Introduction

Magnetic refrigeration based on MCE is an environment-friendly, high-efficiency technology compared to the traditional gas-cycle refrigeration1,2,3,4. After the discovery of the first magnetic refrigeration prototype near room temperature5 and the giant MCE in Gd5(Si2Ge2)6,7, a large MCE has been realized in a lot of materials in the past two decades, such as ReCo2 (Re = Er, Ho and Dy) alloys8,9, manganite oxides (Re, M)MnO3 (Re = Lanthanide, M = Ca, Sr and Ba)10,11, Ni-Mn-X (X = Ga, In and Sn) based Heusler alloys12,13,14,15,16, MnAs based compounds3,17,18,19, La(Fe, Si)13 and related compounds20,21,22,23, as well as rare earth based intermetallic compounds24,25,26,27,28,29,30. Amongst the families of MCE materials, the compounds with first-order magnetic transition (FOMT) have been found promising due to their large and/or sharp changes in magnetization and the strong coupling between crystallographic structure and magnetism, such as Gd5Ge4−xSix7, MnAs1−xSbx18,31, MnFe(As, P, Si, Ge)17,32, LaFe13−xSix(Hδ)14,21,23 and Heusler-type magnetic shape-memory alloys14,16. However, in these materials, the magnetic transitions are frequently accompanied by significant thermal and/or magnetic hysteresis, which would limit the life span of refrigerants or even make the refrigeration cycle impossible3,33. In order to reduce or even eliminate the magnetic hysteresis losses, there have been two strategies. One is to treat the giant MCE materials by special methods, such as microstructure-tuning, as porosity34, fragmentation35,36,37, melt-spun38,39,40, or chemical tuning, as doping41. Another is to search for new high performance compounds with SOMT42. However, the performances in materials with SOMT are rather modest when compared with that with FOMT. It is therefore interesting to search for new FOMT materials with low-level hysteresis and without any additional treatments.

The crystal structure and physical properties of TlFe3Te3 were reported by two groups in 198443,44. TlFe3Te3 crystalizes in a hexagonal structure with space group P63/m, which consists of one-dimensional metallic cluster |Fe3Te3|∞ chains along the hexagonal c-axis, separated by the parallel chains of Tl atoms (see Fig. 1). The authors concluded that the compound undergoes a first-order transition from paramagnetic to ferromagnetic at 220 K based on their physical property measurements. However, neither of them observed discernible thermal and field hysteresis. Since the absence of hysteresis is appealing for magnetic refrigerant, in this report, we recheck the type of the magnetic transition and elucidate the MCE of TlFe3Te3 by performing resistivity and magnetization measurements. We found that this compound exhibits a large MCE with a small magnetic field change, ΔH and with a low-level thermal and field hysteresis, thus identifying it to be another class of solids for the magnetic refrigerants.

Figure 1
figure 1

The powder X-ray diffraction (XRD) pattern (black star: observed data; red line: calculated curve; green line: background; blue line: difference; wine bar: Bragg positions) and the crystal structure of TlFe3Te3 viewed along c-axis (red ball: Tl; dark yellow ball: Te; green ball: Fe).

Full size image

Results and Discussion

Figure 1 presents the powder x-ray diffraction (XRD) pattern of TlFe3Te3 and its Rietveld refinement. All the diffraction peaks could be indexed by a hexagonal structure with space group P63/m. The lattice parameters a = 9.355(1) Å and c = 4.224(5) Å were obtained by the refinement, which are in good agreement with previous reports44. The electron probe micro-analyzer (EPMA) experiments performed on several single crystals verified that the sample composition (the average atomic ratio) is of Tl : Fe : Te = 0.99(1) : 2.95(2) : 3.00(1), which is in consistent with the nominal composition. The temperature dependence of electrical resistivity along c-axis, ρ(T), for a TlFe3Te3 crystal is shown in Fig. 2(a). In the whole measuring temperature range, the positive resistivity-temperature coefficient of ρ(T) indicates its metallic behavior. The resistivity has a very sharp drop at 220 K with detectable thermal hysteresis [see the inset of Fig. 2(a)], which is associated with the first-order ferromagnetic transition. The resistivity at 300 K and 1.8 K are of 120 μΩ cm and 1.8 μΩ cm, respectively. The small resistivity should be viewed as a merit since a good thermal conductivity is required for a high performance magnetic refrigerant material45. Both a rather low residual resistivity and a considerable large residual resistivity ratio (RRR) = 67 indicate that our crystals are of high quality.

