Shiqi Yina,
Vinit Sharmab,
Austin McDannaldc,
Fernando A. Reboredob and
Menka Jain*ad
aDepartment of Physics, University of Connecticut, Storrs, CT 06269, USA. E-mail: menka.jain@uconn.edu
bMaterials Theory Group, Materials Science & Technology Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA
cMaterials Science and Engineering Department, University of Connecticut, Storrs, CT 06269, USA
dInstitute of Materials Science, University of Connecticut, Storrs, CT 06269, USA
First published on 11th January 2016
In this work, HoCrO3 and Fe substituted HoCrO3 and DyCrO3 (i.e. HoCr0.7Fe0.3O3 and DyCr0.7Fe0.3O3) powder samples were synthesized via a solution route. The structural properties of the samples were examined by Raman spectroscopy and X-ray diffraction techniques, which were further confirmed using the first-principle calculations. The dc magnetic measurements indicate that the Cr3+ ordering temperatures for the HoCrO3, HoCr0.7Fe0.3O3, and DyCr0.7Fe0.3O3 samples were 140 K, 174 K, and 160 K, respectively. The ac magnetic measurements not only confirmed the Cr3+ ordering transitions in these samples (obtained using dc magnetic measurements), but also clearly showed the Ho3+ ordering at ∼10 K in the present HoCrO3 and HoCr0.7Fe0.3O3 samples, which to our knowledge, is the first ac magnetic evidence of Ho3+ ordering in this system. The effective magnetic moments were determined to be 11.67μB, 11.30μB, and 11.27μB for the HoCrO3, HoCr0.7Fe0.3O3, and DyCr0.7Fe0.3O3 samples, respectively. For the first time, the magnetocaloric properties of HoCrO3 and HoCr0.7Fe0.3O3 were studied here, showing their potential for applications in magnetic refrigeration. In an applied dc magnetic field of 7 T, the maximum values of magnetic entropy change were determined to be 7.2 (at 20 K), 6.83 (at 20 K), and 13.08 J kg−1 K−1 (at 5 K) and the relative cooling power were 408, 387, and 500 J kg−1 for the HoCrO3, HoCr0.7Fe0.3O3, and DyCr0.7Fe0.3O3 samples, respectively.
Recently, another ME MF oxide system based on rare-earths, rare-earth chromite (RCrO3), has been explored for its MCE properties and suitability for MR. For example, in DyCrO3 (DCO), large MCE value of 8.4 J kg−1 K−1 and relative cooling power of 217 J kg−1 at 15 K and 4 T was first reported, which was attributed to the low-temperature ordering of Dy3+ at ∼2.16 K.15,20 This renders DCO useful for MR in the temperature range from 5 K to 30 K. These RCrO3 materials stabilize in orthorhombically distorted perovskite structure in which the exchange coupling between the Cr3+ nearest neighbors is predominantly antiferromagnetic (G-type) and these ions order magnetically at a Néel temperature (TCrN) from 113 to 140 K depending upon the R-ion.21 Additionally, RCrO3 systems are of great interest as these exhibit spin-reorientation, rare-earth ordering,22 metamagnetic transition, or temperature induced magnetization reversal in some cases at low temperatures (<50 K).23,24 In a similar system – rare-earth ferrite, for example, DyFeO3, the Dy3+ ordering has been reported to occur at 4.5 K17,25 and a giant entropy change at 5 K (around Dy3+ ordering) was reported to be 16.62 J kg−1 K−1 under field change of 2 T.10 Among the rare-earth ions, Ho3+ has the second highest magnetic moment after Dy3+ (10.4μB for Ho3+ as compared to 10.6μB for Dy3+).9 In HoFeO3 bulk powder, Ho3+ ordering has been reported to occur at 3.3 K or at 6.5 K,12,26 while in HoFeO3 single crystal, Ho3+ ordering was reported at 4.1 K and a spin reorientation was reported ∼50–60 K; so, large MCE value of 19.2 J kg−1 K−1 was obtained at the Ho3+ ordering temperature.27 Yin et al. reported Dy3+ ordering temperature at 14 K in DyCr0.5Fe0.5O3 and the maximum MCE value was improved to 10.5 J kg−1 K−1 at 5 K and 4 T.16 From above discussion, it is clear that HoCrO3 (HCO) is likely to show large MCE values in slightly higher temperature range than DCO due to slightly higher Ho3+ ordering temperature. Further, by Fe3+ substitution at the Cr-site, the ordering temperature of Ho3+ is expected to increase and correspondingly its MCE value may maximize at a higher temperature compared to that in pure HCO, rendering it applicable for magnetic refrigeration in slightly higher temperature range than those for DCO. As in RMnO3 system, Cr–O–Cr bond angle in RCrO3 system would play an important role in its magnetic properties (and hence MCE properties) that would be modified with either R-site or Cr-site substitutions.28,29 Therefore, in order to understand the structure–property correlations, it is of great importance to utilize first-principle to calculate the lattice parameters of the stable structure and density of states (DOS) complementary to the experimental work.
