High silicon non-oriented electrical steel sheets fabricated by hot dipping and diffusion annealing

Abstract

High silicon (∼6.5 wt%) non-oriented electrical steel sheets were produced by hot dipping cold rolled 3.5 wt% Si steel sheets into a hypoeutectic Al-8 wt% Si melt followed by diffusion annealing. The microstructure and element diffusion of the processed sheets were characterized and simulated. It is demonstrated that the Al and Si contents across the sheet thickness are largely increased, and the elements are fairly uniformly distributed, which significantly reduced the core loss of the steel sheets, i.e., close to that of the 6.5 wt% silicon steel manufactured by chemical vapor deposition. This manufacturing route features low capital investment and easy process control and has great potential to become a viable technique to produce high silicon electrical steels.

Keywords

non-oriented electrical steel high silicon steel hot dipping diffusion annealing core loss Al-Si melt

Introduction

Non-oriented electrical steels (NOES) are widely used in electrical machines, power generators, automobiles, etc., to enhance the magnetic flux during energy conversion.^1–3 They are characterized by relatively low coercivity, high magnetic permeability and relatively high saturation magnetization.^4,5 NOES with lower than about 3.5 wt% Si are the common products produced by steel companies, although NOES with higher silicon contents show lower core losses,^6,7 especially at high frequencies. NOES with 6.5 wt% Si exhibit the highest maximum permeability, near-zero magnetostriction, and very low core losses, and are particularly suitable for high-frequency applications such as traction motors in electric vehicles.^5,7 However, when the Si content is over ∼5.0 wt%, the ductility of the steel drops to nearly zero⁸ due to the formation of the ordered B₂ (FeSi/Al) and DO₃ (Fe₃Si/Al) phases,^1,9 which makes it nearly impossible to cold roll the steel.^10–13 Hence, commercial production of 6.5 wt% Si electrical steels using conventional routes has not been realized, although numerous studies in laboratories have been reported in the literature.^6,8,14–17

Various unconventional methods have been developed to produce 6.5 wt% Si steel, e.g., rapid quenching,¹⁸ directional solidification,^5,19,20 spray forming,²¹ melt spinning,^22,23 strip casting,^7,24 unconventional alloying,^7,8,21,25 etc., which, however,²⁶ have not been accomplished in mass production. Chemical vapor deposition (CVD) is a technique that has been developed and adopted by the industry^27,28 to commercially produce 6.5 wt% Si steel from cold rolled, thin 3.0 wt% Si steel sheets, which involves a chemical reaction, SiCl₄ + 5Fe → Fe₃Si + 2FeCl₂, and requires special equipment and precise process control. In addition, it poses environmental hazards due to the use of harmful chemicals like SiCl₄.²⁹

Hot dipping is a common process used in the steel industry to produce galvanized or other coated steel sheets for corrosion resistance. In this process, high-melting-point steel sheets are dipped into low-melting-point molten alloys (e.g., Zn, Al, etc.) to form a coating layer on the surfaces of the steel sheets, which acts as a protective layer for the steel substrate. Similar processes have been used to dip silicon steel sheets into Al-Si alloys to produce high Si/Al steel.^10–13^,26,30–32 Danzo et al.^10,26,31 used a hypoeutectic Al-1.0 wt% Si alloy bath to perform hot dipping followed by diffusion annealing at high temperatures and studied the formation of various Fe-Al-Si intermetallic phases. Due to the low Si content in the melt, the overall silicon in the substrate after diffusion annealing is not high (<4.5 wt%).²⁶ On the other hand, the effect of Si and Al concentrations on the core loss of the developed material was not reported. Ros-Yãnez et al.^30,32 performed hot-dipping of an Fe-3.3 wt% Si substrate in a hypereutectic Al-27 wt% Si melt followed by high temperature annealing for different times. However, the reported core losses (measured up to 1.4 T at 50 Hz) were still relatively high (2.11 W/kg) as compared to those of the 6.5 wt% Si steel manufactured by CVD. It has been shown that using high Si hypereutectic Al-Si melt for dipping leads to the formation of the DO₃ type intermetallic phases in the coating layer and forms voids along the coating/substrate interface which drastically impact the magnetic properties.³³ In addition, using a higher Si melt for dipping also requires a very high annealing temperature (around 1200 °C) to diffuse the Si into the steel substrate, causing high energy consumption.

The current work is based on the same concept of these previous studies, but uses an Al-Si alloy melt containing a moderate amount of Si (8 wt%) that is close to the eutectic point (12.6 wt% Si) of the Al-Si system, which not only reduces the dipping and annealing temperatures, but also avoids issues associated with the very high (27 wt%) or very low Si (1 wt%) Al-Si alloys mentioned above. The intermetallic phases formed in the coating layer, the distribution of elements in the substrate and the magnetic properties of the hot-dipped and diffusion-annealed steel sheets were characterized. It is shown that the use of hypoeutectic Al-8 wt% Si melt can produce electrical steel sheets with Si close to 6.5 wt%. With high enough annealing temperature (1000°C) and long enough diffusion time (8 h), a core loss of ∼0.62 W/kg at 50 Hz and 1.0 T is achieved, which is close to that (0.58 W/kg) of the 6.5 wt% Si steel manufactured by CVD. The results illustrate that the hot-dipping and diffusion-annealing route has great potential to become a viable technique to cost-effectively produce high silicon electrical steels.

Experimental

The substrate material used in the present study was a 3.5 wt% Si NOES. The chemical composition (wt%) of the steel was: 0.002 C, 3.5 Si, 0.5 Mn, 0.5 Al, 0.0019 S, 0.0029 N. The steel was melted in a vacuum induction furnace and cast into ingots of 200 mm × 200 mm. The ingot was then homogenized at 1200 °C for 2 h and roughed to 25 mm in a 2-high reversing rolling mill between 1050 and 900 °C. The roughed plates were reheated to 1040 °C and held for 1.5 h, and then rolled to a final thickness of 3.1 mm. The hot-rolled plates were subject to annealing at 750 °C for 1 h followed by grit blasting to remove the scales. The plates were then cold rolled to a final thickness of 0.5 mm. The cold-rolled sheets were cut into strips (substrates) of 200 mm (rolling direction) × 30 mm (transverse direction) for hot dipping.

