Observation of a Broadened Magnetocaloric Effect in Partially Crystallized Gd60Co40 Amorphous Alloy

To investigate the effect of crystallization treatment on the structure and magnetocaloric effect of Gd60Co40 amorphous alloy, the melt-spun ribbons were annealed at 513 K isothermally for 20, 40 and 60 min. The results indicate that, with increasing annealing time, the Gd4Co3 (space group P63/m) and Gd12Co7 (space group P21/c) phases precipitated from the amorphous precursor in sequence. In particular, in the samples annealed for 40 and 60 min, three successive magnetic transitions corresponding to the phases of Gd4Co3, Gd12Co7 and remaining amorphous matrix were detected, which induced an overlapped broadened profile of magnetic entropy change (|ΔSM|) versus temperature. Under magnetic field changing from 0 to 5 T, |ΔSM| values of 6.65 ± 0.1 kg−1·K−1 and 6.44 ± 0.1 J kg−1·K−1 in the temperature spans of 180–196 K and 177–196 K were obtained in ribbons annealed for 40 and 60 min, respectively. Compared with the fully amorphous alloy, the enhanced relative cooling power and flattened magnetocaloric effect of partially crystallized composites making them more suitable for the Ericsson thermodynamic cycle.


Introduction
As a potential alternate of conventional vapor compression refrigeration, magnetic refrigeration on the basis of magnetocaloric effect (MCE), has attracted a great deal of attention with the advantages of compactness, higher energy efficiency, environmental friendliness and less noise [1,2]. At constant pressure, the total entropy S of a magnetic material is composed of magnetic entropy S M , lattice entropy S L and electronic entropy S E , among which the S M depends on both temperature and applied magnetic field strongly, while the S L and S E usually can be considered temperature dependent only [3].
When a ferromagnetic substance is magnetized isothermally, the alignment of magnetic moment causes enhanced magnetic order and lower S M , and then the system releases heat to the surrounding environment since both S L and S E remain constant. If the magnetizing process is adiabatic, to maintain the total entropy unchanged, the S L and S E increase and the temperature becomes higher [4]. The above mentioned is the principle of MCE, and it is reversible for the demagnetization process. Furthermore, the isothermal magnetic entropy change ∆S M and adiabatic temperature change ∆T ad are important parameters to characterize the MCE of magnetic refrigerants [4].
Compared with the reference Carnot cycle, the Ericsson cycle consists of two isothermal and two iso-field processes and was proposed to be utilized for temperature ranges higher than 20 K. The influence of S L and S E can be neglected in the two iso-field processes by adding a regenerator to the magnetic refrigeration system [3]. As an ideal material for the Ericsson cycle, its magnetic entropy change |∆S M | should be a constant value in the of 17.62/81.46 under a Ti gettered argon atmosphere. The alloy ingots were turned over and remelted four times to ensure the homogeneity. Secondly, the ingots were broken into small pieces of 3-4 g, and then the ribbon samples were manufactured by a single-roller melt spinning technique with a surface linear speed of 50 m/s under high-purity argon atmosphere.
The width and thickness of the ribbons were 2-3 mm and 20-50 µm, respectively. In this work, the partially crystallized samples were obtained by annealing the as-spun ribbons at 513 K for 20, 40 and 60 min, and the fully crystallized counterpart was produced through thermal treating at 653 K (the termination temperature of crystallizing exothermic peak on the heat flow curve) for 20 min.
X-ray diffraction (XRD, Bruker D8 Advance, Karlsruhe, Germany) measurements were performed at room temperature using Cu Kα radiation (λ = 0.154178 nm) and a 2θ range of 20-90 • with operation conditions of 40 keV and 150 mA. The thermal properties of the ribbons were characterized using differential scanning calorimetry (DSC, Netzsch STA 449F3, Selb, Germany) under the protection of an argon atmosphere with a heating rate of 0.33 K/s. The magnetic properties of the ribbons were detected by a physical property measurement system (PPMS, Quantum Design PPMS-9 T system, San Diego, CA, USA).
