Study on the recrystallization microstructure and texture evolution of a rolled 0.075 mm ultra-thin grain-oriented silicon steel

This paper investigated the recrystallization behavior of a rolled 0.075 mm ultra-thin grain-oriented silicon steel, and microstructure and texture evolutions were studied. At the early stage, dominant {hk0}<001> recrystallized grains were ascribed to the preferential {hk0}<001> nucleation, then their growth was reduced by texture pinning effect. In contrast, a few ‘random’ grains grew excessively to consume surrounding {hk0}<001> fine-grained matrix, meanwhile nuclei originating from initial non-Goss grains also showed size advantage and further grew. Despite of quasi-two-dimensional microstructure of ultra-thin sheet, the anisotropic surface energy did not show marked effect on grain growth difference. Those coarse grains, which were 10 times larger than the matrix grains, were surrounded by high energy (HE) boundaries, indicating that their growth was driven by high grain boundary mobility. With the increasing annealing time, the formation of {hk0}<001> oriented assemblies and the growth of non-{hk0}<001> oriented grains contributed to the overall grain size increase. Regarding the relationship between microstructure, texture and magnetic properties, {hk0}<001> texture was weakened and reduced magnetic induction accordingly, and the combination of growing while heterogeneous grain size and changing texture made iron loss a first decrease and then increase tendency. It is essential to control annealing time to obtain suitable combination of microstructure and texture.


Introduction
Ultra-thin grain-oriented silicon steel (thickness ≤ 0.10 mm) is an important soft magnetic material for its application in intermediate and high frequency devices [1][2].The superior magnetic induction can be pursued by improving {hk0}<001> recrystallization texture, and the iron loss is affected by both texture and microstructure [3].To optimize the magnetic properties of ultra-thin grain-oriented silicon steel, it is essential to control the recrystallization microstructure and texture.
The production of ultra-thin grain-oriented silicon steel usually applies cold rolling and annealing processes on commercial grain-oriented silicon steel sheets.Much of the researches, including our previous work, have examined the deformation and recrystallization behaviors of ultra-thin grainoriented silicon steel [4][5][6].Considering the sharp Goss texture in raw material, it is suggested that the recrystallization mainly occurs on {111}<112> deformed matrix and develops strong {hk0}<001> recrystallization texture.However, regarding the grain growth process in this ultra-thin silicon steel, especially the heterogeneous microstructure evolution, much is still unclear.Moreover, the relationship between microstructure and magnetic properties has not been clarified.
In this study, a commercial glassless grain-oriented silicon steel was cold rolled and annealed, then the whole nucleation and grain growth process was systematically investigated.Special attention was paid to understanding the inhomogeneous microstructure evolution and corresponding texture as well as magnetic properties change.The results will help to design a more suitable procedure to produce ultra-thin silicon steel.

Experimental methods
The starting material was a 0.337 mm glassless grain-oriented silicon steel with strong Goss texture.The steel sheets were cold-rolled to 0.075 mm by multiple rolling passes, then the ultra-thin sheets were cut to pieces of 30 mm (along transverse direction, TD) × 300 mm (along rolling direction, RD) and were annealed at 800-900 °C for different periods.In the end, the magnetic induction (B8) and iron loss (P1.5/400) of fully recrystallized ultra-thin sheets were evaluated by a single sheet tester.
For simultaneous microstructure and texture analysis, electron backscattering diffraction (EBSD) data were collected in rolling plane of annealed silicon steel sheets using a Zeiss GeminiSEM500 field emission scanning electron microscope (SEM) equipped with EBSD system.Lab-Based X-ray Diffraction Contrast Tomography (Lab-based DCT) was applied to observe full three-dimensional microstructure characteristic and neighboring grains' distribution.To avoid statistical error, the EBSD measurements for each condition were repeated in ~10 samples and typical results were shown in this paper.All the Lab-based DCT measurement regions have dimensions of 4 mm (RD) × 2 mm (TD) × 0.08 mm (rolling direction, ND).