Prior to hot dipping, the substrates were prepared by grit blasting to remove grease or oxides on the surfaces. The substrates were subsequently soaked in a 5% soap solution to remove any residual grease and rinsed in ethanol followed by air drying. The prepared substrates were then dipped in an Al-Si alloy melt with 8 wt% Si maintained at 740 °C for 30 s using a setup shown in Figure 1 followed by air cooling to room temperature. The hot-dipped samples were then subject to diffusion annealing in a tube furnace at 800, 900, and 1000 °C for 0.5, 1, 2, and 8 h in argon atmosphere. The samples were inserted into the pre-heated furnace and held for the designated times at the designated temperatures. They were subsequently cooled in the furnace to 200 °C before being taken out. Examples of the samples before and after hot dipping are shown in Figure 2.

Figure 1.

Hot dipping setup used in the present study along with the schematic diagram showing different components.

Figure 2.

Pictures of the substrates: a) before dipping, b) after dipping in an Al-8 wt% Si melt at 740 °C for 30 s.

The hot-dipped and annealed samples were cut on the rolling direction (RD)-normal direction (ND) plane for microstructure characterization and composition analysis. The samples were ground using various grits of silicon-carbide paper and polished using diamond pastes and colloidal silica suspension. SEM-BSE (scanning electron microscopy-backscattered electron) imaging was performed to distinguish (elemental contrast) the intermetallics formed during the hot dipping process. The elemental distribution across the thickness of the samples and the intermetallic phases formed during diffusion annealing were analyzed using energy dispersive spectroscopy (EDS). Magnetic properties of the diffusion-annealed samples were measured using a single sheet tester, which takes small samples of 50 mm (length) × 30 mm (width) for the measurements. AC core losses at 1.0 T and 1.5 T were measured at 50, 60, 100, 200, and 400 Hz, and the testing parameters used are listed in Table 1. Samples processed at different annealing conditions were measured and the results were compared to those of the 6.5 wt% Si steel sheets produced using the CVD method.³⁴

Table 1.

Parameters used for magnetic property measurements

Strip dimension (length × width)	50 mm × 30 mm
H coil turns	360
B coil turns	30
H coil resistance	4.5 Ω
B coil resistance	1.4 Ω
B step size	0.01 T
Frequency	50–400 Hz
AC waveform	Sinusoidal

Results and discussion

Microstructure of the dipped strips

The typical cross-section microstructure of the hot-dipped strip is illustrated in Figure 3. There is an interdiffusion layer near the coating-substrate interface where the reaction between the electrical steel sheet and the Al-Si alloy takes place. As a result, needle (τ₆) and massive (τ₅) type intermetallic phases are formed in this region.³⁵ For reference, Table 2 lists the crystal symmetry and lattice parameters of the various intermetallic phases observed in this work. The presence of the τ₆-Fe₂_{Al₉Si₂ (Location B in Figure 3(b)) and τ₅-Fe₂Al₇Si phases (Locations C and D in Figure 3(b)) is confirmed from the EDS measurements, and the compositions of the material at various locations shown in Figure 3(b) are listed in Table 3.}

Figure 3.

Microstructure of the 3.5 wt% Si steel hot dipped at 740 °C for 30 s: a) cross-section SEM image showing the coating, the interface, and the substrate, b) high magnification BSE image of the dashed region in (a) showing the locations (A-F) for the EDS measurements.

Table 2.

Crystal symmetry and lattice parameters of the intermetallic phases formed during hot-dipping and diffusion annealing

Phase	Stoichiometry	Crystal type	Space group	Lattice parameters (Å)	Ref.
Fe₄Al₁₃	Fe₄Al₁₃	Monoclinic	C2/m	a = 15.492, b = 8.078	³⁶
Fe₄Al₁₃	Fe₄Al₁₃	Monoclinic	C2/m	c = 12.471, β = 107. 7°	³⁶
Fe₂Al₅	Fe₂Al₅	Orthorhombic	Cmcm	a = 7.656, b = 6.415	³⁷
Fe₂Al₅	Fe₂Al₅	Orthorhombic	Cmcm	c = 4.218	³⁷
FeAl	FeAl	Cubic	Pm $\bar{3}$ m	a = b = c = 2.85	³⁸
Fe₃Si	Fe₃Si	Cubic	Fm $\bar{3}$ m	a = b = c = 5.63	³⁹
τ₁	Fe₃Al₅Si₅	Triclinic	P $\bar{1}$	a = 4.65, b = 6.32	⁴⁰
				c = 7.49, α = 101.4°
				β = 105.9°, γ = 101.2°
τ₅	Fe₂Al₇Si	Hexagonal	P6₃	a = 12.345, c = 26.21	⁴¹
τ₆	Fe₂Al₉Si₂	Monoclinic	A2/a	a = 6.161, b = 6.175	⁴²
τ₆	Fe₂Al₉Si₂	Monoclinic	A2/a	c = 20.813, β = 90.42°	⁴²

Table 3.

The contents of Fe, Si, and Al (EDS analysis) at the locations shown in Figure 3(b).

Location	Fe (at%)	Si (at%)	Al (at%)	Possible phase
A	0.22	6.17	93.60	-
B	11.02	16.50	72.48	τ₆
C	17.53	12.13	70.33	τ₅
D	17.69	12.38	69.92	τ₅
E	89.64	8.79	1.57	Ferrite
F	90.67	8.06	1.27	Ferrite

The formation of the specific intermetallic phases mainly depends on the chemical composition of the Al-Si melt and the composition of the substrate material.¹⁰ As reported in previous studies,^10,26 the intermetallic phases formed during hot dipping are mostly iron aluminides if the melt contains low Si. With the increase of the Si content in the molten alloy, Fe-Al-Si intermetallic phases can also form due to the incorporation of Si in the iron aluminides, which occurs through the replacement of some Al atoms by Si atoms.

