones [12, 19]. However, detailed explanation of this effect is yet to be established. In this study, we further expand our investigation on a series of Mn 50 Ni 40- xin 10 Co x alloys, with an emphasis on analyzing the magnetic moment interactions between Mn-Mn and Mn-Co atoms in our proposed model. 2. Experimental Procedures Polycrystalline Mn 50 Ni 40-x In 10 Co x (x=0, 1, 2 and 3) alloy ingots were prepared by means of arc melting in argon atmosphere using high purity (99.99 at.%) elemental metals. The samples are referred to as Co0, Co1, Co2, and Co3, respectively. The button shaped ingots were heat treated at 1173 K for 24 hours in vacuum followed by quenching into water for homogenization. Transformation behaviour of the alloys was studied by means of differential scanning calorimetry (DSC) using a TA Q10 DSC instrument with a cooling/heating rate of 10 K/min. Phase identification and crystal structures were determined by means of X-ray powder diffraction using Cu-Kα radiation. The compositions were determined by means of quantitative X-ray energy dispersive spectrometry (EDS) equipped on a Zeiss 1555 field-emission scanning electron microscope (FESEM). The magnetic properties were studied using a superconducting quantum interference device magnetometer (SQUID). CHAPTER 2 50

57 3. Experimental Results 3.1 Crystal structure Fig. 1 shows XRD spectra of powder samples of Mn 50 Ni 40-x In 10 Co x alloys measured at room temperature. (111) (200) (220) A (222) A (400) A (422) A (440) A Co3 X-ray Intensity Co0 (220) M (044) M (022) M (004) M (400) M (224) M (422) M (022) (022) M M (220) A (220) M (220) M (004) M (004) M (400) A (400) M (400) M (224) (224) M M (422) A (422) M (422) M (044) M (044) M Co2 Co ( o ) Fig. 1. X-ray diffraction spectra of Mn 50 Ni 40-x In 10 Co x alloys. The diffraction peaks of Co0 and Co1 alloys are indexed to body-centered tetragonal non-modulated martensite structure, which is also observed in Mn 2 NiGa alloys [8]. The Co2 alloy has a mixed structure of body-centered cubic austenite and tetragonal martensite. This indicates that the addition of 2 at.% of Co lowers the martensitic CHAPTER 2 51

58 transformation temperatures to below the room temperature. The Co3 alloy shows a pure austenite structure with bcc fundamental lattice reflections of (220), (400), (422) and (440) and superlattice reflections of (111), (200), (311) and (222). The superlattice structure can be determined by comparing the relative intensities of (111) and (200). It is evident that I 111 /I 200 >1, as shown in the inset of Fig. 1, implying that the superlattice is of the Hg 2 CuTitype, consistent with other Mn 2 NiZ(Z=Ga,Sn,Sb) alloys [20-22]. Fig. 2 shows the effect of Co addition on the lattice parameters and unit cell volumes for the austenite and martensite at room temperature. It is seen that the transformation from the cubic austenite to tetragonal martensite is realized by an expansion in the c direction and equal contractions in the a and b directions (a=b), which is consistent with Mn 2 NiGa alloy [8] Martensite Austenite Lattice Constants (nm) c M V A V V M a A a M Unit Cell Volume (nm 3 ) Co Addition (at%) Fig. 2. Effect of Co addition on lattice parameters and unit cell volume of Mn 50 Ni 40- xin 10 Co x alloys. The lattice distortion can be estimated to be (c M -a A )/a A =15.7 % along the c direction and (a M -a A )/a A = -8.2 % for the a and b directions for alloy Co2. Both the expansion in c direction and contractions in a and b directions are larger than those in Ni 2 MnGa alloy, which are 8.4 % and -6.6 % respectively [2]. This large lattice deformation implies higher frictional resistance to the propagation of transformation interfaces, leading to large CHAPTER 2 52

59 transformation hysteresis. The unit cell volumes of both the austenite and martensite are found to slightly increase with increasing substitution of Co for Ni, obviously related to the slightly larger size of Co atom relative to Ni. It is also evident that the transformation from the austenite to martensite is accompanied by a volume contraction, of -2.4%. The large volume change may induce cracking in the material during transformation cycles. 3.2 Alloy composition All these alloys show single phase microstructure, as confirmed by SEM observation. The compositions of these alloys were determined by quantitative EDS analysis, as summarized in Table 1. The Mn contents for all four alloys are approximately 49 at.%, indicating a volatilization loss of ~1 at.% of Mn during the arc-melting process. The continuous reduction of Ni is compensated well by the addition of Co, as the designed nominal compositions. The content of In remained nearly unchanged for all four alloys, at between 9.9 at.% and 10.5 at.%. The valence electron concentrations of the alloys (e/a ratio) are calculated using the compositions obtained from the EDS analysis based on the sum of s, p and d electrons for Mn (7), Ni (10), Co (9) and In (3). It is seen that the e/a ratio of the alloys decreased from to with increasing Co substitution for Ni from 0 to 3 at.%, obviously due to the smaller number of valence electrons of Co (9) relative to that of Ni (10). Table 1. Composition and e/a ratio of Mn 50 Ni 40-x In 10 Co x alloys. Mn at.% Co at.% Ni at.% In at.% e/a ratio x= x= x= x= CHAPTER 2 53

60 3.2 Martensitic transformation Fig. 3 shows DSC curves of the Mn 50 Ni 40-x In 10 Co x alloys. It is seen that the martensitic transformation behaviour evolves progressively, to lower temperatures, with increasing the Co content of the alloys. Co3 Co2 Heat Flow 0.1 w/g Co1 Co Temperature (K) Fig. 3. DSC curves of the martensitic transformation of Mn50Ni40-xIn10Cox alloys. The transformation thermal parameters, including starting, finishing and peak temperatures (M s, M f, M p, A s, A f and A p ) for the forward and reverse transformation, transformation hysteresis (ΔT=A p -M p ), enthalpy change (ΔH) and entropy change (ΔS), of the alloys are summarised in Table 2. H is obtained directly from the DSC measurement, and S is calculated as H 1 S, where T0 ( M p Ap ). T 2 0 Table 2. The martensitic and austenitic transformation starting, finishing and peak temperatures (M s, M f, M p,a s, A f, A p ), transformation hysteresis (ΔT= A p - M p ), enthalpy change (ΔH) and entropy change (ΔS) of Mn 50 Ni 40-x In 10 Co x alloys. M s M f M p A s A f A p ΔT ΔH ΔS (K) (K) (K) (K) (K) (K) (K) (J/g) (J/K kg) CHAPTER 2 54

61 x= x= x= x= Fig. 4 shows the effect of Co substitution for Ni on phase transformation temperatures (M p and A p ) and transformation hysteresis (ΔT) of the alloys. Transformation Tempearture (K) A p M p T Co Addition (at%) Transformation Hysteresis (K) Fig. 4. Effect of Co addition on phase transformation peak temperatures (M p and A p ) and transformation hysteresis (ΔT). It is seen that the transformation temperatures decreased progressively with increasing the Co content. This is in good agreement with the effect of Co doping in Ni- Mn-Ga [23] and Ni-Mn-Sb [24] alloys. This is obviously related to the e/a ratio decrease with the increase of Co substitution for Ni. The ΔT remained practically unchanged, between 20 and 26 K for the alloys of different Co content. It is known that the transformation hysteresis generally corresponds to the frictional resistance to the martensitic transformation, stemming largely from the lattice mismatch, distortion and volume change of the transformation. Generally, a larger lattice distortion means the martensitic transformation requires higher energy to overcome the friction during the motion of the phase boundaries, thus leading to larger transformation hysteresis. It is seen CHAPTER 2 55

62 in Fig. 2 that the lattice distortions and the volume change are practically the same for all the four alloys, thus resulting in nearly constant transformation hysteresis for the transformation. Fig. 5 shows the effects of Co addition on the transformation enthalpy (ΔH) and entropy (ΔS) changes of the alloys, as functions of transformation temperature T o in (a) and e/a ratio in (b). It is to be noted that for the martensitic transformation both ΔH and ΔS are negative values and the plot customarily neglects this. It is seen that both the enthalpy and entropy changes increased continuously with increasing T o and with e/a ratio, caused by Co addition. The influence of e/a ratio on the entropy change of martensitic transformation has been reported for Ni 50+x Mn 25-x Ga 25 [25], Ni 50 Mn 50-x In x [7], and Ni 50 Mn 50-x Sn x [6] alloys. In these alloy systems, ΔS increases with increasing transformation temperatures and e/a ratio, which is in good agreement with the findings of this study. Similar phenomenon has also been observed in Ni 50 Mn 40-x Sn 10 Fe x and Ni 50 Mn 37 (In,Sb) 13 alloys in our previous studies [26, 27]. Enthalpy Change (J/g) a S H increase Co addition Entropy Change (J/K-kg) T o (K) CHAPTER 2 56

63 Enthalpy Change (J/g) b S H increase Co addition e/a Ratio Entropy Change (J/K-kg) Fig. 5. Effect of Co addition on enthalpy and entropy changes, (a) as function of transformation temperature T o =(M p +A p )/2, and (b) as function of e/a ratio. 3.3 Thermomagnetization behaviour Fig. 6 shows the thermomagnetization behaviour of the four alloys. The sample was first cooled down to 10 K in a zero magnetic field prior to the measurement. A magnetic field was applied at 10 K and then the measurement was taken upon heating to 395 K at a rate of 10 K/min and cooling back again to 10 K in the same field. Fig. 6(a) shows the M(T) curves of the alloys between 10 and 395 K in a field of 50 Oe. It is seen that the Co0 alloy showed a mild decrease of magnetization at between 320 and 340 K upon heating, owing to the Curie transition of the alloy. Based on the DSC measurement (Fig. 3), the M s temperature of this alloy is 381 K. However, the M(T) data shows that the hysteresis between the heating and cooling curves prevailed at between 320 and 340 K, and continued to present down to 100 K, as shown in the inset of Fig.6 (a). This implies that the martensitic transformation is not complete and the austenite coexists in this alloy at very low temperature. Therefore, the Curie temperature corresponds to that of the remaining austenite at below M s temperature, denoted A T C =322 K of Co0. CHAPTER 2 57

64 Magnetization (emu/g) Co0 A 0.8 T C 0.4 a H=50 Oe Co3 Co0 Co2 Co Temperature (K) T C A T C A 120 b Co3 H=70 koe Magnetization (emu/g) Co2 Co1 Co Temperature (K) Fig. 6. Thermomagnetization behaviour of Mn 50 Ni 40-x In 10 Co x alloys under a field of (a) H=50 Oe and (b) H=70 koe. In the Co1 alloy, the martensite is antiferromagnetic-like at low temperatures, as evidenced by the nil magnetization. It is also worth noting the cooling curve retraced the heating curve at the entire low temperature regime below the martensitic transformation. Normally a splitting phenomenon between the zero field cooled and field cooled M(T) curves is observed at low temperatures in Ni-Mn based alloys [6, 7], which indicates the coexistence of ferromagnetic and antiferromagnetic ordering at the martensitic state. However, the M(T) data of Co1 suggests that the ferromagnetic structure is vanished and CHAPTER 2 58

65 the antiferromagnetic exchange is dominant at the martensitic phase. The antiferromagnetic martensite started transforming to ferromagnetic austenite at 320 K upon heating, followed immediately by the Curie transition of the austenite at T A C =345 K. Upon cooling, the magnetization of the austenite increased rapidly through its Curie transition, followed by a short magnetization plateau before demagnetization rapidly at 325 K upon further cooling via the transformation from the ferromagnetic austenite to antiferromagnetic martensite. The martensitic transformation in Co2 alloy can be clearly observed, shown as the magnetization change upon both heating and cooling, as evidenced by the obvious transformation hysteresis. The Curie transition for the austenite occurred at 378 K. Similar to Co1 alloy, the nil magnetization and superposition of M(T) curves at below the martensitic transformation suggest that the martensite is mainly antiferromagnetic ordered. Co3 showed clear martensitic transformation as the abrupt magnetization change upon heating and cooling in the temperature range between 170 and 195 K. The A T C of Co3 is determined to be 393 K. Unlike Co1 and Co2, the separation between the heating and cooling curves appeared at below 50 K in Co3, suggesting the coexistence of ferromagnetic ordering and antiferromagnetic ordering at its martensitic state. It is seen that the martensitic transformation shifted to lower temperatures whereas the Curie transition shifted to higher temperatures with increasing Co content in these alloys. The increased is attributed to the fact that the exchange interaction between Co-Mn is stronger than that between Ni-Mn [12]. Fig. 6(b) shows the M(T) curves of the four alloys between 10 and 395 K in a field of 70 koe. The magnetization of the Co0 alloy did not change much during the heating and cooling cycle, at between 12 and 17 emu/g. The minor increase of the magnetization from 12.5 to 15.5 emu/g at around 380 K upon heating corresponds to the partial occurrence of the martensitic transformation. With increasing the Co content, it is clear that the martensitic transformation shifted to lower temperatures. More notably, for alloys through Co1 to Co3, the magnetization of the austenite increased steadily with increasing Co content at given temperatures. For example, the magnetization increased from 52 to 70 A T C CHAPTER 2 59

