Effect of Minor Ce Substitution for Pr on the Glass Formability and Magnetocaloric Effect of a Fe₈₈Zr₄Pr₄B₄ Metallic Glass

Abstract

In the present work, Fe₈₈Zr₄Pr₃B₄Ce₁ metallic glass (MG) was successfully prepared by minor Ce substitution for Pr, and compared with Fe₈₈Zr₄Pr₄B₄ MG in terms of glass forming ability (GFA), magnetic and magnetocaloric properties. The GFA, T_c and the maximum magnetic entropy change (−ΔS_m^peak) of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG were found to decrease slightly. At the same time, the possible interaction mechanism of minor Ce replacing Pr was also explained. The critical exponents (β, γ and n) obtained by the Kouvel–Fisher method indicate that Fe₈₈Zr₄Pr₃B₄Ce₁ MG near T_c exhibits typical magnetocaloric behavior of fully amorphous alloys. The considerable maximum magnetic entropy change (−ΔS_m^peak = 3.84 J/(kg × K) under 5 T) near its Curie temperature (T_c = 314 K) as well as RCP (~ 646.3 J/kg under 5 T) make the Fe₈₈Zr₄Pr₃B₄Ce₁ MG a better candidate as a component of the amorphous hybrids that exhibit table-shape magnetic entropy change profiles within the operation temperature interval of a magnetic refrigerator.

1. Introduction

As is known, traditional gas compression–expansion refrigeration technology, dependent on fluorine-containing refrigerants, has many disadvantages such as greenhouse gas emission, destroying the ozone layer, low refrigeration efficiency, and so on. Therefore, magnetic refrigeration (MR) technology using magnetocaloric refrigerant has received a lot of attention because of its high efficiency, lower energy loss, environmental friendliness and structural compactness [1,2,3,4,5,6]. Magnetocaloric refrigerants are the magnetic materials that emit/absorb heat adiabatically when a magnetic field is applied/removed, which is called magnetocaloric effect (MCE) [4,5,6]. The magnetocaloric effect of the magnetic materials is induced by the reduction in magnetic entropy upon magnetization, and, as a consequence, the magnetocaloric properties of a magnet are usually evaluated by the change in magnetic entropy (−ΔS_m) under a certain magnetic field. As such, early research focused on the MCE of first-order magnetic phase transition (FOMPT) materials that exhibit a sharp −ΔS_m profile with rather high maximum −ΔS_m (−ΔS_m^peak) [7,8,9]. However, the narrow working temperature intervals of these FOMPT materials make them difficult to match the requirements of magnetic refrigerants working in an Ericsson cycle; that is, a fattened −ΔS_m curve over the range of operating temperatures in a magnetic refrigerator [10]. In addition, the FOMPT materials inevitably show some disadvantages such as high magnetic and thermal hysteresis [11]. In contrast, amorphous magnetocaloric alloys that experience a second-order magnetic phase transition (SOMPT) exhibit several characteristics superior to the FOMPT materials, such as low energy loss induced by their negligible coercivity and high electric resistance, broadened −ΔS_m curve and tunable −ΔS_m peak temperature that make them easily composed to achieve the fattened −ΔS_m curve [12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]. Unfortunately, although rare earth (RE)-based metallic glasses (MGs) exhibit rather high glass formability (GFA), excellent −ΔS_m^peak at low temperature and ultrahigh refrigeration capacity (RC), their formability and −ΔS_m^peak get worse when their Curie temperature (T_c) increases to or above the ambient temperature [12,13,14,15,16,17,18]. Thus, RE-based amorphous magnetocaloric alloys are more likely to be applied in low temperature refrigeration instead of room temperature (RT) refrigeration. The Fe-based amorphous magnetocaloric alloys exhibit good glass formability when their T_c is near the ambient temperature, but their −ΔS_m^peak is usually very low [19,20,21,22,23]. For example, Fe-Zr-B MGs with T_c ranging from the cold end (T_Cold) to the hot end (T_Hot) of domestic cooling equipment can be easily fabricated, but their −ΔS_m^peak under 5 T is less than 3.34 J/(kg × K) [21,22,23], which is far from enough for them to be utilized as cooling agents in domestic cooling appliances. More recently, by microalloying the Fe-Zr-B MGs with other transition metals or RE metals, we successfully adjusted the T_c and improved the −ΔS_m^peak of the Fe-Zr-B MGs [24,25,26,27,28,29,30,31]. For instance, the −ΔS_m^peak under 5 T reaches 3.55 J/(kg × K) at 336 K in the Fe₈₅Co₃Zr₅B₄Nb₃ amorphous ribbon [24] and at 333 K in the Fe₈₅Zr₈B₄Sm₃ amorphous ribbon [25]; it reaches 4.0 J/(kg × K) at 323 K in the Fe₈₈Zr₄Pr₄B₄ amorphous ribbon [26] and 4.10 J/(kg × K) at 335 K in the amorphous Fe₈₈Zr₄Nd₄B₄ ribbon [27].

