Accepted Manuscript Enhanced coercivity of Nd-Ce-Fe-B sintered magnets by adding (Nd, Pr)-H powders Tianyu Ma, Bo Wu, Yujing Zhang, Jiaying Jin, Kaiyun Wu, Shan Tao, Weixing Xia, Mi Yan PII:
S0925-8388(17)31874-1
DOI:
10.1016/j.jallcom.2017.05.257
Reference:
JALCOM 41988
To appear in:
Journal of Alloys and Compounds
Received Date: 6 April 2017 Revised Date:
21 May 2017
Accepted Date: 25 May 2017
Please cite this article as: T. Ma, B. Wu, Y. Zhang, J. Jin, K. Wu, S. Tao, W. Xia, M. Yan, Enhanced coercivity of Nd-Ce-Fe-B sintered magnets by adding (Nd, Pr)-H powders, Journal of Alloys and Compounds (2017), doi: 10.1016/j.jallcom.2017.05.257. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT Enhanced coercivity of Nd-Ce-Fe-B sintered magnets by adding (Nd, Pr)-H powders Tianyu Ma a, Bo Wu a, Yujing Zhang a, Jiaying Jin a, Kaiyun Wu a, Shan Tao a, Weixing Xiab,*, and Mi Yana,* a
School of Materials Science and Engineering, State Key Laboratory of Silicon Materials,
Zhejiang University, Hangzhou 310027, China b
RI PT
Key Laboratory of Novel Materials for Information Technology of Zhejiang Province,
CAS Key Laboratory of Magnetic Materials and Devices, Ningbo Institute of Materials
SC
Technology and Engineering, Chinese Academy of Sciences, Ningbo 315201, China Abstract
M AN U
Incorporating the highly abundant rare earth (RE) Ce into Nd-Fe-B sintered magnets has attracted considerable interest recently. The inferior anisotropic field (HA) of Ce2Fe14B to Nd2Fe14B, however leads to low coercivity of the Nd-Ce-Fe-B magnets. To further enhance the coercivity, in this work, (Nd, Pr)-H powders were used as intergranular additive to
TE D
restructure the grain boundaries of the low cost (Nd,Pr)22.3Ce8.24FebalB1 (wt.%) sintered magnets. When added with only 2 wt.% (Nd, Pr)-H, the Hcj can be enhanced from 10.6 kOe to 12.7 kOe with slight reduction in remanence and maximum energy product. The
EP
dehydrogenation of (Nd, Pr)-H during sintering promotes the diffusion of Nd and Pr towards the 2:14:1 phase grains and the smoothing of grain boundaries. The formation of (Nd, Pr)-rich
AC C
shell with locally enhanced magnetocrystalline anisotropy and the magnetic isolation between adjacent 2:14:1 phase grains result in obvious coercivity enhancement, which was supported by magnetic domain structure characterizations and micromagnetic simulations. It suggests that through grain boundary restructuring, the coercivity of Nd-Ce-Fe-B magnets can be enhanced to be comparable of commercial Nd-Fe-B magnets, which may shed new insights into the fabrication of low cost RE permanent magnets. Keywords: Nd-Fe-B permanent magnets, coercivity, microstructure.
1
*
ACCEPTED MANUSCRIPT Corresponding authors. E-mail addresses:
[email protected] (Prof. W. Xia) and
[email protected] (Prof. M. Yan).
1. Introduction
RI PT
From an economic point of view, incorporating Ce into Nd-Fe-B sintered magnets has attracted renewed interest recently due to its rich natural abundance, relatively low price and full solubility in the 2:14:1 phase [1-12]. However, high performance and high Ce substitution
SC
level in the Nd-Ce-Fe-B magnets cannot be obtained simultaneously because the intrinsic magnetic properties of Ce2Fe14B (saturation magnetic polarization JS = 1.17 T and
M AN U
magnetocrystalline anisotropic field HA = 26 kOe at room temperature) are much inferior to Nd2Fe14B (JS = 1.60 T and HA = 73 kOe) [2]. For instance, when 20 wt.% Nd is replaced by Ce, coercivity of the (Nd,Pr)24.8Ce6.2FebalB1 (wt.%) magnet sintered for 2 h at 1010 degrades seriously to as low as 7.7 kOe [9]. Unfortunately, the obtained coercivity is much
TE D
lower than the commercial grade magnets, for which the coercivity should not be less than 12 kOe at room temperature [13]. Consequently, there is still a large coercivity gap between the commercial Nd-Fe-B magnets and the Nd-Ce-Fe-B magnets with Ce content over 20wt.%.