Figure 2
figure 2

(a) The temperature dependence of resistivity with a current applied parallel to c-axis and the expansion near the transition temperature (inset). (b) The temperature dependence of magnetization, M(T), for both H || c-axis and H ⊥ c-axis. The M(T) near the transition temperature for (c) H || c-axis, (d) H ⊥ c-axis, the arrows show the cooling and heating process during measurements.

Full size image

Figure 2(b) shows the magnetization as a function of temperature, M(T), measured from 2 to 300 K in an applied magnetic field H = 1000 Oe, aligned both || and ⊥ the c-axis, with a field cooling process. A sharp increase of M for both directions at the Curie temperature, TC ~ 220 K, confirms the occurrence of a ferromagnetic transition. Larger magnetization along c-axis suggests that the easy axis of magnetization is in the c axis. As discussed by Uhl et al.43 and Pelizzone et al.44, the strong magnetic anisotropy observed in the ferromagnetic state is certainly related to its peculiar structure being composed of |Fe3Te3|∞ chains, whose central part is a column of edge-sharing octahedral Fe clusters. The Fe-Fe distance of 2.6 Å within the clusters are comparable to the interatomic distance in metallic iron, while the nearest two Fe atoms belong to different |Fe3Te3|∞ chains are 6.7 Å apart. Thus, a strong anisotropy of the exchange coupling is to be expected. As shown in Fig. 2(c,d), it is clear that the M(T) curves near TC exhibit a small thermal hysteresis for both directions, which is in contrast to that reported by Uhl et al.43 and Pelizzone et al.44, who did not observe any hysteresis in their measurements. We observed a distinguishable but very small hysteresis, (i.e., the hysteresis temperature ΔThy = 0.2 K for H || c-axis and 0.1 K for H ⊥ c-axis), which suggests that a first-order ferromagnetic transition occurs at ~220 K.

In order to further identify the type of the transition and to explore the MCE, we performed the isothermal magnetization measurements near the TC. Figure 3 shows the magnetization as a function of magnetic field, M(H), measured at various temperatures around TC with both H || c-axis and H ⊥ c-axis and with both increasing and decreasing magnetic field. A small magnetic hysteresis was again observed. The maximum hysteresis is 50 Oe for H || c-axis [see Fig. 3(a)], while for H ⊥ c-axis, the hysteresis is rather small and becomes even indiscernible [see Fig. 3(b)]. The M(H) curves for both H || c-axis and H ⊥ c-axis exhibit a different behavior, which is associated with the large anisotropy of magnetization discussed above. The M2 versus H/M curves for both directions are shown in Fig. 3(c,d), respectively. According to the Banerjee criterion46, the curves at some temperatures have a negative slope and a inflection, which confirms further the occurrence of the first-order ferromagnetic transition around 220 K. The small hysteresis in M(H) curves enables us to use the Maxwell equation to estimate the isothermal magnetic entropy change (ΔSM). The ΔSM is calculated by a formula:

Figure 3
figure 3

The isothermal magnetization near TC as a function of magnetic field, M(H), measured with a temperature step of 1 K for H (a) || and (b) ⊥ the c axis. The arrows indicate the measurements with increasing and decreasing magnetic field process. The corresponding M2 vs H/M curves for H (c) || and (d) ⊥ the c axis.

Full size image

which is an approximation of the integral form of the Maxwell equation.

Figure 4(a,b) present the temperature dependence of −ΔSM with the magnetic field changes ΔH up to 0–5 T, for both H || c-axis and H ⊥ c-axis. For H || c-axis, the −ΔSM(T) curve with ΔH = 0–1 T shows a pronounced peak around TC and a table-like behavior can be observed in the −ΔSM(T) curves with ΔH = 0–2 T and 0–3 T, i.e., there is a temperature range corresponding to the maximum value of magnetic entropy change, which is beneficial for application. With ΔH = 0–1, 0–2, 0–3, 0–4 and 0–5 T, −Δ = 5.9, 7.0, 8.2, 8.5 and 8.9 J/kg K, respectively, which increases continuously with the increasing field change and tends to almost saturate at higher magnetic field change. It is known that a “table-like” behavior and no strong ΔH dependence of −Δ value are the typical behaviors for FOMT materials2,45. Although −Δ values are smaller than that for the some giant MCE materials (see Table 1), these values of TlFe3Te3 are comparable with the most potential magnetic refrigerant materials with the a first-order ferromagnetic transition (see Table 1). For the H ⊥ c-axis case, all the −ΔSM(T) curves with different ΔH values exhibit a peak around TC without table-like behavior and the maximum value of magnetic entropy change −Δ is smaller than that for the H || c-axis. The anisotropy of MCE may origin from the peculiar magnetic structure, as discussed above.