Wang et al.³⁵ reported the formation of Fe₄Al₁₃, Fe₂Al₉Si₂ (τ₆), and Fe₂Al₇Si (τ₅) phases when dipping a mild steel in an Al-7.5 wt% Si melt. These intermetallic phases are also formed in this study. Due to the very high solubility of Al in the iron lattice (32.6 at%), an Al rich ferrite is also formed near the interface. Location E (5.20 µm from the interface, Figure 3(b)) shows slightly higher Al and Si contents than Location F (19.60 µm from the interface), but the Al and Si contents of both locations are about 2 times higher than the original Al and Si contents of the substrate, indicating apparent diffusion of these elements into the substrate during hot dipping. The coating layer thickness after hot dipping is 56.62 ± 2.70 µm, which is about 1/9 of the substrate thickness.

Microstructure of the diffusion-annealed strips

The microstructures of the samples after diffusion annealing at 800 °C for different times are shown in Figure 4. When the annealing time is short (0.5, 1, and 2 h), there is a clear interface between the coating layer and the substrate, and the formation of intermetallic phases (as confirmed by EDS analysis), e.g., Fe₄Al₁₃, Fe₂Al₅, Fe₃Si, FeAl, and τ₁, is clearly seen. The Fe₄Al₁₃ phase forms at the outer edge of the coating layer, the Fe₂Al₅ phase is the bulk of the coating layer (matrix), and the τ₁ phase is present within the matrix as a separate layer between the coating/substrate interface and the outer edge. When the annealing time is increased to 8 h, the coating layer is completely dissolved, and no compositional variation is noticed across the thickness. Nevertheless, some small, disconnected Fe₂Al₅ particles can still be detected at the outer edge.

Figure 4.

SEM images of the hot-dipped sample after diffusion annealing at 800 °C for: a) 0.5 h, b) 1 h, c) 2 h, and d) 8 h. The coating/substrate interface is shown by yellow lines.

It should be noted that the Fe₄Al₁₃ phase is only present in the sample after 0.5 h annealing, and it is absent in all the other cases (Figure 4). Wang et al.³⁵ have shown that this phase forms at the beginning of the phase transformation in the coating, which is then transformed to Fe₂Al₅ as the annealing continues. The sequence of phase formation as reported in³⁵ is: liquid→τ₆→τ₅→Fe₄Al₁₃→Fe₂Al₅→τ₁→ferrite. Figure 4 only shows the later part of this chain of transformations, and it is not clear if the Fe₂Al₅→τ₁ transformation has occurred or not since these two phases coexist from 0.5 to 2 h, and the phase remained (only a very small amount at the edge though) after 8 h is Fe₂Al₅. In addition, a part of the Fe₂Al₅ phase transforms to FeAl (solid solution of aluminum in α-iron) when the annealing time is 2 h. This is due to the migration of the Al atoms in the coating into the substrate and the reverse migration of the Fe atoms from the substrate to the coating.

The presence of Fe₃Si is also detected (near the coating/substrate interface) during annealing at 800 °C from 0.5 to 2 h. This phase has been observed in the sample after hot dipping, and it has also been reported to exist in siliconized steel using the CVD method.⁴³ However, with the increase of the annealing temperature and time, this phase disappears due to the diffusion of Si into the substrate. The Si content in Fe₃Si is 25 at%, and the rest is Fe; thus, the difference in concentration between the Fe₃Si layer and the substrate layer facilitates the diffusion of Si into the substrate. Fe₃Si is a stoichiometric phase which results in high inter-diffusivity of Si and Fe. Hence the phase disappears early in the diffusion annealing process.⁴⁴

Figure 5 shows the microstructure and intermetallic phases of the samples after annealing at 900 °C, as confirmed by EDS analysis. In this case, the coating layer is dominated by the FeAl and Fe₂Al₅ phases after annealing for 0.5 to 2 h (Figures 5(a), 5(b), and 5(c)), and the τ₁, Fe₄Al₁₃ and Fe₃Si phases are not seen. In contrast to the samples after annealing at 800°C where the matrix is Fe₂Al₅ and the τ₁ (Fe₃Al₅Si₅) phase is distributed between the coating/substrate interface and the outer edge, here the matrix is FeAl and the Fe₂Al₅ phase appears as layers between the interface and the outer edge or at the outer edge. Again, when the annealing time is increased to 8 h, the intermetallic phases disappear, and a homogenous sheet is formed (Figure 5(d)).

Figure 5.

SEM images of the hot-dipped samples after diffusion annealing at 900 °C for: a) 0.5 h, b) 1 h, c) 2 h, and d) 8 h. The coating/substrate interface is shown by yellow lines.

If the annealing temperature is increased to 1000 °C (Figure 6), 1 h is sufficient to dissolve all the intermetallic phases, as shown in Figure 6(b). The Fe₂Al₅ and FeAl phases are still present after 0.5 h (Figure 6(a)), but to a lesser extent as compared to the samples annealed at lower temperatures (Figure 4(a) and Figure 5(a)). The voids present along the coating/substrate interface in samples annealed at 1000 °C are known as Kirkendall voids, which are observed during high temperature annealing due to the difference in diffusivity of the Fe and Al elements at high temperatures. The voids are created due to the coalescence of vacancies which are generated during the diffusion process.¹⁰ The coalescence of vacancies is linked to diffusion along the c-axis of the Fe₂Al₅ phase, which has an orthorhombic crystal structure and has more vacancies along its c-axis.

Figure 6.

SEM images of the hot-dipped samples after diffusion annealing at 1000 °C for: a) 0.5 h, b) 1 h, c) 2 h, and d) 8 h. The coating/substrate interface is shown by a yellow line.