In the present work, the structural, magnetic (ac and dc), and MCE properties of the HCO, HoCr0.7Fe0.3O3, and DyCr0.7Fe0.3O3 bulk powder samples have been examined. In addition, the lattice distortions and density of states are studied using the first principle calculations based on density functional theory (DFT), which can provide crucial information that can lead to the design of materials with enhanced MCE properties. To our knowledge, this is the first work on the exploration of ac magnetic properties and MCE properties of HCO and Fe-substituted HCO system. Also, RCP value of Fe substituted DCO is reported for the first time in addition to the density of state calculations in the RCrO3 system. Given that experimental study of such complex systems is not only time consuming and costly but also require sophisticated experimental techniques. A combined experimental and first-principles computational methods based study is capable to provide crucial insights about the physicochemical properties resultant form defects/impurities that complements experiments. To our knowledge, this is the first combined experimental and computational attempt to explore ac magnetic properties, MCE properties and electronic structure of HCO and Fe-substituted HCO, which can potentially enhance the efforts towards synthesis and design of new ME MF materials.
In order to further understand the crystal and magnetic structure, DFT based spin-polarized first-principles calculations are performed using the projector augmented wave method as implemented in the Vienna ab initio simulation package.30–32 In present calculations, the exchange correlation interaction is treated within the generalized gradient approximation (GGA) using the Perdew–Burke–Ernzerhof (PBE) functional.33 The electronic wave functions were expanded in a plane wave basis with a cut off energy of 500 eV. It is noteworthy that due to the errors associated with the on-site Coulomb and exchange interactions,34 DFT based methods are known to fail to reproduce an accurate description of the electronic structure for strongly correlated systems such as transition metal oxides35–37 and rare-earth compounds.28 In such cases, the accuracy of DFT can be improved by incorporating a Hubbard-model-type correction (U), which accounts for localized d and f orbitals. Hence, in the present work to describe the localized nature of the f states, in all the calculations U values ∼3.7 eV and 3.9 eV are used for Ho and Dy, respectively.28,38,39 A Monkhorst–Pack k-point mesh of 5 × 5 × 4 is employed to produce converged results within 0.1 meV per formula unit. In doped cases, one Cr atom was substituted with dopants Fe. It should be mentioned that 25 at% was chosen for Fe concentration in DFT calculations not only to be close to experimental results but also to keep a reasonable super cell size. We expect the trend in physical properties of 25 at% doped samples to be similar as for 30 at% doped samples.
Parameter | HCO | HCFO | DCFO | |
---|---|---|---|---|
a (Å) | DFT (PBE+U) | 5.269 | 5.247 | 5.229 |
EXP | 5.248 | 5.259 | 5.280 | |
b (Å) | DFT (PBE+U) | 5.543 | 5.600 | 5.5808 |
EXP | 5.525 | 5.540 | 5.536 | |
c (Å) | DFT (PBE+U) | 7.585 | 7.567 | 7.555 |
EXP | 7.545 | 7.564 | 7.577 | |
V (Å3) | DFT (PBE+U) | 221.52 | 222.37 | 220.50 |
EXP | 218.79 | 220.40 | 221.48 | |
Crystallite size (nm) | 104.1 ± 13.3 | 142.9 ± 26.1 | 119.1 ± 14.7 | |
Strain | (6.8 ± 3.5) × 10−4 | (17.0 ± 3.6) × 10−4 | (10.3 ± 3.3) × 10−4 |
The crystallite size and strain of the present powder samples were estimated using Williamson–Hall (W–H) analysis, expressed by the formula:44
(1) |
Fig. 2 Scanning electron microscopy images of the (a) holmium chromite (HCO), (b) 30% iron substituted holmium chromite (HCFO), and (c) 30% iron substituted dysprosium chromite (DCFO) samples. |
Complementary to XRD, Raman spectra provides useful data of the phonon spectra and structural distortion of RCrO3.48 The room temperature Raman spectra of the three samples are shown in Fig. 3. RCrO3 with orthorhombic Pbnm structure possess 24 Raman active modes (7Ag + 5B1g + 7B2g + 5B3g).16,48 In Fig. 3, the mode assignments were done following the work by McDannald et al. and Yin et al.15,16 It should be noted that all the observed peaks in Raman spectra of the present samples could be assigned for the RCrO3 system. Strong peaks at ∼693 cm−1, 676 cm−1, and 676 cm−1 were observed for HCO, HCFO, and DCFO, respectively and were not reported in most cases for RCrO3, which are attributed to the antisymmetric stretching of FeO6 or CrO6 octahedra in RCrO3.16 Raman peaks are sensitive to impurities and structure of the material. In the present samples, no extra peaks of impurities (such as Fe2O3, Cr2O3, or Fe3O4, etc.) were observed in the Raman spectra. Thus, it is concluded that the present samples are phase pure, further corroborating the XRD results.