In this study, the temperature dependence of the zero-field cooling magnetization (M-T) curve was collected under an applied field of 0.02 T during the heating process from 50 to 350 K. The isothermal initial magnetization (M-H) curves were collected under the applied magnetic field change from 0 to 5 T at selected temperatures from 108 to 248 K. The temperature interval of 4 and 10 K were chosen for the region in vicinity of and far away from T C , respectively, and the scanning speed of the magnetic field was slow enough to ensure accurate recording of the data in the isothermal process. Then, the magnetic entropy change |∆S M | can be calculated by the M-H curves based on the Maxwell relation [2]: where S M , T, H and M indicate the magnetic entropy, temperature, applied magnetic field and magnetization of the material, respectively. To derive the temperature dependence of |∆S M |, the numerical approximation of the integral is usually utilized as follows [30,31]: where M i and M i+1 are experimental values of magnetization at temperatures T i and T i+1 under external magnetic field H i .

Sturctural and Thermal Characterization
From the XRD pattern of the as-spun Gd 60 Co 40 alloy shown in Figure 1a, there is only one typical diffuse broaden peak without any detectable crystalline peaks, indicating its amorphous structure. This feature was verified by the two exothermal crystallization peaks on DSC curve of the Gd 60 Co 40 as-spun ribbon exhibited in Figure 1b. Additionally, the onset crystallization temperature T x and the end temperature of crystallization peak with values of 523 and 653 K can be obtained, respectively. Based on these, the annealing temperatures of 513 K (10 K lower than T x ) and 653 K were chosen. As displayed in Figure 1b, the absence of an exothermal peak on the DSC curve of the sample annealed at 653 K for 20 min demonstrates that it is fully devitrified. Then, according to the XRD result (in Figure 1a), two types of crystalline phases Gd 4 Co 3 (space group P6 3 /m) and Gd 12 Co 7 (space group P2 1 /c) were identified. From Figure 1a, the sharp peaks in XRD patterns of the samples annealed at 513 K for 20, 40 and 60 min indicate the existence of crystals. Combined with the exothermic crystallization process on DSC curves illustrated in Figure 1b, we deduced the microstructure of the alloys as crystals embedded in an amorphous matrix. In the 513 K/20 min annealed ribbon, a broad hump overlapped by one obvious crystalline peak was detected, demonstrating that the precipitation amount is slight, and the crystallites can be identified as Gd4Co3 phase in comparison with the fully crystallized sample. In the 513 K/40 min and 513 K/60 min counterparts, complicated diffraction peaks reveal that the Gd4Co3 and Gd12Co7 phases in-situ crystallized from the amorphous precursor.
Although the annealing temperature is lower than the Tx, the precipitation of Gd4Co3 and Gd12Co7 crystalline phases likely corresponds to the first and second crystallization peaks, respectively, due to heat fluctuation or thermal inertia during crystallization treatment [11]. The DSC curves of the annealed ribbons confirm this assumption, since the primary crystallization peak becomes broader (513 K for 20 min) and then disappears (513 K/40 min and 513 K/60 min) [32,33]. Moreover, with increasing annealing time, the area under the peak gradually decreases, indicating a reduction of the relative content of the amorphous phase in the whole composite material [33].
Utilizing the Scherrer formula [34], grain sizes of ~23 ± 3 nm were estimated for the crystallites in these amorphous-nanocrystalline composites, which were almost unaffected by the annealing time. Furthermore, the similar XRD patterns and DSC curves of the 513 K/40 min and 513 K/60 min annealed ribbons imply their analogous microstructure. In another word, the grain size and transformation volume tend to saturate after certain annealing time, which is attributed to the metastable equilibrium between the remaining amorphous matrix and the crystallites [35].  From Figure 1a, the sharp peaks in XRD patterns of the samples annealed at 513 K for 20, 40 and 60 min indicate the existence of crystals. Combined with the exothermic crystallization process on DSC curves illustrated in Figure 1b, we deduced the microstructure of the alloys as crystals embedded in an amorphous matrix. In the 513 K/20 min annealed ribbon, a broad hump overlapped by one obvious crystalline peak was detected, demonstrating that the precipitation amount is slight, and the crystallites can be identified as Gd 4 Co 3 phase in comparison with the fully crystallized sample. In the 513 K/40 min and 513 K/60 min counterparts, complicated diffraction peaks reveal that the Gd 4 Co 3 and Gd 12 Co 7 phases in-situ crystallized from the amorphous precursor.