It is thus seen that, to completely dissolve the intermetallic phases formed during hot dipping, diffusion annealing can be performed at a high temperature, which needs shorter annealing times, or at a relatively low temperature, but it requires longer annealing times. The thickness of the coating layer (as determined by the coating/substrate interface in SEM) is an indication of the extent to which the Si and Al diffuse into the substrate: a smaller coating thickness after diffusion annealing means a higher Al + Si content in the substrate. With the increase of the annealing time and temperature, the coating thickness is reduced (Table 4) due to the dissolution of the intermetallic phases into the substrate. For diffusion annealing at 800 °C and 900 °C, the intermetallic phases are completely diffused after 8 h, while for annealing at 1000°C, the intermetallic phases are entirely diffused after only 1 h.

Table 4.

The coating thickness (μm) after annealing at different temperatures for different times

Annealing Temperature (°C)	Annealing Time (h)
Annealing Temperature (°C)	0.5	1	2	8
800	46.44 ± 3.56	41.45 ± 2.68	36.91 ± 3.19	-
900	45.83 ± 1.72	38.72 ± 1.96	26.49 ± 2.75	-
1000	29.35 ± 1.91	-	-	-

The Fe-Al-Si ternary system

The formation of intermetallic phases in the Fe-Al-Si system is governed by the migration of the Fe, Al, and Si atoms, which is highly dependent on the temperature, composition, and grain boundary characteristics. To better understand the formation of intermetallic phases during the diffusion annealing process, the Fe-Al-Si ternary phase diagrams are calculated using the Thermo-Calc⁴⁵ software with the TCAl8⁴⁶ database. Three sections at 800, 900, and 1000°C are shown in Figure 7. Superimposed on these figures are the intermetallic phases observed in the experiments. It should be noted that the same phases (e.g., Fe₂Al₅ and FeAl) may be present at all the annealing temperatures, but their compositions are different (as seen in different locations of the phase diagram). As mentioned earlier, the predominant intermetallic phases at 800 °C are Fe₂Al₅, Fe₄Al₁₃, Fe₃Si, FeAl and τ₁. For samples annealed at 900 °C, the number of intermetallic phases is reduced, i.e., only Fe₂Al₅ and FeAl are detected if the annealing time is 0.5–2 h. For annealing at 1000 °C, Fe₂Al₅ and FeAl are also the intermetallic phases formed, but they are only observed in the sample after 0.5 h annealing.

Figure 7.

Sections of the Fe-Al-Si ternary phase diagram at: a) 800 °C, b) 900 °C, and c) 1000 °C. The observed phases are superimposed on the diagrams to illustrate their compositions. Some phases have been observed at different temperatures and times.

Distribution of Al and Si in the strips

To characterize the composition across the thickness of the dipped and annealed samples, EDS line scans are performed up to the half thickness of the sheets and the average values (two measurements) of Al and Si are reported, as shown in Figure 8. In all cases except when the annealing time is 8 h, there is a sharp drop of the Al concentration at the coating-substrate interface (about 50 to 75 μm from the surface). When the annealing temperature is relatively low (800 and 900°C) or the annealing time is short (0.5–2 h), apparent fluctuations in the Al and Si concentration are observed up to the interface. These fluctuations are due to the Fe-Al-Si intermetallic phases which show varying amount of Al in decreasing order as follows: Fe₄Al₁₃, Fe₂Al₅, τ₅, and τ₆.³⁵ At these temperatures, if the annealing time is within 0.5–2 h, the concentrations of Al and Si in the substrate are almost the same across the thickness, i.e., ∼0.8 wt% and 4.6 wt% for Al and Si, respectively (see the inset in Figures 8(a) and 8(b)), and they are only moderately higher than the initial concentrations in the steel. When the annealing time is increased to 8 h, the distribution of Al becomes smoother, and no sharp drop or fluctuation is observed. The highest concentration in the coating layer is reduced to about 15–20 wt%, while the lowest concentration in the substrate is increased to 1–2%. The Si content in the substrate is increased to ∼5.5 wt% (800°C) and ∼6.0 wt% (900°C), respectively, although there are apparent variations across the thickness.

Figure 8.

Profiles of the Si and Al concentrations after diffusion annealing at: a) 800 °C, b) 900 °C, and c) 1000 °C; d) average Al and Si contents vs annealing time and temperature.

When the annealing temperature is 1000°C (Figure 8(c)), a sudden drop of the Al content is also noticed (except for the sample after 8 h annealing), but it is moved to about 75 μm. Again, for 0.5–1 h annealing, the overall Al content is ∼0.8 wt% (from 120 μm to the center), while the overall Si content is ∼4.6 wt% (from 150 μm to the center), almost the same as those after annealing at 800 and 900°C. After annealing for 2 h, the Al content gradually drops from about 25 wt% at the surface to about 0.8 wt% at the center, while the Si content maintains about 5.8 wt% from the interface (75 μm) until about 150 μm and then gradually decreases to about 4.6 wt% at the center. After annealing for 8 h, the variations of both the Al and Si across the thickness are relatively small. While the Al content decreases from about 10 wt% at the surface to about 4 wt% at the center, the Si content gradually increases from about 4 wt% at the surface to about 6.4 wt% at the center. This distribution of Si across the thickness is opposite to what is expected (as has been seen in CVD processed high silicon steel⁴⁷), i.e., the center should have lower Si content than the surface due to the higher Si content at the coating layer. This unusual behavior is explained later in the Diffusion Simulation section.

It is thus seen that it is possible to increase the silicon content in the steel sheets through hot dipping and diffusion annealing. With the increase of annealing time and temperature, the average Al and Si contents increase gradually (Figure 8(d)). The increase of the Al content is more significant than the increase of the Si, due to the significantly larger amount of Al in the melt. When the annealing temperature is 1000°C and the annealing time is 8 h, the highest Al content is 10 wt% (surface) while the highest Si concentration reaches 6.4 wt%. The highest average Al and Si contents across the thickness are 5.4 wt% and 6.01 wt% (Table 5), respectively, about 980% and 71% increases as compared to those of the original steel sheets.

