The working environment of permanent magnets in high-speed motors is harsh, because they are seriously affected by multi-physical fields such as temperature, stress, weak magnetic control field, etc., and irreversible demagnetization is easy to occur under the extreme working conditions. In this work, a model of high-speed Interior Permanent magnet Motor is established, the easy demagnetization area of the permanent magnet is simulated by electromagnetic and thermal simulations to determine, and its local anti-demagnetization optimization is performed. Dy/Tb co-infiltration is carried out in the easy demagnetization region of the permanent magnets, and the process of selective zone diffusion is analyzed based on the principle of selective zone grain boundary diffusion. Through microscopic observation and macroscopic magnetic performance test, the effect of local anti-demagnetization performance enhancement of the optimized permanent magnets is verified. The results show that the magnetic properties of the optimized permanent magnets are obviously improved after the optimization of the selected zones. The present approach is important for the follow-up less heavy-rare-earth research of permanent magnet and its application in high-speed motors.

NdFeB permanent magnet (PM) is widely used in traction motors due to their excellent room temperature magnetic properties.1,2 The PM needs to be continuously operated under high temperature, weak magnetic control field and stress environment, which leads to the increasingly prominent demagnetization problem of NdFeB and affects the stable operation of the motor.

Grain boundary diffusion (GBD) is a method proposed in recent years to improve the coercivity of PM with less HRE, which has been proposed to increase the coercivity (Hcj) while maintaining the remanence magnetic properties.3–5 However, the coating effect of the whole surface is verified to be difficult to produce the expected effect on the NdFeB magnet. Selected area grain boundary diffusion (SAGBD) is a GBD approach that mainly diffuses the source from the edge and corner areas of the PM, which can achieve efficient utilization of diffusion sources.6 However, in practical motor applications, it is difficult to improve the regional anti-demagnetization ability of PMs only for edge diffusion. At present, there are few studies on the anti-demagnetization capability of PM carried out in reverse from the perspective of motor application.

In this work, the demagnetization characteristics of PM during high-speed operation of electric motor is investigated, the SAGBD is carried out in the easily demagnetized area. This method can accurately improve the anti-demagnetization performance of the easy demagnetization region of the PM while saving the amount of HRE. This work provides a new idea for the design of high-speed and high-performance motor and provides a reference for more accurate GBD processes.

The main parameters of the established Interior Permanent magnet Motor (IPM) model are shown in Table I. The operating temperature of the PM is set to 20 °C (corresponding to short-term high-current operation, demagnetization current 400 A) and 120 °C (corresponding to long-term small-current operation, demagnetization current 60 A). The selection of PMs should be considered comprehensively. It is necessary to have certain magnetic properties and local optimization potential. Finally, commercial sintered NdFeB, brand N38H is selected. The selected grade of core material is 20SW1200H. Its magnetization curve (B-H curve) at power frequency (50 Hz) and iron loss curve (B-P curve) at different frequencies are shown in Fig. 1.

TABLE I.

Main parameters of IPM.

ParameterValueParameterValue
Peak power 120 kW Air gap 0.7 mm 
Peak speed 10 000 rpm Stator outer diameter 220 mm 
DC voltage 540 VDC Rotor outer diameter 144.6 mm 
Slots 72 PM size 21*20*4.5 mm 
Length 100 mm Core material 20SW1200H 
ParameterValueParameterValue
Peak power 120 kW Air gap 0.7 mm 
Peak speed 10 000 rpm Stator outer diameter 220 mm 
DC voltage 540 VDC Rotor outer diameter 144.6 mm 
Slots 72 PM size 21*20*4.5 mm 
Length 100 mm Core material 20SW1200H 
FIG. 1.

Magnetization curve (a) and iron loss curve (b) for 20SW1200H.

FIG. 1.

Magnetization curve (a) and iron loss curve (b) for 20SW1200H.