Multi-Magnetic Phase Transition
Although the annealing temperature is lower than the T x , the precipitation of Gd 4 Co 3 and Gd 12 Co 7 crystalline phases likely corresponds to the first and second crystallization peaks, respectively, due to heat fluctuation or thermal inertia during crystallization treatment [11]. The DSC curves of the annealed ribbons confirm this assumption, since the primary crystallization peak becomes broader (513 K for 20 min) and then disappears (513 K/40 min and 513 K/60 min) [32,33]. Moreover, with increasing annealing time, the area under the peak gradually decreases, indicating a reduction of the relative content of the amorphous phase in the whole composite material [33].
Utilizing the Scherrer formula [34], grain sizes of~23 ± 3 nm were estimated for the crystallites in these amorphous-nanocrystalline composites, which were almost unaffected by the annealing time. Furthermore, the similar XRD patterns and DSC curves of the 513 K/40 min and 513 K/60 min annealed ribbons imply their analogous microstructure. In another word, the grain size and transformation volume tend to saturate after certain annealing time, which is attributed to the metastable equilibrium between the remaining amorphous matrix and the crystallites [35]. From the magnified image of the 210-230 K part on dM/dT-T curves (displayed in the inset of Figure 2b), there is a weak peak for 513 K/20 min annealed ribbon, which reveals first magnetic phase transition at 219 K correlated to the magnetic transition of the Gd 4 Co 3 phase (T C = 220 K) [21]. Its second magnetic transition obtained at 196 K is sharp and associated with the amorphous ferrimagnetic phase (T C = 193 K) [17], which is in good agreement with the microstructure consisted by slight amount of crystallites and predominant amorphous matrix, as discussed in Section 3.1.

Multi-Magnetic Phase Transition
Curie temperature TC was defined as the temperature corresponding to the minimum of the derivative dM/dT-T curves shown in the Figure 2b. It is evident that all the composite materials experienced two or more magnetic phase transitions during heating.
From the magnified image of the 210-230 K part on dM/dT-T curves (displayed in the inset of Figure 2b), there is a weak peak for 513 K/20 min annealed ribbon, which reveals first magnetic phase transition at 219 K correlated to the magnetic transition of the Gd4Co3 phase (TC = 220 K) [21]. Its second magnetic transition obtained at 196 K is sharp and associated with the amorphous ferrimagnetic phase (TC = 193 K) [17], which is in good agreement with the microstructure consisted by slight amount of crystallites and predominant amorphous matrix, as discussed in Section 3.1. Owing to the analogous microstructure of the 513 K/40 min and 513 K/60 min annealed samples, their multi-magnetic transition behaviors are similar to each other and three transitions can be observed at temperatures 174 K/194 K/219 K and 176 K/195 K/219 K. Compared with the magnetic transition behavior in the 513 K/20 min annealed sample, an extra transition related to the Gd12Co7 crystalline (TC = 160.8 K) appeared [36]. The deviation of the TC is possibly ascribed to the different microstructure (e.g., crystal size and surrounding phase structure) between the bulk crystalline material and the in-situ precipitated crystallites, in addition, similar phenomena and values of TC were reported in Gd12Co7 melt-spun ribbon (TC = 179 K) [22].
For the fully devitrified ribbon annealed at 653 K for 20 min, the dM/dT-T curve manifested three magnetic phase transitions at 162 K and 218 K associated with Gd12Co7 and Gd4Co3, respectively, which is consistent with the XRD results.

Magnetocaloric Properties
The isothermal initial magnetization M-H curves under the magnetic field change ΔH = 5 T of partially crystallized Gd60Co40 ribbons with different annealing time are displayed in Figure 3. With raising temperature, the magnetization of all the samples exhibits apparent transition from easy-saturated to linear-field-dependent. The type of the magnetic phase transition was estimated through Arrott plots (M 2 vs. H/M) according to Banerjee criteria [37], which is based on the mean-field theory and derived from the M-H isotherms [38]. As shown in Figure 4, the positive slope of all the curves indicates every magnetic transition in the multi-phase alloys is second order magnetic phase transition (SOMT). In comparison with the materials of first order magnetic phase transition (FOMT), such as Owing to the analogous microstructure of the 513 K/40 min and 513 K/60 min annealed samples, their multi-magnetic transition behaviors are similar to each other and three transitions can be observed at temperatures 174 K/194 K/219 K and 176 K/195 K/219 K. Compared with the magnetic transition behavior in the 513 K/20 min annealed sample, an extra transition related to the Gd 12 Co 7 crystalline (T C = 160.8 K) appeared [36]. The deviation of the T C is possibly ascribed to the different microstructure (e.g., crystal size and surrounding phase structure) between the bulk crystalline material and the in-situ precipitated crystallites, in addition, similar phenomena and values of T C were reported in Gd 12 Co 7 melt-spun ribbon (T C = 179 K) [22].