66 emu/g with the increase of Co content from 1 to 3 at.% at 350 K, giving rise to an average increase of 9 emu/g per 1 at.% Co. It is also seen that the magnetization behaviour of Co1 and Co2 were completely reversible after a complete transformation cycle under 70 koe. In contrast, the magnetization loop of alloy Co3 did not close at 10 K. It is due to the kinetic arrest of the austenite phase under the influence of high magnetic field. This effect has also been observed in several Ni-Mn-In alloys [28-30]. The magnetization at the finishing point on the cooling curve comprises of the contributions of the newly formed martensite and the retained austenite. The enhanced magnetization of Co3 at the end of the cooling implies that more austenite has been retained by the high magnetic field. This is reasonable given the significantly lowered martensitic transformation temperature, i.e., reduced thermodynamic driving force for the transformation, of this alloy. 3.4 Magnetization A The maximum magnetizations of the austenite and martensite ( M and taken at A f and M M ) are As from the heating M(T) curves under 70 koe. The magnetization difference between the austenite and martensite is obtained from A M A The M, M, M, and T C are summarized in Table 3. A M M M M. Table 3. Effect of Co addition on maximum magnetizations of the austenite and A M martensite obtained at A f and A s ( M and M ), magnetization difference of the transformation ( M ), and Curie temperatures of the austenite ( T ). A C M A M M M A T C Co Co Co Co CHAPTER 2 60

67 A M A Fig. 7 shows M, M and M as functions of Co content. It is seen that M increased greatly with the increase of Co content, from 15.5 emu/g in Co0 to 118 emu/g in M Co3, meanwhile M first decreased from 12.5 emu/g in Co0 to 3 emu/g in Co1, and then it increased to 29 emu/g in Co3 alloy. ΔM shows a steady increasing trend with increasing Co content in the alloys, giving rise to a maximum value of 89 emu/g in Co3 alloy. The greatly increased M is beneficial for obtaining large magnetic driving force for metamagnetic transformation, i.e. Zeeman Energy, in these alloys. Magnetization (emu/g) Co Addition (at%) M A M M M Fig. 7. The maximum magnetization of the austenite and martensite (M A and M M ) and magnetization difference between the phases (ΔM) as a function of Co addition. The giant magnetization difference across the martensitic transformation is obviously due to the distinct magnetic states between the austenite and martensite. To further examine the magnetic configurations in the austenitic and martensitic phases, M(H) curves were carried out at 5 and 350 K for the Co0, Co1, Co2 and Co3 alloys, respectively. Fig. 8 shows the magnetization of the alloys at (a) 5 K and (b) 350 K. It is known that all the alloys are at martensitic state at 5 K based on the thermomagnetization measurements (Fig. 6). CHAPTER 2 61

68 Magnetization (emu/g) a T=5 K Co3 Co0 Co2 Co1 Magnetization (emu/g) Magnetic field (koe) b T=350 K Co3 Co2 Co Magnetic field (koe) Fig. 8. Magnetization curves at (a) 5 K and (b) 350 K of Mn 50 Ni 40-x In 10 Co x alloys. Fig. 8(a) shows that Co0 has a relatively quick magnetizing behaviour at the beginning of M(H) curve, indicating the existence of ferromagnetic ordering at its martensitic state. However, based on the form of M(H) curve and the low magnetisation of 15 emu/g at 50 koe, the antiferromagnetic exchange is expected to coexist with ferromagnetic structure at 5 K. Co3 alloy shows a similar M(H) behaviour, but with stronger magnetic correlations than that in Co0, evidenced by the higher magnetisation of 26 emu/g at 50 koe. The M(H) curves of Co1 and Co2 are nearly linear, particularly for Co1, strongly suggesting the existence of long-range antiferromagnetic ordering at the martensitic state of these alloys. CHAPTER 2 62

69 Fig. 8(b) shows the magnetization behaviour of the austenitic phase at 350 K of Co1, Co2 and Co3 alloys. The absence of M(H) data of Co0 is due to its martensitic state at 350K, which is irrelative for comparison with other alloys at the austenitic state. The austenite of Co1 showed a gradual magnetization growth and maximized at 42 emu/g upon magnetizing. It is known that the T A C (345 K) of Co1 is very close to the magnetizing temperature (350 K), therefore, the shape of the M(H) data indicates the existence of magnetic short-range correlations in the paramagnetic austenitic state. Co2 and Co3 presented very typical ferromagnetic behaviour due to the rapid increase of magnetization at the initial portion of M(H) curves and the high magnetization magnitude of 59 and 70 emu/g, respectively. 4. Discussion 4.1 Entropy change It is seen in Fig. 5 that the value of the entropy change of the transformation decreased significantly with Co doping, by 36 % reduction with addition of 3 at.% Co. Entropy change (ΔS) of a martensitic phase transformation is a measure of the difference of degree of order between the austenite and martensite. In addition to being a function of temperature itself, ΔS is generally considered to have three contributions, including crystal structural ordering (ΔS latt ), magnetic structure ordering (ΔS mag ) and electronic structure ordering (ΔS el ). In Ni 2+x Mn 1-x Ga and X 2 MnSn (X=Co, Ni, Pd, Cu) alloys, it has been shown that the electronic contribution ΔS el to ΔS is small [25, 31]. The crystal structural contribution ΔS latt to ΔS depends on the crystal structures of the transformation. For the present four alloys, the structural change is the same and the magnitudes of lattice distortions of the transformation are similar according to the XRD measurements. Thus, the crystal structural contribution to the entropy change is also expected to be unchanged for these alloys. In this regard, the increase of total entropy change ΔS with Co content is attributed to the magnetic ordering contribution, neglecting the temperature effect. CHAPTER 2 63

70 Fig. 9 shows a schematic of the contributions of ΔS latt, ΔS mag to ΔS as functions of Co content in the alloys. ΔS ΔS=ΔS latt +ΔS mag ΔS mag Co content ΔS ΔS latt Fig. 9. Illustration of the effect of Co addition on entropy change of the alloys. It is known that S A latt M A S, thus the S S S 0 for the forward A M M latt latt latt latt transformation. In the Figure ΔS latt remains a constant negative value irrespective of Co content. On the other hand, it is known that S A mag M A S, thus S S S 0. This M mag mag mag mag is because of the higher magnetic ordering in the austenite relative to the martensite. The introduction of Co enhances the ferromagnetic ordering of the austenite [13, 16, 19], thus decreasing the magnetic entropy of the austenite. Meanwhile, the magnetic interaction between the atomic constituents in the martensite is not significantly affected, relative to the austenite, as evident in Fig. 7. This results in positive increase of magnetic entropy change of the transformation with increasing Co content. Consequently, the total entropy change for the A M transformation becomes less negative with increasing Co content, as observed in Fig Magnetic moment interactions Structure of stoichiometric Mn 2 NiIn CHAPTER 2 64

71 The crystal structure of the austenite is Hg 2 CuTi-type super lattice cubic structure. This structure is commonly observed in Mn-rich Heusler alloys, such as Mn 2 NiGa [20] and Mn 2 CoZ(Z=Al,Ga,Ge,In,Sn,Sb) [32]. In this structure, Mn atoms occupy A (0,0,0) site and B (1/4,1/4,1/4) site, leaving C (1/2,1/2,1/2) site to Ni atoms and D (3/4,3/4,3/4) site to the third element atoms. This structure is illustrated in Fig. 10, showing the unit cell models for both the austenite in (a) and martensite in (b) of a stoichiometric Mn 2 NiIn alloy. Such structure can be expressed in a stacking order of Mn-Mn-Ni-X ( F43m the diagonal direction of the unit cell. space group) along Fig. 10. Atomic configuration in the unit cell of Mn 2 NiIn alloy: (a) unit cell of the austenite with Mn-Mn-Ni-In stacking order (Hg 2 CuTi structure) and (b) unit cell of the martensite with Mn-Mn-Ni-In tetragonal structure. Based on the calculation by Chakrabarti et al [9], the spin magnetic moments of Mn(A), Mn(B) and Ni in Mn 2 NiIn alloy are -3.08, 3.42 and 0.13 µ B, respectively. The magnetic moment of In is very small and is neglected in the present discussion. The spin magnetic moments of Mn atoms are symbolically depicted using arrows on the atoms shown in Fig. 10. The length of the arrows roughly represents the magnitude of the magnetic moment of the atom. The exchange interaction between Mn atoms is known to CHAPTER 2 65

72 depend strongly on Mn-Mn distance in the lattice. Early studies have shown that the magnetic interaction between Mn atoms changes from antiferromagnetic to ferromagnetic when the Mn-Mn distance is increased to above a critical value of approximately 0.30 nm [33]. For the austenite with Mn-Mn-Ni-In stacking order in the unit cell (Fig. 10(a)), the distance between the nearest neighboring Mn at A site and B site is AB 3 a/4= nm and the distance between the second nearest neighboring Mn at two adjacent A sites is AA 2 a/2= nm using the lattice constant of a= nm for the Co3 alloy. This implies that the moments of Mn(A)-Mn(B) form antiparrallel coupling and Mn(A)-Mn(A) form parallel coupling. It is seen that the spin directions of Mn magnetic moments at A site are antiparrallel with those of Ni at C site, and they are also opposed to those of Mn at B site. This forms a ferrimagnetic structure in this atomic configuration, with antiparrallel aligned magnetic moments between (A,C) and (B,D) sub-lattices. Assuming that the stoichiometric Mn 2 NiIn alloy also undergoes the same structural transformation to a tetragonal martensite as for the present alloys, the distance between A site and B site changes very little by the transformation, at AB 3 a/4= nm in the martensite, and the distance between two adjacent A sites is shortened to AA 2 a/2= nm as shown in Fig. 10(b). Therefore, Mn(A)-Mn(B) still forms antiferromagnetic interaction and Mn(A)-Mn(A) forms ferromagnetic interaction in the martensite. This implies that the magnetic exchange interactions in the martensite are similar with those in the austenite, which is ferrimagnetic Atomic configuration in off-stoichiometric Mn 2 Ni 1+x In 1-x In Ni-rich off-stoichiometric Mn 2 Ni 1+x In 1-x alloys, the magnetization has been found to increase greatly relative to its mother alloy Mn 2 NiIn, as evidenced by the magnetization of ~75 emu/g (at 230 K) in Mn 50 Ni 40 In 10 [10] comparing to 9.27 emu/g in Mn 2 NiIn [9]. The drastic increase of the magnetization cannot be solely attributed to the magnetic moment contribution from the extra Ni which substitutes for In, as the magnetic moment of Ni is small, typically ~0.13 µ B. This suggests that the increase may originate from the biggest magnetic moment contributor Mn atoms. This implies that the atomic configuration must have changed after the Ni substitution for In. CHAPTER 2 66

73 Fig. 11 shows the atomic configuration in the unit cell of the austenite in (a) and the martensite in (b) of the Co0, with the nominal composition of Mn 50 Ni 40 In 10. Fig. 11. Atomic configurations in the unit cell of Co0 alloy (Mn 50 Ni 40 In 10 ): (a) unit cell of the austenite and (b) unit cell of the martensite. The number of displaced atoms does not represent the actual proportion of substitution, which is only for qualitative interpretation. It is seen that some portion of Mn atoms at A site have been replaced by Ni atoms, and these new Mn atoms share D site with In atoms. This hypothesis is based on the rule of preferential site occupation in Mn 2 YZ (Y: 3d elements; Z: III-V A group elements) alloys reported by Liu et al. [32]. They observed that Y elements on the right hand side of Mn in the Periodic Table of Elements prefer to occupy (A,C) sites, whereas Y elements to the left of Mn have strong preference for B site occupancy. In Mn 2 YZ (Y = V, Cr, Mn, Fe, Co and Ni; Z =Al, Ga, In, Si, Ge, Sn and Sb) Heusler alloys, this rule of atomic occupancy has been shown to be well obeyed [22, 32, 34-36]. According to this principle, Ni substitution for In will have the priority to take A site in preference to Mn. The distance between the new Mn atoms at D site and Mn atoms at B site is nm, which favors ferromagnetic exchange interaction between the Mn atoms. Therefore, the magnitude of antiferromagnetic CHAPTER 2 67

74 alignment between Mn(A) and Mn(B) is reduced, and the new Mn atoms at D site form ferromagnetic interaction with the Mn atoms at B site. In Fig. 11(a), the spin magnetic moments of the new Mn(D) align parallel with those of Mn(B). At the meantime, the new (A,C) sub-lattice after the replacement of Ni for Mn at A site also forms ferromagnetic interaction with Mn(B) and Mn(D), thus creating a local ferromagnetic structure of Mn(B)-Ni-Mn(D) in the alloy. This explains the increase of the magnetization of the austenite in off-stoichiometric Mn 2 Ni 1+x In 1-x alloy after Ni substitution for In compared to the stoichiometric Mn 2 NiIn alloy. In the martensite unit cell (Fig. 11(b)), the distance change between the nearest Mn(A)-Mn(B) atoms is very small, from to nm, which suggests that the coupling between Mn(A)-Mn(B) does not change after the transformation, still showing antiferromagnetic interaction between them. However, along the a and b directions in the tetragonal unit cell, the distance between A and C sites (same for B and D sites) is shortened to AC=BD= nm, which strongly favours antiferromagnetic coupling between Mn(B)-Mn(D) [37, 38]. This explains the presence of antiferromagnetic interaction at the martensitic phase obtained from the M(H) data shown in Fig. 8 (a). In this regard, it is reasonable to attribute the disappearance of the local ferromagnetic structure of Mn(B)-Ni-Mn(D) in the martensite to the significant decrease of the Mn(B)-Mn(D) distance Magnetic moment contribution from Co Fig. 12 shows the atomic configuration in the unit cells for the austenite (a) and the martensite (b) after Co doping in off-stoichiometric Mn 50 Ni 40 In 10 alloys. Followed by the rule of selectivity of atomic configuration in Heusler alloys as aforementioned, Co substitution for Ni should just replace Ni at either A site or C site, and no new atomic configuration is formed. By assuming that Co atoms replace the Ni atoms at A site, a stronger local ferromagnetic structure of Mn(B)-Co-Mn(D) will be formed, shown in Fig. 12(a). Each Co atom at the site of Ni contributes a larger magnetic moment (~1.2 µ B ) relative to Ni (0.13 µ B ), based on the calculation of magnetic moments in Mn 2 NiCo x Ga 1-x CHAPTER 2 68