However, preliminary results show that the excellent MCE of the iron-based MGs appear near or above the T_Hot of a domestic refrigerator. It is known that the high −ΔS_m^peak at temperatures higher than T_Cold but lower than T_Hot is also required for the construction of fattened −ΔS_m curves suitable for the Ericsson refrigeration cycle. Thus, it is necessary to develop a new type of metallic glass with excellent MCE at RT. As such, it is critical to decrease the −ΔS_m peak temperature of the iron-based MGs without dramatically deteriorating their −ΔS_m^peak. In the present work, according to our preliminary results on the effect of Ce substitution on the T_c and −ΔS_m^peak of the Fe-Zr-B amorphous alloys [32], we add minor Ce to replace the Pr element in the Fe₈₈Zr₄Pr₄B₄ amorphous alloy for the purpose of obtaining good magnetocaloric properties at a temperature slightly lower than the T_Hot of a domestic refrigerator. The mechanism for the influence of minor Ce substitution on the magnetic as well as magnetocaloric properties of the Fe₈₈Zr₄Pr₄B₄ metallic glass was also investigated. The research results provide a feasible path for the Fe-Zr-B-RE amorphous alloy to reduce T_c and avoid significant deterioration of magnetocaloric properties while reducing costs.

2. Materials and Methods

The Fe₈₈Zr₄Pr₃B₄Ce₁ ingot was manufactured by arc-melting the high purity raw materials more than five times to ensure uniformity of composition [33]. Ribbons were fabricated by spraying the Fe₈₈Zr₄Pr₃B₄Ce₁ melt from a quartz tube to the surface of a copper roller rotating at a linear velocity of 55 m/s. The whole sample preparation process is protected by a high purity Ar atmosphere. The cross-sectional morphology of the Fe₈₈Zr₄Pr₃B₄Ce₁ as-spun ribbon was characterized through a Hitachi tungsten filament scanning electron microscope (SEM, model SU-1500). The ~40-μm-thickness as-spun ribbons were selected for structural analysis by X-ray diffraction (XRD) using a Cu K_α radiation with a scanning speed of 1 °/min on a PANnalytical spectrometer. Under program-controlled temperature conditions, the glass transition behavior, melting and crystallization of Fe₈₈Zr₄Pr₃B₄Ce₁ ribbons were distinguished by measuring the power difference between the sample and the reference material as a function of temperature (i.e., the thermal effect information related to heat absorption and release). Hence, the thermodynamic parameters, including glass transition temperature (T_g), initial crystallization temperature (T_x) and liquidus temperature (T_l), were derived from the differential scanning calorimetry (DSC) curve measured by a NETZSCH DSC-404 C calorimeter at a heating speed of 20 K/min to evaluate the formability of the MG ribbon. The temperature dependence of the heat capacity (C_p(T)) curve of the glassy sample was measured by a Perkin-Elmer DIAMOND calorimeter. The magnetic measurements of the amorphous ribbons, including magnetization vs. temperature (M-T) curve, isothermal magnetization (M-H) curve and hysteresis loop, were performed on the vibrating sample magnetometer (VSM) module of a physical property measurement system (PPMS, model 6000, Quantum Design) after applying an oscillating magnetic field to a fully amorphous ribbon to eliminate residual magnetism. The sample for magnetic measurement was prepared by sticking several ribbons together using non-magnetic cement. To minimize the impact of demagnetization, the magnetic field is applied parallel to the length of the sample.