Table 5.

Average Al and Si concentrations of the samples after diffusion annealing

Time (h)	Al (wt%)			Si (wt%)
Time (h)	800°C	900°C	1000°C	800°C	900 °C	1000 °C
0.5	0.91	1.03	1.64	4.78	4.88	5.15
1	1.06	1.29	2.7	4.84	5.16	5.18
2	1.76	1.72	4.13	5.05	5.06	5.19
8	3.83	4.58	5.4	5.58	5.99	6.01

Diffusion simulation

To better understand the above-observed element distribution, the process of diffusion annealing was simulated using the DICTRA software^45,48 available in the Thermo-Calc package, which can simulate diffusion-controlled transformations in multicomponent alloy systems. DICTRA simulations are performed by solving multicomponent diffusion equations in one dimension utilizing the flux balancing of components across a moving planar interface and considering the gradient in chemical potential as the driving force for diffusion,⁴⁹ which remains in a state of local equilibrium. The present problem is like a diffusion couple, where the diffusion is considered to proceed via a vacancy migration mechanism that involves atom jumping to the nearest vacancy site.

The DICTRA simulations in the present study were carried out using the TCFE12 and MOBFE7 databases, where a diffusion couple was set up with a coating layer attached to a substrate layer. The thickness of the coating layer was taken to be 50 µm and the substrate 250 µm, which resembles the system in this study. A linear grid was chosen for the current system with 250 and 50 points for the substrate and coating layer, respectively. A ternary system was chosen for simplicity where the composition of the coating represented the composition of the solidified Al-Si melt (87 wt% Al, 9 wt% Si and 4 wt% Fe) and the substrate composition was 96 wt% Fe, 3.5 wt% Si and 0.5 wt% Al. The presence of Fe (4 wt%) in the Al-Si melt is due to the interfacial reaction between the melt and the steel substrate during dipping.²⁶ The simulations were carried out at 800, 900 and 1000 °C for 0.5 h and 8 h. The homogenization model available in DICTRA was used to simulate the system. The results obtained along with the EDS measurements for samples subject to the same diffusion annealing conditions are shown in Figure 9.

Figure 9.

Simulated element distribution profiles as compared to those measured by EDS for samples diffusion annealed at: (a, b) 800 °C, (c, d) 900 °C, and (e, f) 1000 °C. a), c), e) annealed for 0.5 h, and b), d), f) annealed for 8 h. Note the coating layer starts from distance = 0 µm.

Generally, DICTRA simulations can roughly predict the trend of diffusion for most of the conditions, and the simulations are more accurate when the annealing time is 8 h. The diffusion of Si can be more accurately predicted than Al and Fe in all cases. The simulated profiles of Fe, Al, and Si at 900 and 1000 °C for 8 h match quite closely to the EDS measurements, but the simulation results for 800 °C and for 0.5 h deviate largely from the measurements in the coating layer. The large difference between the simulated and measured concentration profiles in the coating layer can be attributed to the Fe-Si-Al intermetallic phases and the interdiffusion layer, which largely affect the diffusion of Fe/Al/Si atoms,^50,51 but the DICTRA simulations cannot catch it as the software relies on accurate diffusion data (which is absent) for individual components in a multicomponent system and in different thermodynamically stable phases. In addition, the DICTRA simulation is a very simplistic representation of the system, which does not consider the presence of grain boundaries or dislocations, which also largely affect the diffusion process.⁵¹

It is noted that almost every Si profile in Figure 9 (except Figure 9(f)) shows a clear hump at the coating-substrate interface, and the position of this hump depends on the annealing temperature and time. This may be attributed to the diffusion of the Si atoms in the intermetallic phases (mainly τ₅ and τ₆) formed in the coating layer (Figure 3). τ₅ and τ₆ phases have a higher Si concentration (Table 3) than the substrate, which tend to diffuse Si into the substrate. As the annealing time increases, other intermetallic phases with lower Si content than the substrate, e.g., Fe₂Al₅, Fe₄Al₁₃ and FeAl,,⁵² may form, which causes the diffusion of the Si from the substrate towards these phases and generates a peak Si near the interface (Figure 9). Xia et al.⁵³ have attributed the aggregation/hump in the Si concentration (as observed in Figure 9) to the reduced chemical potential of Si in the Fe-Al intermetallic (in the Fe-Al-Si ternary system) as compared to the steel substrate. Hence, the uphill diffusion of Si as observed in the present study (Fe-Al-Si multi-component system) may be attributed to the chemical potential reduction, which is the driving force for the diffusion to continue, unlike in the binary alloy where the diffusion proceeds due to concentration gradient. In other words, the diffusion of Fe, Al, and Si across the coating-substrate takes place in a way to minimize the overall free energy of the system.^54,55 The formation of such a peak in the Si concentration has also been observed during the diffusion annealing after cladding of Fe-Al onto steel by Křivská et al.⁵⁶ and by Barros et al.⁵⁷ during diffusion annealing after hot dipping a 3.0 wt% Si steel substrate in an Al-25 wt% Si melt.

As mentioned before, the sample annealed at 1000 °C for 8 h (Figure 9(f)) shows a higher Si content in the center than in the surface, which is opposite to the distribution observed in the CVD processed 6.5 wt% Si electrical steels (with higher Si concentration at the surface). This is due to the same reason discussed above. The dissolution and formation of intermetallic phases during the annealing process forms a local hump of Si and the location of the hump depends on the annealing temperature and time. The higher the temperature, the closer the hump to the center and the smoother the hump (smaller height). After annealing at 1000°C for 8 h, the hump is located at the mid-thickness of the substrate (300 μm from the surface, Figure 9(f)), which shows a slightly higher Si concentration than its neighboring regions (Figure 9(f) only shows the left side). Due to the flattening of the hump at high temperature, it appears as an almost smooth Si profile if only the left side of the hump is looked (Figure 9(f)). It is expected that, when the annealing time is further increased, the Si content (and the Al and Fe as well) will eventually become uniform across the entire thickness when final equilibrium is reached.