Magnetic properties
Figure1 gives the magnetic properties of annealed 0.075 mm silicon steel sheets.It is found that the sample annealed at 850 °C for 20 min manifests optimum properties, namely a fairly high magnetic induction of B8≥1.85 T in figure 1(a) and a low iron loss of P1.5/400<12 W/kg in figure 1(b).When the annealing temperature is set to 900 °C, a rapid decrease of magnetic induction is depicted, meanwhile the iron loss increases highly.Although the iron loss may reach a lower value between 10 min and 15 min, the magnetic induction can be deduced to be much lower than the samples annealed at 850 °C.Compared with annealing at 850 °C, high magnetic induction and low iron loss cannot be realized at the same heating time at 800 and 820 °C.Therefore, we trace the microstructure and texture evolution at 850 °C, and aim to find a suitable processing route for magnetic properties optimization.Besides, the iron loss shows a first decrease and then increase trend at this annealing temperature, allowing more comprehensive analysis.

Microstructure and texture evolutions
The magnetic properties of silicon steel are highly affected by the microstructure and texture.In figure 2, the orientation maps are colored by specific texture components, and the colors of these texture components are listed below the maps correspondingly.A predominant {hk0}<001> recrystallization texture can be observed in every sample.As discussed in our earlier work, when the rolled ultra-thin grain-oriented silicon steel is subjected to annealing, preferential {hk0}<001> nucleation occurs either inside shear bands or at in-grain deformation-induced boundaries, and the preferential {hk0}<001> nucleation leads to strong {hk0}<001> recrystallized texture [6].In particular, when no marked advantageous grain growth occurs and the grain size is fairly uniform, as depicted in figure 2(a-b), the area fraction of {hk0}<001> grains are overwhelmingly high.The varied preferential {110}<001> or {210}<001> texture components in different regions are related to the deviation angle of initial grain orientation from exact {110}<001> in raw material.With the elongation of annealing time, grain growth behaviors can be mainly divided into two types: The dominant {hk0}<001> recrystallized grains meet each other in short time, a large number of low angle boundaries of low energy appear accordingly and inevitably decrease driving force for these grains' growth.That is, the growth of {hk0}<001> grains are inhibited by texture pinning effect, and a multitude of {hk0}<001> fine grains form large-sized {hk0}<001> assemblies.In contrast, a few 'random' grains grow excessively to consume fine {hk0}<001> matrix, as shown in figure 2(c).
As annealing periods are added to 25 min and 30 min, more grains manifest obvious growth advantage.Among these grown grains, non-{hk0}<001> grains occupy the majority, so beneficial {hk0}<001> texture is further weakened correspondingly.In figure 3(a), {hk0}<001> grain size differ in the upper and lower regions, this can be interpreted by distinct nucleation rates in two regions.The nucleation rate of more accurate {hk0}<001> grains in the lower region is much higher, leading to stronger pinning effect and smaller {hk0}<001> grain size.The growing abilities of non-{hk0}<001> grains are stronger in the areas with sharper {hk0}<001> texture.It needs to note that {hk0}<001> grain growth is inhibited, while these small grains cluster to the assembly having extremely large size.To elucidate the growth behaviors of different oriented grains, both {hk0}<001> assemblies and sub-grains, as well as other oriented grains are taken into consideration.It needs to note that, grain distribution characteristic is quite dependent on the measured area, so the EBSD measurements were repeated in several samples, then average grain size values are calculated and given in figure 4. In figure 4, all types of grains show size increase with the elongation of annealing time, while the formation of {hk0}<001> recrystallization texture in the beginning results in nonuniform boundary mobility thereby invalidating grain growth theory to some extent.The growing rate of {hk0}<001> sub-grains, with the angle tolerance of 2°, is rather low due to texture pinning effect.In comparison, the sizes of other oriented grains and {hk0}<001> assemblies show similar increasing trend, and they possess much higher speed than inhibited {hk0}<001> sub-grains.The formation of {hk0}<001> oriented assemblies and the growth of non-{hk0}<001> oriented grains contribute to the overall grain size increase.However, with the increase of annealing time, the growing rate of {hk0}<001> assembly is reduced and turns to be lower than that of non-{hk0}<001> grains.Lab-based DCT technique can provide full three-dimensional grain information in large measuring volumes.Figure 5 display the grain maps of samples annealed for 10 min, 20 min and 30 min respectively.The maps of each sample are colored by IPF with respect to ND and RD.Similar to that in EBSD data, a fairly uniform fine microstructure is obtained after annealing for 10 min.To combine IPF ND and IPF RD grain maps, {110}<001>, {100}<001> orientation preference are observed in different regions, corresponding to respective recrystallization characteristic on different oriented deformed substrate [6].The occurrence of γ-fiber grains suggests the existence of initial grain boundaries in raw material, in other words, figure 5(a) displays the recrystallized microstructure of initial two coarse grains.When annealing time is 20 min, {hk0}<001> oriented fine grains are found to be consumed by several other oriented grains in figure 5(b).The measured region is assumed to originate from one coarse initial grain.The orientations of fine recrystallized grains are between {110}<001> and {210}<001>, and the grown grains are shown to have crystal orientations of deviated Goss, γ-fiber, λ-fiber, and other non-{hk0}<001>.For the sample annealed for 30 min, the measured region in figure 5(c) originates from initial two coarse grains.{110}<001> fine grains are overwhelmingly dominant in the lower region, while no specific preferred orientation is observed among the relatively large grains in the upper region, corresponding to the nucleation behavior of initial non-Goss grains.The fraction of non-Goss grains in raw material is relatively low, while its effect cannot be neglected.Compared with fairly uniform microstructure in the upper region, several grains grow excessively among fine {110}<001> grains.Meanwhile, larger grains originating from initial non-Goss grains show faster growing rate and cross initial grain boundary to swallow those fine grains in the lower region.These grains' superiority in growth shows the effect of size advantage forming in earlier stage, and the requirement of orientation gradient for higher boundary mobility is met.There are still quite a few small grains not been consumed by the growing large grains, at the lower left corner and along initial grain boundary.
To sum up, based on EBSD and lab-based DCT data, the recrystallization behaviors stemming from different oriented initial grains in raw material are analyzed.It is shown that the recrystallization process of ultra-thin grain-oriented silicon steel is mainly determined by the predominant advantage of {hk0}<001> grains at earlier stages, then {hk0}<001> grains are gradually consumed by grown other oriented grains or form large assemblies during grain growth process.

