Exploring Ho substituted Y-Fe-B nanocrystalline alloys and hot worked magnets

Aiming to balance the utilization of rare earth (RE) resources and develop Y-Fe-B based permanent magnets, Ho is employed as strategic substitution for enhancing the magnetic properties and thermal stability of nanocrystalline Y-Fe-B alloys. Ho substituting Y can enhance the coercivity of Y-Fe-B alloys while maintaining their excellent thermal stability. 30 at.% Ho substitution leads to an abnormal increase of remanence J r and (Y0.7Ho0.3)2Fe14B alloy exhibits good magnetic properties with remanence J r = 0.73 T, intrinsic coercivity H cj = 303 kA m−1, and maximum energy product (BH)max = 66 kJ m−3. High thermal stability with temperature coefficient of remanence α = −0.124%/K and temperature coefficient of coercivity β = −0.245%/K were obtained between 300–400 K. The results for RE-rich (Y1−xHox)2.5Fe14B alloys also show that the magnetic properties change with Ho content are similar to those of (Y1−xHox)2Fe14B alloys, but the coercivity is higher. In addition, nanocrystalline (Y0.5Ho0.5)2.5Fe14B magnets were prepared by hot-pressing and hot deformation process. Due to the lack of low melting point RE-rich phase, this alloy is difficult to be densified and deformed. The formation of high temperature RE2O3 and RE6Fe23 phases and the lack of continuously distributed RE-rich grain boundary phase are responsible for the poor texture of hot deformed magnet. The hot deformed magnet has the magnetic properties of J r = 0.50 T, H cj = 739 kA m−1, and (BH)max = 40 kJ m−3 together with high thermal stability. The micro-analysis demonstrated the chemical segregation of Y and Ho elements. Higher proportion of Ho than Y existed in main phase and grain boundary phase indicate excess Y were precipitated as Y-rich oxides.


Introduction
Nd-Fe-B based rare earth (RE) permanent magnets with excellent magnetic properties have been widely employed in various industries.However, the excessive consumption of critical RE elements such Pr, Nd, Dy and Tb in Nd-Fe-B magnets has triggered tremendous attempt to develop permanent magnets based on high abundance RE elements of La, Ce and Y [1][2][3][4].The present research on the abundance RE based magnets are mainly focused on the Ce-Fe-B based alloys [5][6][7] or the Nd-Fe-B based alloys with partial incorporation of La, Ce and Y [8,9].The Y-Fe-B based alloys have not been well investigated.Although Y 2 Fe 14 B compound also exhibits inferior intrinsic properties than Nd 2 Fe 14 B, it has more potential than La 2 Fe 14 B or Ce 2 Fe 14 B for preparing permanent magnets with high performance/cost ratio due to its higher saturation magnetization (J s ), comparable anisotropic field (H A ) and higher Curie temperature (T c ) [10,11].In addition, the H A of Y 2 Fe 14 B exhibit the weak temperature dependence [12], which is beneficial to the high thermal stability of Y-containing RE-Fe-B magnets.Sun et al [13] designed several series of melt-spun Y m Fe 100-m-n B n alloys and found that the 2:14:1 hard magnetic phase could be detected in the alloys with 13 m 19 and 6 n 18.The Y 2 Fe 14 B/YFe 2 bi-phase exchange coupling interaction existed in Y 16 Fe 78 B 6 alloy with optimal magnetic properties remanence J r = 0.61 T, coercivity H cj = 255 kA m −1 and maximum energy product (BH) max = 33 kJ m −3 after crystallization at 700 °C.Liu et al [4] studied the phase precipitation behavior of amorphous Y 2 Fe 14 B alloy and found that the precipitation temperature of RE 2 Fe 14 B (2:14:1) main phase was close to 606 °C.By substituting Y with 50 at.%Nd, the H cj of (Y 1−x Nd x ) 2 Fe 14 B alloys increased from 149 kA m −1 to 416 kA m −1 , and the exchange coupling between grains was enhanced.However, its excellent thermal stability could not be maintained due to the incorporation of large amounts of Nd.Liao et al [14] found that La-substitution is unfavorable to the magnetic properties of (Y 1−x La x ) 2 Fe 14 B alloys (x = 0-0.5),but the better thermal stability can be obtained due to the unusual increased H cj with increasing temperature.(Y 1−x Ce x ) 2 Fe 14 B alloys and (Y 1−x La x ) 2 Fe 14 B alloys exhibit opposite changes of the room-temperature magnetic properties or elevated temperature behavior with Ce and La concentration, respectively.Tang et al [15] also reported lower temperature coefficients of (Y 1−z Dy z ) 2.2 Fe 14 B alloys than Nd-Fe-B alloys.