In order to investigate the interaction of orbital and magnetic ordering, we examine the electronic structure of the present samples by analyzing the density of states (DOS). Computed total and atom projected DOS of the pure and doped HCO (or DCO) samples are plotted in Fig. 4, where Fermi level is aligned to zero for convenience. As it can be seen in Fig. 4(a) and (b), both pure HCO and pure DCO are found to be insulator with energy gap of about 3.1 eV and 2.7 eV, respectively. It should be noted that the DFT calculated band gap here is close to the recently reported experimentally obtained energy-gap values for HCO (3.26 eV)49 and DCO (2.8 eV).50 The small difference between experimental and computed band-gap values is due to the well-known deficiency of conventional DFT methods in predicting band-gaps. Fig. 4 also clearly suggests that: (i) the valence band of the total DOS has contributions from both rare-earth and transition metal elements and (ii) in the valence band, the majority of the DOS in the vicinity of the Fermi-level arises from the d-states of the Cr/Fe atoms. The notable point is that the in valence band, highest occupied level shows O 2p character, while in conduction band the lowest unoccupied level has Cr 3d character. While the DOS in the conduction band can be explained in terms of optical conductivity spectra where the first peak mainly an attribute of the first optical transition as observed in the earlier optical conductivity spectra measurements.50 Furthermore, in the conduction band the DOS have contributions from both Cr and O atoms but mainly dominated by Cr (3d) orbitals. On the other hand in Fe doped HCO/DCO, the DOS in the vicinity of the Fermi-level are largely contributed by Fe, which result into a shift in valence band maximum. This shift can be explained on the basis of the hybridization of d-orbitals of Fe and Cr with p-orbitals of oxygen in the valance band.28 The O (2p) states and Cr/Fe (3d) states further enhance the strong hybridization between the orbital and spin order resulting in the magnetic and structural modulations, consistent with the Jahn–Teller mechanism.28
The temperature dependence of the dc magnetization (mass) with an applied magnetic field (H) of 50 Oe measured in both zero-field cooled (ZFC) and field cooled (FC) mode are exhibited in Fig. 5. The Néel temperature (Cr3+ ordering temperature, TCrN) was observed at 140 K, 174 K, and 160 K for HCO, HCFO and DCFO samples, respectively. As it can be seen that the TCrN of HCFO and DCFO were higher than those of pure HCO and DCO, respectively,15 which is attributed to the effect of Fe substitution. It is worth noting that the present DCFO sample shows a lower TCrN than 261 K reported recently for DyCr0.5Fe0.5O3.16 This indicates that the TCrN is tunable in DyCr1−xFexO3 (similarly for HoCr1−xFexO3) solid-solution by controlling the Cr3+/Fe3+ ratio. In addition to the Néel temperature, another transition at 10 K was observed for DCFO, which can be attributed to the ordering of Dy3+.16 However, HCO or HCFO samples did not show the Ho3+ ordering in the temperature dependent dc magnetic data. The magnetization (M) for the present HCO sample (max ∼ 30 emu g−1) is consistent with that reported by Tiwari et al.,40 but higher than the present HCFO sample (max ∼ 21 emu g−1). It should be noted that Shao et al. reported that in HoFeO3 single crystal, the maximum magnetization value was ∼4.5 emu g−1 at 100 Oe, which is much lower than that of HCO.27 Thus, the reduction in magnetization of the present HCFO samples could be due to the iron substitution. The magnetization value of DCFO was in good agreement with the report of 50% Fe substituted DCO and the magnetic susceptibility of both are ∼0.02 emu (g Oe)−1 {calculated using χ = M/H, where M is the magnetization (mass) and H is the applied magnetic field}.16 However, the maximum magnetization of the present DCFO is much smaller than those of HCFO or HCO samples because of the ordering of Dy3+ ions (see Fig. 5(c)).
