A Study on Powder Metallurgy Process for x Electric Vehicle Stator Core

Abstract

The powder metallurgy process of manufacturing the motor core and inductor core using SMC greatly changes formability depending on the process variables. Therefore, this study explored the optimal process conditions of the powder metallurgy of the SMC stator core using Fe-6.5 wt.%Si by applying the Taguchi method, and selected deviations between the maximum and minimum relative densities as characteristic values; selected the formation pressure, molding temperature, and heating time as control factors; and derived the process conditions with the maximum SNR. As a result, the molding pressure was 120 MPa, the molding temperature was 500 °C, and the heating time was 120 s, and the material properties of the electrical properties’ core loss, saturation flux density, and bulk conductivity were measured and analyzed. After that, a prototype was produced, the analysis was verified, the mechanical properties were verified by performing density and SEM analysis at 15, 9, and 3 mm points based on the press vertical direction, and a motor was manufactured to verify the electrical properties.

1. Introduction

Recently, research on the production of motor cores and inductor cores using soft magnetic composites (SMCs) has been actively conducted [1]. Unlike laminated steel sheets, which have excellent magnetic properties on the laminated surface, SMCs can exhibit isotropic magnetic properties, be formed into complex shapes, and enable efficient operation at high frequencies by mitigating eddy current losses, as the individual particles are insulated from each other. Additionally, the production costs and time are relatively low, as SMCs can be manufactured via near-net-shape processes [2].

The application range of SMCs is extensive and growing, encompassing various domains such as the automotive, aerospace, and renewable energy sectors. SMCs are particularly advantageous in the design and manufacture of electric motors, transformers, and inductors due to their ability to be shaped into complex geometries and their superior performance at high frequencies. For instance, in the automotive industry, SMCs are increasingly being used in the development of more efficient and compact electric motors, contributing to the advancement of electric and hybrid vehicles. Similarly, in the aerospace industry, SMCs are being utilized to produce lightweight and efficient components that can withstand harsh operating conditions. Their use in renewable energy applications, such as wind turbine generators, is also notable for enhancing energy conversion efficiency and reliability [3,4].

Axial Flux Permanent Magnet (AFPM) machines, known for their compact and efficient design, often employ SMCs in their construction. AFPM machines benefit from the isotropic magnetic properties and high-frequency efficiency of SMCs, which allow for more compact and lightweight designs compared to traditional radial flux machines. The use of SMCs in AFPM stator cores contributes to reduced eddy current losses and improved thermal management, thereby enhancing overall machine performance. These properties make SMCs highly suitable for advanced electric motor applications, including those in electric vehicles and renewable energy systems in which space and efficiency are critical considerations [5].

SMCs are produced using materials such as pure iron, Fe–Si, Fe–Ni, Fe–Co, and ferrite. Fe–Si alloys exhibit high hardness and electrical resistivity; among them, Fe-6.5 wt.%Si is known to have the best electrical properties. Fe–Ni alloys have the highest permeability and lowest coercivity. Fe–Co alloys have a high magnetization saturation and are often used in high-power applications. Ferrite has a lower electrical conductivity than other alloys, but is chemically stable and corrosion-resistant, thus being suitable for use in harsh environments [6,7,8]. In addition, various methods are used to prepare composite powders, such as water atomization, gas atomization, mechanical alloying, and electrolysis. The formability of SMCs is highly dependent on the properties of the powder produced by the powder metallurgy process [9,10,11,12]. The formability of SMCs is a crucial factor affecting their electrical performance, which is strongly related to the air gaps and density of the material. In low-density SMCs, the internal pores hinder magnetization, thereby reducing permeability and increasing hysteresis loss and iron loss [13,14,15,16,17,18]. To address these problems, studies have focused on bimodal powders.

However, adequate research is yet to be conducted on the powder metallurgy and process parameters for SMCs manufactured with Fe-6.5 wt.%Si powder to be used as stator cores of motors. Therefore, this study aimed to determine the powder metallurgy process conditions that can minimize the density deviation of such SMCs, for which the Fe-6.5 wt.%Si powder is prepared via 50% hybrid (water + gas) atomizing and 50% gas atomizing (gas atomizing powder; 3 h milling time) [19]. The flows of this study were as follows: setting control factors and levels, analysis of powder metallurgy and the signal-to-noise ratio (SNR), and experimental verification.

2. Theoretical Background

2.1. Powder Metallurgy and Key Factors

Powder metallurgy is a method of manufacturing products using metal and ceramic powders. Compared with traditional processing methods, such as casting and forging, the press-sintering method in powder metallurgy offers better material utilization, precision, porosity control, sintering reinforcement, and energy efficiency, as well as the ability to manufacture complex shapes using unique material combinations. These advantages allow SMCs to replace laminated electrical steel sheets as materials for manufacturing the stator cores of motors.