For the fully devitrified ribbon annealed at 653 K for 20 min, the dM/dT-T curve manifested three magnetic phase transitions at 162 K and 218 K associated with Gd 12 Co 7 and Gd 4 Co 3, respectively, which is consistent with the XRD results.

Magnetocaloric Properties
The isothermal initial magnetization M-H curves under the magnetic field change ∆H = 5 T of partially crystallized Gd 60 Co 40 ribbons with different annealing time are displayed in Figure 3. With raising temperature, the magnetization of all the samples exhibits apparent transition from easy-saturated to linear-field-dependent. The type of the magnetic phase transition was estimated through Arrott plots (M 2 vs. H/M) according to Banerjee criteria [37], which is based on the mean-field theory and derived from the M-H isotherms [38]. As shown in Figure 4, the positive slope of all the curves indicates every magnetic transition in the multi-phase alloys is second order magnetic phase transition (SOMT). In comparison with the materials of first order magnetic phase transition (FOMT), such as Gd 5 (Si 2 Ge 2 ) and LaFe 13−x Si x compounds [30,39], the magnetic refrigerants with SOMT possess advantages of negligible thermal and magnetic hysteresis, which make them more suitable for practical application, although their magnetic entropy change is lower [40]. Gd5(Si2Ge2) and LaFe13−xSix compounds [30,39], the magnetic refrigerants with SOMT possess advantages of negligible thermal and magnetic hysteresis, which make them more suitable for practical application, although their magnetic entropy change is lower [40].  The correlation between magnetic entropy change |ΔSM| and temperature was determined by using Equation (2) to calculate the data of M-H isotherms, and the |ΔSM| vs. T curves of the annealed Gd60Co40 ribbons under the field change from 0 to 5 T are illustrated in Figure 5. The achieved values of maximum magnetic entropy change (|ΔSM pk |) were 7.73, 6.75 and 6.54 J·kg −1 ·K −1 at temperatures Tpk of 194, 190 and 190 K for samples annealed at 513 K for 20, 40 and 60 min, respectively, which are smaller than those of asspun Gd60Co40 amorphous alloy (8.3 J·kg −1 ·K −1 ) [17]. Gd5(Si2Ge2) and LaFe13−xSix compounds [30,39], the magnetic refrigerants with SOMT possess advantages of negligible thermal and magnetic hysteresis, which make them more suitable for practical application, although their magnetic entropy change is lower [40].  The correlation between magnetic entropy change |ΔSM| and temperature was determined by using Equation (2) to calculate the data of M-H isotherms, and the |ΔSM| vs. T curves of the annealed Gd60Co40 ribbons under the field change from 0 to 5 T are illustrated in Figure 5. The achieved values of maximum magnetic entropy change (|ΔSM pk |) were 7.73, 6.75 and 6.54 J·kg −1 ·K −1 at temperatures Tpk of 194, 190 and 190 K for samples annealed at 513 K for 20, 40 and 60 min, respectively, which are smaller than those of asspun Gd60Co40 amorphous alloy (8.3 J·kg −1 ·K −1 ) [17]. The correlation between magnetic entropy change |∆S M | and temperature was determined by using Equation (2) to calculate the data of M-H isotherms, and the |∆S M | vs. T curves of the annealed Gd 60 Co 40 ribbons under the field change from 0 to 5 T are illustrated in Figure 5. The achieved values of maximum magnetic entropy change (|∆S M pk |) were 7.73, 6.75 and 6.54 J·kg −1 ·K −1 at temperatures T pk of 194, 190 and 190 K for samples annealed at 513 K for 20, 40 and 60 min, respectively, which are smaller than those of as-spun Gd 60 Co 40 amorphous alloy (8.3 J·kg −1 ·K −1 ) [17].
