Enhancement in magnetocaloric properties of ErCrO₃ via A-site Gd substitution Free

Magnetic refrigeration (MR) based on magnetocaloric effect (MCE) has been known for more than a century and is considered to be an attractive alternative cooling technique compared to the conventional gas compression (CGC) refrigeration.^1,2 There are several advantages of MR over CGC: (i) ecologically friendly (no ozone depleting or direct greenhouse gases like chlorofluorocarbons in CGC), (ii) higher cooling efficiency (30%–60% of the Carnot cycle), and (iii) refrigerators based on MR technology can be built compactly as it involves adiabatic demagnetization of ‘solid’ magnetically ordered materials.^3–5 Thus, MR can also be referred to as solid state refrigeration.⁶ In the last decade, MR has found wide commercial and industrial applications, such as in large-scale air-conditioners, chemical processing, large refrigerators (like those in supermarkets), and gas liquification.² Generally, there are two temperature regimes where MR shows great promise: near room temperature and at cryogenic temperature.^6,7 Furthermore, in the cryogenic temperature region (<30 K), MR can avoid the use of costly helium refrigerant.^6,8 In practice, MR requires the combination of a strong magnetic field and a material with sufficiently large MCE.⁴ Apart from the magnetic field source, particular attention has been given to the magnetocaloric materials due to their crucial role in the development of MR.^4,9,10 A broad spectrum of magnetocaloric alloys and compounds, such as binary and ternary intermetallic compounds, Gd₅(Si_1-xGe_x)₄, Mn-based alloys, La(Fe_13-xMn_x)-based compounds, and manganites, have been carefully designed and investigated to meet the needs of MR.^3,11–14 It has been found that the coexistence of multiple ferroic phases facilitates large entropy change, which makes materials with multiferroic properties strong candidates for giant MCE.^15,16

Recently, the rare earth chromites, RCrO₃ (R = Y, Ho, Er, Yb, Lu), have been reported to exhibit multiferroic properties due to the local structural inhomogeneity or exchange striction between the rare earth and the Cr³⁺ cation.^17–21 These materials crystallize in an orthorhombically distorted perovskite structure with four formula units per unit cell (space group Pbnm).^17,22 RCrO₃ orders antiferromagnetically (G type) due to the exchange coupling of Cr³⁺-Cr³⁺ below the N$e´$el temperature.²³ The other two types of antiferromagnetic (AFM) exchange couplings, Cr³⁺-R³⁺ and R³⁺-R³⁺, account for numerous fascinating magnetic phenomena, such as spin reorientation, large MCE, and temperature induced magnetization reversal.²³ To date, some prototype RCrO₃ with impressive MCE properties have been reported.^15,24,25 As far as ErCrO₃ (ECO) is concerned, only a few papers have reported its magnetocaloric properties. For example, Yin et al. have reported the large entropy change of 16.9 J kg⁻¹ k⁻¹ for ErCrO₃ crystal at 8.5 K under the field strength of 4.5 T.²⁴ However, no attempts have been made to tailor the MCE properties of ErCrO₃ bulk powders through A- or B-site substitution.

In this work, Gd is selected as an Er-site dopant for the ErCrO₃ compound. Selection of Gd as the A-site dopant mainly stems from the fact that Gd or Gd based compounds are often present as magnetic refrigerant in MR with excellent MCE properties due to its large spin ground state, weak superexchange interaction, and quenched orbital momentum.²⁶ Thus, effects of Gd doping on the structural, magnetic, and magnetocaloric properties of ErCrO₃ are explored in detail here. The composition studied here is Er_0.33Gd_0.67CrO₃ (EGCO). The doping ratio was chosen based on the previous work on selective rare-earth doping in chromites by our group.²⁷ 

ErCrO₃ (ECO) and Er_0.33Gd_0.67CrO₃ (EGCO) bulk powder samples were prepared by citrate solution route. High-purity Er(NO₃)₃, Gd(NO₃)₃, and Cr(NO₃)₃ precursors were dissolved in water stoichiometrically and then mixed together with citric acid. The solution was continuously stirred, heated, and dried on a hot plate. The resultant powder was then annealed at 900 °C for 2 h in oxygen atmosphere to obtain ECO and EGCO bulk powder samples. The crystal structure and phase purity of the as prepared samples were examined by x-ray diffraction (XRD, Bruker D2 Phaser diffractometer with Cu-Kα radiation). The diffraction patterns were refined by the Rietveld refinement method using Fullprof software.¹⁸ Room temperature Raman spectra were recorded with a Renishaw System 2000 using 514 nm Ar-ion laser. Scanning Electron Microscopy (SEM) images were obtained using JEOL JSM-6335F system with a field emission gun operating at 10 kV. Dc and ac magnetic measurements were performed using a vibrating sample magnetometer (VSM) and ac magnetic susceptibility option attached to the Evercool Physical Property Measurement System from Quantum Design.