To make the Y-Fe-B magnets more useful, it is important to enhance their coercivity.Tb and Dy are the most effective elements for improving the anisotropy field of RE-Fe-B alloys.However, considering the strategic perspective of balancing the utilization of RE resources, it is very meaningful to reduce the usage of Dy or Tb.Similar to Dy 2 Fe 14 B and Tb 2 Fe 14 B compounds, Ho 2 Fe 14 B exhibit the intrinsic properties of high H A , high T c and low J s .In addition, Ho and Nd elements have similar chemical characteristics.Di et al [16] found that the substitution of Nd by Ho can improve the coercivity, thermal stability and corrosion resistance of sintered Nd-Fe-B magnets.Liang et al [17] increased the H cj of Nd-Fe-B magnets significantly by adding Ho 63.4 Fe 36.6 alloys.Li et al [18] found that the incorporation of Ho can inhibit the formation of 1:2 phase and optimize the H cj and thermal stability of (Ce 1−x Ho x ) 14 Fe 80 B 6 alloys.Hence, it is feasible to improve the H cj and thermal stability of Y-Fe-B alloys by the substitution of Ho.However, the detailed investigations on the effects of Ho has not been carried out yet.
In this work, melt-spun Y-Fe-B based alloys with good magnetic properties and thermal stability were obtained by partial Ho substitution.Y-Ho-Fe-B magnets were also prepared by hot pressing and hot deformation process.The microstructure, element segregation behavior, and magnetic properties of magnets were investigated in detail.

Experimental
Y-Ho-Fe-B alloy ingots with nominal compositions of (Y 1−x Ho x ) 2 Fe 14 B and (Y 1−x Ho x ) 2.5 Fe 14 B (x = 0.1-0.5)were prepared by arc-melting the mixture of Y, Ho, Fe, and Fe-B metals, with purities higher than 99.8% under Ar atmosphere.Each ingot was re-melted 5 times for homogenization.The 5 g polished ingot was selected as the precursor for melt-spinning process.The ribbon samples were obtained by melt spinning in a sealed insert Ar environment, where the ejection pressure was 0.09 MPa and pore diameter of the quartz tube was 0.7 mm.The optimal magnetic properties of ribbons were obtained by adjusting the copper wheel velocity in the range of 15~36 m s −1 .Using melt-spun (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B powder with particle size of 100~200 μm as the precursors, the nanocrystalline hot-pressed (HPed) and hot-deformed (HDed) magnets were prepared by hot pressing (HP) or HP followed by hot deformation (HD) in vacuum.HP was performed at 700 °C for 30 min at 400 MPa, and HD was carried out with a deformation rate range of 0-0.02 mm s −1 at 700 °C for 60 min with ~70% height reduction.
The phase constitutions of the samples were characterized by x-ray diffractometer (XRD, X' Pert Pro, PANalytical, Netherlands) with Cu-K α radiation.The phase analysis was performed using the Rietveld refinement with Maud software.The microstructure was investigated by scanning transmission electron microscope (STEM, FEI Talos F200, USA) equipped with an energy dispersive spectrometer (EDS).The magnetic properties were obtained by a 5 T vibrating sample magnetometer (VSM) in the physical property measurement system (PPMS-9, Quantum Design).The magnetization-temperature (M-T) curves of samples were measured under a magnetic field of 500 Oe in the temperature range of 300-700 K.