3. Results and Discussion

The Fe₈₈Zr₄Pr₃B₄Ce₁ as-spun ribbon is amorphous according to its XRD pattern shown in Figure 1. The cross-sectional morphology (the upper left inset of Figure 1a) and the prepared samples (the upper right inset of Figure 1a) of the Fe₈₈Zr₄Pr₃B₄Ce₁ as-spun ribbon, indicate a thickness of ~40 μm and a width of ~2 mm. The glassy characteristic of the Fe₈₈Zr₄Pr₃B₄Ce₁ ribbon is also illustrated by the upward glass transition hump before the downward crystallization peak on its DSC trace, as shown in Figure 1b. The onset of T_g and T_x of the amorphous ribbon determined from its DSC trace is ~795 K and ~856 K, respectively. The T_l of Fe₈₈Zr₄Pr₃B₄Ce₁ alloy obtained from its melting curve, which is illustrated in the inset of Figure 1b, is determined to be ~1545 K. Therefore, we can assess the GFA of the Fe₈₈Zr₄Pr₃B₄Ce₁ amorphous sample by calculating the reduced glass transition temperature (T_rg = T_g/T_l = 0.515) [34] as well as the parameter γ (= T_x/(T_g + T_l) = 0.366) [35]. The T_rg of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG sample is slightly higher than that of the Fe₈₈Zr₄Pr₄B₄ MG [26], while the γ parameter is slightly decreased by the Ce substitution. Therefore, it seems that the Ce addition does not obviously change the glass formability of the Fe₈₈Zr₄Pr₄B₄ metallic glass. On the other hand, although the Fe₈₈Zr₄Pr₃B₄Ce₁ as well as Fe₈₈Zr₄Pr₄B₄ MGs do not show T_rg and γ values comparable to the bulk metallic glasses, their T_rg and γ values are still larger than most other Fe-Zr-B MGs [21,22,23], indicating that the Fe₈₈Zr₄Pr₃B₄Ce₁ and Fe₈₈Zr₄Pr₄B₄ alloys can be easily prepared into MG ribbon.

The M-T curve under 0.03 T of Fe₈₈Zr₄Pr₃B₄Ce₁ sample was measured after a zero-field-cooling process from RT. Figure 2a depicts the M-T curve under 0.03 T of the Fe₈₈Zr₄Pr₃B₄Ce₁ glassy sample. The ferromagnetic materials exhibit strong magnetism when magnetized. However, as the temperature increases, the intensification of thermal motion will affect the ordered arrangement of magnetic moments of the magnetic domain. When the temperature reaches enough to disrupt the orderly arrangement of magnetic moments of the magnetic domain, the magnetic domain is disrupted, the average magnetic moment becomes zero and the magnetism of ferromagnetic materials disappears and becomes paramagnetic. As seen in the (dM/dT)-T plots of the sample in the inset, T_c of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG is thus determined at the minimum value of the dM/dT to be 314 K, which is about 9 K lower than that of the Fe₈₈Zr₄Pr₄B₄ MG [26]. The decreased T_c caused by the replacement of Ce for Pr may be closely related to the antiferromagnetic coupling of the Ce atom with the Fe atom [32]. The hysteresis loops of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG ribbon, as displayed in Figure 2b, suggest that the MG is paramagnetism at 380 K and soft magnetism at 200 K. The Fe₈₈Zr₄Pr₃B₄Ce₁ MG exhibits excellent soft magnetic properties with almost zero hysteresis and high magnetic susceptibility at 200 K, both of which are typical characteristics of fully amorphous alloys and are essential for magnetocaloric materials. The saturation magnetization (M_s) of the Fe₈₈Zr₄Pr₃B₄Ce₁ alloy (~129 Am²/kg at 200 K) is slightly lower than that of the Fe₈₈Zr₄Pr₄B₄ MG (~137 Am²/kg at 200 K [26]), indicating that the magnetocaloric properties of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG may be not as high as Fe₈₈Zr₄Pr₄B₄ MG because both the M_s and the −ΔS_m of the amorphous alloys are primarily determined by the ordering of their magnetic moments.

