Large magnetocaloric effect in manganese perovskite La0.67−xBixBa0.33MnO3 near room temperature

La0.67−xBixBa0.33MnO3 (x = 0 and 0.05) ceramics were prepared via the sol–gel method. Structural, magnetic and magnetocaloric effects have been systematically studied. X-ray diffraction shows that all the compounds crystallize in the rhombohedral structure with the R3̄c space group. By analyzing the field and temperature dependence of magnetization, it is observed that both samples undergo a second order magnetic phase transition near TC. The value of TC decreases from 340 K to 306 K when increasing x from 0.00 to 0.05, respectively. The reported magnetic entropy change for both samples was considerably remarkable and equal to 5.8 J kg−1 K−1 for x = 0.00 and 7.3 J kg−1 K−1 for x = 0.05, respectively, for μ0H = 5 T, confirming that these materials are promising candidates for magnetic refrigeration applications. The mean-field theory was used to study the magnetocaloric effect within the thermodynamics of the model. Satisfactory agreement between experimental data and the mean-field theory has been found.


Introduction
Over the past few years, the perovskite manganites with ABO 3type compounds Tr 1Àx M x MnO 3 (where Tr stands for a trivalent rare-earth element such as Bi 3+ , La 3+ or Pr 3+ , and M for the divalent alkaline earth ions such as Sr 2+ , Ca 2+ or Ba 2+ ), have been extensively studied due to their extraordinary magnetic and electronic properties as well as their promise for potential technological applications. 1,2 A prominent feature of the mixedvalence perovskite manganite materials is an insulator-metal (IM) transition accompanied simultaneously by the paramagnetic-ferromagnetic (PM-FM) transition giving rise to a colossal magnetoresistance (CMR) effect. 3 The existence of the observed CMR near the transition temperature was due to the mixed valence state of Mn, evolving from Mn 3+ (t 3 2g [e 1 g [, S ¼ 2) in the parent atom LaMnO 3 to Mn 4+ (t 3 2g [e 0 g , S ¼ 3/2) to the doped element SrMnO 3 . 4 The double exchange interaction of the neighboring spin moment of (Mn 3+ , Mn 4+ ) coupled through oxygen ions (O 2+ ), the small polaron theory and the Jahn-Teller (JT) effect have been proposed to explain the CMR phenomenon near the transition temperature. 5 In addition, when a eld is applied to this material, the unpaired spins are aligned parallel to the eld. Since the total entropy of spins plus the lattice remains constant, the magnetic entropy change (ÀDS M ) is removed from the spin system and goes into the lattice, which lowers the magnetic entropy and produces a net heat. On the contrary, when an applied eld is removed from a magnetic sample, the spin tends to become random, leading to increment of the entropy and causing the material to cool down. As well known, the maximum of the magnetic entropy change in this kind of material always occurs around its magnetic ordering temperature (i.e., Curie temperature, T C ). Nowadays, there is a need of new advanced magnetics materials with a second order magnetic phase transition, showing a large reversible (ÀDS M ) at low applied elds. Some theoretical works have focused on this subject, for second order phase transition via the molecular mean eld theory. 6 For this it is important to know the eld dependence of a given magnetic refrigerant sample. The study of the magnetocaloric effect is not only important from the point of view of potential applications; it also provides a tool to understand the intrinsic properties of a material. In Bi based manganites, the lone pair electrons of Bi 3+ ion hybridize with oxygen 2p orbitals, which in turn reduces the bond length of d Bi-O and bond angle of q Mn-O-Mn and increases the bond length of d Mn-O . 7 From this viewpoint, this paper reports the structural, magnetic and magnetic entropy change of Bi-substituted perovskite manganites La 0.62 Bi 0.05 Ba 0.33 MnO 3 . It is found that these materials show quite large magnetic entropy changes induced by low magnetic eld changes. Ba(NO 3 ) 2 , Bi(NO 3 ) 2 $5H 2 O and Mn(NO 3 ) 2 $6H 2 O precursors, all with purity of 99.9%, were weighed in the desired proportions and dissolved with small amounts of water. Ethylene glycol (EG) and citric acid (CA) were used as polymerization/complexation agents, respectively, forming a stable solution. 100 cm 3 of metallic salts solution was added to 300 cm 3 of a solution containing a mixture of citric acid (60 g) and ethylene glycol (13 mL). This solution was then heated on a thermal plate under constant stirring, where polymerization occurs in the liquid solution and leads to a homogeneous sol. When the sol is further heated to remove the excess of solvent, an intermediate resin is formed. Calcination of the resin at 573 K in air was performed and sintering at 1073 K for 10 hours. These procedures are outlined in the ow chart of Fig. 1.