Magnetic properties

The core losses of the hot-dipped and diffusion-annealed samples measured at 50 and 60 Hz (1.0 and 1.5 T) are shown in Figure 10. In most cases, increasing the annealing temperature or time decreases the core loss; when the annealing time is increased to 8 h, the core loss is significantly decreased at all annealing temperatures. However, when the annealing time is 8 h, increasing the annealing temperature from 800°C to 900°C slightly increases the core loss at 1.0 T (Figures 10(a) and 10(c)). Also, when the annealing temperature is 1000°C and the annealing time is 0.5–2 h, the core losses at 1.5 T are quite close to each other, and a longer annealing time may not lead to a lower core loss (Figures 10(b) and 10(d)).

Figure 10.

Core losses of hot-dipped and diffusion-annealed samples measured at 50 and 60 hz: 1 T (a, c) and 1.5 T (b, d).

The decrease of the core loss by diffusion annealing is mainly caused by the increase of the Si and Al contents in the steel, which increases the resistivity of the material,^21,33,58 hence reducing the eddy current loss. Diffusion annealing releases the residual stress accumulated in the material during cold rolling and also leads to the recrystallization of the microstructure, both significantly reducing the core loss. As has been shown before, when the annealing temperature is relatively low (800 and 900°C) and the annealing time is short (0.5–2 h), the Al and Si contents in the steel are moderately increased to about 0.8 wt% and 4.6 wt% from 0.5 wt% and 3.5 wt%, respectively, which leads to the moderate decrease of core losses. The samples annealed for 8 h show much lower core losses than those annealed for shorter times, due to the largely increased Al and Si contents in the steel (5.4 wt% and 6.01 wt% respectively). As shown in Table 6, the core loss (0.62 W/kg) of the sample after annealing at 1000°C for 8 h (50 Hz/1.0 T) is comparable to that (0.58 W/kg) manufactured using the CVD method (the same thickness of 0.5 mm).⁴³ It is also noted that this core loss is much lower than that of the commercial 3.2 wt% Si steel (1.0 W/kg) and that (0.92 W/kg) of the steel produced by Ros-Yanez et al.³⁰ using a similar hot dipping and diffusion annealing process (even with a smaller thickness of 0.35 mm). At 400 Hz/1.0 T, the core loss (15.3 W/kg) of the sample annealed at 1000 °C for 8 h is even lower than that (15.6 W/kg) of the 6.5% Si steel produced by CVD (0.5 mm thickness).

Table 6.

Comparison of the core losses of different materials tested under AC conditions at different frequencies

Material	Thickness (mm)	Core loss (W/kg)
Material	Thickness (mm)	W_10/50	W_15/50	W_10/100	W_10/200	W_10/400
Current study (1000 °C, 8 h)	0.5	0.62	1.31	1.67	4.76	15.3
Current study (900 °C, 8 h)	0.5	0.92	1.35	2.36	6.7	20.62
Current study (800 °C, 8 h)	0.5	0.84	1.44	2.12	5.79	17.36
Hot dipped + diffusion annealed³⁰	0.35	0.92	1.79	-	-	-
6.5 wt% Si (CVD)³⁴	0.3	0.49	-	-	-	10.00
6.5 wt% Si (CVD)³⁴	0.5	0.58	-	-	-	15.6
3.2 wt% Si non-oriented electrical steel⁵⁹	0.35	1.00	-	-	-	-

It is thus seen that hot dipping and diffusion annealing can produce non-oriented electrical steel sheets with very good magnetic properties. Due to its simple processing procedure and low cost, it could become a viable manufacturing technology to produce high silicon electrical steels. However, further studies are still needed to assess the impact of the intermetallic phases formed during the diffusion annealing on the mechanical and magnetic properties of the steel sheets. In addition, the high Al and Si concentration in the sheet is expected to induce brittleness to the steel, which will negatively impact the fabrication of electrical steel cores from the steel sheets, which also needs to be addressed.

Conclusions

Cold rolled 3.5 wt% Si steel sheets are hot dipped in an Al-Si melt and diffusion annealed to increase the silicon content to about 6.5 wt%, which provides a cost-effective route to produce high silicon electrical steels.

Using a hypoeutectic Al-Si alloy with moderate Si content (8 wt%) can reduce the hot-dipping temperature, increase the overall Si in the substrate, and lead to optimal magnetic properties.

Diffusion annealing at 1000°C for 8 h increases the Si content from 3.5 wt% to an average of 6.0 wt% (maximum 6.4 wt%) and the Al content from 0.5 wt% to an average of 5.4 wt%, which increases the electrical resistivity of the steel and reduces the eddy current loss.

The core loss (0.62 W/kg) of the hot-dipped and diffusion-annealed steel sheet at 50 Hz/1.0 T is comparable to that (0.58 W/kg) of the 6.5 wt% Si steel produced by CVD, while the core loss at 400 Hz/1.0 T (15.3 W/kg) is even lower than that (15.6 W/kg) produced by CVD.

With appropriate annealing temperature and diffusion time, the various intermetallic phases formed during hot dipping can be relatively uniformly diffused into the substrate and form high silicon electrical steel sheets.

Footnotes

Acknowledgments

Jian Li and Renata Zavadil are gratefully thanked for SEM and EDS characterization. Maciej Podlesny is thanked for assistance in the hot dipping and diffusion annealing experiments. Peter Newcombe, Bruce Ingram, and David Saleh are thanked for casting the 3.5 wt% electrical steel and Michael Attard is acknowledged for rolling the steel.

ORCID iDs

Gyanaranjan Mishra

Youliang He

Author contribution(s)

Gyanaranjan Mishra: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Software; Validation; Visualization; Writing – original draft.

Youliang He: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Project administration; Resources; Supervision; Validation; Visualization; Writing – review & editing.