The temperature dependence of −ΔS_m (−ΔS_m-T curve) can be derived from the M-H curves measured at various temperatures. Figure 3a displays the M-H curves of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG from 200 K to 380 K under 5 T. On the basis of these M-H curves, the M²-H/M plots, namely the Arrott plots of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG, can be established accordingly, as illustrated in Figure 3b. The Arrott plots (M²-H/M) at each temperature show a positive slope and are almost parallel to each other from 200 K to 380 K, both of which indicate the typical feature of the materials experiencing a SOMPT according to the Banerjee criterion [36]. The second-order magnetic transition allows the alloy to undergo a continuous phase transition in a broad temperature range and hence leads to a better overall cooling capacity. The −ΔS_m-T curves under various external magnetic fields of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG obtained from its M-H curves are depicted in Figure 4a. The −ΔS_m^peak of the Fe₈₈Zr₄Pr₃B₄Ce₁ ribbon reaches 1.15 J/(kg × K) under 1 T, 1.57 J/(kg × K) under 1.5 T, 1.94 J/(kg × K) under 2 T, 2.63 J/(kg × K) under 3 T, 3.26 J/(kg × K) under 4 T and 3.84 J/(kg × K) under 5 T at 312.5 K. The −ΔS_m^peak of the Fe₈₈Zr₄Pr₃B₄Ce₁ ribbon is marginally lower than that of the Fe₈₈Zr₄Pr₄B₄ MG [26], probably because of the lower magnetic moment of the Ce atom than the Pr atom due to there being only one up-paired electron in the 4f shell of Ce atom. The minor Ce atom substitution for Pr atom reduces the total magnetic moment of the Fe₈₈Zr₄Pr₄B₄ MG, which is confirmed by the effective magnetic moment (μ_eff). As shown in Figure 4b, the temperature dependence of H/M of the Fe₈₈Zr₄Pr₄B₄ and Fe₈₈Zr₄Pr₃B₄Ce₁ ribbons were obtained from their M-T curves. According to the Curie–Weiss law [37], the slopes of the lines above their T_c are correlated to the μ_eff, and, thus, the μ_eff of the two MGs are calculated to be about 8.89 μ_B for Fe₈₈Zr₄Pr₄B₄ and 7.74 μ_B for Fe₈₈Zr₄Pr₃B₄Ce₁. Apparently, the μ_eff of the alloy is reduced with the addition of the Ce atom, resulting in a decrease in −ΔS_m^peak.

Although the −ΔS_m^peak of the Fe₈₈Zr₄Pr₃B₄Ce₁ ribbon is not as high as that of the Fe₈₈Zr₄Pr₄B₄ MG, it is still higher than the −ΔS_m^peak near 310 K of other amorphous alloys and even high entropy alloys (HEA) reported in the literature [25,26,38,39,40,41]. For example, its −ΔS_m^peak under 5 T is about 234% higher than that of the Al₂₀Mn₂₀Fe₂₀Co_15.5Cr_24.5 HEA (1.15 J/(kg × K) at 314 K [38]), 193% higher than that of the Mn₂₀Al₂₀Co₁₄Fe₂₃Cr₂₃ HEA (1.31 J/(kg × K) at 310 K [39]), 22.3% higher than that of the Fe₈₇Zr₇B₄Dy₂ MG (3.14 J/(kg × K) at 308 K [40]), 17.4% higher than that of the Fe₈₇Zr₈B₄Sm₁ MG (3.27 J/(kg × K) at 308 K [25]), 5.5% higher than that of the Fe₈₆La₇B₅Ce₂ MG (3.64 J/(kg × K) at 313 K [41]) and 6.67% larger than that of the Fe₈₈Zr₆Pr₂B₄ MG (3.6 J/(kg × K) at 306 K [26]). Figure 4c displays the −ΔS_m-T curves of several iron-based MGs under 5 T. The Fe₈₈Zr₄Pr₃B₄Ce₁ MG ribbon shows a rather high −ΔS_m^peak near 310 K. On the other hand, the relative cooling power (RCP = −ΔS_m^peak × ΔT_FWHM, where ΔT_FWHM is the full width at the half of −ΔS_m^peak [42]) of the Fe₈₈Zr₄Pr₄B₃Ce₁ MG, can be calculated as 164.7 J/kg under 1.5 T and 646.3 J/kg under 5 T according to the −ΔS_m-T curve, both of which are similar to the values of amorphous alloys and much higher than those of the first-order magnetic transition alloys or compounds [26,41,43,44]. Since the Fe₈₈Zr₄Pr₃B₄Ce₁ MG experiences an SOMPT, it exhibits large value of magnetic entropy changes over a wide temperature range, which may be caused by the coupling interaction between RE-RE and RE-TM. Therefore, it can be predicted that Fe₈₈Zr₄Pr₃B₄Ce₁ MG has a good magnetocaloric effect over a large temperature range.

