Effects by the microstructure after hot and cold rolling on the texture and grain size after final annealing of ferritic non-oriented FeSi electrical steel

The magnetic properties of fully processed non-oriented FeSi electrical steel are characterized by their magnetization behavior and specific magnetic losses. The magnetic properties are determined by texture and microstructure. Less gamma fiber intensity and a high intensity of preferable texture components, especially cube fiber texture, are desirable to obtain an excellent magnetizing behavior. Furthermore, large grain sizes are necessary to reach low values of the specific magnetic losses. This is in contrast to conventional steels, where for high values of the mechanical strength high values of the gamma texture and very low grain sizes are necessary.

The fabrication route of the fully processed non-oriented electrical steel comprises a heavy cold rolling of the hot rolled material before final annealing. This is necessary to obtain the desired final thickness. During annealing of the cold rolled material at first the softening process (recovery, recrystallization) is going on before grain growth is becoming the main process. To fulfill the requirements on large grain size for low loss materials, grain growth plays an important role. The inhomogeneous deformation structure after cold rolling across the thickness of the material as well as temperature gradients during annealing lead to rather inhomogeneous structural changes across the thickness at final annealing.

Not so much investigation was done of the effect of the different processing steps during fabrication of fully processed non- oriented electrical steel: i.e. the effect of hot rolling, hot band annealing (optional), cold rolling and final annealing on the evolution of the microstructure and magnetic texture.^1–8 This includes partly also new technologies for hot band fabrication. The experimental studies prove that there is generally an interaction between the microstructure and texture in the various processing steps.^1,2,5

In this paper we will analyze by optical microscopy and EBSD the influence of different microstructures of the hot strip and the resulting microstructure after cold rolling on the appearance of recrystallization and grain growth after final annealing and the resulting texture and microstructure of ferritic FeSi steel. Different microstructures of the investigated hot strips lead to different deformation structures after cold rolling. This results finally in a different evolution of the microstructure and texture with annealing temperature and time.

Image quality plus rotation angle map and the frequency distribution of misorientation angle are used to access the degree of microstructural changes during processing. The image of the texture is characterized by the orientation distribution functions (ODF).

Results presented in this paper concentrate on selected hot rolled samples of FeSi2.4 with a thickness of 2 mm. Different hot rolled materials were prepared using the four stand pilot line for hot rolling at the Institute of Metal Forming, TU Bergakademie Freiberg (IMF). Thereby, the hot rolling finishing temperature as well as the conditions of coiling after finishing the hot rolling were varied.

In this paper hot rolled material prepared in different way was examined. Sample 1 has a typical microstructure obtained immediately after finishing the hot rolling by rapid cooling after the last pass. Finishing temperature at hot rolling was T _F = 820 $°$C.

The microstructure of sample 2 is obtained by an annealing following immediately after the last pass of the hot rolling process. Temperature after the last pass (T _F) was 860 $°$C. The finishing temperature of sample 3 was lower than for sample 1 (T _F = 770 $°$C). After the hot rolling was finished the sample was cooled in air. Part of the sample 3 undergoes a hot band annealing (850 $°$C, 20 min.) in a separate process step before cold rolling. The sample is named sample 4.

Cold rolling (CR) of the hot rolled band with a thickness of 2 mm to a thickness of 0.5 mm was realized on a two-high stand at the IMF. Final annealing was carried out in an annealing furnace where it is possible to simulate the mostly used commercial continuously annealing process at temperatures up to 1000 $°$C as well as a box annealing (lower temperatures and longer annealing times). The annealing conditions are chosen similar to the typical applied conditions for ferritic non-oriented FeSi steel, see Ref. 9. The specific annealing conditions are given in connection with the presented results.

Microstructure before and after cold rolling was observed by optical metallography. The resulting deformation structure will be described following the classification given in Ref. 10. EBSD measurements were done to obtain the image quality (IQ) plus the rotation angle map, which give the frequency distribution of misorientation angle for small angle boundaries (2$°$ to 15$°$⁠) and large angle boundaries (≥15$°$⁠). In addition, the frequency distribution of misorientation angles was considered. These quantities are used to characterize the microstructure after cold rolling and the processes of the ongoing softening occurring during final annealing. Thereby, we will analyse the appearance of the various softening processes: recovery, in-situ recrystallization, recrystallization and finally grain growth, as described and defined in Refs. 11 and 12. Recovery comprises cancellation and rearrangement of dislocations, which may lead to the formation of sub grains. Recrystallization is characterized by the formation and motion of new high angle boundaries and the typical deformation texture disappears. At higher annealing temperatures, respectively longer annealing time, grain growth occurs, which is defined as grain coarsening after completed recrystallization (stress free polycrystalline microstructure).