EP
It is generally accepted that coercivity of the 2:14:1-type RE-Fe-B magnets (RE = rare earth) is determined mainly by the nucleation of reverse magnetic domains, which primarily
AC C
occurs at the surface region of 2:14:1 phase grains due to the presence of defects or at an area adjacent to the non-ferromagnetic grains with a high stray field [14-16]. Many efforts have been taken to enlarge the nucleation field HN to enhance the coercivity of Nd-Fe-B sintered magnets through optimizing microstructure. In comparison with the way of increasing HA of the 2:14:1 phase by direct alloying with expensive heavy rare-earth (HRE) Dy/Tb [17], the approaches of post-sinter annealing [18,19], infiltrating low-melting point RE-containing alloys [20-23], and refining grains [24, 25] are more economic. Aim of these approaches is to 2
ACCEPTED MANUSCRIPT obtain the close-to-ideal microstructure, i.e. fine Nd2Fe14B grains are separated by continuously thin nonmagnetic grain boundary (GB) layers so that the short-range exchange coupling between adjacent grains is avoided and/or Nd2Fe14B grain cores are covered by Dy-rich 2:14:1 phase shells so that the nucleation of reverse domains at lower external fields
RI PT
can be suppressed [26]. For instance, in our recent work [27], the coercivity of a (Nd,Pr)22.3Ce8.24FebalB1 (wt.%) sintered magnet can be enhanced to 10.4 kOe by performing two-step annealing for 2h at 890
and 4.5h at 520
to form continuous and smooth GBs.
SC
Alternatively, similar microstructure can also be formed by diffusing low melting point sources, such as Nd-Cu and Nd-Ag to Ce-Fe-B ribbons [28] or Nd-Cu [29] and Nd-Fe [30] to
M AN U
(Nd,Ce)-Fe-B magnets. Accordingly, the coercivity of Ce-Fe-B or (Nd,Ce)-Fe-B magnets can be effectively enhanced through microstructure optimizing.
To further enhance the coercivity, in the present work, infiltrating source (Nd80Pr20)-Hx powders were introduced into the (Nd,Pr)22.3Ce8.24FebalB1 magnet. The RE hydride powders
TE D
are easily prepared by hydrogen decrepitating the Nd80Pr20 alloy, which is commonly used as raw materials in the fabrication of Nd-Fe-B sintered magnets [31]. This intergranular additive is also desirable to aid in-situ diffusing and to enlarge the RE-rich phase volume fraction at
EP
intergranular regions after dehydrogenation, which are beneficial to form magnetically hardening (Nd,Pr)-rich 2:14:1 phase shells and to form continuous nonmagnetic GB layers.
AC C
Coercivity of the restructured Nd-Ce-Fe-B magnet is found to increase linearly with (Nd80Pr20)Hx content, and exceeds 12 kOe with trace doping, i.e. at 2 wt.%. 2. Experimental
Magnetic powders with nominal composition of [(Nd, Pr)73Ce27]30.5FebalM1.3B1.0 (M = Cu, Al, Ga, Zr, in wt.%) were prepared by induction melting, strip casting, hydrogen decrepitation (HD) and jet milling. The raw materials are Nd80Pr20 alloy, Fe80B20 alloy, Ce, Fe, Cu, Al, Ga, and Zr metals with purity above 99 wt.%. Hereafter, these powders are named as Ce-27. (Nd80Pr20)Hx hydrides were prepared by HD treating Nd-Pr alloy consisting of 80 wt. % Nd 3
ACCEPTED MANUSCRIPT and 20 wt.% Pr for 2 hrs at 400
. The as-decrepitated pieces were subjected to high energy
ball milling to prepare fine powders with mean particle size of 3.5 µm. These two kinds of powders were blended together with (Nd80Pr20)-Hx ranging from 0 to 3.0 wt.% under the protection of nitrogen gas. The mixed powders were compacted and aligned under a pressure
RI PT
of 5 MPa and a magnetic field of 1.8 T, followed by isostatic pressing under ~200 MPa. The green compacts were sintered for 2 hrs at 1040 ℃ and subsequently subjected to further annealing for 4 hrs at 600 ℃ under the protection of high purity argon atmosphere. To avoid
SC
possible oxidation, the pressure in the furnace is kept below 2×10-3 Pa during heating, sintering and annealing. Magnetic properties were measured using a NIM-10000H
M AN U
hysteresigraph analyzer system. Microstructure observations were conducted using the scanning electron microscope (SEM, Hitachi S-3400N). Elemental concentration mapping was performed using an electron probe microanalyzer (EPMA) with wavelength dispersive X-ray spectrometer (WDS). Microstructure and electron diffraction were also characterized
TE D