In general, the powder metallurgy process involves the following sequence of steps: (1) powder preparation; (2) mixing; (3) compression; (4) sintering; and (5) post-processing [20,21,22].

Within these steps, molding pressure, molding temperature, heating time, molding speed, sintering temperature, and milling time act as the process variables. The formability of a product also depends on the process conditions in powder metallurgy, which typically include molding density, effective stress, and average stress in compaction. Therefore, minimizing variations in density and the possibility of destruction constitute the main criteria for selecting the optimal process conditions.

2.2. Yield Conditions for Porous Materials

In general, the yield conditions for porous materials follow the Shima/Oyane equation, which is expressed in Equation (1) [23].

$$\mathrm{A}{J}_2^{\prime }+\mathrm{B}{J}_1^2=Y_R^2={\delta Y}_0^2$$

(1)

Here, $J_1$ and $J_2^{\prime }$ are the first-order stress invariant and the second-order deviation stress invariant, respectively. $Y_R$ and $Y_0$ are the yield stresses of porous and non-porous materials, respectively. δ = ($Y_R^2/Y_0^2$) is a function of the relative density R, called “geometric hardening” or “densification hardening”. A and B are also functions of relative density. In Equation (1), when the relative density converges to 1, the material becomes non-porous, and A, B, and δ converge to 3, 0, and 1, respectively, following the von Mises yield function.

2.3. Taguchi Method

Determining the optimal process conditions for powder metallurgy necessitates considerable experimentation and verification. In addition, because of the wide variety of process conditions, an experimental design method that can efficiently evaluate the influence of various factors on the results is required to find the optimal conditions. In this study, obtaining a highly dense powder during the compaction step was selected as the goal. To achieve this goal, the Taguchi method was used, which assumes that the maximum information can be obtained through the minimum number of experiments and that the factors involved do not interact with each other [24,25,26].

The Taguchi method involves using an orthogonal array table and finding the optimal conditions for the control factors through analysis of the loss function and SNR. The loss function represents the cost associated with deviating from the target value. The further the characteristic value from the target value, the greater the loss; when the characteristic value matches the target value, the loss is minimal. The SNR is used to identify control factors that minimize product variation by minimizing the effects of noise. A higher SNR indicates that the selected control factor minimizes the noise factor effect.

The loss function and SNR are categorized into larger-the-better characteristics, nominal-is-best characteristics, and smaller-the-better characteristics, depending on the purpose of the characteristic. For larger-the-better characteristics, larger characteristic values are desirable. For nominal-is-best characteristics, the characteristic value should be as close to the result value as possible. For smaller-the-better characteristics, smaller characteristic values are desirable. The equations for the loss function and SNR corresponding to the aforementioned types of characteristics are shown in Equations (2)–(7).

$${L\left(y\right)}_{larger}=k\left({\displaystyle \frac{1}{y^2}}\right)$$

(2)

$${L\left(y\right)}_{nominal}=k{\left(y-m\right)}^2$$

(3)

$${L\left(y\right)}_{smaller}={ky}^2$$

(4)

Here, L(y) is the loss function, y is the quality characteristic value of the product, m is the target value of y, and k is the quality loss coefficient.

$${S/N}_{larger}=-10\log \left[{\displaystyle \frac{1}{n}}\sum _{i=1}^n{\displaystyle \frac{1}{y_i^2}}\right]$$

(5)

$${S/N}_{nominal}=-10\log \left[{\displaystyle \frac{1}{n}}\sum _{i=1}^n{\left(y_i-m\right)}^2\right]$$

(6)

$${S/N}_{smaller}=-10\log \left[{\displaystyle \frac{1}{n}}\sum _{i=1}^n{y}_i^2\right]$$

(7)

Here, y_i represents the i-th result value, n represents the number of analyses, and m represents the target value of the target characteristic.

3. Material Properties

In this study, 0.7 wt.% of lubricant (MoS₂) and 1 wt.% of binder (H₃PO₄) were added to Fe-6.5 wt.%Si powder produced through 50% hybrid atomization and 50% gas atomization (3 h milling time) [19]. Additionally, SEM images of gas-atomized powder and water-atomized powder are shown in Figure 1.

Figure 2 shows the S-S curve of the Fe-6.5 wt.%Si, which was obtained through a high-temperature compression test. Compression tests were conducted at 450 °C, 600 °C, and 750 °C at a strain rate of 0.1 mm/s. Since the rigid-plastic finite-element method was employed for the powder metallurgy analysis, the elastic region was excluded, and only the plastic region was considered.