After constructing the ln(−ΔS_m)-ln(H) plots at each temperature, we can achieve their slopes (defined as n) by linearly fitting and thus, observe the magnetocaloric behaviors of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG in more detail. Figure 5a represents the temperature dependence of n (n-T curve) of the Fe₈₈Zr₄Pr₃B₄Ce₁ amorphous ribbon. Similar to other amorphous alloys [21,22,24,26,27,40], the n of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG is close to 1 at temperatures well below its T_c, and smoothly drops to the minimum value near its T_c, then dramatically increases with the increasing temperature and approaches 2 at temperatures much higher than its T_c. The minimum n value of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG ribbon, which appears at 312.5 K and is shown in the inset of Figure 5a, is ~0.747 and is close to the predicted value of amorphous alloys proposed by V. Franco et al. based on the Arrott–Nokes equation [45]. Both the n-T curve and the minimum n value of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG indicate typical magnetocaloric behaviors similar to those of fully MGs.

On the other hand, the magnetocaloric behavior near its T_c of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG can also be explored by its critical exponents; that is, n(T_c) = 1 + (β − 1)/(β + γ) [45]. Wherein, β and γ are the exponents related to spontaneous magnetization (M_st) and initial susceptibility (χ₀), respectively, which can be described as follows [46]:

M_st(T) = M₀(−ε)^β, ε < 0, T < T_c

(1)

χ₀(T)⁻¹ = (H₀/M₀)ε^γ, ε > 0, T > T_c

(2)

where M₀ and H₀ are the critical amplitudes, ε = (T − T_c)/T_c is the reduced temperature. Based on Equations (1) and (2), Kouvel and Fisher proposed a method to determine the critical exponents β and γ with high accuracy, namely the Kouvel–Fisher (KF) method [47]. Equations (1) and (2) can be rewritten as:

Y = M_st(T)·(dM_st(T)/dT)⁻¹ = (T − T_c)/β

(3)

X = χ₀(T)⁻¹·(dχ₀(T)⁻¹/dT)⁻¹ = (T − T_c)/γ

(4)

As such, we constructed the modified Arrott plots (M^2.5-(H/M)^0.75) at various temperatures of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG and obtained the temperature dependence of M_st and χ₀⁻¹ from the intersections of the linear extrapolation of high field regions with M^2.5 and (H/M)^0.75 axes, respectively. Figure 5b shows the temperature dependence of M_st(T)·(dM_st(T)/dT)⁻¹ and χ₀(T)⁻¹·(dχ₀(T)⁻¹/dT)⁻¹ of the ribbon. The critical exponents β and γ can be determined to be 0.438 and 1.384 from the slope of the linear fitting of the two plots. The values of β and γ are close to those of other iron-based MGs [45,48]. Therefore, the n value near T_c based on the KF method is calculated to be 0.692, which is slightly lower than the n value based on the Arrott–Nokes equation, but still higher than the theoretical value of the mean field model [45,49,50]. The reason for this may be related to the unique short-range ordered microstructure of MGs.