75 alloys [37]. This results in the further increment of the magnetization of the austenite due to the magnetic moment contribution from Co. Fig. 12. Atomic configurations in the unit cell of Co doped Mn 50 Co x Ni 40-x In 10 (x 1): (a) unit cell of the austenite and (b) unit cell of the martensite. The number of displaced atoms does not represent the actual proportion of substitution, which is only for qualitative interpretation Maximum magnetization of ferromagnetic austenite It is evident in Fig. 6(b) that the magnetization of the ferromagnetic austenite increased progressively with decreasing temperature. The temperature window for ferromagnetic austenite is limited by two temperatures, the Curie temperature of the austenite ( A T C temperature ( ( M A ) as the upper boundary and the (reverse) martensitic transformation A f ) as the lower boundary. The maximum magnetization of the austenite ) is achieved at A f (upon heating), as seen on the M(T) curves (Fig. 6(b)) of these alloys. It is also seen in Fig. 3 that the transformation temperatures decreased significantly with increasing Co content. This means that the CHAPTER 2 69 A M values were actually taken at different

76 A temperatures for these alloys and the increase of M with increasing Co content, shown in Fig. 7 is really due to the widening of the temperature window of the ferromagnetic austenite, instead of purely due to the effect of alloying. The effect of alloying may be estimated by measuring the magnetization of the alloys at a given temperature. As aforementioned, the magnetization at 350 K increased from 52 emu/g for Co1 to 70 emu/g for Co3. On the other hand, the maximum magnetization at A f increased from 55 emu/g for Co1 to 118 emu/g for Co3, corresponding to the magnetization increase of 31.5 emu/g per at.% of Co addition. It is evident that widening of the temperature window is the more prominent factor relative to Co alloying contributing to the large ΔM in these alloys. 4.3 Transformation diagram Fig. 13 shows the effect of e/a ratio, as a result of Co doping, on the martensitic transformation temperatures M p and A p obtained from the DSC measurement and on the Curie transition temperature T A C obtained from the magnetization measurements of the Mn 50 Ni 40-x In 10 Co x alloys. It is seen that M p and A p decreased linearly with decreasing e/a ratio and T A C increased. This observation is consistent with the general observation of positive dependence of martensitic transformation temperatures on e/a ratio reported in the literature for Ni-Mn-Z(Z=Ga,In,Sn,Sb) alloy systems [6, 7, 39]. The linear coefficient is estimated to be 25 K per 0.01 change of e/a ratio, which is comparable to the value determined for Ni 50 Mn 40-x Sn 10 Fe x alloys [26]. CHAPTER 2 70

77 Transformation Temperature (K) T A C A (ferro) A p M p Co3 A (para) Co1 Co2 M (antiferro) Co A 7.84 e/a Ratio Fig. 13. Effect of e/a ratio on M p, A p and A TC temperatures of Mn 50 Ni 40-x In 10 Co x alloys. The temperature-e/a ratio space shown in Fig. 13 can be tentatively divided into three regions representing different crystallographic and magnetic states for the alloys, including austenite (paramagnetic), austenite (ferromagnetic) and martensite (antiferromagnetic). Among the three states, two transformation schemes may occur. In the region to the right of point A, the alloy undergoes a single step transformation between paramagnetic austenite and antiferromagnetic martensite. This transformation is of low interest for magnetic actuation. To the left of point A, the alloy undergoes the transformation sequence from paramagnetic austenite to ferromagnetic austenite and then to antiferromagnetic martensite upon cooling, expressed as A( para) A( ferro) M ( antiferro). In this expression, the single arrow represents magnetic transition and the double arrow represents the martensitic transformation (in this case it is also a concurrent magnetic transition). The phase area of the ferromagnetic austenite opens up with the decrease of e/a ratio, thus providing a wider temperature window for A( ferro) M ( antiferro) transformation. CHAPTER 2 71

78 5. Conclusions The effects of Co substitution for Ni on the martensitic transformation and magnetic behaviour of Mn 50 Ni 40-x In 10 Co x alloys were investigated. The experimental evidences and the discussions lead to the following conclusions: (1) Co substitution for Ni up to 3 at.% greatly decreases the martensitic transformation temperature from 381 K to 175 K in these alloys. The martensite has a non-modulated tetragonal structure, and the crystal structure of the austenite is determined to be Hg 2 CuTi-type superlattice cubic structure. (2) The decrease of the phase transformation temperatures is attributed to the decrease of the e/a ratio for the alloys with increasing Co substitution for Ni. The enthalpy and entropy changes of the transformation are both found to increase with increasing the e/a ratio of the alloys. (3) The maximum magnetization of the austenite (under 70 koe) is significantly increased from 15.5 emu/g in the Co0 alloy to 118 emu/g in the Co3 alloy, whereas that of the martensite shows much less significant change from 12.5 emu/g in the Co0 alloy to 29 emu/g in the Co3 alloy. Consequently, magnetization difference between the austenite and the martensite increases significantly with increasing Co substitution for Ni. The largest ΔM for the martensitic transformation obtained is 89 emu/g in alloy Co3. (4) The increased magnetization of the austenite is attributed to three reasons: (i) formation of ferromagnetic structure of Mn(B)-Ni-Mn(D) in off-stoichiometric Mn 2 Ni 1+x In 1-x, due to the readjustment of atomic configuration in the unit cell caused by Ni substitution for In, (ii) higher magnetic moment contribution of Co relative to Ni, and (iii) widening of the temperature window for ferromagnetic austenite. (5) The low magnetization of the martensite, relative to that of the austenite, is due to the significantly shortened distance between Mn(B)-Mn(D), which leads to the disappearance of the local ferromagnetic structure in a tetragonal martensitic structure. CHAPTER 2 72

79 Acknowledgements The authors wish to acknowledge the financial supports by the Department of Innovation Industry, Science and Research of the Australian Government in ISL Grant CH070136, and by National Natural Science Foundation of China in Grant No References [1] Sutou Y, Imano Y, Koeda N, Omori T, Kainuma R, Ishida K, Oikawa K. Appl Phys Lett 2004;85:4358. [2] Ullakko K, Huang JK, Kantner C, Handley RCO, Kokorin VV. Appl Phys Lett 1996;69:1966. [3] Sasioglu E, Sandratskii LM, Bruno P. Phys Rev B 2004;70: [4] Kanomata T, Fukushima K, Nishihara H, Kainuma R, Itoh W, Oikawa K, Ishida K, Neumann KU, Ziebeck KRA. Mater Sci Forum 2008;583:119. [5] Kanomata T, Yasuda T, Sasaki S, Nishihara H, Kainuma R, Ito W, Oikawa K, Ishida K, Neumann KU, Ziebeck KRA. J Magn Magn Mater 2009;321:773. [6] Krenke T, Acet M, F. Wassermann E, Moya X, Manosa L, Planes A. Phys Rev B 2005;72: [7] Krenke T, Acet M, F. Wassermann E, Moya X, Manosa L, Planes A. Phys Rev B 2006;73: [8] Liu GD, Chen JL, Liu ZH, Dai XF, Wu GH. Appl Phys Lett 2005;87: [9] Chakrabarti A, Barman SR. App Phys Lett 2009;94: [10] Sanchez Llamazares JL, Hernando B, Prida VM, Garcia C, Gonzalez J, Varga R, Ross CA. J Appl Phys 2009;105:07A945. [11] Sanchez Llamazares JL, Sanchez T, Santos JD, Perez MJ, Sanchez ML, Hernando B, Escoda L, Sunol JJ, Varga R. Appl Phys Lett 2008;92: [12] Ma L, Zhang HW, Yu SY, Zhu ZY, Chen JL, Wu GH, Liu HY, Qu JP, Li YX. Appl Phys Lett 2008;92: [13] Wu Z, Liu Z, Yang H, Liu Y, Wu G. Appl Phys Lett 2011;98: [14] Lazpita P, Barandiaran JM, Gutierrez J, Feuchtwanger J, Chernenko VA, Ricgard ML. New J Phys 2011;13: CHAPTER 2 73

80 [15] Kainuma R, Ito W, Umetsu RY, Oikawa K, Ishida K. Appl Phys Lett 2008;93: [16] Yu SY, Cao ZX, Ma L, Liu GD, Chen JL, Wu GH, Zhang B, Zhang XX. Appl Phys Lett 2007;91: [17] Kainuma R, Imano Y, Ito W, Morito H, Sutou Y, Oikawa K, Fujita A, Ishida K, Okamoto S, Kitakami O, Kanomata T. Appl Phys Lett 2006;88: [18] Kainuma R, Imano Y, Ito W, Sutou Y, Morito H, Okamoto S, Kitakami O, Oikawa K, Fujita A, Kanomata T, Ishida K. Nature 2006;439:957. [19] Yu SY, Ma L, Liu GD, Liu ZH, Chen JL, Cao ZX, Wu GH, Zhang B, Zhang XX. Appl Phys Lett 2007;90: [20] Liu GD, Dai XF, Yu SY, Zhu ZY, Chen JL, Wu GH, Zhu H, Xiao JQ. Phys Rev B 2006;74: [21] Luo H, Liu G, Feng Z, Li Y, Ma L, Wu G, Zhu X, Jiang C, Xu H. J Magn Magnc Mater 2009;321:4063. [22] Helmholdt RB, Buschow KHJ. J Less-Common Met 1987;128:167. [23] Cong DY, Wang S, Wang YD, Ren Y, Zuo L, Esling C. Mater Sci Eng A 2008;473:213. [24] Han ZD, Wang DH, Zhang CL, Xuan HC, Zhang JR, Gu BX, Du YW. J Appl Phys 2008;104: [25] Khovailo VV, Oikawa K, Abe T, Takagi T. J Appl Phys 2003;93:8483. [26] Wu Z, Liu Z, Yang H, Liu Y, Wu G, Woodward RC. Intermetallics 2011;19:445. [27] Liu Z, Wu Z, Yang H, Liu Y, Liu E, Zhang H, Wu G. Intermetallics 2010;18:1690. [28] Ito W, Ito K, Umetsu RY, Kainuma R, Koyama K, Watanabe K, Fujita A, Oikawa K, Ishida K, Kanomata T. Appl Phys Lett 2008;92: [29] Sharma VK, Chattopadhyay MK, Roy SB. Phys Rev B 2007;76: [30] Umetsu RY, Ito W, Ito K, Koyama K, Fujita A, Oikawa K, Kanomata T, Kainuma R, Ishida K. Scripta Mater 2009;60:25. [31] Fraga GLF, Brandao DE, Sereni JG. J Magn Magn Mater 1991;102:199. [32] Liu GD, Dai XF, Liu HY, Chen JL, Li YX, Xiao G, Wu GH. Phys Rev B 2008;77: [33] Yamada T, Kunitomi N, Nakai Y, Cox DE, Shirane G. J Phys Soc Jap 1970;28:615. CHAPTER 2 74

81 [34] Luo HZ, Zhang HW, Zhu ZY, Ma L, Xu SF, Wu GH, Zhu XX, Jiang CB, Xu HB. J Appl Phys 2008;103: [35] Ozdogan K, Galanakis I, Sasioglu E, Atkas B. J Phys-Condens Mat 2006;18:2905. [36] Weht R, Pickett WE. Phys Rev B 1999;60: [37] Ma L, Wang WH, Zhen CM, Hou DL, Tang XD, Liu EK, Wu GH. to be published. [38] Kasper JS, Roberts BW. Phys Rev 1956;101:537. [39] Krenke T, Moya X, Aksoy S, Acet M, Entel P, Manosa L, Planes A, Elerman Y, Yucel A, Wassermann EF. J Magn Magn Mater 2007;310:2788. CHAPTER 2 75

82 Paper 3 Martensitic and magnetic transformation behaviours in Mn 50 Ni 42-x Sn 8 Co x polycrystalline alloys Zhigang Wu 1, Zhuhong Liu 2, Hong Yang 1, Yinong Liu 1, Guangheng Wu 3 1 School of Mechanical and Chemical Engineering, The University of Western Australia, Crawley, WA 6009, Australia 2 Department of Physics, University of Science and Technology Beijing, Beijing , China 3 Beijing National Laboratory for Condense Matter Physics, Institute of Physics, Chinese Academy of Science, Beijing , China Abstract This study investigated the effect of Co substitution for Ni in Mn 50 Ni 42 Sn 8 alloy with the aim to increase the magnetic driving force for inducing its martensitic transformation. The martensitic transformation temperatures, enthalpy and entropy changes are found to decrease progressively with increasing the Co content, while the transformation hysteresis increased. Co substitution for Ni also significantly increased the magnetization of the austenite, but with negligible effect on that of the martensite. A large magnetization difference 109 emu/g was achieved across the transformation in a Mn 50 Ni 34 Sn 8 Co 8 alloy. The large magnetization difference between the two phases provides enhanced thermodynamic driving force for the transformation. Consequently, the martensitic transformation was induced by the application of a magnetic field in Mn 50 Ni 36 Sn 8 Co 6 and Mn 50 Ni 34 Sn 8 Co 8 alloys. The effect of Co substitution for Ni on the magnetic interaction among the constituents for the austenite and martensite was clarified in this study, which provides a guide for alloy design for magnetoactuation applications. CHAPTER 2 76