Table 1. Measured thermal expansion coefficient as a function of temperature.

Temperature [°C]	Coefficient of Thermal Expansion
100	1.13
200	1.19
300	1.26
400	1.31
500	1.35
600	1.42
700	1.47

and

Table 2. Thermal conductivity test results as a function of temperature.

Temperature [°C]	Thermal Diffusivity [mm2/s]	Specific Heat [J/gK]	Thermal Conductivity [W/mK]
25	4.20	0.45	14.2
100	4.34	0.54	17.6
200	4.50	0.57	19.0
300	4.59	0.58	20.0
400	4.60	0.60	20.6
500	4.51	0.64	21.6
600	4.21	0.91	28.6
700	3.76	1.07	30.2

list the thermal expansion coefficient, thermal diffusion coefficient, and thermal conductivity for temperatures up to 700 °C.

Additionally, the core loss, saturation magnetic flux density, and bulk conductivity are shown in

Table 3. Core loss of Fe-6.5 wt.%Si.

Jmax [T]	Frequency [Hz]
	250	500	1000
0.01	0.019 W/kg	0.038 W/kg	0.076 W/kg
0.05	0.095 W/kg	0.189 W/kg	0.378 W/kg
0.1	0.189 W/kg	0.378 W/kg	0.756 W/kg

,

Table 4. Saturation flux density of Fe-6.5 wt.%Si.

Definition	Ms [emu/g]	Saturation Flux Density [T]
Sample 1	188.28	1.72
Sample 2	190.84	1.74
Average	189.56	1.73

and

Table 5. Bulk conductivity of Fe-6.5 wt.%Si.

Definition	Bulk Conductivity [S/m]
Sample 1	0.44
Sample 2	0.55
Average	0.495

, respectively. The core loss was measured using an AC BH analyzer and LCR meter, the saturation magnetic flux density was measured using a Vibrating Sample Magnetometer, and the bulk conductivity was measured using a resistance meter. The specimen used for the measurements was toroidal in shape and had a density of 7.26 g/cm³.

4. Simulation

4.1. Shape of Stator Core

Figure 3a shows the shape of the stator core to be manufactured with the Fe-6.5 wt.%Si powder. Because a very high-capacity press would be needed to produce the entire stator core, the model was divided into six portions. Moreover, unlike in existing models for the sheet-stacking process, the stress was distributed by applying a fillet value of 0.5 mm to each corner [27].

4.2. PM Analysis Model

Figure 3b shows the initial molded body, assumed to be in powder form, which had a relative density of 0.5. Further, in terms of initial dimensions, it had an outer diameter of 36 mm, an inner diameter of 18.3 mm, and a height of 29.12 mm. Figure 3c shows the meshed body; the number of mesh elements was 310,196. The powder metallurgy analysis was performed using DEFORM-3D (SFTC, ver. 13.1).

4.3. DOE Parameters and Level Settings

To homogenize the distribution of voids within the molded body, the deviation between the maximum and minimum relative densities was set as a characteristic value; since this deviation must be minimal, it was designated as a smaller-the-better characteristic. Relative density is based on the volume of the material alone, i.e., excluding the volume of the voids from the entire volume of the body, and is related to molding pressure, molding temperature, and heating time. In terms of molding pressure, 150 MPa, which was the capacity limit of the holding press, was set as the maximum, and 120, 135, and 150 MPa were selected as the levels. In order to compress at high temperatures due to the brittleness of Si, the molding temperatures were set at 400, 500, and 600 °C, and the temperature rise times were set as 60, 90, and 120 s. The control factors and their levels are summarized in

Table 6. Design of control factors and their levels.

Factor	Description	Level
		1	2	3
A	Molding pressure [MPa]	120	135	150
B	Molding Temperature [°C]	400	500	600
C	Heating time [s]	60	90	120

, and the experimental plan based on the factor settings is summarized in

Table 7. L₉(3³) orthogonal array.

Simulation No.	A	B	C
1	1	1	1
2	1	2	2
3	1	3	3
4	2	1	2
5	2	2	3
6	2	3	1
7	3	1	3
8	3	2	1
9	3	3	2

. With regard to the process conditions, excluding the control factors of the experimental design method, while the die and lower punch were fixed, the upper punch was moved 10.92 mm in the −z direction to reach the target height of the final molded body, i.e., 18.20 mm. Additionally, in all cases, the initial temperature of the powder was 20 °C.

4.4. Motor Performance Analysis

The motor used in this study has 4 poles and 12 slots, and its model is as shown in Figure 4. Additionally, the specifications of the motor are shown in

Table 8. Motor model specification.