The phase purity and structure of sample were identied by X-ray powder diffraction at room temperature using a Siemens D5000 X-ray diffractometer with a graphite monochromatized CuKa radiation (l CuKa ¼ 1.5406Å) and 20 # 2q # 90 with steps of 0.02 and a counting time of 18 s per step. According to our measurements, this system is able to detect up to a minimum of 3% of impurities. The structure analysis was carried out using the Rietveld method with FULLPROF soware (version 0.2-Mars 1998-LLB-JRC). 8 Scanning electron microscopy (SEM) using a Philips XL30 equipped with a eld emission gun at 20 kV was used to characterize La 0.67Àx Bi x Ba 0.33 MnO 3 morphologies.
Magnetization (M) versus temperature (T) and magnetization versus magnetic eld (m 0 H) were performed by using BS1 and BS2 magnetometers developed in Louis Neel Laboratory at Grenoble. The isothermal curves were determined in the magnetic eld range of 0-5 T. The temperature interval is xed to 2 K in the vicinity of the Curie temperature (T C ). The temperature steps were smaller near T C and larger further away.

X-ray diffraction and microstructure analysis
The X-ray diffraction pattern for the samples (x ¼ 0.00 and 0.05) is shown in Fig. 2. The samples of La 0.67Àx Bi x Ba 0.33 MnO 3 are a single phase without detectable secondary phase, within the sensitivity limits of the experiment (a few percent). The Rietveld renements was successful considering the R 3c (no. 167) rhombohedral and centro symmetric space group (inset (a) of Fig. 2, for x ¼ 0.0 for example). Standard hexagonal setting of the R 3c space group (with a H and c H cell parameters) was used. The manganite structure (LaAlO 3 type) is described in this hexagonal setting, with (La/Bi/Ba) atoms at 6a (0, 0, 1/4) position, Mn at 6b (0, 0, 0) and O at 18e (x, 0, 1/4) position. This distorted manganite is characterized by a a a antiphase oxygen octahedral tilt system (Glazer notation 9 ) corresponding to rotations along the three pseudo cubic directions of the manganite. Detailed results of the structural renements are regrouped in Table 1. It can be observed from the inset (b) in Fig. 2 that the position of the most intense peak shows a slight shi towards low angles with the increase of Bi, indicating that the cell volume of the La 0.67Àx Bi x Ba 0.33 MnO 3 specimens increases.