using a JEM-2100F transmission electron microscope (TEM). Magnetic domain structures and electron holography were conducted using the microscope equipped with a electron biprism. The under- and over-focused images are also analyzed with the transport-of-intensity
EP
equation (TIE) using the QPt software package (HREM Co.) to obtain the magnetization mapping [32, 33]. TIE is usually used to detect the phase from the intensity of electron wave,
AC C
therefore the distribution of magnetization vector of the sample can be imaged. To further investigate the possible influences of (Nd,Ce)2Fe14B core - Nd2Fe14B shell structure and magnetism of the grain boundaries on the coercivity, a model was constructed according to the microstructure derived from SEM and Lorentz-TEM. In this model, 8 cuboid-shaped matrix phase grains with side L = 100 nm are considered, each of them is separated by 2-nm-thick GB phase. Cubic meshes with a size of 2nm were applied to the simulation. The Landau-Lifshitz-Gilbert (LLG) equation at each node was solved by the 3D 4
ACCEPTED MANUSCRIPT NIST OOMMF software [34]. Material parameters of Nd2Fe14B and Ce2Fe14B are tracked from literatures [2, 35, 36], i.e. the saturation magnetization (µ0Ms), the magnetocrystalline anisotropy constant (K1), and the exchange stiffness (A) are 1.61 T, 4.3 MJ/m3 and 7.7 PJ/m for Nd2Fe14B, 1.17T, 1.5MJ/m3 and 5PJ/m for Ce2Fe14B, respectively. The material
RI PT
parameters for the (Nd,Ce)-Fe-B were obtained by using a linear interpolation between the values of these two extreme compositions, similar to those reported in [37]. 3. Results and Discussion
SC
Figure 1 shows the room temperature demagnetization curves for the starting magnet (Ce-27) and the magnets added with 1, 2 and 3 wt.% (Nd80Pr20)Hx, respectively. The inset
M AN U
shows the magnetic properties Hcj and Br as a function of (Nd80Pr20)Hx amount. All the magnets have high squareness factor over 90% and large density above 7.56 g/cm3. The starting magnet possesses coercivity Hcj of 10.6 kOe, remanence Br of 12.8 kG and maximum energy product (BH)max of 39.3 MGOe. With the increase of (Nd80Pr20)Hx amount, Hcj
TE D
increases gradually to 12.0 kOe at 1wt.%, 12.7 kOe at 2 wt.% and 12.9 kOe at 3 wt.%, respectively. The coercivity increment is accompanied with very slight reduction in Br and (BH)max. For instance, Br is 12.4 kG and (BH)max is 37.8 MGOe when added with 2 wt.%
EP
(Nd80Pr20)Hx. It suggests that the Nd-Pr hydrides bring a significant coercivity enhancement, and the magnetic properties are comparable to 38MGOe commercial magnet. The cost of rare
AC C
earth raw materials is estimated for the present Nd-Ce-Fe-B magnet and a 38MGOe commercial magnet, respectively. Identified by inductively coupled plasma (ICP) analysis, the commercial magnet contains ~ 22 wt.% Nd, ~5.2 wt.% Pr and ~4.3 wt.% Gd, whose cost is ~$14.7/kg (as of April 2017 [38]). The cost of RE raw materials in our 2 wt.% (Nd80Pr20)Hx added (Nd,Pr)22.3Ce8.24FebalB1 magnet is just ~$12.1/kg, 17.7% lower than that for the commercial magnet. Consequently, the present Nd-Pr-Ce-Fe-B magnet added with (Nd80Pr20)-Hx has much lower material cost than the commercial ones. The Ce-27 magnets added with different (Nd80Pr20)-Hx amounts show typical bi-phase 5
ACCEPTED MANUSCRIPT morphology, as illustrated in Fig. 2, where the dark contrast is the matrix 2:14:1 phase and the bright contrast is the intergranular RE-rich phase. For the starting magnet, the RE-rich phase appears to be thin and discontinuous along the GBs (Fig. 2a). In some areas, adjacent 2:14:1 phase grains contact directly with each other. Short-range exchange coupling could exist for
RI PT
such local grains, which is not beneficial to obtain high coercivity [14, 15, 18]. When (Nd80Pr20)Hx is added, the distribution of RE-rich phase changes apparently (Figs. 2b, c, and d). The intergranular RE-rich phase becomes clearer and more continuous when compared to
SC
the starting magnet, isolating well the adjacent grains. Such distribution of RE-rich GB phase can eliminate the surface defects and is beneficial to reduce the possible reversal nucleation
M AN U
sites [24, 26], which is one important contribution to the enhanced coercivity. Further element distribution analysis confirms that Nd and Pr elements are enriched at the surface regions of 2:14:1 phase grains, as shown in Figs. 2 e and f. Because both HA (87 kOe) and JS (1.56 T) of Pr2Fe14B are also larger than those for Ce2Fe14B [2], the enrichments of Nd and Pr in the outer