Phase constitution and microstructure of melt spun Ho substituted Y-Fe-B alloys
The XRD refinement patterns for melt-spun (Y 1−x Ho x ) 2 Fe 14 B (x = 0.1-0.5)alloys are shown in figure 1(a).The RE 2 Fe 14 B (2:14:1) phase with the tetragonal structure (space group P4 2 /mnm) is formed for all samples.No secondary phase is detected, indicating that the substitution of Ho for Y will not change the phase constitution of melt-spun (Y 1−x Ho x ) 2 Fe 14 B alloys.The lattice parameters of (Y,Ho) 2 Fe 14 B phase in all samples calculated from the Rietveld refinement are shown in table 1.With the increase of Ho content, the lattice parameters of 2:14:1 phase remain basically unchanged, which is attributed to the similar radius of Ho 3+ ion (0.901 Å) and Y 3+ ion (0.900 Å) [19].Figure 1(b) shows the XRD refinement patterns for melt-spun (Y 1−x Ho x ) 2.5 Fe 14 B (x = 0.1-0.5)alloys.Except 2:14:1 phase, the REFe 2 (1:2) phase with the cubic structure (space group Fd-3m) can be detected.Unlike CeFe 2 phase (T c ≈ 235 K), which exhibits paramagnetic properties at room temperature, YFe 2 and HoFe 2 phases exhibit soft magnetic properties due to their higher T c of 545 K and 608 K, respectively [20][21][22].The presence of a small amount of REFe 2 phase acts as soft magnetic phase to enhance the remanence through the exchange coupling with 2:14:1 main phase.In addition, the small size of YFe 2 hard particles is usually existed in the Y-Fe-B ribbons, and is generally considered beneficial for the hard magnetic properties due to dispersion strengthening effect [13].However, the formation of high temperature YFe 2 phase is not beneficial to the deformation and densification of Y-Fe-B magnets during hot deforming.The precipitation of YFe 2 phase also consumes excess RE in the magnet that is essential for the construction of RE-rich grain boundary phase required for high coercivity.The high-resolution (HRTEM) images in figures 2(d), (g) and Fast Fourier transformation (FFT) pattern in figure 2(g) inset also identified the main grain of hard magnetic 2:14:1 phase.Compared with (Y 0.5 Ho 0.5 ) 2 Fe 14 B alloy, the grain boundaries in (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B alloy are obvious, as shown in figures 2(b) and (c), which is one of the main reasons for significantly enhanced coercivity by isolating the 2:14:1 grains.In order to clarify the element distribution in the alloys, figures 2(a)-(c) and table 2 show the atomic compositions of 2:14:1 phase and grain boundary phase analyzed by EDS.The RE/Fe atomic ratios of grains 1, 3 and 4 are close to that of RE 2 Fe 14 B, and the Ho/RE atomic ratios are 0.66, 0.59 and 0.62, respectively, which are higher than the nominal Ho/RE atomic ratio of 0.50.The result indicates that Ho has more tendency to enter into 2:14:1 phase than Y, which is beneficial to high coercivity.The RE/Fe atomic ratios of points 2 and 5 are 1:3.52 and 1:2.81,In addition, the J r of (Y 1−x Ho x ) 2 Fe 14 B alloys decrease not obviously with the increase of Ho content from x = 0.2 to x = 0.3.In table 3, J r /J 5T > 0.5 for all alloys suggests the existence of intergranular interaction.J r /J 5T value for x = 0.3 alloy is 0.61, higher than those of other (Y 1−x Ho x ) 2 Fe 14 B alloys, further indicating the significant remanence enhancement effect in the alloy with 30 at.% Ho substitution.