Furthermore, the field dependence of the refrigeration capacity is also controlled by the critical exponent, as follows [51]:

$$RC \propto H^N, N=1+\frac{1}{\delta }$$

(5)

where δ is the critical magnetization isotherm at T_c, that is:

$$M=D{H}^{\frac{1}{\delta }}$$

(6)

D is the critical amplitude. According to the Widom scaling relation (δ = 1 + γ/β) [52], the exponent δ can be determined to be 4.160. Moreover, the determination of the exponent δ can also be obtained by the modified Equation (6), as follows [46]:

$$\ln M=\ln D+\frac{1}{\delta }\ln H$$

(7)

Taking into account the measurement increments of 5 K in the M-H curves from 300 K to 330 K and the T_c of 314 K for the Fe₈₈Zr₄Pr₃B₄Ce₁ MG, the M-H curve at 315 K was selected to construct the ln(M) vs. ln(H) plot, as shown in Figure 5c. The linear fitting is rather accurate, with a regression coefficient (Adj. R-Square) of up to 0.9999. The value of the exponent δ is derived from the slope of the linear fitting to be 4.093 ± 0.003, which is close to the result based on the Widom scaling relation. Therefore, according to Equation (5), we can obtain that RC is roughly proportional to H^1.24. The ln(RCP)-ln(H) plot of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG constructed from its RCP under different fields is illustrated in Figure 5d. The plot also fits well linearly and the slope is determined to be 1.130 ± 0.005. Clearly, the N value obtained from the ln(RCP)-ln(H) plot is slightly lower than that obtained from the KF method and the modified Equation (6), but these values are around the range of iron-based MGs. The deviation of n and N are supposed to be due to the error in obtaining M_st and χ₀⁻¹ from the modified Arrott plots.

A phenomenological universal behavior for the −ΔS_m of the SOMPT materials has been proposed by V. Franco et al. [53]. The −ΔS_m-T curves under all magnetic fields are normalized with their respective −ΔS_m^peak; that is, $∆S^{’}\left(T, H_{max}\right)=∆S_m\left(T, H_{max}\right)/∆S_m^{peak}\left(T, H_{max}\right)$. The temperature axis is divided into upper and lower parts with T_c as the boundary, and rescaled in different ways, as follows:

$$\theta =\left\{\begin{array}{c}-\left(T-T_c\right)/\left(T_{r1}-T_c\right),T\le T_c\\ \left(T-T_c\right)/\left(T_{r2}-T_c\right),T>T_c\end{array}\right.$$

(8)

where T_r1 and T_r2 are the starting and ending temperatures corresponding to ΔT_FWHM under different magnetic fields, respectively. Figure 6a shows ΔS_m/ΔS_m^peak-θ curves under different magnetic fields of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG. We can find that the normalized ΔS_m curves under each magnetic field can collapse onto the same universal curve. This uniformity indicates the typically magnetocaloric behavior of the SOMPT Fe₈₈Zr₄Pr₃B₄Ce₁ MG.

To evaluate the magnetocaloric properties of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG in a more direct way, we calculated the adiabatic temperature rise (ΔT_ad) of the amorphous ribbon, as follows:

$$\mathsf{\Delta }{T}_{ad}(T, 0\to H)=-\frac{T}{C_p\left(T\right)}\mathsf{\Delta }{S}_m(T, 0\to H)$$

(9)

Figure 6b illustrates the temperature dependence of ΔT_ad (ΔT_ad-T curve) of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG obtained from its −ΔS_m-T curves and C_p(T) curve (shown in the inset). The ΔT_ad reaches a maximum value of ~1.05 K under 1.5 T and ~2.64 K under 5 T, respectively. These well-known magnetocaloric materials (such as Gd [54], Gd₅(Si₂Ge₂) [7], MnAs [8] and Fe₄₉Rh₅₁ [55]) undergo a first-order magnetic phase transition, exhibiting a giant magnetocaloric effect. Therefore, the magnetic entropy change curve shows extremely high sharp peaks within a narrow temperature range, and the magnetic entropy change peak value is much higher than that of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG with a second-order magnetic phase transition in this study. Therefore, the RCP of Fe₈₈Zr₄Pr₃B₄Ce₁ MG is larger than that of famous magnetocaloric materials, but ΔT_ad is smaller than of these materials. Considering its relatively high −ΔS_m^peak at 312.5 K, RCP and ΔT_ad, the Fe₈₈Zr₄Pr₃B₄Ce₁ MG is a prospective candidate for an intermediate component of magnetic refrigerants with a table-shape −ΔS_m curve within the interval between T_Cold and T_Hot of a domestic refrigerator.