Clodualdo Aranas: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Project administration; Resources; Supervision; Validation; Visualization; Writing – review & editing.

Funding

The authors disclosed receipt of the following financial support for the research: This work was supported by the Program of Energy Research and Development, Natural Resources Canada [grant number CMAT-23-108]; Natural Sciences and Engineering Research Council of Canada Alliance Program [grant number ALLRP-568545]; Natural Resources Canada's Energy Innovation Program [grant number EIP-OTD-249].

Declaration of conflicting interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and publication of this article.

Data availability

Some experimental data can be made available on request

References

Ouyang

Chen

Liang

, et al. Review of Fe-6.5 wt%Si high silicon steel—A promising soft magnetic material for sub-kHz application. J Magn Magn Mater 2019; 481: 234–250.

Sugimoto

. History and future of soft and hard magnetic materials. Springer: Singapore, 2019, pp.261–277.

Moses

. Electrical steels. Past, present and future developments. IEE Proceedings A: Physical Science Measurement and Instrumentation Management and Education Reviews 1990; 137: 233–245.

Fiorillo

Bertotti

Appino

, et al. Soft magnetic materials. Wiley Encyclopedia of Electrical and Electronics Engineering 2016: 1–42.

Zheng

Liang

, et al. Formation of columnar-grained structures in directionally solidified Fe-6.5 wt.%Si alloy. Intermetallics (Barking) 2011; 19: 165–168.

Fang

Zhang

, et al. Evolution of microstructure and texture in grain-oriented 6.5%Si steel processed by strip-casting. Mater Charact 2017; 126: 125–134.

H-Z

Liu

H-T

Liu

Z-Y

, et al. Microstructure, texture evolution and magnetic properties of strip-casting non-oriented 6.5 wt.% Si electrical steel doped with cerium. Mater Charact 2015; 103: 101–106.

Cheng

Liu

Xiang

, et al. Effects of Nb content on the microstructure and magnetic properties of Fe-6.5 wt.%Si alloy. Optoelectronics and Advanced Materials-Rapid Communications 2016; 10: 82–86.

Gonzalez

Houbaert

. A review of ordering phenomena in iron-silicon alloys. Revista de Metalurgia 2013; 49: 178–199.

10.

Danzo

Houbaert

Verbeken

. Diffusion driven columnar grain growth induced in an Al–Si-coated steel substrate. Surf Coat Technol 2014; 251: 15–20.

11.

Ros-Yanez

Barros

Houbaert

, et al. Crystallographic texture evolution during processing of hot dipped and diffusion annealed high silicon electrical steel. In: INTERMAG 2006 - IEEE International Magnetics Conference, 2006, pp. 901–901.

12.

Houbaert

Ros-Yáñez

Monsalve

, et al. Texture evolution in experimental grades of high-silicon electrical steel. Physica B Condens Matter 2006; 384: 310–312.

13.

Ros-Yanez

Ruiz

Barros

, et al. Study of deformation and aging behaviour of iron–silicon alloys. Mater Sci Eng A 2007; 447: 27–34.

14.

Fang

Liang

, et al. Cold rolled Fe-6.5 wt. % Si alloy foils with high magnetic induction. J Appl Phys 2012; 111: 094913.

15.

Cheng

Liu

Xiang

, et al. Effect of Cu addition on microstructure, texture and magnetic properties of 6.5 wt% Si electrical steel. J Magn Magn Mater 2021; 519: 167471.

16.

Kim

Pan

Lin

, et al. The effect of boron content on the processing for Fe-6.5 wt% Si electrical steel sheets. J Magn Magn Mater 2004; 277: 331–336.

17.

Lin

Zhao

, et al. Effect of annealing on microstructure, texture and magnetic properties of Fe-6.5 wt%Si-0.03 wt%Nb alloy. J Magn Magn Mater 2020; 504: 166699.

18.

Liang

Lin

, et al. Microstructure and mechanical properties of rapidly quenched Fe–6.5 wt.% Si alloy. J Alloys Compd 2010; 504: S476–S479.

19.

Zhang

Chen

Bei

, et al. Directional recrystallization and microstructures of an Fe–6.5wt%Si alloy. J Mater Res 2009; 24: 2654–2660.

20.

Zhang

Yang

, et al. Strain-softening behavior of an Fe–6.5wt%Si alloy during warm deformation and its applications. Mater Sci Eng A 2011; 528: 1391–1395.

21.

Bolfarini

Silva

MCA

Jorge

, et al. Magnetic properties of spray-formed Fe–6.5%Si and Fe–6.5%Si–1.0%Al after rolling and heat treatment. J Magn Magn Mater 2008; 320: e653–e656.

22.

Liang

Wang

, et al. Fabrication of Fe-6.5wt%Si ribbons by melt spinning method on large scale. Advances in Materials Science and Engineering 2015; 2015: 1–5.

23.

Wang

Jiang

Liang

, et al. Magnetic properties and core loss behavior of Fe-6.5 wt.%Si ribbons prepared by melt spinning. Advances in Materials Science and Engineering 2015; 2015: 1–6.

24.

, et al. Influence of hot rolling reduction rate on the microstructure, texture and magnetic properties of a strip-cast Fe-6.5 wt% Si grain-oriented electrical steel. J Magn Magn Mater 2020; 494: 165755.

25.

Zhang

, et al. Effects of boron on microstructure and mechanical properties of Fe-6.5 wt.%Si alloy fabricated by directional solidification. Intermetallics (Barking) 2013; 35: 67–72.

26.

Danzo

Verbeken

Houbaert

. Microstructure of hot dip coated Fe–Si steels. Thin Solid Films 2011; 520: 1638–1644.

27.

Abe

Takada

Murakami

, et al. Magnetic properties of commercially produced Fe-6.5wt% Si sheet. J Mater Eng 1989; 11: 109–116. 1989; 11.

28.

Okami

Abe

Yamaji

, et al. Continuous siliconizing technology of 6.5% silicon steel sheet. Tetsu-to-Hagane 1994; 80: 777–782.