Figure 3(c) shows the hysteresis loops for melt-spun (Y 1−x Ho x ) 2.5 Fe 14 B (x = 0.1-0.5)alloys with RE-rich composition at room temperature.The (Y 1−x Ho x ) 2.5 Fe 14 B alloys show similar behavior of composition dependent magnetic properties as (Y 1−x Ho x ) 2 Fe 14 B alloys.However, the 2nd quadrant demagnetization curves (figure 3(c) inset) exhibit not uniform, indicating the weak exchange coupling between magnetic phases.The above analysis indicated that the soft magnetic RE 6 Fe 23 and REFe 2 phase are precipitated in the (Y 1−x Ho x ) 2.5 Fe 14 B alloys, so the uncoupled RE 6 Fe 23 or REFe 2 phase will become the nucleation center of the reverse magnetization, which is not beneficial to H cj .The similar reports have been found in Nd-Fe-B nanocomposites [25].For example, the x = 0.3 alloy shows no significant increase in H cj compared with x = 0.2 alloy due to its more uneven demagnetization process.Figure 3(d) shows the Ho content dependent magnetic parameters for (Y 1−x Ho x ) 2.5 Fe 14 B alloys.The (Y 0.9 Ho 0.1 ) 2.5 Fe 14 B alloy has J r , H cj and (BH) max values of 0.69 T, 332 kA m −1 and 56 kJ m −3 , respectively, and its H cj is close to that of (Y 0.7 Ho 0.3 ) 2 Fe 14 B alloy.With Ho content increases to 0.2, the alloy has the best combination of magnetic properties.The J r , H cj and (BH) max are 0.68 T, 672 kA m −1 and 69 kJ m −3 , respectively, much better than (Y 0.5 Ho 0.5 ) 2 Fe 14 B alloy.Despite the great difference in the intrinsic properties (J s and H A ) of Y 2 Fe 14 B and Ho 2 Fe 14 B phases, the J r of the alloys maintains a high level at Ho contents of x = 0.1 and x = 0.2.The isolation of hard magnetic phases by RE-rich grain boundaries is the main reason for high H cj in low Ho-substitution alloys.The H cj of (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B alloy reaches 1119 kA m −1 , which is even higher than Ho 2 Fe 14 B alloy of 1075 kA m −1 .The significant increased H cj should be caused by the RE-rich grain boundary phase and the substitution of Ho in the 2:14:1 phase.
Table 4 shows the comparison of melt-spun (Y 1−x Ho x ) 2.5 Fe 14 B in this work and Ce-Fe-B alloys with excellent magnetic properties by others.Compared with (Ce 0.7 La 0.3 ) 2.5 Fe 14 B alloy with J r = 0.60 T, H cj = 345 kA m −1 , and (BH) max = 50 kJ m −3 , (Y 1−x Ho x ) 2.5 Fe 14 B alloys with the same RE content show the more excellent magnetic properties, where the magnetic properties of x = 0.2 and 0.3 alloys are J r = 0.68 T, H cj = 627 kA m −1 , (BH) max = 69 kJ m −3 and J r = 0.58 T, H cj = 671 kA m −1 , (BH) max = 53 kJ m −3 , respectively.In addition, the coercivity of Ce 17 Fe 78 B 6 alloy (16.8 at.% of RE) is only 491 kA m −1 , and the coercivity of Ta or Co(Zr)substituted Ce-Fe-B alloys are about 437~553 kA m −1 , even lower than that of (Y 1−x Ho x ) 2.5 Fe 14 B (x = 0.2-0.5)alloys with lower RE ratios (14.3 at.% of RE).Ho substitution can significantly optimize the coercivity of Y-Fe-B alloys.
Figure 4(a) shows the M-T curves for melt-spun (Y 1−x Ho x ) 2 Fe 14 B (x = 0.1-0.5)alloys.An obvious magnetic phase transition was observed in the temperature range of 300-700 K, which should be corresponding to the T c of the 2:14:1 phase.With the increase of Ho content from x = 0.1 to x = 0.5, T c increases from 546 K to 564 K, which can be explained by the higher T c of Ho 2 Fe 14 B than that of Y 2 Fe 14 B. The higher T c is beneficial to the thermal stability of (Y 1−x Ho x ) 2 Fe 14 B alloys.The temperature dependences of remanence and coercivity are shown in figure 4(b).With the increase of temperature, J r and H cj decrease monotonously.The alloys with low Ho-substitution (x = 0.1 and 0.2) show relatively low temperature dependence due to the weak temperature sensitivity of anisotropy field of Y 2 Fe 14 B compound.Figure 4(c) shows the remanence temperature coefficients (α) and coercivity temperature coefficients (β) in 300-400 K (T 0 -T), defined by equations (1) and (2) [14].