83 Keywords: A: magnetic intermetallics; B: alloy design; B: shape-memory effects; B: martensitic transformations; B: magnetic properties. 1. Introduction Magnetomartensitic transformations in certain alloys have attracted extensive research interest in the past 15 years, since the discovery of giant magnetic-field-induced strains (MFIS) of 5-10 % in near stoichiometric Ni-Mn-Ga alloys in 1996 [1-3]. The large MFIS is associated with the rearrangement of martensite variants, driven by the magnetocrystalline anisotropy of these variants. The magnetic driving force for this type of actuation, which is derived from the magnetic anisotropy of the martensite variant, is generally small, of the order of kj/m3 [4]. Given that the shape change is typically 6%, this yields a magnetically generated stress of 5-6 MPa. Such stress is barely enough to overcome the mechanical resistance for martensite variant detwining [2, 4]. The low force output has been proven to be a main limitation for the application of these materials for mechanical actuation. An intrinsic solution to this problem is to increase the magnetic driving power for the martensitic transformation. To increase power density, a new group of off-stoichiometric Heusler Ni-Mn-Z(Z=In,Sn,Sb) alloys have been developed. These alloys present concurrent martensitic and magnetic transformation, in which one phase (the martensite) has much lower magnetization compared to the other (the austenite) [5]. In Ni 50 Mn 34 In 16 alloy, the magnetization difference between the two transforming phases is around 70 emu/g, which gives rise to large magnetic power of 700 kj/m 3 at 1 Tesla field. This large magnetization difference provides the necessary thermodynamic diving force, thus the opportunity for obtaining a magnetic-field-induced reverse martensitic transformation. In the recent few years, much effort has been put into increasing the magnetic driving force for martensitic transformation in Ni-Mn-Z(Z=In,Sn,Sb) alloys. This driving force is the Zeeman Energy E Zeeman =µ 0 M H, where µ 0 is the permeability of a vacuum, M is the saturation magnetization difference between the austenite and martensite and H corresponds to the strength of the applied field. Co substitution for Ni has been found CHAPTER 2 77

84 effective for increasing M between the phases in Ni-Mn-Z(Z=In,Sn,Sb) alloys, leading to the successful field induced phase transformation in these alloys [6-10]. In these alloys, it is known that the net magnetic moment mainly comes from the contribution of Mn [11], and the magnetic moment distribution of Mn is very sensitive to interatomic distance. The magnetic interaction between the Mn atoms can change from ferromagnetic to antiferromagnetic alignment when the distance becomes below a critical value, which is ~0.3 nm [12]. With the objective of increasing M for the magnetic-field-induced transformation, understanding of the magnetic moment contribution of Mn in offstoichiometric is essential. In our previous study on Mn 50 Ni 40-x In 10 Co x alloys, we proposed an atomic configuration model in which the mechanism of magnetic exchange interaction between Mn-Mn and Mn-Co was explained. For better understanding of the properties of structural and magnetic transitions in other Mn-rich Mn-Ni-based alloys, it is necessary to extend the studies to a new series of Mn 50 Ni 42-x Sn 8 Co x alloys, with a focus on the effect of Co substitution for Ni on the martensitic transformation and magnetic properties. 2. Experimental Procedures Polycrystalline Mn 50 Ni 42-x Sn 8 Co x (x=0, 2, 4, 6 and 8) alloy ingots were prepared by means of arc melting in argon atmosphere using high purity (99.99 %) elemental metals. The samples are referred to as Co0, Co2, Co4, Co6 and Co8, respectively. The button shaped ingots were heat treated at 1173 K for 24 hours in vacuum followed by quenching into water for homogenization. Transformation behaviour of the alloys was studied by means of differential scanning calorimetry (DSC) using a TA Q10 DSC instrument with a cooling/heating rate of 10 K/min. Phase identification and crystal structures were determined by means of X-ray powder diffraction using Cu-Kα radiation. Microstructures of the samples were studied with optical microscopy and the compositions were determined by means of X-ray energy dispersive spectrometry (EDS). The magnetic properties were studied using a superconducting quantum interference device magnetometer (SQUID). CHAPTER 2 78

85 3. Results and discussion 3.1 Crystal structure Figure 1 shows XRD spectra of Mn 50 Ni 42-x Sn 8 Co x alloys at room temperature. It is seen that from Co0 through to Co6, the alloys show nearly identical diffraction patterns of the martensite with a non-modulated body centered tetragonal structure. The lattice parameters of the martensite are determined to be a=b=0.545 nm, and c=0.697 nm. Alloy Co8 shows a single phase structure with bcc fundamental lattice reflections of (220), (400) and (422) and superlattice reflections of (111), (200) and (311). The superlattice structure can be determined by comparing the relative intensities of (111) and (200) [13]. It is evident that I 111 /I 200 >1, implying that the superlattice is of the Hg 2 CuTi-type, shown in the inset of Figure 1(e). This observation is consistent with those observed in Mn 2 NiZ (Z=In, Sn and Sb) alloys [14, 15] and Mn 50 Ni 37 In 10 Co 3 alloy [16]. In this structure, one Mn sublattice occupies A (0, 0, 0) site (referred to as Mn(A)), the other Mn sublattice is at B site (0.25, 0.25, 0.25) (referred to as Mn(B)), Ni atoms occupy C site (0.5, 0.5, 0.5) and Z atoms occupy D site (0.75, 0.75, 0.75). Such structure can be expressed in a stacking order of MnMnNiX ( F43m space group) along the diagonal [111] direction of the cubic unit cell. The lattice parameter of the austenite in Co8 is determined to be a=0.602 nm. CHAPTER 2 79

86 (e) (111) (200) (220) A (311) A (400) A (422) A Co8 X-ray Intensity (d) (c) (b) Co6 Co4 Co2 (a) (022) M (004) M (400) M (224) M (422) M Co ( o ) Figure 1. X-ray powder diffraction spectra of the Mn 50 Ni 42-x Sn 8 Co x alloys. Inset of (e) shows the comparison between the relative intensities of (111) and (200) reflections. 3.2 Microstructure and alloy composition Figure 2 shows optical micrographs of the microstructures of Co6 and Co8 alloys after homogenization treatment. Both alloys have a single phase throughout the matrix. Alloy Co6 presents evident martensite plates, indicating the martensite state at room temperature. Co8 shows a few martensite plates in the austenite matrix, which may be due to the occurrence of partially stress-induced martensitic transformation. Cracks are visible along the columned grain boundaries for both alloys, indicating the brittleness of these materials. CHAPTER 2 80

87 Figure 2. Optical micrographs for (a) alloy Co6 and (b) alloy Co8. The composition of these alloys was determined by quantitative EDS analysis. The results are summarized in Table 1. For all the alloys, the Mn contents are approximately 49 at.%, indicating that the volatilization loss of Mn is ~1 at.% during the arc-melting process. The continuous reduction of Ni is compensated well by the addition of Co. The Sn content remained nearly unchanged, at between 8.6 at.% and 8.9 at.%. The valence electron concentration per atom (e/a ratio) was calculated using the compositions obtained from the EDS analysis with the sum of s, p and d electrons for Mn (7), Ni (10), Co (9) and Sn (4). It is seen that the e/a ratio decreased from to with increasing Co substitution for Ni from 0 to 8 at.%, obviously due to the smaller number of valence electrons of Co (9) relative to that of Ni (10). Table 1. Composition and e/a ratio of the Mn50Ni42-xSn8Cox alloys. Mn (at%) Co (at%) Ni (at%) Sn (at%) e/a x= x= x= x= x= CHAPTER 2 81

88 3.3 Martensitic transformation Figure 3 shows DSC curves of the Mn 50 Ni 42-x Sn 8 Co x alloys. It is seen that the martensitic transformation behaviour evolves progressively with increasing the Co addition. Co8 Co6 Heat Flow Co4 Co2 1 J/g Co Temperature (K) Figure 3. DSC curves of the Mn 50 Ni 42-x Sn 8 Co x alloys. Both the transformation temperatures and the enthalpy change decreased. The transformation temperatures (T M : the forward transformation peak temperature, T A : the 1 reverse transformation peak temperature and To ( TM TA ) ), transformation hysteresis 2 (ΔT=T A -T M ), enthalpy change ( H) and entropy change ( S) of the alloys are summarized in Table 2. The H values were determined directly from the DSC measurements, and S is estimated based on H S. T o Table 2. Thermal and thermodynamic parameters of the martensitic transformation of the Mn 50 Ni 42-x Sn 8 Co x alloys. T M (K) T A (K) T o (K) ΔT (K) ΔH (J/g) ΔS (J/K kg) CHAPTER 2 82

89 x= x= x= x= x= Figure 4 shows the effect of e/a ratio on phase transformation temperatures (T M, T A and T o ) and hysteresis (ΔT) of the alloys. It is seen that T M and T A increased with increasing e/a ratio (decreasing Co content). This is consistent with the general trend of positive dependence of martensitic transformation temperatures on e/a ratio observed in Ni-Mn- Z(Z=Ga, In, Sn and Sb) alloys [17]. It appears that there are two linear dependences of the transformation temperatures on e/a ratio. At below e/a=7.967, corresponding to 4 at.% of Co, the linear coefficient is 18 K per 0.01 e/a unit for To. At above e/a=7.967, the coefficient is 3.5 K per 0.01 e/a unit. Similarly, ΔT also shows two distinct dependences on e/a ratio. It increases with more Co content at below and remains independent of e/a ratio at above. CHAPTER 2 83

90 Transformation Temperature (K) T A T o T M T increasing Co centent Transformation Hysteresis (K) e/a Ratio Figure 4. Effect of Co addition on phase transformation temperatures and hysteresis expressed as functions of e/a ratio. Figure 5 shows the effects of Co addition on the transformation enthalpy and entropy changes of the alloys, shown as functions of transformation temperature T o in (a) and of e/a ratio in (b). It is seen that both the enthalpy and entropy changes increased continuously with increasing T o and with e/a ratio, caused by Co addition. The influence of e/a ratio on the entropy change of martensitic transformation has been reported for Ni 50+x Mn 25-x Ga [18, 19], Ni 50 Mn 50-x In x [20], and Ni 50 Mn 50-x Sn x [21] alloys. In these alloy systems, ΔS increases with increasing transformation temperatures and e/a ratio, which is in good agreement with the findings in the present study. The change of ΔS is mainly attributed to the change of the magnetic component of the total ΔS caused by increasing the Co addition. With increasing the Co content, the magnetic entropy change increases in the alloys. This argument is supported by the evidence that the magnetization of the austenite increases while that of the martensite remains unchanged with increasing the Co content in the alloys, as shown in Figure 7(b). For the forward transformation, ΔS A-M, the (positive) increase of magnetic entropy change reduces the (negative) lattice entropy change, thus resulting in the decrease of the total (negative) ΔS for these alloys. CHAPTER 2 84

91 a Entalpy Change (J/g) H S increasing Co content T o (K) Entropy Change (J/K-kg) b Entalpy Change (J/g) H S increasing Co content e/a Ratio Entropy Change (J/K-kg) Figure 5. Effect of Co addition on enthalpy and entropy changes of the Mn 50 Ni 42-x Sn 8 Co x alloys, (a) as function of transformation T o =(T M +T A )/2, and (b) as function of e/a ratio. 3.4 Thermomagnetization Figure 6 shows the magnetization of the alloys during a heating-cooling cycle between 200 and 395 K in a field of 50 Oe (Figure 6(a)) and 70 koe in (Figure 6(b)). The sample was first cooled down to 200 K in a zero magnetic field prior to the measurement. A magnetic field was applied at 200 K and then the sample was heated at a rate of 10 K/min up to 395 K and cooled back again to 200 K in the same field. CHAPTER 2 85

92 a Magnetization (emu/g) Co Temperature (K) H=50 Oe Co0 Co Magnetization (emu/g) Co6 Co4 b Magnetization (emu/g) Temperature (K) M=109 emu/g Co8 Co6 H=70 koe Co4 Co2 Co Temperature (K) Figure 6. Thermomagnetization curves in a magnetic field of (a) H=50 Oe and (b) H=70 koe for Mn 50 Ni 42-x Sn 8 Co x. At a low field of 50 Oe, as shown in Figure 6(a), alloys Co4, Co6 and Co8 undergo a structural transformation between a ferromagnetic austenite and a ferrimagnetic martensite with obvious transformation hysteresis between the heating and cooling curves. For Co4, the magnetization drop of the austenite upon heating to ~380 K is due to the Curie A transition of the austenite, denoted as T 380K. The Curie transition was not observed in C Co6 and Co8 within the testing temperature range, suggesting higher Curie transition CHAPTER 2 86

93 temperatures at above 400 K for these two alloys. In the inset of Figure 6(a), alloys Co2 and Co4 present similar martensitic transformation, but with much smaller magnitude of magnetizations of the austenite compared to those of Co4, Co6 and Co8. The Curie temperatures of the austenite of Co0 and Co2 alloys seem close to 400 K, which are higher than that of Co4. Normally, with more Co substitution for Ni, the Curie temperature is expected to increase, due to the stronger exchange interaction between Co-Mn than that between Ni-Mn [22]. The anomaly of A T C in these alloys is not clear at this stage. It is evident that the martensitic transformation temperature decreased with increasing Co content in these alloys, consistent with the observation from DSC measurement (Figure 3). The Curie temperatures of these alloys are obviously higher than those in Ni 2 Mn 1+x Sn 1-x [21, 23, 24] alloys, which are around 320 K. It has been reported that T C is 588 K in Mn 2 NiGa alloy [25], suggesting the generally higher T C in Mn-rich Mn 2 NiZ alloys relative to Ni-rich Ni 2 MnZ alloys. It is also worth noting that the transformation hysteresis is around 40 K for Co8, which is much larger than the value determined from DSC measurement (26 K). The large hysteresis usually means higher frictional resistance to the propagation of the transformation interfaces. This may lead to large difficulty for two-way magnetic-field-induced martensitic transformation in Co8 alloy. Figure 6(b) shows the thermomagnetization behavior of the five alloys between 200 and 395 K at a field of 70 koe. The martensitic transformation temperatures were lowered under the influence of the higher magnetic field. For alloy Co8, the transformation temperatures are * * T 255 K and T 285 K at 70 koe, which are approximately 30 K lower than those at 50 Oe. M A Figure 7 shows magnetization of the austenite at the martensitic transformation starting temperature M s ( M A Ms ), magnetization of the martensite at the martensitic transformation finishing temperature M f ( M M M f ), magnetization difference across the A M transformation ( M M M ), at 70 koe as functions of Co content in the alloys. It is M s M f CHAPTER 2 87