The φ₂=45$°$ section of the orientation distribution functions (ODFs), see Fig. 1, is used to characterize the intensity of the significant texture components after rolling and annealing. This comprises (see Figure 1) α- and γ – fibers as well as the α^* - fiber. The most relevant magnetic texture components and fibers are: {100} <001> cube and {100} <110> rotated cube, {110} <001> Goss, and the texture fiber theta ({100} <uvw>). For the assessment of different magnetic texture components in the φ₂=45$°$ section on the magnetic properties see Ref. 13. The EBSD measurements were done using a field emmision scanning electron microscope of Zeiss. The software was TSL OIM7 from AMETEK.

To characterize the through thickness variation, the ODF section φ₂=45$°$⁠, IQ and misorientation obtained from EBSD measurements were also analyzed for different layers across the thickness. To this purpose the thickness was divided in 6 layers (1/6 surface region, 2/6 transition section, 3/6 region near the center). Grain size d after final annealing was determined by the linear intercept method.

The microstructure of the hot band depends sensitively on the parameters of the fabrication of the hot band: preheating temperature, temperature at pre-hot rolling, starting and finishing temperature of final hot rolling and the resulting temperature-time at coiling and cooling. The different processing conditions during fabrication of the hot band result in different softening processes (recovery, recrystallization). In addition, the gradients of the stress as well as of temperature at rolling give rise to an inhomogeneous microstructure at hot rolling. This holds for the conventional route of hot rolling as well as for new innovative techniques.^6,7 We selected typical examples of hot bands (sample 1 to 3) for our investigations, which were prepared following the conventional route, see Figure 2. In addition, we also examined one sample, (sample 4), which experienced an additional hot band annealing before cold rolling. As can be seen from Figure 2, an appropriate thermal treatment after the last pass (sample 2) as well as after a separate thermal annealing before cold rolling (sample 4) may result in a partly recrystallized microstructure.

Such an additional hot band annealing in a separate processing step, like for sample 4, is used commercially for the fabrication of high permeability grades, see for instance.^14,15 In our case the hot band annealing was done at a temperature below 900 $°$C, which results in an incomplete recrystallized microstructure. Commercially hot band annealing at higher temperatures, above 900 $°$C, results in a more or less perfect recrystallized microstructure. The effect of the grain size in this case on the resulting texture during final annealing was studied in Refs. 16 and 17. However, the ideal microstructure after complete recrystallization during hot band annealing is quite different from that in a real fabrication process for hot band, which exhibits a complex microstructure and results finally in inhomogeneous deformation substructures after cold rolling across the thickness, as demonstrated in Figure 3. Fig. 3 illustrates that the microstructure of the hot band effects remarkable the resulting microstructure after cold rolling. The microstructure after cold rolling may be classified according to Ref. 10. After the cold rolling sample 1 and sample 3 exhibit pronounced well-defined deformation bands, which are small. Some deformation bands show deformation substructures. The optical micrographs after cold rolling of the hot rolled sample 2 and 4 on the other hand manifest deformation bands with remarkable non-uniform straining inside the bands. One observes strain locations, especially in form of shear bands. It should be noted that we observed deformation bands with remarkable shear bands inside the bands for sample 2, which was thermal treated immediately after the last pass at hot rolling, as well as for sample 4, which was thermal treated in a separate processing step before cold rolling.

We have observed shear bands across the thickness only for the hot bands with partly recrystallized microstructure or completely recrystallized microstructure. Shear bands occur specifically for hard cold rolling.¹⁰ In the case that the work hardening is reduced, a concentration of strain in form of strain location may appear. Such a situation seems to appear in our case. Regarding different samples we found¹⁸ that the intensity of shear bands depends on the degree of recrystallization in the hot band, respectively the details of the obtained deformation substructures. The intensity of the shear bands as well as the angle to the rolling direction depend also on the degree of deformation at cold rolling. As noticed in Ref. 10, the shear bands may be the preferable sites for the onset of recrystallization and may give rise to a higher intensity of cube texture at annealing of the cold rolled material.¹⁹ 

The texture after cold rolling is characterized by large intensity of the α - fiber as well as by a remarkable intensity of the γ - fiber, see Figure 4. These are generally the typical texture components at rolling for in-plane strain deformation, see Ref. 20.