The values of α and β for x = 0.1 alloy are −0.138%/K and −0.088%/K, respectively.With the substitution of Y by Ho, the opposite changes of α and β are observed.The (Y 0.7 Ho 0.3 ) 2 Fe 14 B alloy show the α and β values of −0.124%/K and −0.245%/K, respectively.The increase of α with increasing Ho substitution may be related to the higher Ho content in 2:14:1phase.

3.3.
Phase composition and magnetic properties of hot worked (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B magnets The preparation of bulk magnets is essential for practical applications.Due to Y-Fe-B is difficult to obtain orientation (see supplementary data), we choose the melt-spun (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B alloys with higher Ho substitution as precursors of magnets.Figure 5 shows the XRD patterns for the HPed and HDed (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B magnets.Here the original powders used for hot pressing and hot deformation are as-melt spun amorphous ribbons prepared at v = 36 m s −1 wheel speed.The reason to use amorphous powder as the precursor is to inhibit the precipitation or growth of high temperature 1:2 phase and 6:23 phase during hot pressing and hot deformation.The XRD patterns for the ribbons annealed at 700 °C for 10 min is also shown in figure 5, and there is almost no 1:2 phase detected in the annealed ribbons.Similarly, no obvious 1:2 phase is precipitated in the HPed or HDed magnets.As we know, 1:2 phase with high melting point is not beneficial to the uniform distribution of RE-rich grain boundary phase [26], which is necessary for isolating the main phase grains and improving the c-axis orientation of hot-deformed RE-Fe-B magnets.However, although the 1:2 phase content is very low in the HDed magnet, the relative strength of (00l) diffraction peaks of 2:14:1 phase in HDed magnet is not significantly increased compared with those of ribbons or HPed magnet, indicating that the HDed magnet is not well oriented, which may be related to the high melting 6:23 phase.Figure 6(a) shows the hysteresis loops for the HPed and HDed (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B magnets.The uniform demagnetization behavior is observed, which is different from that of melt spun ribbons in figure 3(c).The result indicates that there is strong exchange coupling effect in hot worked magnets.Due to the lack of low melting point RE-rich phase, (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B magnets are difficult to deform and orient during hot-deformation process.Consequently, the cracks and the rearrangement of small oriented particles at the broken point result in an abnormal reduction in J r of the magnet after hot-deformation.In addition, for the HDed magnets with high anisotropy, their grains generally change from equiaxial shape to platelet shape, resulting in an increase in demagnetization factor and effective demagnetization field.Hence, generally of HDed magnet has lower H cj than HPed magnet.However, figure 6(a) shows that the H cj of HDed (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B magnet is close to that of HPed magnet, which further indicated that the orientation of HDed magnet is poor.Finally, the values of J r , H cj and (BH) max for HPed and HDed (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B magnets are 0.60 T, 770 kA m −1 , 56 kJ m −3 and 0.50 T, 739 kA m −1 , 40 kJ m −3 , respectively.The results show that the quaternary Y-Ho-Fe-B magnets have potential to achieve good hard magnetic properties compared with high coercivity.
3.4.Microstructure and element distribution behavior of HDed (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B magnet Figure 7 shows the bright field TEM images (a), (d), (j) of different regions of HDed (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B magnet.The roughly equiaxed grains indicate the low orientation and anisotropy of the magnet.There are many small particles are discontinuously distributed at the boundaries of large grains in figure 7(a), which are likely to be the high temperature RE-rich phase, and their high melting point and high deformation resistance are not beneficial to the orientation of the magnet.Figures 7(a) and (d) show different grain size.In figure 7(d), there is clearly two regions with large grains and small grains, which may be attributed at the uneven distribution of intergranular RE-rich phase.The existence of intergranular RE-rich phase can inhibit the grain coalescence and growth during hot deformation, and the coalescence mechanism is usually used to explain the grain growth behavior of hotdeformed Nd-Fe-B magnets [30,31].Figures 7(b) and (e) show the HRTEM images of red square region in figures 7(a) and (d), respectively.The HRTEM and FFT images for region I, II and III in figures 7(b) and (e) are shown in figures (c), (f), (g) and (h), (i), respectively.Except the 2:14:1 main phase, the small grains as RE-rich phases are confirmed to be RE 2 O 3 (region I) and REFe 2 phase (region III).