Close modal

Figure 1 shows that the magnetization curve of the electrical steel material is saturated, and the frequency change has a greater influence on the loss.7 This property is also taken into account in this paper. Figure 2(a) shows the structure of the motor and the magnetic density distribution at peak operating conditions, which indicates the saturation degree of the motor core material. It can be seen that the rotor of the motor has the highest magnetic saturation degree at the magnetic separation bridge position, which reduces the magnetic leakage phenomenon to a certain extent, but the saturation of the magnetic circuit will cause the temperature to rise, and also affect the temperature of the nearby PM to a certain extent. In order to obtain the accurate demagnetization region of PM, PM is set from the ordinary linear ideal demagnetization curve to the nonlinear actual demagnetization curve, which is realized by input N38H discrete demagnetization point. The measured commercial N38H demagnetization curve is shown in Fig. 2(b).

FIG. 2.

Motor structure (a) magnetic density distribution and (b) demagnetization curve (N38H).

FIG. 2.

Motor structure (a) magnetic density distribution and (b) demagnetization curve (N38H).

Close modal

Figure 3 shows the thermal distribution of PM in the motor obtained using the demagnetization characteristics of N38H at 20 and 120 °C. In both cases, the operating temperature of the PMs is near the critical demagnetization temperature, and the temperature of the permanent magnet near the outside of the rotor is the highest, because the influence of the saturation of the magnetic bridge is more significant. However, the temperature rise near the rib is relatively small due to the existence of air gap. However, the overall temperature of the PMs are also near the critical demagnetization temperature, leading to the demagnetization risk of the PM.

FIG. 3.

Heat distribution of PM (a) 20 °C and (b) 120 °C.

FIG. 3.

Heat distribution of PM (a) 20 °C and (b) 120 °C.

Close modal

Considering the high-speed operation condition of the motor, the demagnetization phenomenon of PM is influenced by both temperature and demagnetization current,8 and the demagnetization area is constantly changing at different times. In this paper, the most serious demagnetization moment (the rotor is 158° away from the initial Angle) is selected as the optimization object for research. The demagnetization simulation cloud diagram is shown in Fig. 4.

FIG. 4.

Demagnetization cloud map of PM (a) 20 °C and (b) 120 °C.

FIG. 4.

Demagnetization cloud map of PM (a) 20 °C and (b) 120 °C.

Close modal

The lower the demagnetization coefficient Demag-Coef is the more serious the demagnetization is. It can be seen from the simulation results that the easy demagnetization area of the PM is mainly distributed at the corner of the PM. The infiltration process requirements are put forward in view of the above results: based on the two corners of the same face of the N38H PM, the SAGBD is carried out. Considering that the length of the PM of the high-speed motor is 20 mm after axial segmentation, the size of the diffusion area is preliminarily determined to be 20*2.92*3.11 mm3.

SAGBD refers to an effective way for the diffusion source to diffuse efficiently from the edge or corner area of the PM. The schematic diagram of the traditional GBD and SAGBD principles are shown in Fig. 5.6 

FIG. 5.

Schematic diagram principles of (a) GBD and (b) SAGBD.

FIG. 5.

Schematic diagram principles of (a) GBD and (b) SAGBD.

Close modal

Figure 5(a) shows the schematic diagram of the principle of GBD. Heavy rare earth elements are coated (or sputtered) on the whole surface of the PM, and the coercivity is improved by vacuum heat treatment. Figure 5(b) is a schematic diagram of SAGBD. The diffusion source of the easy demagnetization area (the selected area, i.e.) is HRE coating. The purpose is to reduce the amount of HRE and improve the anti-demagnetization ability of the easy demagnetization area.

Both alloys composed of Dy and Tb elements can improve the mutual infiltration among the Nd2Fe14B main phases to some extent by GBD. The introduction of Cu and Al can modify the main phase for a higher coercivity (Hcj).9–11 

In order to enhance the comparability of the properties of the diffused area and the undiffused area, as well as to explore the characteristics of the diffusion of Dy and Tb elements in the same proportion, this work chooses to coat the (Dy, Tb)80Al10Cu10 alloy in the easy demagnetization region for diffusion. The ratio of Dy:Tb elements is 1:1. Since the HRE diffusion along the easy axis (c-axis) direction is more effective than the diffusion along the direction perpendicular to the c-axis,12 and the coating performance of the entire plane is lower than that of the corner coating.6 Combined with the simulation in Sec. II B, the easy demagnetization area is obtained, and the selected area diffusion coating position is on the easy demagnetization surface of the PM perpendicular to the c-axis, as shown in Fig. 6.