Structures of the as-prepared polycrystalline samples were characterized by XRD patterns along with Rietveld structural refinement, as shown in Figs. 1(a) and 1(b). The refined lattice parameters (a, b, c, bond angles, unit cell volume, and goodness of fit) are summarized in Table I. XRD patterns of both samples fitted well with the orthorhombically distorted perovskite crystal structure with the space group Pbnm. All the peaks are well defined and properly indexed, indicating the high crystallinity and phase purity. The lattice parameters (and hence unit cell volume) of EGCO are larger than that of ECO, which is consistent with the larger ionic radius of Gd³⁺ (1.053 Å) compared with that of Er³⁺ (1.004 Å). The mismatch between R-O and Cr-O bond lengths can be described according to the Goldschmidt tolerance factor $τ=R−O/2(Cr−O)$⁠, where R is the average ionic radii of A-site ions calculated by $0.33rEr2+0.67rGd2$ for the substituted sample.²⁸ Note that $τ$ is less than 1 for both samples indicative of significant cooperative rotations of CrO₆ octahedra and bending of the Cr – O – Cr bond angles as listed in Table I (in plane Cr – O₁ – Cr and out of plane Cr – O₂ – Cr).

The average crystallite sizes of both samples were calculated using the Williamson-Hall (W-H) relation.^9,29 The contributions of crystallite size and strain to the total peak broadening can be expressed as

where β is the full width at half maximum of the diffraction peaks, θ is the diffraction angle, ε is the microstrain parameter, k is the constant that depends on the shape of particles (k = 0.89, for the circular grain),^30,λ is the wavelength of incident X-ray, and d is the value of average crystallite size.^30,31 Here, the instrumental error has been considered by replacing β with β′, where $β′=β2−b2$ (b = 0.042°, for Bruker D2 Phaser used here).¹⁵ Figure 1(c) shows the plot of β cos θ vs. sin θ, from which the average crystallite sizes were extracted (d_ECO=223 nm and d_EGCO=109 nm).

Scanning electron microscopy (SEM) images of ECO and EGCO are displayed in Fig. 2. And the results show that the particle sizes of both materials are around several hundred nanometers. It is evident that the particle sizes of ECO in Fig. 2(a) are larger than those of EGCO in Fig. 2(b). The particle size distributions were shown in the histograms of Figs. 2(c) and 2(d), which were constructed from individual particles in the SEM images for both samples. The histograms were fitted with the lognormal distribution function. From the fitting, the average particle size is estimated to be 212 nm for ECO and 131 nm for EGCO, which are close to the average crystallite sizes obtained by the W-H relation described earlier.

Additional structural characterizations on the bulk powder samples were performed by Raman measurements and Fig. 3 shows the room temperature Raman spectra for the two samples. Both ECO and EGCO have the orthorhombically distorted perovskite structure with the space group Pbnm. The atoms in this structure occupy the 4c (Er, O, Gd), 4b (Cr), and 8d (O) Wyckoff sites.³² These sites contribute to the 7A_g + 8A_u + 5B_1g + 10B_1u + 7B_2g + 8B_2u + 5B_3g + 10B_3u modes of theoretical vibrational representation.³³ Among these modes, 7A_g + 5B_1g + 7B_2g + 5B_3g are Raman active.³⁴ The number of observed modes in Fig. 3 is lower than expected because of the low intensity of other predicted modes or the limited measurement range adopted in this work. The symmetry assignments of Raman modes are also shown in Fig. 3.^35,36 The region below 200 cm⁻¹ is characterized by two sharp peaks, which are dominated by the vibration of the A-site cation and sensitive to the mass of the A-site cation.³⁶ The two modes of ECO and EGCO below 200 cm⁻¹ do not shift much, since the difference of atomic mass between Er (167.26 u) and Gd (157.25 u) is comparatively small. The peak positions for the rest of Raman modes are all decreased. This is consistent with the reported tendency of decreasing some Raman mode position with increasing ionic radius.³⁶ The larger ionic radius (r_Gd: 1.053 Å > r_Er: 1.004 Å) naturally leads to the increased bond length and Cr-O-Cr bond angle for EGCO.³⁵ Thus, room temperature Raman spectra reveal more structural distortion (i.e., CrO₆ octahedra) in ECO compared to EGCO, which is consistent with the XRD results.