Fig. 5 gives the ODF section φ₂=45$°$ (total cross section) of sample 1 and sample 2 after annealing for 20 s at 700 $°$C, 800 $°$C, 920 $°$C and 1000 $°$C. Annealing was done with a heating rate of 5 to 7 $°$C/s and a cooling rate of 11 to 13 $°$C/s. The conditions for the annealing are similar to the commercial practice of continuous final annealing for non-oriented electrical steel. Evolution of the texture may be examined in connection with the evolution of the microstructure at final annealing. Figure 6 gives the frequency distribution of misorientation angle for sample 1 and 2 after annealing at 700 $°$C, 800 $°$C, 920 $°$C and 1000 $°$C for 20 s (total cross section). The Figure indicates that an annealing temperature of 700 $°$C is not sufficient to initiate remarkable structural changes by recrystallization, which is characterized by formation and motion of new high angle grain boundaries. On the other hand, an increase of the intensity of low angle boundaries compared to the cold rolled state is observed. This indicates recovery, which is characterized by the formation of sub grain structures with low angle grain boundaries.^11,12 This recovery is more pronounced for sample 2. Annealing at higher temperatures results already in a different image for the microstructure. When annealing the sample 2 at 800 $°$C for 20 s the recrystallization is going on. The frequency distribution of misorientation angle changes. Evidence for the occurrence of extensive recrystallization is the disappearance of the low misorientation angle, respectively the formation and motion of new high angle boundaries. For sample 1 a large intensity for small misorientation angles is still observed at annealing at 800 $°$C for 20 s. This may point to the fact that extensive recovery of sample 1, which exhibits more well-defined deformation bands compared to sample 2, appears at first at higher temperatures. At annealing at 1000 $°$C for 20 s we have for both samples essentially large misorientation angles.

The resulting image for the grain structure can be seen from the Figure 7. Figure 7 shows the image quality (IQ) plus the rotation angle map for sample 1 and 2 after annealing at 800 $°$C and 900 $°$C for 20 s. As can be seen from the Figure 7, the recrystallization is more pronounced for sample 2 compared to sample 1 for annealing at 800 $°$C. More or less regular grains appear for sample 2. Large intensity of small misorientation angles in the center is obtained for sample 1 after annealing at 800 $°$C for 20 s, which indicate recovery. Even for annealing at 920 $°$C a relatively large intensity of small misorientation angles is found in the center for sample 1. The recrystallization seems to be completed for sample 2 already at 800 $°$C, while for sample 1 it needs higher annealing temperatures. Recrystallization itself is discontinuously (secondary recrystallization^11,12) with rather irregular grains. Grain growth, defined as grain coarsening after completed recrystallization (stress less polycrystalline microstructure^11,12)) appears for sample 2 at lower annealing temperatures compared to sample 1. Grain growth finally offers the possibility to reach the desired mean grain size larger than 100 μm for these FeSi electrical steels.

From the observed tendencies of IQ plus rotation angle map, see Figure 7, it can be seen that the ongoing processes, recovery and recrystallization, are different in different regions across the thickness up to the annealing temperature, where the recrystallization is completed. This is also reflected in the image for the texture in different regions across the thickness during annealing. Fig. 9 shows the ODF section φ₂=45$°$ of sample 1 after final annealing at 700 $°$C and 800 $°$C for 20 s in different regions: 1/6, 2/6, 3/6.

In the following we will analyze the evolution of texture for the whole cross section of the samples at annealing with increasing annealing temperature for 20 s, as demonstrated in Figure 5, in connection with the changes of the microstructure and the dominant ongoing processes, recovery and recrystallization and finally grain growth. Thereby, we will study the role played by the different microstructures after cold rolling.

As well-known,²⁰ cold rolling leads mainly to a strong α - fiber texture accompanied by a weaker γ - fiber texture for alloyed steels like FeSi. The image for the texture begins to modify with the onset of recrystallization. Straight recovery is characterized by no change of the texture.^11,12 As known from conventional steels the annealing texture at recrystallization typically reveals a lower intensity of the α - fiber texture and a large intensity of the γ - fiber.