94 seen that M increased rapidly from 13 emu/g in Co0 to 117 emu/g in Co8, while A Ms M remained almost constant (~5 emu/g), thus resulting in an increasing M M M f across the martensitic transformation with increasing the Co content. The maximum in alloy Co8. M is 109 emu/g Magnetization (emu/g) Co Content (at %) A M M s M A M M f Figure 7. The magnetization of the austenite at M s ( M A Ms ), the magnetization of the martensite at M f ( M M M f ), and the magnetization difference across the martensitic A M transformation ( M M M ) at a field of 70 koe as functions of Co content in Mn 50 Ni 42-x Sn 8 Co x. M s 3.5 Magnetic coupling M f Figure 8 shows the atomic configuration in the unit cell of Mn 50 Ni 42-x Sn 8 Co x alloys. Illustration (a) represents the unit cell of the austenite with Mn-Mn-Ni-Sn stacking order (Hg 2 CuTi-type structure) and illustration (b) represents the unit cell of the martensite with Mn-Mn-Ni-Sn tetragonal structure. CHAPTER 2 88

95 Figure 8. Atomic configuration in the unit cell of Mn 50 Ni 42-x Sn 8 Co x alloys: (a) unit cell of the austenite with Mn-Mn-Ni-Sn stacking order (Hg 2 CuTi structure) and (b) unit cell of the martensite with Mn-Mn-Ni-Sn tetragonal structure. For Mn 2 YZ (Y: 3d elements; Z: III-V A group elements) compounds, it has been observed that Y elements with more valence electrons prefer to occupy A and C sites, whereas Y elements with fewer valence electrons have preference for B site occupancy [26]. For the Mn 50 Ni 42-x Sn 8 Co x alloys, Co substitutes Ni at C site, and the excess Ni atoms prefer to occupy A sites, displacing Mn(A) to Mn(D) site, as illustrated in Figure 8(a). Based on this rule, the atomic occupation for Mn 50 Ni 42-x Sn 8 Co x alloys can be written as [Mn(A)Ni(A)] 25 Mn(B) 25 [Co(C)Ni(C)] 25 [Sn(D)Mn(D)] 25. In the austenite (Figure 8(a)), the Mn(A) and Mn (B) are the nearest neighbors with a distance of 0.26 nm ( 3 a /4 ) in between. The nearest distance between Mn(B) and Mn(D) atoms is nm (a/2). It has been found that the magnetic interaction between Mn atoms changes from ferromagnetic to antiferromagnetic alignment when the Mn-Mn CHAPTER 2 89

96 distance reduces to below a critical value of approximately 0.30 nm [12]. Therefore, Mn(A)-Mn(B) is expected to form antiferromagnetic alignment, whereas Mn(B)-Mn(D) forms ferromagnetic alignment in the austenite. After the replacement of Mn(A) by Ni(A), there will be a significant increase of the net magnetic moment for the austenite, which is due to the reduction of the antiferromagnetic Mn(A)-Mn(B) coupling and the formation of the ferromagnetic Mn(B)-Mn(D) coupling. Co substitution for Ni at C site also provides magnetic moment contribution to some extent due to its larger magnetic moment (~1 µ B ) relative to that of Ni (~0.3 µ B ). The significant increase of M A Ms can also be attributed to the enlarged temperature window for the ferromagnetic austenite to develop, caused by Co addition. The M is A Ms measured at M s (upon cooling), as seen on the M-T curves (Figure 6(b)) of these alloys. It is also seen in Figure 3 that the transformation temperatures decreased significantly with increasing Co content. This means that the M values were actually taken at different A Ms temperatures for these alloys. Therefore, the increase of M with increasing Co content is largely due to the widening of the temperature window of the ferromagnetic austenite, in addition to the effect of Co alloying as aforementioned. Upon the martensitic transformation, the crystallographic transformation changes the crystal lattice in the unit cell, consequently altering the exchange coupling of the magnetic atoms. Upon transforming form the cubic austenite (Figure 8(a)) to the tetragonal martensite (Figure 8(b)), a and b axes shrink by 9.5% and c axis elongates by 15.7%. Through the transformation, the nearest distance between Mn(B)-Mn(D) decreases from nm (a/2, parent phase) to nm (a/2, martensite phase) which is below the critical distance for ferromagnetic coupling. This leads to the moments of Mn(B) and Mn(D) to change from in parallel alignment in the austenite to antiparrallel alignment in the martensite. The distance between Mn(A)-Mn(B) changed from nm ( 3 a /4) in the austenite to nm in the martensite, which causes no change to the magnetic alignment between Mn(B) and Mn(D). Exchange interaction between Mn(A) and Mn(B) is still A Ms CHAPTER 2 90

97 antiparrallel alignment. Therefore, the magnetic coupling of the martensite is expected to be ferrimagnetic. 3.6 Magnetic field induced martensitic transformation The magnetic driving force for a magnetic-field-induced martensitic phase transformation arises from the Zeeman Energy E Zeeman =µ 0 M H. The M for the present Mn 50 Ni 42-x Sn 8 Co x alloys has been significantly increased by substituting Co for Ni. To verify this increased M in regards to benefiting the field induced martensitic transformation, alloys Co6 and Co8 were magnetized isothermally at different temperatures, as shown in Figure 9. a Magnetization (emu/g) K 340 K 335 K 5 K Magnetic Field (koe) CHAPTER 2 91

98 b Magnetization (emu/g) 290 K 390 K 280 K 5 K Magnetic Field (koe) Figure 9. Isothermal magnetization behaviours of (a) alloy Co6 and (b) alloy Co8 at different temperatures. Figure 9(a) shows the magnetization behaviour of alloy Co6. At 5 K, the martensite shows low magnetization of 11 emu/g (at 70 koe). At 390 K, the austenite shows a typical soft ferromagnetic behaviour, with a saturation magnetization of 68 emu/g. At 335 K, the martensite shows a very low saturation magnetization (~2 emu/g) at below 20 koe. Upon increasing the magnetic field to above 50 koe, the magnetization increased rapidly, signifying the phase transformation from the martensite to the austenite. The maximum magnetization reached is 30 emu/g at 70 koe. This magnetization is much lower than the saturation magnetization of the austenite (~70 emu/g). This is obviously due to the fact that the transformation is incomplete at 335 K. At 340 K (8 K below T A ), the martensite starts to transform to the austenite at 30 koe and saturates at 72 emu/g at 7 T, indicating the completion of the reverse transformation. Upon removal of the external magnetic field, the magnetization decreases slowly at above 30 koe, and then quickly drops to 20 emu/g, indicating the occurrence of the forward martensitic transformation. Figure 9(b) shows the magnetization behaviour of alloy Co8. The magnetization behaviours of the martensite at 5 K and the austenite at 390 K are similar to those of Co6. At 280 K, the martensite starts to transform to austenite at the field of 30 koe upon magnetizing. The magnetization maximized at 75 emu/g at 70 koe, indicating the CHAPTER 2 92

99 incomplete magnetic-field-induced transformation. At 290 K (14 K below T A ), the martensite starts to transform to the austenite at 10 koe and saturates at ~100 emu/g at 70 koe, indicating the completion of the reverse transformation. Upon removal of the external magnetic field, the austenite remained saturated and did not transform back to the martensite. This is due to the large transformation hysteresis in Co8. The testing temperature of 290 K is well above the forward transformation temperature (T M= 278 K), thus resulting in the retained austenite after demagnetization. It is worth nothing that the complete field-induced transformation can be achieved at lower temperature below T A in Co8 (14 K below T A ) than that in Co6 (8 K below T A ), which is due to the larger ΔM across the transformation of Co8 relative to that of Co6. This indicates that the magnetic driving force is increased with increasing the Co content in Mn 50 Ni 42-x Sn 8 Co x alloys in a certain field, thus easier for obtaining a field-induced transformation. 4. Conclusions In this study, the effects of Co substitution for Ni on the martensitic transformation and magnetic behaviour of Mn 50 Ni 42-x Sn 8 Co x alloys were investigated. The experimental evidences and the discussions lead to the following conclusions: (1) The Mn 50 Ni 42-x Sn 8 Co x alloys exhibit a martensitic transformation from an Hg 2 CuTi-type austenite to a non-modulated tetragonal martensite. The martensitic transformation temperatures were found to decrease significantly with increasing Co substitution for Ni, due to the decreasing e/a ratio in the alloys. The enthalpy and entropy changes of the transformation are both found to decrease with increasing Co addition. (2) The magnetization of the austenite is significantly increased from 13 emu/g in the Co0 alloy to 117 emu/g in the Co8 alloy, whereas that of the martensite remains unchanged at ~5 emu/g. Consequently, magnetization difference between the austenite and the martensite increases significantly with increasing Co substitution for Ni. The largest ΔM for the martensitic transformation obtained is 109 emu/g in alloy Co8. CHAPTER 2 93

100 (3) The increased magnetization of the austenite is attributed to two reasons: (i) higher magnetic moment contribution of Co relative to Ni, and (ii) widening of the temperature window for ferromagnetic austenite to magnetize. (4) The low magnetization of the martensite is due to the significantly shortened distance between Mn(B)-Mn(D), which leads to the antiparallel alignment of the magnetic moments of neighbouring Mn atoms in the tetragonal martensitic structure. (5) The magnetic-field-induced martensitic transformation from ferrimagnetic martensite to ferromagnetic austenite was successfully induced in alloys Co6 and Co8 under a field within the range of 30~70 koe. Acknowledgements The authors wish to acknowledge the financial supports by the Department of Innovation Industry, Science and Research of the Australian Government in ISL Grant CH070136, and by National Natural Science Foundation of China in Grant No CHAPTER 2 94

101 References [1] Murray SJ, Marioni M, Allen SM, O'Handley RC, Lograsso TA. Applied Physics Letters 2000;77:886. [2] Sozinov A, Likhachev AA, Lanska N, Ullakko K. Applied Physics Letters 2002;80:1746. [3] Mullner P, Chernenko VA, Kostorz G. Materials Science and Engineering A 2004; :965. [4] Okamoto N, Fukuda T, Kakeshita T. Materials Science and Engineering: A 2008; :306. [5] Sutou Y, Imano Y, Koeda N, Omori T, Kainuma R, Ishida K, Oikawa K. Applied Physics Letters 2004;85:4358. [6] Kainuma R, Imano Y, Ito W, Sutou Y, Morito H, Okamoto S, Kitakami O, Oikawa K, Fujita A, Kanomata T, Ishida K. Nature 2006;439:957. [7] Kainuma R, Imano Y, Ito W, Morito H, Sutou Y, Oikawa K, Fujita A, Ishida K, Okamoto S, Kitakami O, Kanomata T. Applied Physics Letters 2006;88: [8] Yu SY, Ma L, Liu GD, Liu ZH, Chen JL, Cao ZX, Wu GH, Zhang B, Zhang XX. Applied Physics Letters 2007;90: [9] Ito W, Imano Y, Kainuma R, Sutou Y, Oikawa K, Ishida K. Metallurgical and Materials Transactions A: Physical Metallurgy and Materials Science 2007;38:759. [10] Liu J, Scheerbaum N, Hinz D, Gutfleisch O. Applied Physics Letters 2008;92: [11] Sasioglu E, Sandratskii LM, Bruno P. Physical Review B 2004;70: [12] Yamada T, Kunitomi N, Nakai Y, Cox DE, Shirane G. Journal of the Physical Society of Japan 1970;28:615. [13] Liu GD, Dai XF, Yu SY, Zhu ZY, Chen JL, Wu GH, Zhu H, Xiao JQ. Physical Review B: Condensed Matter 2006;74: CHAPTER 2 95

102 [14] Luo H, Liu G, Feng Z, Li Y, Ma L, Wu G, Zhu X, Jiang C, Xu H. Journal of Magnetism and Magnetic Materials 2009;321:4063. [15] Chakrabarti A, Barman SR. Applied Physics Letters 2009;94: [16] Wu Z, Liu Z, Yang H, Liu Y, Wu G. Applied Physics Letters 2011;98: [17] Krenke T, Moya X, Aksoy S, Acet M, Entel P, Manosa L, Planes A, Elerman Y, Yucel A, Wassermann EF. Journal of Magnetism and Magnetic Materials 2007;310:2788. [18] Khovailo VV, Oikawa K, Abe T, Takagi T. Journal of Applied Physics 2003;93:8483. [19] Wu SK, Yang ST. Materials Letters 2003;57:4291. [20] Krenke T, Acet M, F. Wassermann E, Moya X, Manosa L, Planes A. Physical Review B 2006;73: [21] Krenke T, Acet M, F. Wassermann E, Moya X, Manosa L, Planes A. Physical Review B 2005;72: [22] Ma L, Zhang HW, Yu SY, Zhu ZY, Chen JL, Wu GH, Liu HY, Qu JP, Li YX. Applied Physics Letters 2008;92: [23] Keiichi K, Kazuo W, Takeshi K, Ryosuke K, Katsunari O, Kiyohito I. Applied Physics Letters 2006;88: [24] Wu Z, Liu Z, Yang H, Liu Y, Wu G, Woodward RC. Intermetallics 2011;19:445. [25] Liu GD, Chen JL, Liu ZH, Dai XF, Wu GH. Applied Physics Letters 2005;87: [26] Liu GD, Dai XF, Liu HY, Chen JL, Li YX, Xiao G, Wu GH. Physical Review B 2008;77: CHAPTER 2 96