Improving the −ΔS_m^peak near RT as much as possible seems to be an effective way to improve the magnetocaloric properties of Fe-based MGs. In the previous study of Fe-Ce-B ternary MGs [56], Ce was partially replaced by B, and it was found that with the decrease in Ce atoms, the antiferromagnetic coupling between Ce and Fe atoms was weakened, resulting in the enhancement of 3d-3d interaction between Fe atoms and thus T_c increased. In the study of Fe-Zr-B ternary MGs [32], it was found that by replacing Zr with Ce, T_c decreased from 306 K to 283 K. In summary, the composition dependence of the MCE in the Fe-Zr-Pr-B quaternary amorphous alloys system can be explained by the antiferromagnetic coupling between Ce and Fe atoms and the Ce-Pr interaction caused by the introduction of Ce atoms, which may lead to the weakening of the 3d-3d interaction between Fe atoms. Therefore, the substitution of Ce for Pr will reduce the T_c of Fe₈₈Zr₄Pr₄B₄ MG. At the same time, because the magnetic moment of Ce atoms is lower than that of Pr atoms, the total magnetic entropy of the alloy will be reduced by the substitution of Ce for Pr, so the −ΔS_m^peak will be reduced. There are two reasons for choosing minor Ce to replace Pr in this research: firstly, Ce is cheaper than Pr, which can save costs; secondly, Ce and Pr are in adjacent positions in the periodic table of elements and the 4f shell is different from two electrons, which will not produce a large change. Based on the previous research results, it is expected that the total magnetic moment will not decrease too much while the minor Ce replaces Pr in reducing the T_c of Fe₈₈Zr₄Pr₄B₄ MG, so it is more likely to obtain magnetocaloric materials with good magnetocaloric properties near RT.

4. Conclusions

In summary, minor Ce was selected to replace the Pr atom in a Fe₈₈Zr₄Pr₄B₄ MG, and the Fe₈₈Zr₄Pr₃B₄Ce₁ amorphous ribbon with a thickness of ~40 micrometer was successfully prepared. The influences of the minor Ce substitution for Pr on GFA, magnetic properties and magnetocaloric effect of the Fe₈₈Zr₄Pr₄B₄ MG, as well as their mechanisms, were further studied. The main conclusions are detailed below:

(i)

The T_rg and γ indicate that the minor Ce substitution for Pr does not obviously change the glass formability of the Fe₈₈Zr₄Pr₄B₄ MG, but the glass formability of both the two ribbons is enough to vitrify them into glassy ribbon.

(ii)

The T_c of the Fe₈₈Zr₄Pr₃B₄Ce₁ ribbon decreases by 9 K compared with the Fe₈₈Zr₄Pr₄B₄ MG, which may be closely related to the antiferromagnetic coupling of the Ce atom with the Fe atom. The Fe₈₈Zr₄Pr₃B₄Ce₁ MG ribbon shows typical soft magnetic characteristics of fully amorphous alloys but slightly lower M_s than that of the Fe₈₈Zr₄Pr₄B₄ MG at 200 K. The M²-H/M plots at various temperatures indicate the typical SOMPT feature of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG.

(iii)

According to the Maxwell Equation, the −ΔS_m^peak of the Fe₈₈Zr₄Pr₃B₄Ce₁ ribbon reaches 3.84 J/(kg × K) under 5 T at 312.5 K, which is slightly lower than that of the Fe₈₈Zr₄Pr₄B₄ MG but still higher than the −ΔS_m^peak near 310 K of other amorphous alloys and even high entropy alloys reported in literature.

(iv)

The n-T curve, the minimum n value and the normalized universal curve of the Fe₈₈Zr₄Pr₃B₄Ce₁ MG ribbon also indicate the typical magnetocaloric behaviors of fully amorphous alloys. The values of n and N obtained by the KF method deviate slightly from those obtained by the linear fitting of the field dependence of −ΔS_m^peak and RCP, which may be due to the error in multiple derivation of the KF method.

Consequently, considering the relatively high −ΔS_m^peak near 310 K, RCP (~646.3 J/kg under 5 T) and ΔT_ad (~2.64 K under 5 T), the Fe₈₈Zr₄Pr₃B₄Ce₁ MG ribbon has great potential for application as an intermediate component of magnetic refrigerants with a flattened −ΔS_m curve in a domestic refrigerator.