TE D
region of 2:14:1 phase strongly enhance the local magnetocrystalline anisotropy. This, in turn, can suppress the nucleation of reverse magnetic domains at lower external fields. The formation of (Nd,Pr)-rich shell at the outer region of (Nd,Pr,Ce)2Fe14B grains is like the case
EP
of forming a magnetic hardening Dy-rich shell at the outer region of Nd2Fe14B grains [20-23], thus enhancing the coercivity efficiently. In addition, the higher JS values of Nd2Fe14B and
AC C
Pr2Fe14B than Ce2Fe14B imply that the formation of (Nd,Pr)-rich shell is also beneficial to suppress the degradation of remanence. Note that the volume fraction of RE-rich phase increases when increasing (Nd80Pr20)Hx amount, i.e. the volume fraction of 2:14:1 phase decreases, which will lead to the reduction of remanence. Such effect may exceed the positive role of forming the (Nd,Pr)-rich hardening shell at outer region of the 2:14:1 phase grains. Consequently, a slight reduction in remanence (the inset of Fig.1) is then observed after adding (Nd80Pr20)-Hx. TEM characterizations of typical region in the magnet added with Nd-Pr-H reveal more 6
ACCEPTED MANUSCRIPT clearly the GB feature, as shown in Fig. 3. The low magnification bright field image (BFI) in Fig.3a and high resolution TEM image in Fig.3b were taken with the electron beam parallel to [011] zone axis of the bottom 2:14:1 phase grain. As shown in Fig.3c, the corresponding electron diffraction pattern (EDP) of the selected region (red dashed circle in Fig.3a) close to
RI PT
GB is taken using a very small aperture with the diameter of ~200 nm. To determine the crystal symmetry, this region is tilted to another direction, with the electron beam parallel to [001] zone axis, the EDP is shown in Fig.3d. These two EDPs demonstrate that the outer
SC
region of the matrix phase grain possess the tetragonal crystal symmetry, whose space group is P42/mnm. The HRTEM images with specific crystal plane perpendicular to the electron
M AN U
beam allow one to clearly identify the thickness of GB layers. For the magnet added with 3 wt.% Nd-Pr-H (Fig. 3b), the GB layer’s thickness is ~15 nm, which can ensure good isolation of neighboring 2:14:1 phase grains. Such GB layers play an important role to decouple the adjacent 2:14:1 phase grains during magnetization reversal [14-16].
TE D
Direct evidence of magnetic decoupling is presented in Figs.4 and 5. Figs. 4a and b shows typical magnetic domain structures of the starting magnet with Fresnel mode in the thermally demagnetized state. The sample is prepared with c-axis in plane, and the images are
EP
taken without applying magnetic field. Magnetic domains across grain boundary can be observed as shown in Fig.4a, where the GB is perpendicular to the c-axis. The continuous
AC C
domain wall (the bright line) across the GB indicates that the adjacent grains (G1 and G2) are not magnetically isolated. Off-axis holograph was used to further illustrate the magnetization distribution within the region highlighted with dashed blue rectangle in Fig. 4a. The corresponding reconstructed holography image is clearly displayed in Fig. 4b. It also demonstrates that the lines of magnetic flux are passing through the adjacent hard magnetic grains. The lines orient almost along the same direction, which indicates that short-range exchange coupling exists between them [39]. In this case, i.e. adjacent 2:14:1 phase grains are magnetically coupled, the magnetization reversal will be cascaded rapidly across the GBs 7
ACCEPTED MANUSCRIPT once nucleation of reverse domains occurs at the weakest point in the bulk magnet. Fig. 5 shows the magnetic domain structures with c axis of the 2:14:1 phase out of the observed plane, where the over-focused Fresnel Lorentz images are taken in zero magnetic field (demagnetized state) for both the starting magnet (Fig. 5a) and the magnet added with 3
RI PT
wt.% (Nd80Pr20)Hx (Fig. 5c). Maze-like domain structure can be observed from this view direction, which is in agreement with previous observations [40]. To further illustrate clearly the magnetic domain configurations, calculated magnetization maps are illustrated in Figs. 5b
SC
and d, respectively, which are obtained by under-, in-, and over-focused Fresnel Lorentz images and TIE method. There exists clear difference in domain structure between these two