103 CHAPTER 3. Increasing ductility of Ni- Mn-based alloys Paper 4 Effect of Fe addition on the martensitic transformation behaviour, magnetic properties and mechanical performance of Ni 50 Mn 38-x In 12 Fe x polycrystalline alloys Zhigang Wu 1, Zhuhong Liu 2, Hong Yang 1 and Yinong Liu 1 1 School of Mechanical and Chemical Engineering, The University of Western Australia, Crawley, WA 6009, Australia 2 Department of Physics, University of Science and Technology Beijing, Beijing , China Abstract CHAPTER 3 97

104 This study investigated the effect of Fe substitution for Mn on the transformation, magnetic and mechanical behaviours of Ni 50 Mn 38-x In 12 Fe x (x=0, 3, 4, 5, 6) alloys. These alloys show a martensitic transformation from a B2 austenite to an orthorhombic martensite at above the room temperature. Substitution of Fe for Mn at above 3 at.% introduced a fcc γ phase in the microstructure, the amount of which increased with increasing the Fe addition. The formation of γ phase influenced the composition of the matrix phase, particularly the Mn and In contents, leading to a series of changes in alloy properties. The e/a ratio of the matrix phase decreased rapidly with increasing Fe addition, resulting in the decrease of martensitic transformation temperature and enthalpy change. Fe addition also effectively weakens the antiferromagnetic ordering of the austenite in the matrix phase, leading to the increase of magnetisation difference across the martensitic transformation. The compressive strength and ductility appear to optimise at 4~5 at.% Fe addition, reaching 770 MPa and 14.3 %, respectively. The relative shape memory effect decreased from 94 % to 37 % after 4 at.% Fe addition. Keywords: Shape memory alloy; martensitic transformation; magnetic properties; intermetallics 1. Introduction Ferromagnetic shape memory alloys Ni-Mn-Z (Z=In, Sn, Sb) have been widely investigated in the past few years as potential candidates for magnetic actuation. The large difference in magnetic state between the austenite and martensite produces high magnetic driving force for magnetic-field-induced martensitic transformation [1]. Substitution of Ni by Co is found to increase the magnetic ordering of the austenite and to decrease that of the martensite, further increasing the magnetisation difference of the martensitic transformation in Ni-Mn-Z (Z=Ga, Al, In, Sn, Sn, Sb) alloys [2-8]. As a result, >1% strain has been realised in Ni 45 Co 5 Mn 36.7 In 13.3 and Ni 43 Co 7 Mn 39 Sn 11 alloys [2, 3], demonstrating the promise of the alloys for actuation applications in smart systems. Unfortunately, the intrinsic brittleness of these intermetallic compounds severely hinders their engineering application. To date, it has not been possible to process polycrystalline Ni-Mn-Z (Z=In, Sn, Sb) ferromagnetic shape memory alloys using CHAPTER 3 98

105 conventional methods. It is known that the introduction of a ductile second phase is helpful in improving the ductility of the alloys, as initially discovered in Ni-Fe-Ga [9], Co-Ni-Ga [10] and Co-Ni-Al [11] alloys. Later, adding Fe and Co was found to form a ductile phase in Ni-Mn-Z (Z=Ga, In, Sn) [12-17], accordingly increasing the ductility of these alloys. Apart from the improved ductility, addition of a fourth element to the ternary Ni- Mn-Z (Z=Ga, In, Sn, Sb) alloys alters the matrix composition, which causes a number of changes in the structure, thermal and magnetic properties. The effect of Fe substitution for Mn in Ni-Mn-Sn alloys has been found to decrease the martensitic transformation temperatures and increase the Curie transition temperatures of both the austenite and the martensite [18, 19]. In our previous study on Ni 50 Mn 40-x Sn 10 Fe x alloys, Fe substitution for Mn changed the composition of the matrix phase in addition to forming the phase [20]. It is revealed that changes in Mn and Sn contents in the matrix phase are the actual reasons for the property changes. This study investigated the effects of the formation of the phase caused by Fe substitution for Mn in Ni 50 Mn 38-x In 12 Fe x. The addition of Fe is expected to alter the composition of the matrix phase, thus affecting the magnetic state and valence electron number of the alloys. Besides concerning on changes of the physical properties influenced by the phase, mechanical performance and shape memory effect were also investigated in Ni 50 Mn 38-x In 12 Fe x alloys. 2. Experimental procedures Bulk ingots of polycrystalline Ni 50 Mn 38-x In 12 Fe x (x=0, 3, 4, 5, 6) alloys were prepared by means of arc melting in argon atmosphere using high purity Ni (99.99 at.%), Mn (99.99 at.%), In (99.99 at.%) and Fe (99.95 at.%). The samples are referred to as Fe0, Fe3, Fe4, Fe5 and Fe6, based on the atomic percentage of Fe addition in the alloys. The button shaped ingots were heat treated at 1173 K in vacuum for homogenisation followed by furnace cooling to room temperature. Transformation behaviour of the alloys was studied by means of differential scanning calorimetry (DSC) using a TA Q10 DSC instrument with a cooling/heating rate of 10 K/min. Phase identification and crystal structures were determined by means of X-ray powder diffraction using a Siemens D5000 instrument with Cu-Kα radiation and transmission electron microscopy (TEM) using a Jeol CHAPTER 3 99

106 2100 instrument. Microstructures of the samples were studied with TEM and scanning electron microscopy (SEM) using a Zeiss 1555 instrument. The compositions were determined by means of X-ray energy dispersive spectrometry (EDS) equipped on SEM. Magnetic properties were studied using a superconducting quantum interference device magnetometer (SQUID). 3. Results and discussion 3.1 Microstructure and alloy composition Figure 1 shows back-scattered electron (BSE) micrographs of the microstructures of the Ni 50 Mn 38-x In 12 Fe x (x=0, 3, 4, 5, 6) alloys after homogenisation treatment. (a) (b) (c) (c) grain boundaries grain boundaries CHAPTER 3 100

107 (e) Figure 1. Back-scattered electron images of the Ni 50 Mn 38-x In 12 Fe x alloys: (a) Fe0, (b) Fe3, (c) Fe4, (d) Fe5, and (e) Fe6 alloys. The BSE microstructures of the alloys were examined without etching. The Fe0 and Fe3 samples (micrograph (a) and (b)) showed uniform single phase structure, without sign of a second phase. The black spots are solidification shrinkage pores formed during ingot casting. The Fe4, Fe5 and Fe6 samples showed a continuous matrix in light contrast and dispersed γ phase particles in dark contrast. The volume fraction of the γ phase increased with increasing Fe addition. It is also seen that the γ phase particles tend to form along the grain boundaries in Fe4 and Fe5 alloys, as shown in micrograph (c) and (d). The alloy Fe6 showed distinctive texture of the γ phase compared to Fe4 and Fe5, with straight and elongated γ phase grains. Table 1 shows compositions of the phases in the samples as determined by quantitative EDS analysis. It is seen that the matrix phase of the Fe-doped alloys contained about 49 at.% Ni. The content of Mn decreased continuously from 37.8 to 32.5 % with increasing Fe addition from 2.9 to 3.8 % in the matrix. The content of In was also found to increase from 12.8 to 14.8 %. The γ phase is effectively a Ni-Mn-Fe alloy containing a small amount of In (~1.1 at.%). The volume fraction of the γ phase is determined by image analysis from the SEM micrographs using Image J. Table 1. Composition, e/a ratio and γ proportion of Ni 50 Mn 38-x In 12 Fe x (x=0, 3, 4, 5, 6) alloys. CHAPTER 3 101

108 Matrix (at.%) Ni Mn In Fe e/a ratio The γ phase (at.%) Ni Mn In Fe γ (%) X= X= X= X= X= The increase of In content in the matrix phase is apparently related to the increase of the fraction of γ phase, which contains very little In. The valence electron concentrations per atom (e/a ratio) of the matrix phase was calculated using the compositions obtained from EDS analysis from the sum of s, p and d electrons for Mn (7), Ni (10), Fe (8) and In (3). It is obvious that the e/a ratio decreases with increasing In and decreasing Mn contents of the alloys.figure 2 shows the effect of Fe addition on the matrix phase composition (graph (a)) and the e/a ratio (graph (b)) of the alloys. With increasing Fe addition, both Fe and In contents in the matrix phase increased, and the Mn content decreased. The e/a ratio of the alloy decreased continuously with increasing Fe addition in the alloys, shown in graph (b). The decrease of the e/a ratio is obviously related to the composition change in the matrix phase caused by the formation of the phase. More specifically, the decrease of Mn (7 valence electrons) and increase of In (3 valence electrons) contents are the main reasons for the decrease of the e/a ratio, though Fe (8 valence electrons) content slightly increased as well in the matrix phase. It is worth noting that there are two different negative dependencies of the e/a ratio on Fe addition, which are per at.% for Fe 4 and for 4 Fe 6. This indicates that the e/a ratio of the matrix phase decreased more rapidly when Fe addition is above 4 at.% in the alloys. CHAPTER 3 102

109 In, Mn Content (at.%) Mn Fe In (a) Fe Content (at.%) Fe Addition (at.%) 7.97 (b) e/a Ratio of Matrix Phase transformation disappears Fe Addition (at. %) Figure 2. Effects of Fe addition on (a) element concentrations and (b) e/a ratio of the matrix phase of the Ni 50 Mn 38-x In 12 Fe x alloys. 3.2 Martensitic transformation behaviour Figure 3 presents DSC curves of the Ni 50 Mn 38-x In 12 Fe x alloys. It is seen that the martensitic transformation behaviour evolved progressively with increasing Fe addition in these alloys. CHAPTER 3 103

110 Fe6 Heat Flow 0.2 w/g Fe5 Fe4 Fe3 T M T A Fe Temperature (K) Figure 3. DSC measurements of martensitic transformation behaviour of the Ni 50 Mn 38- xin 12 Fe x alloys. The martensitic transformation is clearly observed for Fe0, Fe3 and Fe4 alloys, and the transformation temperatures decreased with increasing Fe addition in these alloys. However, no transformation was detected in Fe5 and Fe6. Figure 4 shows the effect of e/a ratio on transformation temperature T o and transformation enthalpy change ΔH. T o is defined as T o =1/2(T M +T A ), where T M and T A are the peak temperatures of the forward and the reverse transformations, and ΔH is obtained from the forward transformation. It is seen that the transformation temperatures of Fe0, Fe3 and Fe4 increased practically linearly with increasing the e/a ratio of the alloys. This observation is consistent with the findings of the effect of e/a ratio on transformation temperatures in Ni-Mn-Z (Al, Ga, In, Sn and Sb) alloys [21-23]. It is also evident that the ΔH increased with increasing e/a ratio of the matrix. CHAPTER 3 104

111 T o (K) 440 H Fe3 320 Fe4 280 Fe e/a Ratio T o H (J/g) Figure 4. Effects of Fe addition on transformation temperature T o =(T M +T A )/2 and transformation enthalpy change ΔH. The arrow pointing to the e/a ratio axis indicates the threshold e/a ratio value below which the martensitic transformation is expected to vanish. Extending ΔH curve to zero tentatively defines the threshold value of e/a ratio below which the martensitic transformation is expected to vanish. The threshold value, as indicated by the arrow in Figure 4, is estimated to be e/a = The e/a ratio value corresponds to an Fe addition of 4.5 at.% estimated from Figure 2 (b). For Fe addition of more than 4.5 at.% in the system, no martensitic transformation is expected. This explains the disappearance of martensitic transformation in Fe5 and Fe Crystal structure Figure 5 shows the crystal structures of the Ni 50 Mn 38-x In 12 Fe x alloys examined by X-ray diffraction at room temperature. The non-modulated orthorhombic martensite structure can be observed in Fe0, Fe3 and Fe4 alloys. The observation of martensitic phase is consistent with the results obtained from DSC (Figure 3), which indicate the martensitic state at room temperature of these alloys. Apart from the martensite structure, diffraction peaks of the fcc phase can also be identified in the spectrum of Fe4 alloy, which is consistent with its microstructure (Figure 1 (c)). Alloy Fe5 exhibits a more complicated case, showing a mixed structure of the orthorhombic martensite, the bcc austenite and the CHAPTER 3 105

112 fcc γ phase. Alloy Fe6 presents a two-phase structure, with the bcc austenite and the fcc γ phase. A M Fe6 X-ray Intensity Fe5 Fe4 Fe3 Fe ( o ) Figure 5. X-ray spectra of the crystal structure of the Ni 50 Mn 38-x In 12 Fe x alloys at room temperature. The phases are identified with labeled symbols: ( ) represents phase, A ( ) represents the austenite phase and M ( ) represents the martensite phase. The lattice parameters of the γ phase in Fe4, Fe5 and Fe6 are very close, with an average value of a=0.366 nm at room temperature. The lattice parameter of the bcc austenite is determined to be a= nm for Fe5 and Fe6 alloys. The lattice parameter of the orthorhombic martensite is determined to be a= nm, b= nm, c= nm for Fe0, Fe3 and Fe4 alloys. Figure 6 shows TEM observation of the microstructure and crystal structure of Fe4 and Fe5 alloys at room temperature. Micrograph (a) shows a bright field image of Fe4. Two phases are present in the microstructure, which are the matrix phase in dark contrast and the phase in light contrast. Selected area diffraction pattern (SADP) from area A of the matrix phase is shown in Figure 6(b), which can be indexed to B2 structure along its [001] zone axis. Presence of (010) reflection is the evidence of the superlattice B2 structure. Figure 6(c) shows the SADP of area B of the phase. The pattern is indexed to fcc system along [011] zone axis. Fe5 also exhibits two phases in the microstructure, including the matrix and the phases. The SADPs obtained from area C CHAPTER 3 106