Definition	Unit	Value
Pole/Slot	-	4/12
Outer diameter/ Inner (Stator)	mm	70/36.6
Outer diameter/ Inner (Rotor)	mm	32/10
Length	mm	49.45

. The analysis was performed at 10,000 rpm using the commercial program Ansys EM (Ansys).

5. Simulation Results

5.1. DOE Results

Based on the powder metallurgy analysis for each combination of levels, the optimal levels of density, effective stress, and mean stress were confirmed.

Table 9. Summary of control factor results across all simulation cases.

Case	Average Relative Density (Absolute Density)	Max.–Min. Relative Density (Absolute Density)	Average Effective Stress [MPa]	Average Mean Stress [MPa]
1	0.800 (7.200 g/cm3)	0.278 (2.502 g/cm3)	753.39	−580.64
2	0.798 (7.182 g/cm3)	0.220 (1.980 g/cm3)	575.75	−435.62
3	0.799 (7.191 g/cm3)	0.240 (2.160 g/cm3)	580.00	−437.99
4	0.800 (7.200 g/cm3)	0.289 (2.601 g/cm3)	748.70	−579.10
5	0.798 (7.182 g/cm3)	0.224 (2.016 g/cm3)	572.10	−432.60
6	0.800 (7.200 g/cm3)	0.229 (2.061 g/cm3)	408.30	−314.72
7	0.797 (7.173 g/cm3)	0.246 (2.214 g/cm3)	738.29	−557.22
8	0.799 (7.191 g/cm3)	0.270 (2.430 g/cm3)	557.72	−433.19
9	0.802 (7.218 g/cm3)	0.283 (2.547 g/cm3)	410.59	−317.03

and Figure 5, Figure 6 and Figure 7 present the relative density, effective stress, and mean pressure values for each combination of levels.

In all cases, it is observed that the relative density varies with the height of the molded body and is not uniform at the joint between the shoe and the partial stator core, which is expected to cause damage during molding. In Case 2, the difference between the maximum and minimum relative densities was found to be the lowest, at 0.220. As the molding pressure increased, the difference between the effective stress and mean stress increased. Moreover, as the molding temperature increased, the effective stress and mean stress decreased.

5.2. DOE Analysis

Since the difference between the maximum and minimum relative densities was a smaller-the-better characteristic, the SNRs were calculated using Equation (7) and are listed in

Table 10. SNRs corresponding to L₉(3³) orthogonal array.

Case	Max.–Min. Relative Density	SNR
1	0.278	11.119
2	0.220	13.152
3	0.240	12.396
4	0.289	10.782
5	0.224	12.995
6	0.229	12.803
7	0.246	12.181
8	0.270	11.373
9	0.283	10.964

. In addition, the average SNRs corresponding to the three levels of each control factor are shown in Figure 8 and

Table 11. SNR analysis for each control factor.

Description		Molding Pressure [MPa]	Molding Temperature [°C]	Heating Time [s]
Level	1	12.22	11.36	11.77
	2	12.19	12.51	11.63
	3	11.51	12.05	12.52
Delta		0.72	1.15	0.89
Rank		3	1	2

.

Based on the SNR analysis, the difference between the maximum and minimum relative densities was found to be most dependent on the molding temperature, followed by the temperature rise time and molding pressure, and is represented by the delta value in Table 11. Since the maximum SNR corresponds to the optimal condition, 120 MPa, 500 °C, and 120 s were selected as the optimal molding pressure, molding temperature, and temperature rise time, respectively.

5.3. Verification of Proposed Optimal Process Conditions

The finite-element analysis results obtained under the optimal process conditions determined through the SNR analysis were compared with the results in Case 2, which had the maximum SNR. Figure 9, Figure 10 and Figure 11 present the comparisons of the relative density, effective stress, and average stress, respectively, and

Table 12. Comparison with proposed process condition and SNR.

Simulation Case	Max.–Min. Relative Density	Relative Density	Effective Stress [MPa]	Mean Stress [MPa]
Optimal process conditions	0.218	0.799	578.69	−441.04
Case 2	0.220	0.798	757.75	−435.62

summarizes all the findings. The difference between the maximum and minimum relative densities under the selected optimal process conditions was observed to be 0.002 lower than that in Case 2, while the average relative density under the optimal conditions was found to be 0.001 higher. In addition, the average effective stress and average mean pressure under the optimal conditions were approximately 179.06 MPa and 5.42 MPa smaller, respectively, than those in Case 2.