The EDS analysis for the grains with different sizes in figures 7(d) and (j) are listed in table 5.The RE/Fe atomic ratio at point 1 is close to 1:2, so the grain 1 can be indexed as REFe 2 phase, which is consistent with the above FFT analysis results of region III.The RE/Fe atomic ratio at point 2 in the RE-rich region around the grains is similar to that of the point 1, and the RE 2 O 3 phase was detected combined with FFT analysis of region I.In addition, the selected small grains (point 5 and 8) and light area (point 4) at grain boundaries in figure 7(j) were confirmed to be RE 6 Fe 23 phase and Fe-rich phase by the EDS analysis in table 4. Interestingly, the Ho/RE atomic ratios in 2:14:1 phase and RE-rich phase gains in different regions are higher than that in the nominal composition of (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B alloy, as shown in table 5, indicating that the additional Y-rich precipitates may be existed in the magnet.The EDS line-scan results from blue lines in figures 7(a) and (d) also show that the same distribution of Y and Ho elements inside the 2:14:1 main phase or RE-rich phase grains, and the proportion of Ho is higher than nominal alloy composition.Figure 7(l) shows that there is large RE content differences in RE-rich region and Fe-rich region between small and large 2:14:1 grains.
Figure 8 shows the selected high-angle annular dark-field (HAADF) STEM-EDS elemental mappings of different regions for the HDed (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B magnet.Ho and Y elements discontinuously distributed at grain boundaries as shown in figure 8(a), indicating that the RE-rich phase mainly surround the 2:14:1 grains in the form of small particles, which is beneficial for high coercivity.The analysis in combination with figure 7 confirmed that the RE-rich phases are REFe 2 phase, RE 6 Fe 23 phase, and RE 2 O 3 phase.Figure 8(b) is STEM-EDS elemental mapping with large magnification, which shows that the Fe-rich phase marked by the arrows also exist  around the large 2:14:1 grains in addition to the discontinuous distribution of RE-rich phase.The precipitation of Fe-rich phase is related to the uneven distribution of rare earth elements, which can also be found between the RE-rich and Fe-rich regions in figure 8(c).

Conclusions
Based on the melt-spun Y-Ho-Fe-B ribbons, an attempt to prepare nanocrystalline Y-Ho-Fe-B magnets by hot pressing and hot deformation process was carried out in this work.The effects of Ho substitution on microstructure, element distribution, magnetic properties and thermal stability of alloys and magnets have been clarified.Influenced by the intrinsic properties of Ho 2 Fe 14 B, the coercivity and the Curie temperature of Y-Ho-Fe-B ribbons increases significantly while the remanence greatly decreases with Ho substitution.However, J r exhibits an abnormal increase when x = 0.3, and the magnetic properties for (Y 1−x Ho x ) 2 Fe 14 B alloy are J r = 0.73 T, H cj = 303 kA m −1 and (BH) max = 66 kJ m −3 , together with excellent temperature coefficients α = −0.124%/K and β = −0.245%/K in 300-400 K.The results for RE-rich (Y 1−x Ho x ) 2.5 Fe 14 B alloys show that the J r is maintained at a high level by low Ho substitution, and the isolation of 2:14:1 phase by RE-rich grain boundaries is the main reason for the increase of H cj .The (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B hot worked magnets were successfully prepared, and the magnetic properties of hot-pressed magnet are J r = 0.60 T, H cj = 770 kA m −1 and (BH) max = 56 kJ m −3 , respectively.The microstructure and element distribution analyses indicated that the Ho/Y ratios in both 2:14:1 grain and RE-rich grain boundaries are higher than that in the nominal alloy composition.The high temperature RE 2 O 3 and RE 6 Fe 23 phases are discontinuously distributed in the grain boundaries which is not beneficial to the deformation and orientation of the hot-deformed magnet.