113 and D are presented in Figure 6(e) and Figure 6(f), respectively. Similarly, the matrix phase shows a B2 structure austenite and the phase is confirmed to be fcc structure. (a) 0.5 μm Matrix phase A phase B (b) _ (c) _ _ [001] [011] (d) Matrix phase C (e) _ phase D (f) _ _ 111 [001] 0.5 μm 200 [011] Figure 6. Room temperature TEM micrographs and selected area diffraction patterns (SADPs) of the matrix and phase of Fe4 and Fe5 alloys. (a) bright field image of Fe4, (b) and (c) SADPs obtained from areas A and B of Fe4, (d) bright field image of Fe5, (e) and (f) SADPs obtained from areas C and D of Fe Magnetic properties CHAPTER 3 107

114 In Ni 50 Mn 25+x In 25-x alloys, it is known that magnetic moments of the excessive Mn atoms occupying the In sites align in parallel formation with respect to those of the Mn atoms at the Mn sites, thus introducing extra ferromagnetic coupling between the two, leading to increase in the saturation magnetisation with increasing Mn substitution for In [24]. In Ni 50 Mn 38-x In 12 Fe x alloys, the concentration of Mn can be up to 38 at.%. That means 13 at.% of Mn occupy the In sites. In this case, it is expected that ferromagnetic interactions of the off-stoichiometric Ni 50 Mn 38-x In 12 Fe x are increased compared to the stoichiometric Ni 50 Mn 25 In 25 alloy. Figure 7 shows the M(T) curves of the Ni 50 Mn 38-x In 12 Fe x alloys in a small magnetic field of 50 Oe. The sample was first cooled down from 390 to 10 K inside the instrument without applying a magnetic field. A 50 Oe field was applied at 10 K and then the magnetisation of the sample was measured upon heating to 390 K. Subsequently, without removing the external field, the measurement was made upon cooling to 10 K. Magnetisation (emu/g) M 0.10 T C (a) Fe0 T C A Magnetisation (emu/g) T C M (b) Fe3 T C A Magnetisation (emu/g) Temperature (K) 6 M 5 T C 4 3 (c) Fe4 T C A Temperature (K) Magnetisation (emu/g) Temperature (K) Temperature (K) (d) Fe5 T C A CHAPTER 3 108

115 Magnetisation (emu/g) (e) Fe6 T C A Temperature (K) Figure 7. Thermomagnetisation behaviour of Ni 50 Mn 38-x In 12 Fe x alloys in a field of 50 Oe. As seen in Figure 7(a), Fe0 presents a very small magnetisation at 10 K when the 50 Oe field was applied, at 0.05 emu/g. The magnetisation of the heating curve showed a broad hump at ~70 K. The abrupt decrease of magnetisation upon heating at ~300 K is attributed to the Curie transition of the austenite, of the small amount of residual austenite in the matrix [22]. This temperature is denoted cooling curve at below A T C. The heating curve did not follow the A T C, apparently due to the application of the magnetic field on this second cooling. The rapid increase of magnetisation at below 70 K upon cooling is attributed to the Curie transition of the martensite, denoted M T C =70 K. The magnetisation behaviour of Fe3 is similar to that of Fe0. The T A temperature is ~355 K (Figure 3), and similar to Fe0, the Curie transition at ~315 K corresponds to that of the remnant austenite in this alloy. The Curie transition of the martensite is determined to be M T C =100 K. Fe4 showed a similar thermomagnetic behaviour to Fe3. The T M C are determined to be about 155 K and 320 K for Fe4 alloy, respectively. Fe5 and Fe6 showed different magnetisation behaviour to the previous three samples. The magnetisation was fairly constant at below or above T A C and A T C. These two samples showed no martensitic transformation within the testing temperature range, thus the absence of the Curie transition of the martensite on the curves. It is seen that the T A C temperature increased slightly from 300 to 320 K with the increase of Fe addition from 0 to 6 at.%. However, the T M C temperature was more CHAPTER 3 109

116 significantly affected by Fe addition, from 70 to 155 K with increasing Fe addition up to 4 at.%. For alloy Fe4, the martensitic transformation temperature T M (319 K) is very close to the Curie transition temperature T A C (320 K). This implies that the martensitic transformation overlaps with the Curie transition of the austenite, i.e., the phase transformation is hidden on the M(T) curve. To reveal the martensitic transformation via the M(T) measurement, a higher magnetic field is applied, which lowers T M without affecting A T C, thus delaying the A M transformation to after the completion of the Curie transition of the austenite [8]. Figure 8 shows the thermomagnetisation behaviour of alloy Fe4 in a high field of 70 koe. It is seen that the magnetisation changed abruptly at ~300 K with a hysteresis of 10 K between the heating and cooling curves. This is obviously due to the martensitic transformation of this alloy. The forward transformation temperature is estimated to be * T M =300 K, which is 10 K below that obtained from the DSC measurement, obviously due to the effect of the applied magnetic field. It is seen that the ferromagnetic interactions in the austenite are much stronger than that in martensitic phase, which leads to 40 emu/g magnetisation difference across the transformation. Magnetisation (emu/g) Fe4 H=7 T T * M Temperature (K) CHAPTER 3 110

117 Figure 8. Thermomagnetisation behaviour of Fe4 alloy in a field of 70 koe. To reveal the magnetic structure of the austenite and martensite, M(H) measurements were carried out of the Ni 50 Mn 38-x In 12 Fe x alloys at 5 K. The measurements are shown in Figure 9. At 5 K, the matrix phase is in martensitic state for Fe0, Fe3 and Fe4, whereas it is in austenitic state for Fe5 and Fe6. It is seen that Fe5 and Fe6 showed typical soft ferromagnetic behaviour, with saturation magnetisation of 105 and 108 emu/g, respectively. The M(H) data of Fe4 also showed the characteristics of ferromagnetic ordering in its martensitic state, but with a much reduced magnetisation of 47 emu/g. The initial slope of M(H) curve of Fe3 indicates the short range ferromagnetic correlations together with antiferromagnetic exchange in the martensitic phase. The coexistence of the ferromagnetic and antiferromagnetic structures can also be seen as the splitting between the heating and cooling M(T) curves (Figure 7b). In Ni 50 Mn 38-x In 12 Fe x alloys, extra Mn atoms occupy In sites, forming a ferromagnetic coupling between the Mn(Mn site) and Mn(In site) atoms in the austenitic phase [24]. Through the martensitic transformation, the distance between the Mn(Mn site)-mn(in site) decreases and favours antiferromagnetic interaction in the martensitic phase. However, the Mn atoms at the Mn site still form ferromagnetic interaction in the martensitic phase. Therefore, the inhomogeneous magnetic structure is common to observe in the martensitic state of the Mn-rich Ni-Mn- Z(Z=In,Sn,Sb) alloys [21, 22, 25]. The magnetisation of Fe0 showed nearly linear dependence on the applied magnetic field up to 40 koe. The linearity of M(H) data suggests that it is antiferromagnetic in the martensitic state of this alloy. The saturation magnetisations are determined to be 13.6 and 27.7 emu/g for Fe0 and Fe3, respectively. CHAPTER 3 111

118 Magnetisation (emu/g) T=5 K Fe6 Fe5 Fe4 Fe3 Fe0 A M Magnetic Field (koe) Figure 9. Magnetisation measurements of the Ni 50 Mn 38-x In 12 Fe x alloys at 5 K. A represents austenitic state and M represents martensitic state. It is seen that the saturation magnetisation increased with increasing the Fe addition in these alloys. Based on the composition determination, it is known that the Mn concentration decreased in the matrix phase with more Fe addition. With high concentration of Mn, antiferromagnetic exchange is expected to dominate, as the composition is close to the antiferromagnet Ni 50 Mn 50. This explains the linearity of M(H) data in the martensitic state of Fe0 which has the highest Mn concentration (37.8 at.% Mn) among the Ni 50 Mn 38-x In 12 Fe x alloys. With the decrease of Mn content (increase of Fe addition), the antiferromagnetic ordering weakens, and gradually the long-range ferromagnetic ordering forms in the matrix phase. 3.5 Mechanical properties Figure 10 shows compressive deformation behaviour of the alloys, with (a) showing the stress-strain curves and (b) showing the effect of Fe content on the maximum stress and strain. CHAPTER 3 112

119 Fe5 Fe4 Stress (MPa) Fe0 Fe6 100 Fe Strain (%) Maximum Compressive Stress (MPa) Maximum Compressive Strain (%) Fe Content (at.%) Figure 10. Compressive deformation behaviour of Ni 50 Mn 38-x In 12 Fe x alloys; (a) stress-strain curves; (b) maximum compressive stress and strain as functions of Fe content. Fe0 and Fe3 showed obvious stress plateau corresponding to the reorientation of martensite variants. However, stress-strain curves for Fe4, Fe5 and Fe6 suggest that the deformation was mainly due to the dislocation mechanisms rather than the martensite detwining process, since the stress plateau corresponding to the reorientation of martensite variants did not appear in the stress-strain curves. It is also seen that the compressive strength first increased and then decreased with increasing Fe content, reaching a maximum of 770 MPa at 5 at.% Fe. The compressive strain showed similar tendency with increasing Fe addition. The maximum strain reached was 14.3 % at 3 at.% Fe. CHAPTER 3 113

120 These observations are in contrast to those made in Ni 50 Mn 34 In 16-y Fe y and Ni 50 Mn 34 In 16-y Co y alloys, which show continuous increase of strength and strain with increasing the amount of Fe or Co (or the amount of the phase) in the alloys [17, 26]. The decrease of the compressive strength and strain at high levels of Fe addition in this case is attributed to the particular morphology of the phase. It is seen in Figure 1 that the phase in Fe6 exists as thin and elongated grains in parallel arrays in the matrix phase. This morphology is detrimental for the strength and ductility of this material. More importantly, the grain boundaries of matrix phase of Fe6 are not covered nicely by phase grains like those in Fe4 and Fe5 alloys [27], evidenced by the microstructural observations of these alloys shown in Figure 1. Therefore, Fe6 showed more intergranular cracking under compression tests (Figure 11(e)), thus presenting relative low strength and ductility compared to Fe4 and Fe5. It should be noted that the ductility of Fe0 and Fe3 is associated with the martensite variant reorientation, and there is not much real plastic deformation before failure. Good ductility is actually given by Fe4 and Fe5, which contain 6.8 % and 11.5 % of the phase, respectively. Figure 11 shows the fracture morphologies of the Ni 50 Mn 38-x In 12 Fe x alloys after compressive testing. It is seen from micrographs (a) and (b) that Fe0 and Fe3 fractured via typical intergranular cracking. With the increase of Fe addition, the fracture becomes a mixture of intergranular cracking and transgranular cracking. Some pull-out holes are also evident due to the pulling out of the phase particles, implying improvement of the ductility of Fe4 and Fe5 alloys, shown in micrographs (c) and (d). The cracks in Fe6 mainly formed along the elongated grains in the microstructure, as seen in micrograph (e), leading to the decrease in strength and ductility. CHAPTER 3 114

121 (a) (b) 100 µm 100 µm (c) (d) 20 µm 20 µm (e) 20 µm Figure 11. SEM micrographs of the fractured surfaces of Ni 50 Mn 38-x In 12 Fe x alloys after compressive testing. (a) Fe0, (b) Fe3, (c) Fe4, (d) Fe5 and (e) Fe6. Figure 12 shows the compressive stress-strain curves at room temperature and the strain recovery after heating to above the T A temperature of Fe3 and Fe4 alloys. Fe3 was deformed to 7.3 % strain. It exhibited a spontaneous recovery of 2.5 % upon unloading, CHAPTER 3 115

122 leaving a residual strain of 4.8 %. The arrowed curves below the x-axis represent the strain recoveries upon heating to 500 K for 5 min. The recovered strain is 4.5 % and the recovery ratio is 94 % for Fe3. Fe4 had a compressive strain of 6.7 %, and generated a residual strain of 2.4 % after unloading. After heating to 500 K for 5 min, the strain recovery was 0.9 %, giving the total recovery ratio of 37 % for Fe4. It is also worth noting that the critical stress to initiate the detwining of the martensite variants is about 130 MPa in Fe3 which is much lower than that of 330 MPa in Fe4. The disappearance of stress plateau in Fe4 is due to the strain hardening effect caused by the phase. Accordingly, it can be envisaged that the shape memory effect of the two-phase Fe4 alloy is poor due to the presence of the phase, which does not participate in the reversible martensitic transformation. 200 (a) Fe3 500 (b) Fe Stress (MPa) 100 Stress (MPa) SME SME Strain (%) Strain (%) Figure 12. Shape memory effect of Fe3 and Fe4 alloys. 4. Conclusions This study investigated the effects of Fe addition for Mn on the properties of Ni 50 Mn 38 In 12. The main findings may be summarised as following: (1) Fe substitution for Mn in Ni 50 Mn 38-x In 12 Fe x alloys at above 3 at.% causes formation of an fcc phase. The phase is a Ni-Mn-Fe solid solution phase with small amount of In dissolved. Formation of phase results in the decrease of Mn and increase of In contents of the matrix phase. Consequently, the e/a ratio of the matrix phase decreases with increasing Fe addition. CHAPTER 3 116