M AN U
samples. For the starting magnet (Figs. 5a and b), three grains are separated by a small triple junction and two GB layers. The magnetic domains of grain 1 and grain 2 are almost continuous across their grain boundary (GB1) (being consistent with Fig. 4), while the domain structures of grain 2 and grain 3 are separated by GB2. On the contrary, for the
TE D
magnet added with 3 wt.% (Nd80Pr20)Hx (Figs.5c and d), each grain forms its own individual domain structure and they are separated by the GBs and domains crossing different grains are not observed. If the GB is ferromagnetic or thin enough (less than 2.1 nm in [18]), strong
EP
exchange coupling between grains make the domain structure continuous as shown by GB1, which were rarely found in specimens added with (Nd80Pr20)Hx. If the GBs are thick, the
AC C
exchange coupling can be weakened and thus domain structure is isolated by GBs as shown in GB2 in Figs. 5a and b and GBs in Figs. 5c and d. As mentioned above, the microstructural modification and the formation of Nd/Pr element grade with (Nd80Pr20)Hx addition can weaken the exchange coupling effect between adjacent grains and strengthen the local magnetocrystalline anisotropy at the grain outer regions, hence enhancing the coercivity of Nd-Ce-Fe-B magnet. To understand the role of RE hydrides on the microstructure of Nd-Ce-Fe-B sintered magnets, the related reactions during sintering and post annealing should be addressed. It is noted that the hydrogen release events 8
ACCEPTED MANUSCRIPT
in RE hydrides are associated with the low temperature desorption REH3 → REH2 (approximate formulae) and the high temperature desorption REH2 → RE [31, 41]. Upon heating in vacuum, desorption reaction finishes at 800
for Nd hydride and 730
for Pr
hydride, respectively [41]. In the present work, the magnets are heated and sintered with very
sintering temperature 1040
RI PT
low pressure that is close to the vacuum condition. Consequently, upon heating to the , the hydrogen is desorbed gradually, the powders become very
reactive and aid in-situ diffusing from the liquid RE-rich phase towards the unmelt 2:14:1
SC
phase particles. The excess Nd/Pr will stay at the intergranular regions, enlarging the volume fraction of liquid phase. It is beneficial to increase the direct contact area between GBs and
M AN U
the 2:14:1 phase grains and to improve the wettability. Consequently, continuous and homogeneous intergranular GBs are formed. Post annealing will further smooth GBs and promote their continuity and thickness. In addition, the released hydrogen can react with the oxygen of REOx phase at the intergranular regions and form metallic RE-rich phase [31],
2:14:1 phase.
TE D
which is also beneficial to improve the wettability between the intergranular phase and the
Our micromagnetic simulation results further support the positive contributions of grain
EP
boundary restructuring on the coercivity of Nd-Ce-Fe-B magnets. Shown in Fig. 6 are the simulated results for four different cases, (I) (Nd0.75Ce0.25)2Fe14B grains separated by
AC C
ferromagnetic GB layer (µ0MsGB = 0.8T), (II) (Nd0.75Ce0.25)2Fe14B grains separated by nonmagnetic GB layer (µ0MsGB = 0 T), (III) (Nd0.75Ce0.25)2Fe14B core is covered by 2-nm-thick Nd2Fe14B shell, each grain is separated by ferromagnetic GB layer (µ0MsGB = 0.8T), and (IV) (Nd0.75Ce0.25)2Fe14B is covered by 2-nm-thick Nd2Fe14B shell, but each grain is separated by nonmagnetic GB layer (µ0MsGB = 0 T). For all cases, 8 cuboid-shaped grains with a size of 100×100×100 nm3 separated by 2-nm-thick GB layer are considered, where the cross-sections are shown in Figs. 6 (a) and (b), respectively. From the simulated 9
ACCEPTED MANUSCRIPT demagnetization curves (Fig. 6c), it can be clearly seen that the coercivity becomes larger if magnetization of the GB is weakened to zero (comparison between case I and case II). The coercivity can also be enhanced once the Nd2Fe14B shell is formed in the surface region of (Nd,Ce)2Fe14B core, as revealed by the difference between case I and case III, or the
RI PT
difference between case II and case IV. Although the coercivity obtained from simulation is larger than the one obtained from the real magnet, it indeed demonstrates that both magnetic isolation and strengthening of local magnetocrystalline anisotropy play positive roles on
SC
enhancing coercivity.