In order to quantitatively discuss the ionic match between A and B sites in perovskite compounds, a geometrical quantity, noted Goldschmidt tolerance factor (t), is usually introduced and is dened as: 10 Here r La+Bi+Ba ; r Mn and r O are the average ionic radii of A, B and oxygen, respectively in the perovskite ABO 3 structure. Manganite compounds have a perovskite structure if their tolerance factor lies in the limits of 0.75 < t < 1 and in an ideal case when the value must be equal to unity. In the present work, the tolerance factor of La 0.62 Bi 0.05 Ba 0.33 MnO 3 is calculated from Shannon's ionic radii (r La 11,12 and it is found to be t ¼ 0.9595 and 0.9599 for x ¼ 0 and x ¼ 0.05, respectively, which is within the range of stable perovskite structure. The value of average crystallite size was estimated from the full width at half maximum (FWHM) of X-ray diffraction peaks.  The effects of synthesis, instruments and processing conditions were taken into consideration while making the calculation of crystallite size. The dependence of the size effect is given by , where l is the wavelength of CuKa radiation (l ¼ 1.5406Å), K is grain shape factor (¼0.89) and D s is the thickness of the crystal. Using the Williamson-Hall (W-H) method, 13 the average values of both D w and lattice strain (3) can be obtained from the intercept and the slope of the following relation, respectively, where b is the full-width at half-maximum of an XRD peak, q is the Bragg angle, and K ¼ 0.9 is the shape factor. The values of average crystallite size D S , D W and micro-strain of La 0.67Àx Bi x -Ba 0.33 MnO 3 compounds are tabulated in Table 1. The particle size, calculated in the present system using Williamson-Hall technique, is larger than the particle size obtained from Debye-Scherrer method because the broadening effect due to strain is completely excluded in Debye-Scherrer technique. 14 Fig. 3 shows the SEM photograph of the compounds. The samples contained connected particles with hexagonal shape and clear grain boundaries. These particles are largely agglomerated with a broad size distribution. The average value of thickness of both compounds is listed in Table 1.
Aer measuring the diameters of all the particles in SEM image, the size distribution histogram is tted with the lognormal function expressed as: Here D 0 and s are the average diameter obtained from the SEM results and the data dispersions, respectively. The inset of Fig. 3 shows the dispersion histogram. The mean diameter hDi and standard deviation s D were determined using these relations: 15,16 hDi The results analysis showed hDi $ 397.48 mm and s D ¼ 291.53 mm.

Bulk magnetization
Low-eld magnetization (M) versus temperature was rst measured for the samples, in order to have an estimation of the transition temperature (T C ). The result is presented in Fig. 4. The M(T) curve reveals that when increasing temperature, the samples exhibit a magnetic transition from paramagnetic (PM) state to ferromagnetic (FM) one. This transition occurs at the Curie temperature (T C ) which is obtained from the peak of dM/ dT curve. The Curie temperature decreases from 340 K to 306 K when increase x from 0.00 to 0.05, respectively for m 0 H ¼ 0.05 T. The Curie temperature of the Bi-doped compound was found to be lower than that of the undoped sample. This indicates that Bi substitution appears to weaken the magnetic interaction in the sample. Theoretical calculation has shown that off-center shis of the ions with ns 2 electronic conguration results in the structural distortion and minimization of the Coulombic energy. 17 An orientation of the 6s 2 lone pair toward a surrounding anion (O-2p) can produce a local distortion or even hybridization between Bi-6s-orbitals and O-2p orbitals, 18 leading to the block of the movement of e g electrons through the Mn-O-Mn bridges (stronger localization).