The temperature dependent dc magnetization data were measured under zero-field cooled (ZFC) and field cooled (FC) conditions with an applied magnetic field of 50 Oe between 5 and 300 K, as shown in Fig. 4. There are substantial differences between the measured magnetization of ECO [Fig. 4(a)] and EGCO [Fig. 4(b)] as a function of temperature. The Gd substitution decreases the magnitude of magnetization above 12 K in both ZFC and FC conditions when compared to that of ECO sample. This could be attributed to the slightly smaller magnetic moment of Gd³⁺ ions (7.94 $μB$⁠) than that of Er³⁺ (9.58 $μB$⁠) ions. There are two distinct magnetic transitions in both FC and ZFC curves for ECO, as shown in Fig. 4(a). With decreasing temperature, the ECO orders antiferromagnetically below $TNCr ∼ 132$ K, which could be attributed to the antiferromagnetic (AFM) ordering of the Cr³⁺ moments, and then the magnetization reaches a maximum at around T_SR (spin-reorientation temperature) $∼$ 20 K due to the combined effect of long range spin reorientation of Cr³⁺ sublattice at ∼ 10.2 K and reorientation of Er³⁺ moments at 34.7 K.^17,24,37 However, for EGCO, only one magnetic transition (Cr³⁺ ordering) is observed, as shown in Fig. 4(b).^17,23 Gd substitution might lower the spin reorientation temperature for EGCO. In order to accurately determine the magnetic ordering temperature $TNCr$⁠, χT vs. T, and d(χT)/dT vs. T are plotted, as shown in Figs. 4(c) and 4(d).^38,39 From the position of the peak in the d(χT)/dT vs. T plot, $TNCr$ values for ECO and EGCO are determined to be 131 K and 154 K, respectively.

The dc magnetic susceptibility, χ, could be fitted well to the Curie-Weiss law in the paramagnetic region above $TNCr$⁠. However, the deviation from Curie-Weiss behavior near the $TNCr$ could be ascribed to the onset of canted antiferromagnetic order. This deviation was modeled by Moriya to account for the Dzyaloshinskii-Moriya (DM) antisymmetric exchange interaction⁴⁰ and has been applied in similar compounds to show the deviation near $TNCr$⁠.^39,41 Therefore, the susceptibility is modelled by the modified Curie-Weiss equation [Eq. (2)] resulting from the DM interaction^39,41

where χ is the susceptibility, T is the temperature, C is the Curie constant, Θ is the Weiss temperature, $TNCr$ is the ordering temperature of Cr, and $T0$ is a fitted parameter. The fitting results are shown in Fig. 5 and listed in Table II. The negative value of Θ indicates the AFM order in both ECO and EGCO. The calculated effective magnetic moment values for the ECO and EGCO are 10.84 $μB$ and 9.91 $μB,$ respectively, which are close to that obtained via sum of magnetic moments of the free rare-earth-ions and Cr-ion (ECO: Er³⁺ = 9.58 $μB$⁠, Cr³⁺ = 3.87 $μB$⁠, ECO = $9.582+3.872=10.33μB$⁠; EGCO: Gd³⁺=7.94 $μB$⁠, EGCO=$0.33×9.582+0.67×7.942+3.872=9.35μB$⁠).²⁵ The spin-orbit coupling effects might cause a slight deviation in the effective magnetic moment from the sum of magnetic moments of free ions (spin only values).¹⁷ It should be noted that the above modified Curie-Weiss equation provides information about the magnetic interaction between the Cr³⁺ ions by Eqs. (3) and (4), where Z is the coordination number of Cr³⁺, S is the spin quantum number of Cr³⁺, k_B is Boltzmann's constant, and $Je$ and D are the strength of symmetric and antisymmetric exchange interactions between Cr³⁺, respectively,^39,40

Other magnetic interactions are not likely to contribute to the $TNCr$ and $T0$⁠, since the Cr³⁺ ordering temperature is well above the ordering temperature of rare earth ions. The estimated magnitudes of D are 19.1% and 12.5% of the magnitudes of $Je$ for ECO and EGCO, respectively.