Finally, the above findings demonstrate that restructuring the GB phase is very effective
M AN U
to enhance the coercivity of the low cost Nd-Ce-Fe-B magnets. As pointed out by Kronmüller [42], the magnetocrystalline anisotropy of the 2:14:1 phase grains surface region is always lower than that of inside the grains, which lead to an obvious reduction in the magnetic reversal nucleation field HN and the coercivity of the Nd-Fe-B magnet. The HN can be
TE D
enlarged effectively by constructing the so-called core-shell structure, i.e. the surface regions of the Nd2Fe14B grains were enriched with heavy rare earth elements, Tb, Dy or Ho [14-17, 22, 23, 31, 43-45]. In our case, the Nd2Fe14B shell has much larger K1 than the (Nd,Ce)2Fe14B
EP
core, it then can enlarge HN effectively like the case of Nd2Fe14B core - (Nd, Dy)2Fe14B shell structure. In addition, previous investigations also demonstrated that the formation of
AC C
continuous nonmagnetic GB thin layers surrounding Nd2Fe14B grains is the key microstructural feature of high-coercivity Nd-Fe-B magnets [14, 24]. The nonmagnetic GB layers can weaken the short-range exchanging coupling between adjacent grains, and subsequently avoid the rapid propagation of magnetization reversal from one grain to another. As proved by our SEM and L-TEM images, such key microstructural feature indeed forms through incorporating extra Nd-Pr hydrides into the Nd-Ce-Fe-B magnets. 4. Conclusions Coercivity of the Nd-Ce-Fe-B sintered magnet is enhanced by incorporating Nd-Pr 10
ACCEPTED MANUSCRIPT hydrides to restructure the grain boundaries. With adding a slight amount of 2wt.%, the coercivity of Nd-Ce-Fe-B with 27 wt.% Ce substitution for Nd is above 12 kOe, being comparable to commercial permanent magnets. The enhanced coercivity is attributed to the weakened exchange coupling between adjacent grains by forming continuous grain boundary
RI PT
layer and the strengthened local magnetocrystalline anisotropy at the outer region of the grains where Nd/Pr are diffused from the active intergranular additive after dehydrogenization. The above findings highlight the importance of grain boundary structure on the coercivity of
SC
Nd-Ce-Fe-B sintered magnets. Proper modification of grain boundary structure may further increase the magnetic properties as well as Ce substitution level, which is helpful to extend
M AN U
the utilization of the naturally abundant rare earth, cerium. Acknowledgement
This work was supported by the National Natural Science Foundation of China (Nos. 51622104, 51590881, 51571176 and 51401180), the National Key Research and Development
TE D
Program (No. 2016YFB0700902), the Key Research and Development Program of Zhejiang Province (No. 2017C01031), the Fundamental Research Funds for the Central Universities, the “973” program (No. 2014CB643702), and the Instrument Developing Project of the
AC C
EP
Chinese Academy of Sciences (No. YZ201536).
11
ACCEPTED MANUSCRIPT References [1] M. Sagawa, S. Fujimura, N. Togawa, H. Yamamoto, and Y. Matsuura, J. Appl. Phys. 55 (1984) 2083-2087. [2] J.F. Herbst, Rev. Mod. Phys. 63 (1991) 819-898.
RI PT
[3] A. Alam, D.D. Johnson, Phys. Rev. B 89 (2014) 235126. [4] S.L. Huang, H.B. Feng, M.G. Zhu, A.H. Li, Y. Zhang, and W. Li, AIP Adv. 4 (2014) 107127.
SC
[5] C.J. Yan, S. Guo, R.J. Chen, J. Liu, D. Lee, A.R. Yan, IEEE Trans. Magn. 50 (2014) 2102605.
M AN U
[6] A.K. Pathak, M. Khan, K.A. Gschneidner, R.W. McCallum, L. Zhou, K. Sun, K.W. Dennis, C. Zhou, F.E. Pinkerton, M.J. Kramer, V.K. Pecharsky, Adv. Mater. 27 (2015) 2663-2667. [7] Z.B. Li, B.G. Shen, M. Zhang, F.X. Hu, J.R. Sun, J. Alloys Compd. 628 (2015) 325-328. [8] Z. Li, W. Liu, S. Zha, Y. Li, Y. Wang, D. Zhang, M. Yue, J. Zhang, and X. Huang, J. Magn.
TE D
Magn. Mater. 393 (2015) 551-554.
[9] M.G. Zhu, R. Han, W. Li, S.L. Huang, D.W. Zheng, L.W. Song, and X.N. Shi, IEEE Trans. Magn. 51 (2015) 2104604.
EP
[10] A.K. Pathak, M. Khan, K.A. Gschneidner, R.W. McCallum, L. Zhou, K. Sun, M.J. Kramer, V.K. Pecharsky, Acta Mater. 103 (2016) 211-216.