Driving force for grain growth
It has long been recognized that in ultra-thin sheets, grain growth of the texture component showing lower surface energy than neighboring grains is favored.For BCC structured silicon steel, {110} crystalline plane is believed to have relatively lower surface energy.However, in the EBSD orientation maps and Lab-based DCT grain maps, no obvious preferential orientation can be found in the grown grains.This implies that the anisotropic surface energy does not exert marked effect on grain growth difference in this study.In particular, {110}<001> grains' growth are highly inhibited by texture pinning effect.
In table 1, we analyze the grain boundary frequency with different misorientation in A20, A25 and A30, representing annealed samples at 850 °C for 20 min, 25 min and 30 min respectively.The boundaries surrounding those coarse grains with diameters of 10 times greater than the matrix are paid special attention.For A20, the majority of surrounding boundaries of grown grains are 20°-45° high energy (HE) boundaries, and half of these HE boundaries are 35°-45° boundaries.In comparison, when considering all grains, the frequencies of HE boundaries and 35°-45° boundaries decrease to 46% and 21.6% respectively.For A25 and A30, the frequencies of 35°-45° boundaries are much higher for those grown grains, reaching 48.2% and 50.4%.Therefore, it is deduced that those coarse grains' growth is driven by high grain boundary mobility.In addition, the frequency of >45° grain boundary also increase to a high level in A25 and A30, and this is not helpful for further grain growth, while the reason behind it has not been clear yet.To conclude, despite of quasi-two-dimensional microstructure of ultra-thin sheet, the driving pressure for grain growth does not come from the orientation dependence of the surface energy, while mainly connected to high grain boundary mobility.