The T pk is near to the T C of the amorphous matrix in each partially crystallized alloy, meaning that the amorphous phase makes the predominant contribution to magnetocaloric effect. With increase of the annealing time, the Gd 4 Co 3 and Gd 12 Co 7 phases successively precipitated from the amorphous matrix and resulted in a reduction of the amorphous phase content; therefore, the |∆S M pk | of the multi-phase alloys decreases. However, due to the synergistic effects of multi-magnetic phase transition, the |∆S M |−T curves of the annealed Gd 60 Co 40 alloys in this study were broadened, usually accompanied by wide full temperature width at half maximum (∆T FWHM ) and large relative cooling power (RCP, another parameter to evaluate the MCE as heat transferred between the hot and cold reservoirs in an ideal refrigeration cycle) with expression of RCP = |∆S M pk | × ∆T FWHM [1]. The RCP of the 513 K/20 min, 513 K/40 min and 513 K/60 min annealed samples were 726.6, 789.8 and 797.9 J·kg −1 , respectively. In comparison with that of as-spun Gd 60 Co 40 amorphous alloy (713.8 J·kg −1 ) [17], the results reveal that the RCP increases with elongation of the annealing time. The Tpk is near to the TC of the amorphous matrix in each partially crystallized alloy, meaning that the amorphous phase makes the predominant contribution to magnetocaloric effect. With increase of the annealing time, the Gd4Co3 and Gd12Co7 phases successively precipitated from the amorphous matrix and resulted in a reduction of the amorphous phase content; therefore, the |ΔSM pk | of the multi-phase alloys decreases.
However, due to the synergistic effects of multi-magnetic phase transition, the |ΔSM|-T curves of the annealed Gd60Co40 alloys in this study were broadened, usually accompanied by wide full temperature width at half maximum (ΔTFWHM) and large relative cooling power (RCP, another parameter to evaluate the MCE as heat transferred between the hot and cold reservoirs in an ideal refrigeration cycle) with expression of RCP = |ΔSM pk | × ΔTFWHM [1]. The RCP of the 513 K/20 min, 513 K/40 min and 513 K/60 min annealed samples were 726.6, 789.8 and 797.9 J·kg −1 , respectively. In comparison with that of as-spun Gd60Co40 amorphous alloy (713.8 J·kg −1 ) [17], the results reveal that the RCP increases with elongation of the annealing time. As discussed in [23,36], the ΔTFWHM values of the materials in this study are much larger than the temperature span of any real magnetocaloric refrigerator so that the RCP may overestimate their performance in practical applications. In comparison with RCP, the temperature averaged entropy change (TEC) can properly reflect the merit of materials with a broad magnetocaloric response but small magnetic entropy change [41]. This is calculated over a range of temperatures ΔTlift that a material can reasonably support in response to a given field change ΔH, as follows: The value of the temperature at the center of the average, Tmid, is chosen by sweeping over the available ΔS(T)ΔH,T data and selecting the value that maximizes TEC(ΔTlift) for the given ΔTlift, similar to the evaluation of the maximum energy product of a permanent magnet. In this study, the ΔTlift of 10 K and ΔH of 1 T were chosen, and the TEC(10 K) of the 513 K/20 min, 513 K/40 min and 513 K/60 min annealed samples were 2.34, 1.69 and As discussed in [23,36], the ∆T FWHM values of the materials in this study are much larger than the temperature span of any real magnetocaloric refrigerator so that the RCP may overestimate their performance in practical applications. In comparison with RCP, the temperature averaged entropy change (TEC) can properly reflect the merit of materials with a broad magnetocaloric response but small magnetic entropy change [41]. This is calculated over a range of temperatures ∆T lift that a material can reasonably support in response to a given field change ∆H, as follows: The value of the temperature at the center of the average, T mid , is chosen by sweeping over the available ∆S(T) ∆H,T data and selecting the value that maximizes TEC(∆T lift ) for the given ∆T lift , similar to the evaluation of the maximum energy product of a permanent magnet. In this study, the ∆T lift of 10 K and ∆H of 1 T were chosen, and the TEC(10 K) of the 513 K/20 min, 513 K/40 min and 513 K/60 min annealed samples were 2.34, 1.69 and 1.71 J·kg −1 ·K −1 , respectively, at T mid of 192, 178 and 188 K. The values are lower than that of Gd and higher than that of La 0.813 K 0.16 Mn 0.987 O 3 , indicating their magnetocaloric performance is not very good.