123 (2) The critical temperature (T o ), the enthalpy change (ΔH) of the martensitic transformation decreased with increasing Fe addition, due to the decrease of e/a ratio of the matrix phase caused by the formation of the phase. A threshold value of e/a ratio is identified at ΔH=0 to be 7.948, below which no martensitic transformation is expected in Ni 50 Mn 38-x In 12 Fe x alloys. This value corresponds to Fe substitution of 4.5 at.% for Mn. (3) The Curie temperature of the martensite ( T M C ) increased rapidly from 70 to 155 K with Fe addition to 4 at.%, whereas that of the austenite ( T ) increased slightly from 300 to 320 K with Fe addition to 6 at.%, The austenite shows much stronger ferromagnetic characteristic relative to that of the martensite. The ferromagnetic ordering of the martensite was enhanced with increasing Fe addition, due to the reduced content of antiferromagnetically coupled Mn in the martensitic phase. (4) Compressive stress and strain did not simply increase with increasing amount of the phase. Good compressive strength and ductility are exhibited by Fe4 and Fe5 alloys. (5) The shape memory effect decreased significantly with the introduction of phase. The alloy without phase in the microstructure (Fe3) showed a shape memory effect of 94 %, while the alloy containing 6.8 % phase (Fe4) presented a shape memory effect of 37 %. Acknowledgement A C The authors wish to acknowledge the financial supports by the Department of Innovation Industry, Science and Research of the Australian Government in ISL Grant CH070136, and by National Natural Science Foundation of China in Grant No CHAPTER 3 117

124 Reference [1] Sutou Y, Imano Y, Koeda N, Omori T, Kainuma R, Ishida K, Oikawa K. Applied Physics Letters 2004;85:4358. [2] Kainuma R, Imano Y, Ito W, Morito H, Sutou Y, Oikawa K, Fujita A, Ishida K, Okamoto S, Kitakami O, Kanomata T. Applied Physics Letters 2006;88: [3] Kainuma R, Imano Y, Ito W, Sutou Y, Morito H, Okamoto S, Kitakami O, Oikawa K, Fujita A, Kanomata T, Ishida K. Nature 2006;439:957. [4] Kainuma R, Ito W, Umetsu RY, Oikawa K, Ishida K. Applied Physics Letters 2008;93: [5] Yu SY, Cao ZX, Ma L, Liu GD, Chen JL, Wu GH, Zhang B, Zhang XX. Applied Physics Letters 2007;91: [6] Yu SY, Ma L, Liu GD, Liu ZH, Chen JL, Cao ZX, Wu GH, Zhang B, Zhang XX. Applied Physics Letters 2007;90: [7] Wu Z, Liu Z, Yang H, Liu Y, Wu G. Applied Physics Letters 2011;98: [8] Ito W, Imano Y, Kainuma R, Sutou Y, Oikawa K, Ishida K. Metallurgical and Materials Transactions A: Physical Metallurgy and Materials Science 2007;38:759. [9] Masdeu F, Pons J, Santamarta R, Cesari E, Dutkiewicz J. Materials Science and Engineering: A 2008; :101. [10] Liu J, Xie H, Huo Y, Zheng H, Li J. Journal of Alloys and Compounds 2006;420:145. [11] Tanaka Y, Oikawa K, Sutou Y, Omori T, Kainuma R, Ishida K. Materials Science and Engineering: A 2006; :1054. [12] Yang S, Ma Y, Jiang H, Liu X. Intermetallics 2011;19:225. [13] Ma Y, Yang S, Liu Y, Liu X. Acta Materialia 2009;57:3232. [14] Xin Y, Li Y, Chai L, Xu H. Scripta Materialia 2007;57:599. [15] Wang HB, Sui JH, Liu C, Cai W. Materials Science and Engineering: A 2008;480:472. [16] Feng Y, Jiehe S, Zhiyong G, Wei C. International Journal of Modern Physics B 2010;23:1803. CHAPTER 3 118

125 [17] Feng Y, Sui JH, Gao ZY, Zhang J, Cai W. Materials Science and Engineering A 2009;507:174. [18] Passamani EC, Xavier F, Favre-Nicolin E, Larica C, Takeuchi AY, Castro IL, Proveti JR. Journal of Applied Physics 2009;105: [19] Fukushima K, Sano K, Kanomata T, Nishihara H, Furutani Y, Shishido T, Ito W, Umetsu RY, Kainuma R, Oikawa K, Ishida K. Scripta Materialia 2009;61:813. [20] Wu Z, Liu Z, Yang H, Liu Y, Wu G, Woodward RC. Intermetallics 2011;19:445. [21] Krenke T, Acet M, F. Wassermann E, Moya X, Manosa L, Planes A. Physical Review B 2005;72: [22] Krenke T, Acet M, F. Wassermann E, Moya X, Manosa L, Planes A. Physical Review B 2006;73: [23] Krenke T, Moya X, Aksoy S, Acet M, Entel P, Manosa L, Planes A, Elerman Y, Yucel A, Wassermann EF. Journal of Magnetism and Magnetic Materials 2007;310:2788. [24] Kanomata T, Yasuda T, Sasaki S, Nishihara H, Kainuma R, Ito W, Oikawa K, Ishida K, Neumann KU, Ziebeck KRA. Journal of Magnetism and Magnetic Materials 2009;321:773. [25] Khan M, Dubenko I, Stadler S, Ali N. Journal of Physics-Condensed Matter 2008;20: [26] Feng Y, Sui JH, Gao ZY, Dong GF, Cai W. Journal of Alloys and Compounds 2009;476:935. [27] Ishida K, Kainuma R, Ueno N, Nishizawa T. Metallurgical and Materials Transaction A 1991;22:441. CHAPTER 3 119

126 Paper 5 Metallurgical origin of the effect of Fe doping on the martensitic and magnetic transformation behaviours of Ni 50 Mn 40-x Sn 10 Fe x magnetic shape memory alloys Zhigang Wu 1, Zhuhong Liu 1,2, Hong Yang 1, Yinong Liu 1, Guangheng Wu 3 and Robert Woodward 4 1. School of Mechanical Engineering, The University of Western Australia, Crawley, WA 6009, Australia 2. Department of Physics, University of Science and Technology Beijing, Beijing , China 3. Beijing National Laboratory for Condensed Matter Physics, Institute of Physics, Chinese Academy of Sciences, Beijing , China 4. School of Physics, The University of Western Australia, Crawley, WA 6009, Australia Abstract This study investigated the metallurgical origin of the effects of Fe substitution for Sn on the martensitic and the magnetic transformation behaviours of Ni 50 Mn 40-x Sn 10 Fe x (x=0, 3, 4, 5, 6) alloys. Substitution of Fe for Mn at above 3 at% introduced an fcc γ phase in the microstructure. Formation of the γ phase influenced the composition of the bcc/b2 matrix, leading to decrease in martensitic transformation temperatures and transformation entropy change. The Curie temperature of the parent phase increased slightly, whereas the Curie temperature of the martensite increased rapidly with increasing Fe addition. Changes in the temperatures of the martensitic and magnetic transformations are confirmed to directly relate to the e/a ratio of the matrix caused by formation of γ phase. The minimum e/a ratio value for the occurrence of the martensitic transformation is estimated to be for the alloy system studied. A narrow e/a ratio range of 8.113~8.137 is estimated for the occurrence of metamagnetic transformation M ( para) A( ferro). This metamagnetic CHAPTER 3 120

127 reverse transformation was induced by a magnetic field at 225 K within a range of 3~7 T in the Ni 50 Mn 35 Sn 10 Fe 5 alloy. The magnetic work required to induce the transformation is estimated to be ~176 J/kg, comparable to the thermodynamic energy deficit for the transformation at the testing temperature estimated from thermal measurement. These findings clarify the origin of the effects of Fe doping in Ni 50 Mn 40-x Sn 10 alloys and provides reference on alloys design for this system. Keywords: A. magnetic intermetallics; B. alloy design; B. shape-memory effects; B. martensitic transformations; B. magnetic properties 1. Introduction Ni-Mn-X (In, Sn and Sb) alloys have attracted much attention since the discovery of magnetic field induced reverse martensitic transformation by Sutou et al. in 2004 [1]. In these alloy systems, the martensitic transformation coincides with the magnetic transformation from a L2 1 structure ferromagnetic austenite to an orthorhombic paramagnetic martensite. The difference in magnetisation between the two phases provides a driving force for the structural transformation under the influence of magnetic field. In 2006, Kainuma et al. achieved shape recovery accompanying the martensitic transformation in Ni-Co-Mn-In single crystalline and Ni-Co-Mn-Sn polycrystalline alloys [2, 3], demonstrating the promise of the alloys for actuation applications in smart systems. In addition to magnetoactuation, these alloy systems also exhibit several other interesting properties. The electrical resistance of the parent phase exhibits a typical metallic behaviour, whereas it is semimetal-like for martensite in such alloy systems [4]. Since magnetic transition coincides with a first order martensitic phase transformation, the giant magnetoresistance effect [4-6] and giant magnetocaloric effect [7-9] have also been discovered. Beside the investigation on Ni-Co-Mn-In (Sn) alloys, many other alloys with similar compositions have also been extensively studied recently in a number of aspects, including the thermal and stressed induced martensitic transformation behaviours [10, 11], phase separation and magnetic properties [12, 13], martensitic transformation characteristics [14], time effect [15] and aging effect [16]. CHAPTER 3 121

128 One main hindrance to engineering application of these materials is the intrinsic brittleness associated with the intermetallic compound nature of the alloys. Similar problem is also found in Ni-Mn-Ga alloys. To improve ductility, a second ductile γ phase has been introduced by introducing Co or Fe into Ni-Mn-Ga alloys [17-20]. Recently, Feng et al. reported that substitution of Fe for In in Ni-Mn-In alloys introduces the γ phase and enhances the ductility of the alloys [21]. Whereas the purpose is to improve ductility, addition of a fourth element to the ternary Ni-Mn-X (In, Sn and Sb) alloys inevitably alters the matrix composition, hence the structure and thermal and magnetic properties. Passamani and Fukushima have recently investigated the effect of Fe substitution for Mn in Ni-Mn-Sn alloys on their magnetic properties. They found that the martensitic transformation temperatures decrease rapidly whereas the Curie transition temperatures of both austenite and martensite increase with the increasing Fe substitution [22, 23]. It was also found that the addition of Fe leads to the enhancement of FM exchange interaction in the austenitic and martensitic phases, and the magnetic exchange bias effect was detected in the samples with Fe substitution below 10 at.% [22]. Whereas much attention has been given to the influences of fourth element addition on the magnetic and transformation properties of these alloys, given the level of complexity associated with the quaternary systems, much less is understood of the metallurgical origins of these influences. This study is concerned with this fundamental issue by investigating the effects of Fe substitution for Mn in Ni 50 Mn 40-x Sn 10 Fe x. Fe bears much resemblance to Mn in this alloy system, including magnetic state and valence electron number, thus providing an opportunity to examine the metallurgical influence of the addition to the properties of the alloys, in addition to being a selected element for ductility improvement for some common ferromagnetic shape memory alloys. 2. Experimental Procedures Polycrystalline Ni 50 Mn 40-x Sn 10 Fe x (x=0, 3, 4, 5, 6) alloy ingots were prepared by means of arc melting in argon atmosphere using high purity (99.99 %) elemental metals. The samples are referred to as Fe0, Fe3, Fe4, Fe5 and Fe6, respectively. The button shaped ingots were heat treated at 1173 K in vacuum for homogenisation followed by furnace CHAPTER 3 122

129 cooling to room temperature. Transformation behaviour of the alloys was studied by means of differential scanning calorimetry (DSC) using a TA Q10 DSC instrument with a cooling/heating rate of 10 K/min. Phase identification and crystal structures were determined by means of X-ray powder diffraction using Cu-Kα radiation. Microstructures of the samples were studied with optical microscopy and scanning electron microscopy (SEM) and the compositions were determined by means of X-ray energy dispersive spectrometry (EDS). The magnetic properties were studied using a superconducting quantum interference device magnetometer (SQUID). 3. Results and discussion 3.1 Microstructure and crystal structure Figure 1 shows back-scattered SEM micrographs of the microstructures of the Ni 50 Mn 40-x Sn 10 Fe x (x=3, 4, 5, 6) alloys after homogenization treatment. The Fe3 sample (micrograph (a)) showed a uniform single phase structure, without any sign of a second phase. A few black spots presented in alloy Fe3 are solidification shrinkage pores formed during ingot casting. These pores are also presented in the other alloy samples. The microstructure of the F0 sample is essentially identical to that of F3, except the actual chemical composition of the matrix. The Fe4, Fe5 and Fe6 samples showed a continuous matrix in light contrast and dispersed γ phase particles in dark contrast. The volume proportion of γ phase obviously increased with more Fe content from alloy Fe4 to Fe6. The Fe6 alloy showed much smaller γ phase particles and a different texture compared to the other two alloys. CHAPTER 3 123

130 (a) (b) 100 µm 100 µm (c) (d) 100 µm 100 µm Figure 1. Back-scattered electron images of Ni 50 Mn 40-x Sn 10 Fe x alloys: (a) x=3; (b) x=4; (c) x=5 and (d) x=6. Figure 2 shows XRD spectra of the Ni 50 Mn 40-x Sn 10 Fe x alloys. Alloys Fe0 and Fe3 showed an orthorhombic martensite crystal structure. Alloy Fe4 exhibited a more complicated case, showing a mixed structure of the orthorhombic martensite, the bcc austenite and the γ phase. Alloy Fe5 presented a two-phase structure, with the fcc γ phase and the bcc austenite. Alloy Fe6 also showed a two-phase structure, but with the fcc γ phase and a B2 austenite. The lattice parameters of the martensite are similar for Fe0, Fe3 and Fe4 alloys, and the values are listed in Table 1. The lattice parameters of the γ phase in Fe4, Fe5 and Fe6 are very close, with an average value of a=0.365 nm at room temperature. The lattice parameter of the bcc austenite was determined to be a= nm for Fe4 and nm for Fe5. The lattice parameter of the B2 phase in Fe6 was determined to be a= nm. The phases present in the samples are summarised in Table 1. It is evident that the intensity of the characteristic peaks of the γ phase increased with increasing Fe content, which is consistent with the microstructure observations presented above. CHAPTER 3 124