AC C
[11] J.Y. Jin, T.Y. Ma, Y.J. Zhang, G.H. Bai, and M. Yan, Sci. Rep. 6 (2016) 32200. [12] B.X. Peng, T.Y. Ma, Y.J. Zhang, J.Y. Jin, and M. Yan, Scr. Mater. 1 (2017) 11-14. [13] Super Magnet Co., Ltd. (2016) Available at: http://www.spmagnet.cn/m_page/26-m aterial-grade.html.
[14] K. Hono, H. Sepehri-Amin, Scr. Mater. 67 (2012) 530-535. [15] T.G. Woodcock, Y. Zhang, G. Hrkac, G. Ciuta, N.M. Dempsey, T. Schrefl, O. Gutfleisch, D. Givord, Scr. Mater. 67 (2012) 536-541. [16] O. Gutfleisch, M.A. Willard, E. Brück, C.H. Chen, S.G. Sankar, and J.P. Liu, Adv. Mater. 12
ACCEPTED MANUSCRIPT 23 (2011) 821-842. [17] M. Sagawa, S. Fujimura, H. Yamamoto, Y. Matsuura, and K. Hiraga, IEEE Magn. Mag. 20 (1984) 1584-1589. [18] W.F. Li, T. Ohkubo, and K. Hono, Acta Mater. 57 (2009) 1337-1346.
RI PT
[19] T.H. Kim, S.R. Lee, M.W. Lee, T.S. Jang, J.W. Kim, Y.D. Kim, and H.J. Kim, Acta Mater. 66 (2014) 12-21.
[20] K. Löewe, C. Brombacher, M. Katter, and O. Gutfleisch, Acta Mater. 83 (2015) 248-255.
SC
[21] F.M. Wan, Y.F. Zhang, J.Z. Han, S.Q. Liu, T. Liu, L. Zhou, J.B. Fu, D. Zhou, X.D. Zhang, J.B. Yang, Y.C. Yang, J. Chen, and Z.W. Deng, J. Appl. Phys. 115 (2014) 203910.
M AN U
[22] T.Y. Ma, X.J. Wang, X.L. Liu, C. Wu, and M. Yan, J. Phys. D: Appl. Phys. 48 (2015) 215001.
[23] M. Yue, W.Q. Liu, D.T. Zhang, Z.G. Jian, A.L. Cao, and J.X. Zhang, Appl. Phys. Lett. 94 (2009) 092501.
TE D
[24] Y. Une and M. Sagawa, Paper Presented at the 21st Workshop on Rare-Earth Permanent Magnets and Their Applications, Bled, Slovenia, 2010. [25] H. Sepehri-Amin, Y. Une, T. Ohkubo, K. Hono, and M. Sagawa, Scr. Mater. 65 (2011)
EP
396 - 399.
[26] H. Sepehri-Amin, T. Ohkubo, and K. Hono, Acta Mater. 61 (2013) 1982 - 1990.
AC C
[27] Y.J. Zhang, T.Y. Ma, J.Y. Jin, J.T. Li, C. Wu, B.G. Shen and M. Yan, Acta Mater. 128 (2017) 22-30.
[28] W.B. Cui, T.B. Zhang, X.Q. Zhou, D.D. Yu, Q. Wang, X.G. Zhao, W. Liu and Z.D. Zhang, Mater. Lett. 191 (2017) 210 - 213. [29] M. Ito, M. Yano, N. Sakuma, H. Kishimoto, A. Manabe, T. Shoji, A. Kato, N. M. Dempsey, D. Givord, and G. T. Zimanyi, AIP Advances 6 (2016) 056029. [30] K. Chen, S. Guo, X.D. Fan, G.F. Ding, J.H. Di, R.J. Chen, D. Lee and A.R. Yan, AIP Advances 7 (2017) 025213. 13
ACCEPTED MANUSCRIPT [31] P. Liu, T.Y. Ma, X.H. Wang, Y.J. Zhang, and M. Yan, J. Alloys Compd. 628 (2015) 282 286. [32] S. Bajt, A. Barty, K.A. Nugent, M. McCartney, M. Wall, D. Paganin, Ultramicroscopy 83 (2000) 67-73.
RI PT
[33] K. Ishizuka, B. Allman, J. Electron Microsc. 54 (2005) 191-197. [34] R. McMichael, M. Donahue, OOMMF software package. http://math.nist.gov/oommf, in. [35] T. Schrefl, R. Fischer, J. Fidler, and H. Kronmüller, J. Appl. Phys. 76 (1994) 7053 - 7058.