FIG. 6.

Coating diffusion area (a) 3D coating model of the PM (b) single-face coating (c) double-faces coating.

FIG. 6.

Coating diffusion area (a) 3D coating model of the PM (b) single-face coating (c) double-faces coating.

Close modal

The PM coating diffusion is divided into two schemes. Scheme 1 is single face coating, which is only the surface of the vertical c-axis plane in the blue area with severe demagnetization in Fig. 4 is coated, and the coating size h*w = 20*2.92 mm2, thickness t = 0.2 mm, as shown in Fig. 6; Scheme 2 is double faces coating, which is two surfaces perpendicular to the c-axis are coated at the same area as the scheme 1, as shown in Fig. 6(c). The thermal diffusion process is carried out in a vacuum environment, vacuum sintering furnace (VR-600), diffusion for 20 h, diffusion temperature 900 °C.

Based on the electric spark line cutting process, the diffused PM is cut along the middle position perpendicular to the c-axis direction. After coarse grinding and fine grinding, the scanning electron microscope (FEI Nova NanoSEM450) and energy spectrometer (X-Max50) were used for observation. Figure 7 shows the distribution and variation trend of Dy and Tb elements at different depths from the diffusion surface. Figure 7(b) is the position identification of line scan and area scan. The line scan starts from the position of the diffusion surface, 0.5 mm away from the edge, along the c-axis direction to 2500 μm. The regional scanning positions are 100 and 2200 μm (about the middle position of the PM), respectively. The results of line scan and area scan in Figs. 7(a) and 7(c) show that the concentration of Dy and Tb in the middle of the PM by double faces diffusion is greater than that by single face diffusion. In addition, with the increase of the diffusion depth, the trend of the gradual decrease of Dy and Tb content in the two diffusion schemes is more obvious. In the region of 2000–2500 μm, the content of Dy is slightly more than that of Tb, which may be caused by the fact that the atomic radius of Dy is smaller than that of Tb, and the Dy atoms are more likely to move within the crystalline interface during diffusion. The results confirms that the diffusion concentration at the edge is greater than the middle of the PM.

FIG. 7.

Dy/Tb variation trend with the distance from diffusion surface (a) single face diffusion of line scan (a1) and (a2) and area scan (a3) and (a6) (b) The position identification of line scan and area scan (c) double faces diffusion of line scan (c1) and (c2) and area scan (c3)–(c6).

FIG. 7.

Dy/Tb variation trend with the distance from diffusion surface (a) single face diffusion of line scan (a1) and (a2) and area scan (a3) and (a6) (b) The position identification of line scan and area scan (c) double faces diffusion of line scan (c1) and (c2) and area scan (c3)–(c6).

Close modal

The temperature performance of the PM was tested in the range of 80–160 °C. The high temperature demagnetization performance was characterized by testing the change rate of magnetic flux in the closed magnetic circuit [ΔΦ/Φ*100% = (Φ − Φ′)/Φ*100%, where Φ is the actual measured value of the magnetic flux of the closed magnetic circuit provided by the PM at room temperature 20 °C, and Φ′ is the measured value of the magnetic flux of the closed magnetic circuit at the variable temperature to be measured]. The results are shown in Fig. 8.

FIG. 8.

Temperature-flux change rate curve.

FIG. 8.

Temperature-flux change rate curve.

Close modal

As can be seen from Fig. 8, the rate of change of magnetic flux of the three kinds of PM increases with the increase of temperature, indicating that the magnetic flux of the closed magnetic circuit provided by the PM decreases. The rate of change of magnetic flux is basically stable below 90 °C, and the decreasing trend of traditional commercial PM N38H is accelerated from about 95 °C.

The double faces diffusion PM is relatively stable, and the rate of change of the commercial PM reaches 9.91% at 120 °C, the single PM is 5.47%, and the double faces diffusion PM is 3.56%. The difference is even more obvious as the temperature increase to 160 °C. The test results show that double faces coated diffusion PM has higher stability of comprehensive temperature magnetic properties.

The magnetic properties of the diffusion part of the PM is tested. Firstly, the PM samples is cut in the same way as in Sec. III A, and then the magnetic properties of the coated parts were tested to compare the changes in magnetic property parameters before and after diffusion (single face and double faces coating). The results are shown in Figs. 9 and 10 and Table II.