Relationship between microstructure and magnetic properties
Ultra-thin grain-oriented silicon steel attracts much attention for its low iron loss under intermediate and high frequency applications.In this work, more and more grown grains deviate away from preferred {hk0}<001> orientation with increasing annealing time, thus the weakening of {hk0}<001> recrystallization texture leads to the decrease of magnetic induction.Regarding the iron loss, the microstructure and texture exert different influence on each iron loss component, namely hysteresis loss, eddy current loss and residual loss, and the combined effect determines the first decrease and then increase tendency of iron loss.
It is generally acknowledged that growing grain size could reduce the hysteresis loss.The sum of eddy current loss and residual loss is proportional to the magnetic domain wall spacing, which is directly connected to grain size, so the increasing grain size is unfavorable for the eddy current loss and residual loss.During the annealing periods between 10 min and 20 min, the reduced iron loss is attributed to the increasing grain size, and this phenomenon indicates that the decrease of hysteresis loss exerts main effect; Between 20 min and 30 min, the assembly size demonstrates a reduced growth rate, while non-{hk0}<001> oriented grains grows more rapidly.In consequence, {hk0}<001> texture is weakened at a higher speed, causing the hysteresis loss less reduce or even increase.Meanwhile the eddy current loss and residual loss further increase, thus the whole iron loss increases accordingly.It needs to note that, in terms of the advantageous effect of grain size on hysteresis loss, average grain size is considered in this paper.However, the effect of highly heterogeneous microstructure, as well as commonly existing low-angle misorientation boundaries inside the {hk0}<001> oriented assemblies, may have complicated influences, such as resulting in uneven magnetic properties, this needs to be studied in the future.Considering both the magnetic induction and iron change with increasing annealing time, it is essential to control annealing time to obtain suitable combination of microstructure and texture.

Conclusions
1) EBSD and lab-based DCT data illustrate the recrystallization behaviors of ultra-thin grain-oriented silicon steel.At the early stage, dominant {hk0}<001> recrystallized grains are ascribed to the preferential {hk0}<001> nucleation, then their growth is reduced by texture pinning effect.In contrast, a few 'random' grains grow excessively to consume surrounding{hk0}<001> fine-grained matrix, meanwhile nuclei originating from initial non-Goss grains show size advantage and further grow.The formation of {hk0}<001> oriented assemblies and the growth of non-{hk0}<001> oriented grains contribute to the overall grain size increase.2) Despite of quasi-two-dimensional microstructure of ultra-thin sheet, the anisotropic surface energy not exert marked effect on grain growth difference.Those coarse grains 10 times larger than the matrix are surrounded by HE boundaries, and their growth is driven by high grain boundary mobility.
3) With the increase of annealing time, the weakening of {hk0}<001> texture reduces magnetic induction accordingly.Regarding the iron loss, the combination of growing while heterogeneous grain size and deteriorating texture makes it a first decrease and then increase tendency.It is essential to control annealing time to obtain suitable combination of microstructure and texture.

Figure 1 .
Figure 1.Magnetic properties of 0.075 mm silicon steel sheets under different annealing methods.(a) Magnetic induction; (b) iron loss.

Figure 4 .
Figure 4.The size changes of various oriented grains with increasing annealing time.

Figure 5 .
Figure 5. 3D grain maps colored by IPF ND and RD of the annealed ultra-thin silicon steel sheets by Lab-based DCT.

Table 1 .
Grain boundary distribution of annealed ultra-thin silicon steel sheets.