The inset of Fig. 4 shows the temperature dependence of the inverse magnetic susceptibility of x ¼ 0 and x ¼ 0.05. It could be tted to the Curie-Weiss law just above T C (the PM region): c ¼ C/T À q CW , where q CW is the Weiss temperature and C is the Curie constant dened as: where m 0 ¼ 10 7 H m À1 is the permeability, g is the Landé factor, m B ¼ 9.27 Â 10 24 J T À1 is the Bohr magneton, k B ¼ 1.38 Â 10 23 J K À1 is the Boltzmann constant, J ¼ L + S is the total moment and m eff is the effective paramagnetic moment. We can determine the effect of paramagnetic moment (m exp eff ) from the curie constant. The theoretical m calc eff is estimated using the following expression:  (3) 13.5023 (1) Table 2. It is found that the m exp eff is greater as compared to m calc eff . This discrepancy validates the formation of ferromagnetic spin clusters within the paramagnetic state. 19 A linear t yields positive Curie-Weiss temperature q CW ¼ 312 K (x ¼ 0.05). This result conrms a mean FM interaction between spins for all samples (Table 2). Moreover, this value is higher than T C , which may be due to the existence of short range FM ordering. 19 The structure analysis shows that the unit cell becomes slightly larger as the 6s 2 lone pair character becomes dominant, it has been shown that the Bi-O bond is shorter than the La-O, despite of the similar ionic radius of La 3+ and Bi 3+ ions. 20 This can be interpreted as arising from the rather covalent character of the Bi-O bonds. The electronegativity of Bi enhances Paper hybridisation between 6s 2 of Bi 3+ orbitals and 2p of O 2À orbitals and this hybridisation produces a local distortion. It is observed that transition temperature T C decreases with increase in Bi ratio. This is presumably due to tilts the MnO 6 octahedra, resulting in a reduced overlap between the Mn-3d and O-2p orbitals. 21 It should also be noted that the La 0.67 Ba 0.33 MnO 3 sample is ferromagnetic while Bi 0.67 Ba 0.33 MnO 3 is antiferromagnetic, indicating a competition between the double exchange and the antiferromagnetic super exchange in these compounds can decrease T C . This phenomenon has been observed in the compound Bi 0.6Àx La x Ca 0.4 MnO 3 . 22

Effect of Bi on magnetocaloric properties
The change of magnetic entropy of magnetic compounds has the largest value near a phase transition. According to the classical thermodynamic theory, the isothermal magnetic entropy change (-DS M ) produced by the variation of a magnetic eld from zero to m 0 H max is given by: 23 In the case of magnetization measurement in small discrete magnetic elds and temperature interval DT, DS M can be approximated to:  Fig. 5. The compounds exhibit large changes in magnetic entropy around Curie temperature (T C ), which is a characteristic property of simple ferromagnets due to the efficient ordering of magnetic spins at the temperature induced by magnetic eld. 25 Large magnetic entropy changes DS max M are reported for all the samples and are summarized in Table 3. The magnitude of (ÀDS max M (T)) for all samples increases with increasing the applied magnetic eld (inset of Fig. 5). For example, the maximum magnetic-entropy value increases from 2.37 J kg À1 K À1 for x ¼ 0.00 to 2.8 J kg À1 K À1 (2T) and 5.8 J kg À1 K À1 for x ¼ 0.00 to 7.3 J kg À1 K À1 for x ¼  0.05 respectively (5T). Guo et al. 26 indicated that the large magnetic entropy change in perovskite compounds could originate from the spin-lattice coupling in the magnetic ordering process. Strong coupling between spin and lattice is corroborated by the observed signicant lattice change accompanying magnetic transition in perovskite manganites. 27 The lattice structural change in the Mn-O bond distance as well as in the hMn-O-Mni bond angle would in turn favor the spin ordering. Thus a more abrupt variation of magnetization near Curie temperature (T C ) occurs, resulting in a large magnetic entropy change as a large MCE.
The change of magnetic entropy can be also calculated from the eld dependence of the specic heat by the following integration: undergoes a sudden change from positive to negative around T C with a positive value above T C and a negative value below T C . The maximum/minimum value of DC p (m 0 H, T) observed at 320/ 300 K, exhibits an increasing trend with applied eld and is obtained to be 122.4/À115.43 J kg À1 K À1 for x ¼ 0.05 at 5 T.