FIG. 9.

Demagnetization curve.

FIG. 9.

Demagnetization curve.

Close modal
FIG. 10.

Magnetic hysteresis loop.

FIG. 10.

Magnetic hysteresis loop.

Close modal
TABLE II.

Main parameters of IPM.

ParameterBefore diffusionSingle faceDouble faces
Hcj (kA/m) 1417 1949 2061 
Br (T) 1.274 1.253 1.247 
(BH)max (kJ/m3310.5 302.0 297.9 
Hcb (kA/m) 977.2 957.8 950.4 
ParameterBefore diffusionSingle faceDouble faces
Hcj (kA/m) 1417 1949 2061 
Br (T) 1.274 1.253 1.247 
(BH)max (kJ/m3310.5 302.0 297.9 
Hcb (kA/m) 977.2 957.8 950.4 

In Figs. 9 and 10, the red line and the green line represent the magnetic characteristic curves of double faces coating and single face coating, respectively. The blue line and the black line represent the magnetic characteristic curves of commercial PM N38H at 20 and 120 °C, respectively. At room temperature(20 °C), the Hcj = 1949 kA/m (37.54% higher than that before diffusion), the remanence Br = 1.253 T (1.65% lower than that before diffusion), and the maximum magnetic energy product (BH)max = 302.0 kJ/m coercivity (2.74% lower than that before diffusion). After double faces coating, Hcj = 2061 kA/m (5.75% higher than that of single face coating), Br = 1.247 T (0.48% lower than that of single face coating), and maximum magnetic energy product (BH)max = 297.9 kJ/m3 (1.36% lower than that of single face coating). It can be seen that single face and double faces coating are within the acceptable range of remanence and magnetic energy product reduction, and the coercivity is obviously improved. Customized diffusion of PM can be carried out for motors with different application requirements.

The demagnetization magnetic field at three different positions of the high-speed IPM model at 20 °C is simulated and compared with the PM demagnetization curve before and after optimization. Figure 11 is the magnetic field intensity distribution map of the coating surface along the x-axis direction of the PM (see Fig. 6 coordinate axis), 1.125 mm from the coating surface and 2.25 mm from the coating surface. Comparing the H value of the marked point with serious demagnetization with the PM Hcj before and after diffusion in Table II, it can be seen that the PM Hcj before diffusion is smaller than the demagnetization magnetic field, resulting in demagnetization in this region. The Hcj after single face or double faces diffusion optimization is significantly larger than the demagnetization magnetic field. Combined with the temperature performance test results in Sec. III B, it can be seen that the local anti-demagnetization ability of the PM is effectively improved.

FIG. 11.

Demagnetization magnetic field distribution of PM.

FIG. 11.

Demagnetization magnetic field distribution of PM.

Close modal

This work aimed at the problem of local demagnetization of PM in high-speed motors, a simulation model is established to determine the demagnetization area, and the local performance of PM is improved based on the SAGBD.

  1. Establish a high-speed motor model, determine the easy demagnetization area of the PM through simulation, and reversely determine the accurate area of material performance enhancement;

  2. Based on the SAGBD, the HRE diffusion in the easy demagnetization area was carried out by single face and double faces schemes with Dy/Tb element ratio of 1:1.

  3. The effect of SAGBD on optimizing the magnetic properties of PM was verified from two aspects of material microscopic observation and macroscopic performance test. Compared with the sample before diffusion, the temperature-flux change rate curve of the double faces diffusion PM grows slowly, and the temperature stability is the best. Hcj of single face and double faces diffusion increased by 27.30% and 31.25%, respectively. Br decreased by 1.68% and 2.17%, respectively. (BH)max decreased by 2.81% and 4.23%, respectively.

  4. From the industrial point of view, while increasing coercivity and reducing the amount of rare earth elements, the regional coating method selected in this paper will increase the process flow to a certain extent. Considering the case of mass production of motors, low grade instead of high grade PM can save more costs.13 In order to further reduce costs and promote industrial production, it can be considered to improve the coating process to a higher degree of automation.