For the FeSi samples the following image appears for the texture at annealing of the cold rolled material, see Figure 5. The textures for sample 1 annealed at 700 $°$C resembles those after cold rolling. The overall texture of sample 1 after annealing at 800 $°$C exhibits smaller intensities of the α - fiber texture compared to annealing at 700 $°$C. In addition, a large intensity of texture in the region of α^* - fiber occurs. The frequency distribution of misoriemtation angles gives no evidence of large recrystallization The image for the distribution of misorientation vs. misorientation angle indicates that recovery for annealing at 800 $°$C/20 s is still dominating, see Figure 6. A α^* - fiber texture, the texture component {411} <148> and its family orientations {h,1,1} <1/h,1,2>, is observed as recrystallization texture after heavy cold rolling of bcc – steels.^21,22

For sample 2 in contrast to sample 1 remarkable recrystallization occurs already after annealing at 800 $°$C for 20 s, as can be identified by the appearance of large grain boundary angles. This may explain that sample 2 exhibits a high intensity of the α - fiber texture for annealing at 700 $°$C, while the intensity of the γ - fiber intensity is high for annealing at 800 $°$C. In addition, Goss texture is observed after annealing at 800 $°$C. Cube texture becomes weaker at 800 $°$C compared to 700 $°$C. Rotated cube texture disappears after annealing at 800 $°$C. The shear bands in the cold rolled state of sample 2 may lead to the earlier onset of recrystallization compared to sample 1 and the occurrence of cube texture.¹⁹ No intensity of texture in the region of α^* - fiber like for sample 1 is found for sample 2 for these annealing temperatures, see Figure 5.

It should be remarked that the inhomogeneity of the deformation structure after cod rolling is also reflected in the inhomogeneity of the evolution of the texture across the thickness, see Fig. 8. The ongoing processes of softening (recovery, recrystallization) differ across the thickness. The further ongoing processes depend on the microstructure of the cold rolled state, which depend on the microstructure of the hot band. For that reason the tendencies for the texture over the whole cross section with the increasing annealing temperatures describe the average.

In the case of complete recrystallization across the thickness at higher annealing temperatures one expects grain growth. As can be seen from Figure 7, grain growth, defined as grain coarsening after completed recrystallization, appears for sample 2 at lower annealing temperatures compared to sample 1. A change of the image of texture at annealing at higher temperatures is connected with the onset of grain growth. For sample 2 one observes, see Figure 5, no longer a large intensity of the gamma fiber. Instead, high intensities in the region of α^* - fiber prevail. The texture nearby the Goss component becomes weaker at 920 $°$C compared to 800 $°$C. For annealing at 1000 $°$C a stronger Goss texture emerge for both samples. The Goss texture is larger for sample 2. In addition, a remarkable cube fiber texture appears for sample 2. With respect to the tendencies of the observed image for the evolution of texture with the annealing temperature we have to notice that there are of course fluctuations along the strip.

While at low annealing temperatures the evolution of texture may be driven by the stored deformation energy, the changes of the image of texture during grain growth may be caused by the dependence of the grain boundary energy and of the mobility on the texture of the grains.^11,12

The relevance of the knowledge on the evolution of the texture at grain growth for the fully processed non-oriented electrical steels is given by the fact that low values of the specific magnetic losses are only reached for large mean grain size. The demand for large grain size is quite different from those for low carbon steels. An enormous grain growth is necessary for the regarded FeSi steel to obtain grain sizes clearly above 100 μm. Figure 9 shows the mean grain size D as a function of the annealing temperature T_a (fixed annealing time of 20 s) for sample 1 and 2. One finds a different behavior of the resulting mean grain size as function of the annealing temperature for sample 1 and 2. The desired mean grain size larger than 120 μm appears for sample 2 at remarkable lower annealing temperatures compared to sample 1. The experimental observations demonstrate clearly the effect of the different deformation substructures after cold rolling for different hot band on the evolution of the grain size at grain growth during final annealing of ferritic FeSi steel. Different deformation structures after cold rolling at the same cold rolled deformation result due to the different microstructure of the hot band (different degree of recrystallization). Similar results were obtained for sample 3 and 4. Sample 4, which exhibit a partly recrystallized microstructure before cold rolling due to an additional thermal treatment before cold rolling, gives also a faster grain growth like sample 2.

The different evolution of the mean grain size with the annealing temperature for the sample 1 and 2 may open a way for a more energy efficient route to reach the desired grain size. A hot band process, which result in a partly recovered and recrystallized microstructure, seems to be an important factor for an optimum deformation substructure after cold rolling and finally for the recrystallization behavior and a fast grain growth at final annealing. Desired optimum microstructure of the hot band may be obtained already by an appropriate annealing after the last pass at hot rolling. In this way the additional processing step, hot band annealing before cold rolling, can be avoided. The preferred microstructure of the hot band also give rise to higher intensity of the preferred texture components.^23,24

In the case of an annealing at a given temperature with different annealing time one obtains a similar image for the evolution of the microstructure and texture as at annealing for a fixed annealing time and different annealing temperature. Figure 10 illustrates the evolution of the texture and grain size for sample 1 after annealing at 750 $°$C with different annealing times of 60 s, 300 s and 1200 s (heating rate 10 K/s; cooling rate 50 K/s). The chronology of the ongoing processes, recovery, recrystallization, grain growth, is like in the case of annealing with increasing annealing temperature.