Nevertheless, the broadened and flatten MCE with |∆S M | values of 6.65 ± 0.1 J·kg −1 ·K −1 and 6.44 ± 0.1 J·kg −1 ·K −1 within the temperature regions of 180-196 K and 177-196 K observed in samples annealed at 513 K for 40 min and 60 min, enable them to be more suitable for the Ericsson thermodynamic cycle [1]. In comparison with the table-like MCE in other alloys at analogous temperature ranges, such as Gd 55 Co 35 Mn 10 annealed at 600 K for 30 min (|∆S M | of 5.46 J·kg −1 ·K −1 with a temperature range from 137 K to 180 K) [14] and fully crystallized Gd 55 Co 35 Ni 10 (620 K/30 min) ribbon (|∆S M | of 5.0 J·kg −1 ·K −1 with a temperature range from 154 K to 214 K) [15], as listed in Table 1, the |∆S M | of the materials in this work are larger, but the temperature width of the plateau of |∆S M |-T is narrower. Table 1. Magnetocaloric properties of present alloys and some representative materials under applied field change of 0-5 T (A and C stand for amorphous and crystalline, respectively). The ∆T plateau denotes the temperature range of the plateau part of the |∆S M | vs. T curves.

Alloys
Structure On another hand, when compared with single amorphous phase alloys, like Gd 75 (Fe 0.25 Co 0.75 ) 25 [24] and Gd 50 (Co 69.25 Fe 4.25 Si 13 B 13.5 ) 50 [25], the combined merits of larger or comparative |∆S M | and broader working temperature range can be observed in these Gd 60 Co 40 annealed samples. The MCE in a temperature range of 160-220 K can be used in the fields of space technology, medicine, biology, life sciences and more [14].
The correlation between |∆S M | and H follows the power law of |∆S M | ∝ H n for the SOMT materials [42], and n is an exponent depending on both applied field and temperature. Particularly, at temperature T = T C or T pk , n is field independent, and n(T pk ) can be extracted from the slope of the linear fit of the rescaled ln|∆S M pk | vs. lnH plots [26,43]. Through the fitting results displayed in Figure 6, the values of 0.76 can be obtained for the 513 K/20 min annealed Gd 60 Co 40 alloy, which is close to~0.75 derived from the experimental data of other soft magnetic amorphous alloys [42], which indicates the prevailing contribution of amorphous matrix and neglectable influence from the slight amount of crystallite in the sample as mentioned above. However, the deviations of 0.90 and 0.91 can be observed in 513 K/40 min and 513 K/60 min annealed counterparts owing to the in-situ precipitated nanocrystalline in the amorphous precursor [44,45].
Assuming that the different magnetic phases are non-interacting, the total magnetic entropy change of the multiphase composites can be computed using a rule-of-mixtures sum of the entropy change in the constituent materials with expression described as [13]: where ∆S M 1 , ∆S M 2 and ∆S M 3 imply the magnetic entropy change, as well as α, β and γ denote the relative weight fractions of the phases 1, 2 and 3 respectively, with the relation of α + β + γ = 1. According to the experimental data of the amorphous Gd 60 Co 40 , crystalline Gd 4 Co 3 and Gd 12 Co 7 [17,21,22], the fitting results of |∆S M |-T curves under field changing from 0 to 5 T for 513 K/20 min and 513 K/60 min annealed Gd 60 Co 40 alloys were depicted in Figure 7. For comparison, the experimental results are also shown. Assuming that the different magnetic phases are non-interacting, the total magnetic entropy change of the multiphase composites can be computed using a rule-of-mixtures sum of the entropy change in the constituent materials with expression described as [13]: where ΔSM 1 , ΔSM 2 and ΔSM 3 imply the magnetic entropy change, as well as α, β and γ denote the relative weight fractions of the phases 1, 2 and 3 respectively, with the relation of α + β + γ = 1. According to the experimental data of the amorphous Gd60Co40, crystalline Gd4Co3 and Gd12Co7 [17,21,22], the fitting results of |ΔSM|-T curves under field changing from 0 to 5 T for 513 K/20 min and 513 K/60 min annealed Gd60Co40 alloys were depicted in Figure 7. For comparison, the experimental results are also shown. It can be seen the calculated results fit the experimental data very well, and