In the case of a slight decrease in the remanence and maximum magnetic energy product performance, the SAGBD applied to the easy demagnetization area of the high-speed motor PM can effectively improve its high temperature performance and coercivity, and the anti-demagnetization ability of the PM is obviously improved.

This research was supported by program of Scholars of the Xingliao Plan (No. XLYC2002113), Shenyang University of Technology Interdisciplinary Team Project (No. 100600453), Central guide to local science and technology development funds (free exploration class basic research) (No. 2023JH6/100100043) and the Scientific and Technological Key Project in Henan Province (No.232102220097).

The authors have no conflicts to disclose.

Weizhou Li: Data curation (equal); Investigation (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). Jungang Cheng: Data curation (equal); Investigation (equal); Methodology (equal); Visualization (equal). Xusheng Lu: Investigation (equal); Software (equal); Writing – original draft (equal). Yangyang Li: Formal analysis (equal); Software (equal). Dong Zhao: Data curation (equal); Resources (equal); Supervision (equal). Wenli Pei: Formal analysis (equal); Methodology (equal); Supervision (equal). Ruilin Pei: Formal analysis (equal); Investigation (equal); Methodology (equal); Supervision (equal); Writing – original draft (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

1.
A.
Trench
and
J. P.
Sykes
,
Engineering
6
(
2
),
115
118
(
2020
).
2.
J.
Hu
,
Y.
Wu
,
J.
He
, and
Z.
Liu
, “
Computer numerical simulations for accelerated design of rare earth permanent magnet motors in renewable energy applications: A review
,”
Eng. Anal. Boundary Elem.
156
,
144
159
(
2023
).
3.
J.
Tao
,
B.
Liu
,
P.
Zhang
,
G.
Xu
,
J.
Lv
,
J.
Huang
,
J.
Yan
,
W.
Sun
,
B.
Li
,
D.
Wang
, and
Y.
Wu
,
J. Rare Earths
41
(
8
),
1203
1210
(
2023
).
4.
J.
He
,
J.
Hu
,
B.
Zhou
,
H.
Jia
,
X.
Liu
,
Z.
Zhang
,
L.
Wen
,
L.
Zhao
,
H.
Yu
,
X.
Zhong
,
X.
Zhang
, and
Z.
Liu
,
J. Mater. Sci. Technol.
154
,
54
64
(
2023
).
5.
T.
Zhou
,
Y.
Guo
,
G.
Xie
,
S. U.
Rehman
,
R.
Liu
,
J.
Liu
,
P.
Qu
, and
M.
Li
,
Intermetallics
138
,
107335
(
2021
).
6.
J.
He
,
W.
Song
,
X.
Liu
,
H.
Yu
,
X.
Zhong
, and
Z.
Liu
,
Appl. Phys. Lett.
120
,
042405
(
2022
).
7.
R.
Pei
,
L.
Zeng
, and
L.
Gao
, in
21st. International Conference on Electrical Machines and Systems (ICEMS)
(
Jeju
,
2018
), pp.
2715
2718
8.
W.
Wang
,
P.
Zheng
,
M.
Wang
,
Y.
Liu
,
Z.
Fu
, and
Y.
Sui
,
IEEE Trans. Magn.
54
,
1
5
(
2018
).
9.
Y. J.
Wong
,
H. W.
Chang
,
Y. I.
Lee
,
W. C.
Chang
,
C. H.
Chiu
, and
C. C.
Mo
,
AIP Adv.
13
,
025226
(
2023
).
10.
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
,
12
21
(
2014
).
11.
H.
Zeng
,
Z.
Liu
,
W.
Li
,
J.
Zhang
,
L.
Zhao
,
X.
Zhong
,
H.
Yu
, and
B.
Guo
,
J. Magn. Magn. Mater.
471
,
97
104
(
2019
).
12.
T. H.
Kim
,
S. R.
Lee
,
S. J.
Yun
,
S. H.
Lim
,
H. J.
Kim
,
M. W.
Lee
, and
T. S.
Jang
,
Acta Mater.
112
,
59
66
(
2016
).
13.
Z.
Wang
,
T. T.
Sasaki
,
Y.
Une
,
T.
Ohkubo
, and
K.
Hono
,
Acta Mater.
248
,
118774
(
2023
).