Figure 1(b)  shows that melt-spun (Y 1−x Ho x ) 2.5 Fe 14 B alloys has very few 1:2 phase precipitation.Based on the calculated from the Rietveld refinement, as the Ho content increases from x = 0.1 to x = 0.5, the mass fraction of 1:2 phase increases only from 1.1 wt.% to 1.8 wt.%.The results suggested that the introduction of Ho does not promote 1:2 phase precipitation.Figures2(a)-(c) shows the bright field TEM images for melt-spun (Y 0.5 Ho 0.5 ) 2 Fe 14 B and (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B alloys.The well-crystallized 2:14:1 phase grains directly contacting with each other are observed in figure 2(a).

Figure 5 .
Figure 5. XRD patterns for the HPed and HDed (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B magnets.And the XRD of original ribbons annealed at 700 °C for 10 min as the control group.

Table 1 .
Phase composition and lattice parameters for melt-spun (Y 1−x Ho x ) 2 Fe 14 B alloys.

Table 2 .
[23]ic Fe 23 and REFe 2 phase, respectively.Sun et al[13]found that except for Y 2 Fe 14 B main phase, YFe 2 secondary phase would appear in melt-spun Y 16 Fe 78 B 6 alloys.Cheng et al[23]discussed the phase transformation and solidification microstructure of Y-Fe-B alloys, and they found that the Y 6 Fe 23 phase was first precipitated in the RE-rich alloy during supercooling process, which is different from the primary Y 2 Fe 17 phase in RE-lean alloys.Combined with XRD results in figure1(b)where the mass fraction of 1:2 phase is relatively low, the excessive RE elements may mainly exist in the form of 6:23 phase.3.2.Magnetic properties and thermal stability of melt-spun Ho substituted Y-Fe-B alloysFigure3(a) shows the 2nd quadrant demagnetization curves for melt-spun (Y 1−x Ho x ) 2 Fe 14 B (x = 0-0.5)alloys measured at room temperature.With increasing Ho content, the remanence (J r ) decreases significantly while the coercivity (H cj ) shows the opposite effect, which could be attributed to the difference in intrinsic magnetic properties between Ho 2 Fe 14 B and Y 2 Fe 14 B compounds.Ho 2 Fe 14 B has lower saturation magnetization (J [14]osition of the marked areas by EDS point-scan in figure 2 (pink point).Phase Point Y (at.%)Ho (at.%)Fe (at.%) O (at.%)respectively, which are confirmed to be RE 6 Fe 23 and REFe 2 , respectively.Moreover, the Ho/RE atomic ratio in point 2 and 5 are 0.73 and 0.68, respectively, higher than that of 2:14:1 main phase, indicating that Ho not only preferably segregates into the 2:14:1 phase, but also more likely enter into the RE-rich grain boundary phase.The red square region in figures 2(b) and (c) are grain boundary regions of (Y 0.5 Ho 0.5 ) 2.5 Fe 14 B alloy, and their HRTEM images are shown in figures 2(e) and (f), respectively.The HRTEM image of selected regions II, III and IV in figures (e) and (f) are shown in figures 2(h), (i), and (k), respectively.FFT image are shown in (h) inset, (j), (l).It is indicated that the main grain is 2:14:1 phase, while the grain boundary regions of region II and III are RE 6 s = 0.8 T) and higher anisotropy field (H A = 7.5 T) of than Y 2 Fe 14 B (J s = 1.41 T, H A = 2.6 T)[24].Thus, the partial substituting Y by Ho could decrease J s and increase H A of (Y 1−x Ho x ) 2 Fe 14 B alloys, resulting in expected high H cj and low J r Similar results have been reported in (Y 1−x Ce x ) 2 Fe 14 B alloys[14].

Table 3 .
The magnetic properties and thermal stability for melt spun (Y 1−x Ho x ) 2 Fe 14 B ribbons.

Table 4 .
The magnetic properties for melt spun (Y 1−x Ho x ) 2.5 Fe 14 B and Ce-Fe-B alloys.

Table 5 .
Atomic composition of the marked areas by EDS analysis in figure 7.