Table 1 Comparison of the MCE properties with some representative materials with a similar magnetic transition temperature.
Full size table
Figure 4
figure 4

The magnetic entropy change as a function of temperature, −ΔSM(T), around TC, with the different field change ΔH = 0–1, 0–2, 0–3, 0–4 and 0–5 T for H (a) || and (b) ⊥ the c axis. Thot −Tcold in (a) represents the full width at half maximum in −ΔSM(T) curve for ΔH = 0–1 T.

Full size image

Another important quality factor of magnetic refrigerant materials is the relative cooling power (RCP) or/and refrigeration capacity (RC), defined29 usually as the product of −Δ and the full width at half maximum in the −ΔSM(T) curve, as an example, i.e., Thot − Tcold for ΔH = 0–1 T in Fig. 4(a). RCP/RC is a measurement of the amount of heat transfer between the cold and hot reservoirs in an ideal refrigeration cycle. Due to the limitation of data measured in our experiments, we only estimated that the RCP values for the ΔH = 0–1, 0–2 and 0–3 T, are of 13, 50 and 74.6 J/kg, respectively. Recently, as a figure of merit for the magnetic refrigerant materials, the dimensionless materials efficiency47,48, η = |Q/W|, is taken into consideration, where electrical or mechanical work, W, is done to drive highly reversible caloric effects in an isothermal body, whose entropy is thus modified such that heat, Q, flows to (Q < 0) or from (Q > 0). Here, we estimated the mass-normalized values of |W| by integrating −μ0MdH0 from the M(H0) data at TC and evaluated the mass-normalized value of heat Q by integrating μ0T0(∂M/∂T)H with respect to H from the M(H0) data at TC, which follows from the Maxwell relation μ0(∂M/∂T)H = (∂S/∂H)T. The materials efficiency η values at TC was estimated to be of 65.7, 32.0, 23.2, 17.1 and 13.9 for ΔH = 0–1, 0–2, 0–3, 0–4 and 0–5 T, respectively.

As a comparison of MCE properties, we choose several compounds with a similar magnetic transition temperature, TM, as well as some typical materials with a near room temperature, TM, focusing on the performence under ΔH = 0–2 T (the maximum magnetic field generated by a permanent magnet is about 2 T). As listed in Table 1, although the −Δ of TlFe3Te3 is less than that in the some pronounced materials with FOMT, such as GdSi2Ge2, MnFeP0.45As0.55, LaFe11.7Si1.3 and 20-LaFe11.57Si1.43 materials, −Δ of TlFe3Te3 is significantly larger than that with SOMT. Both the RCP and η values of TlFe3Te3 are comparable with the most MEC materials, except for some special compounds, such as Tb5Si4, LaFe11.7Si1.3, GdSi2Ge2 and MnFeP0.45As0.55. Besides having a larger ΔSM, TlFe3Te3 has some other advantages, such as a rare-earth-free element, a low synthesis temperature, as well as a low-level hysteresis in the as-grown crystals. But it should be pointed out that the toxicity of Tl element is not so good for the commercial utilization, which may be improved by the replacement of In, Ba, K for Tl in the future.

In summary, after successfully growing TlFe3Te3 single crystals, we carried systematically out the measurements of its resistivity and magnetization to investigate the nature of the magnetic phase transition and the MCE. It was found that TlFe3Te3 is a FOMT metal with TC = 220 K and has a small thermal and field hysteresis near TC. The relative large MCE at a low ΔH makes this compound a promising candidate for magnetic refrigeration around 220 K. Further efforts should be done to substitute Tl by other nontoxic elements in order to utilize this type of materials widely.

Methods

Single crystals of TlFe3Te3 were grown using a self-flux method. A mixture with a ratio of Tl:Fe:Te = 1:3:3 was placed in an alumina crucible, sealed in an evacuated quartz tube, heated at 923 K for 5 days. The product was a black powder from which needle-like single crystals with a typical dimension of ~0.4 × 0.4 × 4 mm3 could be isolated. Powder XRD measurements on crushed single crystals were carried out at room temperature on a PANalytical x-ray diffractometer (Model EMPYREAN) with a monochromatic Cu Kα1 radiation to identify the phase purity and the crystal structure. The composition was confirmed by an electron probe micro-analyzer (EPMA) (Jeol JXA-8100). The magnetic measurements were performed on a Quantum Design Magnetic Property Measurement System (SQUID-VSM, MPMS-5) and the resistivity measurements were carried out on a Physical Property Measurement System (PPMS-9).

Additional Information

How to cite this article: Mao, Q. et al. Large low field magnetocaloric effect in first-order phase transition compound TlFe3Te3 with low-level hysteresis. Sci. Rep. 6, 34235; doi: 10.1038/srep34235 (2016).