The dc susceptibility data in the FC mode of the samples was fitted by the Curie–Weiss law (χ = C/(T − θ)) in the paramagnetic region (above TCrN), as shown in Fig. 6(a–c). Curie constant (C) and Weiss temperature (θ) were obtained for each sample and presented in Table 3. The effective magnetic moment (μeff) was then calculated from C values using:9
(2) |
(3) |
Sample | HCO | HCFO | DCFO |
TCrN (K) | 140 | 174 | 160 |
θ (K) | −36.47 ± 0.60 | −15.31 ± 3.03 | −25.71 ± 2.12 |
C (emu K (Oe−1 mol−1)) | 17.01 ± 0.04 | 15.96 ± 0.19 | 15.88 ± 0.14 |
μeff (μB) | 11.66 ± 0.01 | 11.30 ± 0.07 | 11.27 ± 0.05 |
μ′eff (μB) | 11.23 | 11.35 | 11.44 |
The temperature dependent ac susceptibility data (real part χ′, imaginary part χ′′ in Fig. 7) was measured with an applied ac magnetic field of 10 Oe and frequencies between 100 and 1000 Hz. Both the χ′(T) and χ′′(T) data revealed Cr3+ ordering temperatures at 140 K, 174 K, and 160 K for HCO, HCFO, and DCFO, respectively corroborating the dc magnetic results presented above. In addition, an ordering temperature at ∼10 K was observed for DCFO sample, indicative of the Dy3+ ordering as observed in the dc magnetic data. It should be noted that χ(T) data for HCFO (Fig. 7(a) and (b)) revealed anomaly ∼10 K, which was not observed in dc magnetic data or χ′′(T) data of the sample. This anomaly is indicative of Ho3+ ordering in the samples. To the best of our knowledge, the present data is the first ac magnetic data in literature showing ordering of the Ho3+ moments in pure or doped HCO.
In order to investigate the dependence of magnetic property on magnetic field, isothermal magnetization vs. magnetic field (M vs. H) curves were measured up to 4 T and 160 K, and representative data at 5 K, 50 K, 100 K, and 160 K are shown in Fig. 8. The magnetic behavior of all the samples changes from canted antiferromagnetic (AFM) at low temperature to paramagnetic at high temperature (above their respective TCrN), which can be interpreted as the superposition of three types of magnetic contributions: (i) the weak ferromagnetic contribution that can be attributed to the canting of the AFM order of the transition metals (Fe or Cr), (ii) the strong paramagnetic contribution from the rare-earth sublattice, and (iii) pure AFM contribution from the transition metal (Dy or Ho) sublattice. From the isothermal M–H data of all the samples, the temperature dependence of the coercive magnetic field (Hc) and remnant magnetization (MR) values were obtained and plotted in Fig. 9. As the temperature increases, the coercive field increases initially and maintains at some level. Then it decreases slowly and becomes zero at ∼140 K, 160 K, and 170 K for HCO, HCFO, and DCFO near their TCrN, respectively. The Hc value of DCFO sample is much smaller than those of HCO and HCFO samples, indicating that DCFO has much smaller magnetic hysteresis. Comparatively, the MR (Fig. 9(b)) decreases monotonically with increasing temperature and reaches zero at ∼145 K for all the present samples. All of these features can be interpreted by the competition between the three aforementioned magnetic contributions. When the temperature is near TCrN, the strong paramagnetic signal plays the dominant role in the magnetic behavior, so the samples show no magnetic hysteresis and both Hc and MR are zero.