29.

Shen

Zhang

, et al. Magnetic properties of high silicon iron sheet fabricated by direct powder rolling. J Magn Magn Mater 2004; 281: 135–139.

30.

Ros-Yãez

Houbaert

De Wulf

. Evolution of magnetic properties and microstructure of high-silicon steel during hot dipping and diffusion annealing. IEEE Trans Magn 2002; 38: 3201–3203.

31.

Infante Danzo

Verbeken

Houbaert

. Microstructure and texture evolution of Fe-Si steels after hot dipping and diffusion annealing. Mater Sci Forum 2012; 706–709: 2628–2633.

32.

Ros-Yanez

De Wulf

Houbaert

. Influence of the Si and Al gradient on the magnetic properties of high-Si electrical steel produced by hot dipping and diffusion annealing. J Magn Magn Mater 2004; 272–276: E521–E522.

33.

Ghosh

Chromik

Knight

, et al. Effect of metallurgical factors on the bulk magnetic properties of non-oriented electrical steels. J Magn Magn Mater 2014; 356: 42–51.

34.

Takada

Abe

Masuda

, et al. Commercial scale production of Fe-6.5 wt.% Si sheet and its magnetic properties. J Appl Phys 1988; 64: 5367–5369.

35.

Wang

Chen

, et al. Influence of Si on the intermetallic compound formation in the hot-dipped aluminide medium carbon steel. Mater Charact 2023; 197: 112700.

36.

Grin

Burkhardt

Ellner

, et al. Refinement of the Fe4Al13structure and its relationship to the quasihomological homeotypical structures. Zeitschrift fur Kristallographie–New Crystal Structures 1994; 209: 479–487.

37.

Burkhardt

Grin

Ellner

, et al. Structure refinement of the iron–aluminium phase with the approximate composition Fe2Al5. Structural Science 1994; 50: 313–316.

38.

Stein

Vogel

Eumann

, et al. Determination of the crystal structure of the ɛ phase in the Fe–Al system by high-temperature neutron diffraction. Intermetallics (Barking) 2010; 18: 150–156.

39.

Hafner

Spišák

. Structure and stability of the low-index surfaces of Fe3 Si: ab initio density functional investigations. Phys Rev B Condens Matter Mater Phys 2007; 75: 195411.

40.

Yanson

Manyako

Bodak

, et al. Triclinic Fe3Al2Si3 and orthorhombic Fe3Al2Si4 with new structure types. Crystal Structure Communications 1996; 52: 2964–2967.

41.

Roger

Bosselet

Viala

. X-rays structural analysis and thermal stability studies of the ternary compound α-AlFeSi. J Solid State Chem 2011; 184: 1120–1128.

42.

Rømming

Hansen

Gjønnes

. Crystal structure of β-Al4.5FeSi. Structural Science 1994; 50: 307–312.

43.

Yamaji

Abe

Takada

, et al. Magnetic properties and workability of 6.5% silicon steel sheet manufactured in continuous CVD siliconizing line. J Magn Magn Mater 1994; 133: 187–189.

44.

Mehrer

Eggersmann

Gude

, et al. Diffusion in intermetallic phases of the Fe-Al and Fe-Si systems. Mater Sci Eng A 1997; 239–240: 889–898.

45.

Andersson

Helander

Höglund

, et al. Thermo-Calc & DICTRA, computational tools for materials science. Calphad 2002; 26: 273–312.

46.

Aluminum-based Alloys - Thermo-Calc Software. https://thermocalc.com/products/databases/aluminum-based-alloys/

47.

Hiratani

Namikawa

Ninomiya

, et al. Magnetization mechanism of Si-Gradient steel sheet manufactured by CVD siliconizing process. Electr Eng Jpn 2015; 193: 18–29.

48.

Borgenstam

Engström

Höglund

, et al. DICTRA, a tool for simulation of diffusional transformations in alloys. J Phase Equilib 2000; 21: 269–280.

49.

Ågren

. Thermodynamics and diffusion coupling in alloys - application-driven science. Metall Mater Trans A Phys Metall Mater Sci 2012; 43: 3453–3461.

50.

Barros

Schneider

Houbaert

. Assessment of Si and Al diffusion for the production of high Si and high Si–Al electrical steel. J Magn Magn Mater 2008; 320: e389–e392.

51.

Rabkin

Straumal

Semenov

, et al. The influence of an ordering transition on the interdiffusion in Fe-Si alloys. Acta Metall Mater 1995; 43: 3075–3083.

52.

Maitra

Gupta

. Intermetallic compound formation in Fe–Al–Si ternary system: part II. Mater Charact 2002; 49: 293–311.

53.

Xia

Zhao

Tan

, et al. Effect of Si content on the interfacial reactions in laser welded-brazed Al/steel dissimilar butted joint. J Mater Process Technol 2018; 258: 9–21.

54.

Krishna

. Diffusing uphill with James Clerk Maxwell and Josef Stefan. Curr Opin Chem Eng 2016; 12: 106–119.

55.

Colangeli

Giberti

Vernia

. Uphill diffusions in single and multi-species systems. J Phys A Math Theor 2023; 56: 393001.

56.

Křivská

Šlapáková

Veselý

, et al. Intermetallic phases identification and diffusion simulation in twin-roll cast al-fe clad sheet. Materials 2021; 14: 7771.

57.

Barros

Houbaert

Malengier

, et al. Modeling silicon and aluminum diffusion in electrical steel. J Phase Equilibria Diffus 2005; 26: 417–422.

58.

Machado

Kasama

Jorge

, et al. Evolution of the texture of spray-formed Fe–6.5 wt.% Si–1.0 wt.% Al alloy during warm-rolling. Mater Sci Eng A 2007; 449–451: 854–857.

59.

Mehdi

Hilinski

, et al. Non-oriented electrical steel with core losses comparable to grain-oriented electrical steel. J Magn Magn Mater 2019; 491: 165597.