The question arises, if there is a model by which it is possible to predict the resulting image for the texture at annealing. Regarding only the process of recrystallization, modelling of the texture evolution may be done using models taking into account uniform in-pane strain.²⁵ Thereby, it is assumed that the stored deformation energy determines the resulting recrystallization texture (Taylor model). One obtains a weaker α - fiber texture and large intensity of the γ - fiber, as observed experimentally. However, the appearance of cube fiber texture and Goss texture, as in our case for sample 2, remains unexplained. The deformation structure after cold rolling is not only characterized by uniform straining. Investigation of the effect of strain locations in the form of shear banding on the recrystallization texture proves that recrystallized grains from shear bands in FeSi were mainly Goss,^10,26 as observed for sample 2 in Figure 5.

Another mechanism described in the literature to explain the recrystallization texture is the SIBM (strain induced boundary migration). In this case, grains with lower stored energies grow into ones with higher stored energy, which results in Goss texture and texture near Goss.^10,26 SIBM may also appear as a mechanism in recrystallization parallel to nucleation of grains in regions of high stored energy (Taylor model). As can be seen from Figure 6 and 7, one observes across the cross section of the material the different processes, recovery, recrystallization and grain growth simultaneously at annealing. There is no model for the recrystallization texture, which considers the different types of deformation: uniform straining, strain partitioning and strain location (shear bands), which may appear simultaneously in the cold rolled material, and their effect on the resulting recrystallization. In addition, the kinetic of the different mechanism at recrystallization may depend on the annealing temperature. For all these reasons a prediction of the recrystallization texture at final annealing of heavy cold rolled non-oriented electrical steel fails.

Beside the process of recrystallization one has also to keep in mind that extensively grain growth is necessary to obtain the desired microstructure after final annealing. Thereby, grain growth describe grain coarsening after completed recrystallization (stress less polycrystalline microstructure). There is also no general model for the evolution of the texture at grain growth, which is determined by the grain boundary energy and mobility.^11,12 There is no quantification of the different factors, which determine the grain boundary motion of the regarded FeSi steels. Especially nothing is known on the anisotropy of the mobility and the grain boundary energy anisotropy, which effect the grain growth in different way.^26,27

The image of the experimental observed texture in our case at annealing temperatures, where grain growth is dominant, indicate a selective grain growth with respect to the texture. The intensity of the γ - fiber becomes weaker and the intensity in the region of α^* - fiber becomes stronger. This observation is in agreement with the results in Ref. 28. In addition, we observed a Goss texture and a cube fiber texture after annealing at 1000 $°$C for sample 2. The underlying mechanism remains unclear. This also concerns the fact that we observed a weak Goss component in sample 2 annealed at 920 $°$C, while it is strong after annealing at 800 $°$C and 1000 $°$C.

The microstructure of the hot band before cold rolling plays an important role for the kinetic of the ongoing structural changes at final annealing. It effects also the resulting intensity of the most relevant magnetic texture components.

The evolution of texture reflects the present ongoing softening processes, recovery, recrystallization and finally grain growth at the given annealing conditions.

For the FeSi steels it is indispensable to realize a large grain size at final annealing to get the desired microstructure for low values of the specific magnetic losses. The image of texture at recrystallization is remarkable different from the one during grain growth. Again the microstructure of the hot band gives a different image for the texture at grain growth.

At recrystallization, the intensity of the gamma fiber increases and the intensity of the alpha fiber decreases. At grain growth the intensity of the cube fiber changes and a strong intensity of texture in the region of α^{* -} fiber appears. The microstructure of the hot band before cold rolling also effects the kinetics of grain growth. Grain growth is faster for partly recrystallized microstructure of the hot band, which results in a high intensity of shear bands after cold rolling. The desired large values of the grain size, much larger than 100 μm, at grain growth during final annealing may be obtained in this case under energetically more appropriate conditions.

There is no model for the texture evolution for the studied FeSi steels at final annealing with defined parameters by which the resulting texture may be predicted. Deeper understanding of the relative influence of microstructure after hot and cold rolling and the resulting texture and grain size after final annealing may help to improve the properties of non-oriented ferritic electrical steel. This includes beside experimental investigation, the modelling of the evolution of texture as a function of the strain state and the resulting misoriented substructures after cold rolling.

The authors would like to thank Mr. Y. Houbaert, University Gent and Mr. H. Hermann, TUBAF for critical reading of the manuscript and comments.