In order to further examine the figure of merit of these materials for the evaluation of their applications in MR, the present samples were also evaluated for their MCE behavior, which can be extracted from the isothermal M–H curves exhibited in Fig. 10 (measured up to 7 T field and only in first quadrant as mentioned in Experimental section). The MCE properties can be characterized mainly by two factors: magnetic entropy change ΔSM(T, H) given by:8
(4) |
RCP = |ΔSmax| × ΔTFWHM, | (5) |
RCP = −∫|ΔSM,H|dT | (6) |
In Fig. 11, the MCE values (ΔSM(T, H)) were calculated and determined to be 7.2 J kg−1 K−1 at 20 K for HCO, 6.83 J kg−1 K−1 at 20 K for HCFO, and 13.08 J kg−1 K−1 at 5 K for DCFO at maximum under the magnetic field of 7 T. The MCE values of HCO and HCFO are reported for the first time here and the maximum values were smaller than those of DCO and DCFO.13,36 It is partly because larger magnetic hysteresis exists in HCO or HCFO than in DCO or DCFO (see Fig. 9) and thus more energy is lost in the thermal process, resulting in smaller MCE values. This is supported by the report of Phan et al., in which they propose nearly zero magnetic hysteresis as a criteria to select material for magnetic refrigerant because of energy efficiency.8 Also, the MCE value of HCFO (6.83 J kg−1 K−1) is slightly smaller than that of HCO (7.2 J kg−1 K−1), which indicates that Fe substitution decrease the MCE values in HCO. Conversely, the MCE value of DCFO (10.3 J kg−1 K−1), which is close to the report of DyCr0.5Fe0.5O3 (10.5 J kg−1 K−1),16 was larger than that of pure DCO (8.4 J kg−1 K−1) under the magnetic field of 4 T.15 Therefore, it was inferred that Fe substitution in DCO improves the MCE values. In Table 4, the MCE values, temperature (Tmax) and magnetic field (Hmax) where the maximum MCE values were obtained were summarized and compared to those reported in other references. Tmax is ∼20 K for both HCO and HCFO samples and ∼5 K for DCFO sample. Such difference in the temperature of maximum ΔSM value can be explained by the slightly higher ordering temperature of Ho3+ than that of Dy3+, as presented in Fig. 7. For HCO, another peak in ΔSM value, though much weaker, was observed at 140 K. It was attributed to the ordering of Cr3+, which has much smaller magnetic moment than Ho3+ (seen in Table 2). In Table 4, the MCE properties of the bulk samples were much smaller than HoFeO3 and DyFeO3 single crystals (19.2 and 16.62 J kg−1 K−1). Because the MCE properties of the single crystals were shown to be direction dependent (as measured in other cases),27 the bulk samples show only the average effect and smaller MCE values.
Material | ΔSM,max (J kg−1 K−1) | Tmax (K) | Hmax (T) | RCP (J kg−1) | Reference |
---|---|---|---|---|---|
DyCrO3 | 8.4 | 15 | 4 | 217 | 15 |
DyCr0.5Fe0.5O3 | 10.5 | 5 | 4 | — | 16 |
DyCr0.7Fe0.3O3 | 13.08 | 5 | 7 | 500 | This work |
10.3 | 5 | 4 | 258 | ||
HoCrO3 | 7.2 | 20 | 7 | 408 | |
4.2 | 20 | 4 | 189 | ||
HoCr0.7Fe0.3O3 | 6.83 | 20 | 7 | 387 | |
3.74 | 15 | 4 | 167 | ||
HoMnO3 | 12.5 | 10 | 7 | 312 | 19 |
DyFeO3 (single crystal) | 16.62 | 5 | 2 | 150 | 10 |
18.5 | 5 | 7 | 586 | ||
HoFeO3 (single crystal) | 19.2 | 4.5 | 7 | 220 | 27 |
RCP values of the present samples were calculated and plotted in Fig. 11(d), and also are compared with references at two different fields (4 T and 7 T) in Table 4. At 7 T, the RCP value of HCO sample (408 J kg−1) is larger than those of the present HCFO (387 J kg−1) and previously reported HoMnO3 (312 J kg−1),19 but smaller than the present DCFO (500 J kg−1). Further, at the lower magnetic field of 4 T, DCFO sample shows larger RCP value (258 J kg−1) than that of previously reported DCO sample (217 J kg−1).15 It is worth noting that the RCP values of HCO, HCFO, and DCFO generally follow the same trend as their ΔSM (see Table 4), because RCP value is obtained by the integration of the ΔSM over temperature (see eqn (6)), and larger ΔSM is more likely to result in larger RCP value. However, RCP also depends on the width of the ΔSM versus temperature data and larger value of full width at half maximum is also more likely to result in larger RCP value. That explains why HCO bulk sample shows smaller entropy change than HoMnO3 bulk sample and HoFeO3 single crystal, but still larger RCP value (Table 4).
Interestingly, HCFO showed smaller MCE and RCP values than HCO, while DCFO showed larger MCE and RCP values than DCO, so the effect of Fe substitution on the MCE property of rare-earth chromites varies for different rare-earth ions. From Table 4, it is clear that HCO and HCFO samples show decent MCE and RCP values at slightly higher temperature (20 K) than that in DCO and HoMnO3 (<10 K). Thus, the HCO and HCFO samples are considered suitable for MR application in slightly higher temperature range (∼20 K).
This journal is © The Royal Society of Chemistry 2016 |