SC
[36] D. Liu, T.Y. Zhao, R. Li, M. Zhang, R.X. Shang, J.F. Xiong, J. Zhang, J.R. Sun, and B.G. Shen, AIP Adv. 7 (2017) 056201.
Acta Mater. 99 (2015) 297-306.
M AN U
[37] H. Sepehri-Amin, L. Liu, T. Ohkubo, M. Yano, T. Shoji, A. Kato, T. Schrefl, and K. Hono,
[38] Shanghai Metals Market, Pricing of Rare Earth Metals. (2016) Available at: http://www.metal.com/pricing/rare-earch-metals.
48 (2003) 851 - 856.
TE D
[39] D. Shindo, Y.-G. Park, Y. Murakami, Y. Gao, H. Kanekiyo, and S. Hirosawa, Scr. Mater.
[40] H. Sepehri-Amin, T. Ohkubo, T. Shima, and K. Hono, Acta Mater. 60 (2012) 819 - 830.
128-133.
EP
[41] V. Yartys, O. Gutfleisch, V. Panasyuk, I. Harris, J. Alloys Compd. 253-254 (1997)
AC C
[42] P. Kronmüller, Phys. Status Solidi B 144 (1987) 385 - 396. [43] C. Jin, R. Chen, W. Yin, X. Tang, Z. Wang, J. Ju, D. Lee, and A.R. Yan, J. Alloys Compd. 670 (2016) 72 - 77.
[44] X.L. Liu, X.J. Wang, L.P. Liang, P. Zhang, J.Y. Jin, Y.J. Zhang, T.Y. Ma, and M. Yan, J. Magn. Magn. Mater. 370 (2014) 76 - 80. [45] L.P. Liang, T.Y. Ma, C. Wu, P. Zhang, X.L. Liu and M. Yan, J. Magn. Magn. Mater. 397 (2016) 139 - 144.
14
ACCEPTED MANUSCRIPT Figure Captions Fig. 1 Demagnetization curves for the starting magnet and the magnets added with different amounts of (Nd80Pr20)Hx powders. The inset shows dependences of coercivity and remanence
RI PT
on the (Nd80Pr20)Hx content. Fig. 2 Back-scattered SEM micrographs of the starting magnet (a) and the magnets added with different amounts of Nd-Pr-H: 1 wt.% (b), 2 wt.% (c) and 3 wt.% (d). EPMA mappings
SC
of elemental concentrations of Nd (e) and Pr (f) for the magnet added with 2 wt.% Nd-Pr-H. The black dashed lines illustrate the GBs and are guide for eyes.
M AN U
Fig.3 TEM characterization of typical region for the magnet added with 3 wt.% Nd-Pr-H. (a) Low magnification bright field image, (b) HRTEM image. (c, d) EDPs for the dashed red circle region in (a) with different zone axes.
TE D
Fig. 4. (a) L-TEM Fresnel images of the starting magnet with the c-axis in plane and taken in zero field. (b) Reconstructed holography phase images of the blue dashed region in (a). The red dashed line indicates the grain boundary. Red arrows indicate the magnetization vectors of
EP
domains.
Fig. 5 Over-focused Fresnel Lorentz images and calculated magnetization maps for the
AC C
starting magnet (a, b) and the magnet added with 3 wt.% Nd-Pr-H (c, d). c-axis of the matrix phase grains is out of plane for both samples. The numbers indicate grains with different orientations. The color wheels in the insets of (b) and (d) are used to identify the magnetic induction direction, i.e., the color and brightness indicate the direction and strength of the in-plane magnetization, respectively. Fig. 6
Cross-sections of the block containing (Nd,Ce)2Fe14B grains (a) and the one
containing grains with (Nd,Ce)2Fe14B core covered by Nd2Fe14B shell (b). All the grains are 15
ACCEPTED MANUSCRIPT separated by 2nm-thick grain boundary (GB) layer. (c) Simulated demagnetization curves by
AC C
EP
TE D
M AN U
SC
RI PT
assuming that the GB is ferromagnetic or nonmagnetic, respectively.
16
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Highlights
Coercivity of Nd-Ce-Fe-B sintered magnets can be effectively enhanced by grain boundary restructuring with (Nd, Pr)-H.
local magnetocrystalline anisotropy.
Formation of smooth and continuous grain boundaries promotes the magnetic isolation between adjacent grains.
M AN U
The obtained Nd-Ce-Fe-B magnets have much lower cost than the commercial
EP
TE D
ones with comparable magnetic properties.
AC C
RI PT
Formation of (Nd,Pr)-rich shell surrounding Ce-rich grain core strengthens the
SC