It should be noted that (ÀDS max M ) is not the only parameter deciding about an applicability of material. To estimate if a material can be a good candidate for magnetic refrigeration (MR), Gschneidner and Pecharsky 28 dened the relative cooling power (RCPS), which is the important index which is used to evaluate the cooling efficiency of a magnetic refrigerant. It is dened as the product between the maximum values of the magnetic entropy change (ÀDS max M ) and the full width at half maximum dT FWHM of the magnetic entropy change curve (RCP(S) ¼ ÀDS max M Â dT FWHM ). 29 This parameter corresponds to the amount of heat that can be transferred between the cold and hot parts of the refrigerator in one ideal thermodynamic cycle. The results are summarized in Table 3. Fig. 7 shows the  Table 3 with those reported in the literature for several other magnetic compounds. 31-37

Modeling the magnetic properties
The Weiss molecular mean eld model is a standard model in magnetism. Because of its simplicity, this model is still used in current research for a wide range of magnetic materials, although its limitations are well known. This concept of a molecular eld assumes that the magnetic interaction between magnetic moments is equivalent to the existence of an exchange interaction depending on the magnetization M: 38 where H and H exch are the external magnetic eld, the exchange magnetic eld and l the mean eld exchange parameter respectively. Amaral et al. proposed a model based on mean eld theory and presented an approach of applying this method scenario to isotherm magnetization M(T, H) measurements. 39 In our study, it consider the general mean eld law: 40 M(H, T) ¼ B J [(H + H exch )/T], the Brillouin function B J is written as: , J is the total angular momentum in the lattice, g is the gyromagnetic factor (landé factor), m B is the Bohr magnetron and k B is the Boltzmann's constant. The mean eld exchange parameter l is not predetermined. Then for corresponding values with the same (H + H exch )/T, M is also the same, the value of the inverse B J À1 (M) function, 41 The study of the exchange eld induced by the magnetization change makes it possible to nd the value of the average eld exchange parameter l. Fig. 8 shows H/T versus 1/T for some of the values of M (5 emu per g per step) from 266 K to 342 K for x ¼ 0.05. According to the mean scaling method such H/T versus 1/T curves should show a series of straight lines at different temperature. The linear relationship between H/T and 1/T is kept. Linear ts are then easily made to each isomagnetic line. Typically, the interpolation step was of 1 emu g À1 . The slope of this isomagnetic line, will then give the exchange eld (H exch ). For all compounds in the paramagnetic domain or the materials of domain ordered such as anti-ferromagnetic, it can always expand increasing M in powers of H, or H in powers of M. In this latter approach it stop at the third order and considering that the magnetization is an odd function of eld, it can write: 42 Fig. 9 shows the evolution of the exchange eld versus the magnetization for the La 0.67Àx Bi x Ba 0.33 MnO 3 (x ¼ 0.05 for example). The experimental data should be included for the t by eqn (H exch ¼ l 1 M + l 3 M 3 ). The results show a very small dependence on M 3 (l 3 ¼ À1.3984 Â 10 À5 (T g emu À1 ) 3 ), is found for this second order transition system, thus H exch ¼ l 1 M with l 1 ¼ 1.25 T g emu À1 . Aer obtaining the mean eld exchange parameter the next step of this method consists on building the scaling plot of M vs. (H + H exch )/T (Fig. 10). It has successfully tted the scaled magnetization data with the Brillouin function. From the scaling plot and the subsequent t with the saturation magnetization equal to 72 emu g À1 (this value is close to the experimental one (M s ¼ 69 emu g À1 at 10 K) (inset b of Fig. 4), and the value of the total angular momentum of the manganite is J ¼ 1.9.
The magnetization measurements in the law temperature range show that the saturation magnetization is about M theor    12 shows the evolution of the magnetic entropy change (ÀDS M ) data as a function of temperature at several magnetic applied elds for the La 0.62 Bi 0.05 Ba 0.33 MnO 3 compound, by using the Maxwell relation and that basing on mean eld theory. Both results are in good agreement, except close to T C , here an excepted small difference appears, due to the formation of magnetic domains and critical effects.

Conclusion
In summary, single phase La 0.67Àx Bi x Ba 0.33 MnO 3 (x ¼ 0.00 and 0.05) compounds were prepared by the sol-gel technic. Using the magnetization measurements, magnetic and magnetocaloric effect have been studied. These compounds show second order ferromagnetic-paramagnetic phase transition, with a large magnetic entropy change. It have studied the mean-eld scaling method for these samples. The insight that can be gained from the use of this methodology for a given magnetic system can be of great interest. In a simplistic approach, it can say that if this scaling method does not follow a molecular mean-eld behavior, other methods must be pursued in order to interpret the magnetic behavior of the system. The mean-eld scaling method is able to determine the exchange parameters J, l and g of ours samples. These factors allow estimating some magnetic properties.

Conflicts of interest
There are no conicts to declare.