In order to probe the magnetic transitions in detail, the ac-susceptibility (⁠$χAC$⁠) of ECO and EGCO was recorded under an applied ac magnetic field of 10 Oe with frequencies at 500, 1000, and 5000 Hz in the temperature range 5–200 K. The ac-susceptibility being complex is usually expressed by $χAC=χ′−iχ″$⁠, where $χ′$ is the real component and related to the reversible magnetization process, $χ″$ is the imaginary component and related to the irreversible magnetization process.⁴² Figures 6(a) and 6(b) show the temperature dependence of $χ′$ and $χ″$ for ECO and EGCO, respectively. $TNCr$ is determined by the minimum of $dχ′/dT$ and is shown in Figs. 6(a) and 6(b) insets, which revealed $TNCr$ at 134 K and 156 K for ECO and EGCO, respectively. In addition, another magnetic ordering transition at ∼14 K for ECO sample, not revealed by the dc magnetic data, indicates the spin reorientation resulting from anisotropic magnetic interaction between Er³⁺ and Cr³⁺.¹⁸ The anomalies of the imaginary component (⁠$χ″$⁠) are observed for both samples, which are close to the $TNCr,$ the onset of antiferromagnetic ordering of Cr³⁺. All the magnetic transitions are observed to be frequency independent.

Magnetization vs. magnetic field isotherms were recorded at selected temperatures between 5 and 160 K with a magnetic field up to 40 kOe. Representative hysteresis loops at 5, 25, 50, and 100 K are shown in Figs. 7(a) and 7(b) for ECO and EGCO, respectively. The saturation magnetization values are 14.7 and 6.5 emu g⁻¹ for ECO and EGCO at 25 K and 40 kOe, respectively. The hysteresis loop is symmetric with respect to both magnetization and field axes, indicating the absence of exchange bias effect in both samples. This type of magnetic hysteresis loops is attributed to the existence of ferromagnetic component resulting from the canting of Cr³⁺ spins.^10,15,39

The temperature dependence of remnant magnetization (M_R) and coercive field (H_C), as extracted from the dc magnetization isotherms, is shown in Figs. 8(a) and 8(b), respectively. The temperature dependence of remnant magnetization of both samples follows the same trend as it appears in the field cooled magnetization curves. The remnant magnetization peaks at T_SR ∼ 21 K for ECO and gradually decreases to zero at Néel temperature for both samples. The H_C value first increases with temperature and shows a broad peak and then decreases when the temperature as it goes above $TNCr$⁠. The initial increase in H_C with a decrease in temperature below Néel temperature could be explained by thermal fluctuations of magnetic moments and is described by Kneller's law

where $HC0$ is determined by the ratio of anisotropy constant to saturation magnetization, and T_B is the blocking temperature.⁴³ According to Kneller's law, the value of H_C should keep increasing with a decrease of temperature below $TNCr$⁠. However, the experimental results show a plateau and a reduction of H_C at lower temperatures. The plateau could be attributed to the saturation of weak ferromagnetic signal from the Cr³⁺ sublattice, which might occur at around 100 K.¹⁹ At even lower temperatures, the weak exchange coupling of R³⁺ - Cr³⁺ becomes operative as the thermal energy is lowered with decreasing temperature. Thus, the magnetic moment of rare earth could rotate with the moment of Cr³⁺, leading to decreased H_C value.³⁹ The H_C of EGCO is larger than that of ECO in a wide temperature range. This could be explained by the defects and grain boundaries, which could pin down the domain walls and impede domain rotations when the magnetic field changes.³⁹ In the present case, relatively smaller crystallite (grain) size of EGCO as compared to ECO possibly results in larger H_C for EGCO.

In order to explore the figure of merit of the materials for the MR applications, the curves of isothermal magnetization versus applied magnetic field up to 7 T were recorded at different temperatures for bulk ECO and EGCO, respectively. The temperature dependence of magnetization, as extracted from Figs. 9(a) and 9(c), is shown in Figs. 9(b) and 9(d) for ECO and EGCO, respectively. From these curves, the magnetic entropy change $(−ΔS)$ could be related to the magnetization through the thermodynamic Maxwell relation

where H is the magnetic field, M is the magnetization, and T is the temperature.^2,41 And, $ΔS$ can be calculated by the following equation:

where $∂M(T,H)∂TH$ is the derivative of the curves in Figs. 9(b) and 9(d).⁴⁴ As isothermal magnetization is measured at discrete field and temperature intervals, $ΔS$ can also be approximately calculated by the following equation:² 