5.4. Motor Analysis

The torque was expressed through motor analysis using the manufactured SMCs stator core, as shown in Figure 12. The average torque value was found to be 150.237 mN·m (@10,000 rpm).

6. Experimental Verification

6.1. Mechanical Property

To validate the optimal process conditions, a mold was prepared, and three partial stator cores were molded, as shown in Figure 13. To proceed with the verification, scanning electron microscopy (SEM) analysis was conducted by dividing the center of the teeth of the starter core into 15 mm, 9 mm, and 3 mm from the bottom surface into the top, middle, and bottom, respectively, and the findings were compared with the powder metallurgy simulation results, as shown in Figure 14. Both the SEM results and simulation results showed that the relative density of the product decreased from the top to the middle and then to the bottom.

6.2. Microstructure Evaluation

An XRD test (X-ray diffraction analysis system, XPert Pro MRD, Malvern, Panalytical, Worcestershire, UK), EDS (energy dispersive spectrometer) test to analyze elements, and EBSD (electron backscatter diffraction) test to analyze crystal phases were conducted to confirm the phase information of the stator core, as shown in Figure 15, Figure 16 and Figure 17, respectively. As a result of the XRD test, the peak of the stator core was the same as the generally known peak values of α-Fe of 44.9° (110 plane), 65.4° (200 plane), and 82.8° (211 plane). As a result of EDS, it was confirmed that the composition of the stator core was mainly Fe and Si. In addition, as a result of EBSD, it was confirmed that the phase of the powder was ɑ-Fe, and the crystal structure of the phase was a BCC structure.

6.3. Electrical Property

A test motor was manufactured using the manufactured stator core, as shown in Figure 18. Performance tests were conducted using a dynamometer. As a result, the torque was 150.156 mN·m (@10,107.5 rpm). The error rate between the analysis results and the experimental results was 0.05% and is shown in

Table 13. Comparison of torque analysis and experiment.

Definition	Torque [mN·m]
Analysis (@10,000 rpm)	150.237
Experiment (@10,107.5 rpm)	150.156
Error rate	0.05%

.

7. Discussion and Conclusions

In this study, using Fe-6.5 wt.%Si powder prepared via 50% hybrid atomization and 50% gas atomization (3 h milling time), prototype stator cores were produced under optimal conditions determined via finite-element method analysis based on the Taguchi method. The optimal process conditions were experimentally verified through comparisons with the results from density tests and the SEM analysis. Accordingly, the following observations were noted:

• The Taguchi method was used to select the optimal process conditions for molding the Fe-6.5 wt.%Si powder, and the difference between the maximum and minimum relative densities of the final molded body was set as the characteristic value. In addition, the molding pressure, molding temperature, and temperature rise time, which affect formability, were selected as the control factors.
• Based on powder metallurgy analysis using the Taguchi method, the optimal process conditions were found to be a molding pressure of 120 MPa, a molding temperature of 500 °C, and a temperature rise time of 120 s. Under these conditions, the average relative density was 0.799, and the difference between the maximum and minimum relative densities was 0.218. Additionally, the average effective stress and average mean stress were 578.69 MPa and −441.04 MPa, respectively.
• Core loss, saturation flux density, and bulk conductivity were measured, and motor performance analysis was performed. As a result, the average torque was 150.237 mN·m at 10,000 rpm.
• To validate the optimal process conditions, three prototype stator cores were manufactured, and the densities obtained from the powder metallurgy simulation were compared with those of the prototypes. To analyze the trend of internal density, SEM was analyzed by dividing the 15, 9, and 3 mm sections from the bottom into the top, middle, and bottom. The powder metallurgy analysis showed that the density decreased from the top to the middle to the bottom; this trend was consistent with the SEM analysis.
• XRD tests were performed to confirm the phase information of the stator core, and EDS and EBSD tests were performed to quantitatively characterize the crystal phase and particle distribution. As a result of the test, it was confirmed that the peak value was the same as the peak value of ɑ-Fe, and it had a BCC crystal structure.
• A test motor was manufactured using SMCs stator core, and performance tests were conducted. As a result, the torque was found to be 150.156 mN·m at 10,107.5 rpm. In addition, the reliability of the analysis can be verified, as the error rate between the analysis and test results was 0.05%.

The powder metallurgy analysis in this study was conducted to determine the optimal initial process conditions for the manufacturing of SMCs. We successfully determined process conditions that can reduce defects by minimizing the difference between the maximum and minimum relative densities of the product. In addition, a motor was manufactured using a SMC stator core, and it was confirmed that the error rate between the analysis results and the experimental results was very small.

Additional research is needed to determine optimal process conditions for the manufacturing of stator cores of various shapes using Fe-6.5 wt.%Si powder.