the adopted weight fractions of the phases in 513 K/20 min and 513 K/60 min samples are 85 wt% amorphous matrix + 15 wt% Gd4Co3 and 30 wt% amorphous matrix + 36 wt% Gd4Co3 + 34 wt% Gd12Co7, respectively. Although the relative content of each phase is roughly estimated [46], the fraction of phases is significant to construct the broadened MCE in this kind of multiphase materials [29,47]. In this work, with increasing annealing time, the evolution of microstructure in Gd60Co40 amorphous ribbon achieves an appropriate constituent of different phases, resulting in the enhanced magnetocaloric performance. It can be seen the calculated results fit the experimental data very well, and the adopted weight fractions of the phases in 513 K/20 min and 513 K/60 min samples are 85 wt% amorphous matrix + 15 wt% Gd 4 Co 3 and 30 wt% amorphous matrix + 36 wt% Gd 4 Co 3 + 34 wt% Gd 12 Co 7, respectively. Although the relative content of each phase is roughly estimated [46], the fraction of phases is significant to construct the broadened MCE in this kind of multiphase materials [29,47]. In this work, with increasing annealing time, the evolution of microstructure in Gd 60 Co 40 amorphous ribbon achieves an appropriate constituent of different phases, resulting in the enhanced magnetocaloric performance. by calculating experimental data of constituent materials 1, 2 and 3, which correspond to amorphous Gd60Co40 [17], crystalline Gd4Co3 [21] and melt-spun Gd12Co7 [22], respectively.
On basis of the numerical approach provided by A. Smaïli et al. [48], in this study, when the composite is consisted by 63 wt% Gd4Co3 + 26 wt% Gd12Co7 + 11 wt% amorphous Gd60Co40, a nearly flat-shape |ΔSM|-T profile can be observed between 180 K and 220 K. As discussed in Section 3.1, the Gd4Co3 is probably the primary precipitate. Therefore, the  [17], crystalline Gd 4 Co 3 [21] and melt-spun Gd 12 Co 7 [22], respectively.
On basis of the numerical approach provided by A. Smaïli et al. [48], in this study, when the composite is consisted by 63 wt% Gd 4 Co 3 + 26 wt% Gd 12 Co 7 + 11 wt% amorphous Gd 60 Co 40 , a nearly flat-shape |∆S M |-T profile can be observed between 180 K and 220 K. As discussed in Section 3.1, the Gd 4 Co 3 is probably the primary precipitate. Therefore, the possible method to achieve this composite is as following: at first, rapid thermal annealing in temperature range of 523 K-563 K (the onset and end temperatures of the first crystallization peak on DSC curve) to induce large amount of Gd 4 Co 3 crystallites; and then, annealing at 513 K for 40-60 min (similar to the treatment in this work) to remain certain of the content of the amorphous phase. Further research will be carried out in the next step.

Conclusions
In summary, amorphous-nanocrystalline Gd 60 Co 40 alloys were synthesized by crystallization treatment of the melt-spun amorphous ribbons. With different annealing times (20, 40 and 60 min) at 513 K, Gd 4 Co 3 -type and Gd 12 Co 7 -type phases precipitated from the amorphous matrix in sequence; however, the grain size and transformation volume tended to saturate after certain annealing times due to the metastable equilibrium between the crystallites and the remaining amorphous phase. In the samples annealed for 40 min and 60 min, the multi-phase structure consisted of the Gd 4 Co 3 , Gd 12 Co 7 and amorphous phases, which resulted in three successive magnetic phase transitions at temperatures of 174 K/194 K/219 K and 176 K/195 K/219 K, respectively.
Owing to the overlap of multi-peaks in the |∆S M |-T curves, broadened MCE with |∆S M | values of 6.65 ± 0.1 J·kg −1 ·K −1 and 6.44 ± 0.1 J·kg −1 ·K −1 in the temperature spans of 180 K-196 K and 177 K-196 K under a field change of 0-5 T were obtained in ribbons annealed for 40 min and 60 min, respectively, and could be modeled by considering non-interacting phases. The enhanced relative cooling power and flattened magnetocaloric properties of partially crystallized composites enable them to be more suitable for the Ericsson thermodynamic cycle, in comparison with the single amorphous phase counterparts.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.