The magnetic entropy changes (⁠$−ΔS)$⁠, as calculated using the above equation, are plotted as a function of temperature at various applied magnetic fields in Figs. 10(a) and 10(b) for ECO and EGCO, respectively. It is evident that the $−ΔS$ value increases with increasing magnetic field, as larger magnetization induced by stronger field would lead to larger $−ΔS$ value according to the thermodynamic Maxwell relation. There are two peaks in the vicinity of temperature of spin reorientation for the ECO sample in Fig. 10(a). An abrupt change in magnetization near ordering temperature is usually expected as the orientation of its intrinsic magnetic moments changes, which results in a large $−ΔS$⁠.⁴⁵ The peak at T_SR1$∼$ 9 K could be attributed to the long-range spin reorientation of Cr³⁺ sublattice. The other one at T_SR2$∼$ 15 K could be explained by the anisotropic magnetic interaction between Er³⁺ and Cr³⁺.¹⁸ No anomaly is observed for EGCO in Fig. 10(b), indicating that the ordering temperature might be out of the measured temperature range (5–140 K). The maximum $−ΔS$ for ECO is 10.7 J kg⁻¹ K⁻¹ at 15 K and at 7 T. By comparison, $−ΔS$ for EGCO is much larger (27.6 J kg⁻¹ K⁻¹ at 5 K and at 7 T). At lower magnetic field of 1 T, the maximum magnetic entropy change is 1.4 and 3.2 J kg⁻¹ K⁻¹ for ECO and EGCO, respectively. This large MCE in EGCO is likely due to the Gd³⁺-Gd³⁺ magnetic interaction, which orders antiferromagnetically at 2.3 K, as the magnetization and magnetic entropy change increase rapidly towards the lowest temperature in both Figs. 9(d) and 10(b).²³ Generally, the enhancement is not only from the fact that the ground state of Gd³⁺ will provide the largest entropy per single ion, but also the replacement of Er³⁺ by Gd³⁺ ions will introduce extra magnetic coupling due to different magnetic moment and lattice distortion (change in bond angle, length).²⁶ The extra coupling causes a large change in magnetization resulting in a large magnetocaloric effect (MCE), which is shown in Figs. 9 and 10. The magnetocaloric data are also summarized in Table III, which shows that GdCrO₃ exhibits the largest MCE among all the rare earth chromites listed in the table. For Er_0.33Gd_0.67CrO₃ investigated in this work, the change in entropy and relative cooling power (RCP) values are comparable to those for GdCrO₃. Thus, in this work, we have shown that the Gd substitution in ErCrO₃ was successful in enhancing these properties. The observed large value of $−ΔS$ for EGCO suggests that it could be used as a potential MR material at low temperatures.

Relative cooling power (RCP) is another important factor to evaluate the efficiency of magnetic refrigeration. The RCP is a measure of the amount of heat transferred between the hot and cold reservoirs in one ideal refrigeration cycle and can be calculated by the following equation:

where T₁ and T₂ are the temperatures corresponding to the hot and cold reservoirs, respectively.²⁴ Figure 11 shows the field dependence of RCP values for the two samples studied. The field dependent RCP value is calculated based on the setting that T₁ = 5 K and T₂= 140 K. Note that the RCP value increases with increasing magnetic field for both ECO and EGCO, since stronger magnetic field results in larger $ΔS$ and thus leads to larger RCP value.¹⁵ The RCP value was calculated to be 416.4 and 531.1 J kg⁻¹ at an applied magnetic field of 7 T for ECO and EGCO, respectively, showing that the RCP value of EGCO is much higher than that of ECO. Thus, Gd substitution at the A-site in ECO was found to significantly enhance its magnetocaloric properties. This could be due to the additional $ΔS$ from Gd³⁺ ion as it exhibits the largest spin among all the rare earth elements.⁴⁶ And it is comparable with the RCP values of the most studied magnetocaloric materials mangnites.² With such large values of $ΔS$ and RCP, Gd doped ErCrO₃ might be utilized as refrigerant material for low temperature magnetic refrigeration.

To summarize, bulk powders of ErCrO₃ and Er_0.33Gd_0.67CrO₃ were successfully synthesized via citrate route and their structural, magnetic, and magnetocaloric properties were studied in detail. The orthorhombically distorted perovskite structure with the Pbnm space group and the phase purity was confirmed by the X-ray diffraction for both compositions. Magnetization measurements revealed that Gd substitution leads to an increase in Néel temperature (155 K) as compared to that of the parent ErCrO₃ sample (133 K). The magnetocaloric effect was calculated by measuring the magnetic entropy change (⁠$−ΔS$⁠) from the isothermal dc magnetization data. The maximum values of $−ΔS$ were 10.7 and 27.6 J kg⁻¹ K⁻¹ for the ErCrO₃ and Er_0.33Gd_0.67CrO₃ samples, respectively, at 7 T. The enhancement in the magnetocaloric effect of Gd doped sample could be due to the additional magnetic entropy change from Gd³⁺ ion. In addition, Gd substitution leads to an enhancement of the relative cooling power by 27.5%. The large entropy change and relative cooling power observed in Gd substituted ErCrO₃ suggest that it could be a promising material for low-temperature magnetic refrigeration.

This work was supported by a research grant from the U.S. National Science Foundation (Grant No. DMR-1310149).