NUCLEATION DURING PRIMARY RECRYSTALLIZATION OF RGO ELECTRICAL STEEL SHEET OBSERVED BY THE EBSP-METHOD

RGO electrical steel with low power losses and high permeability to be used in power transformers obtains its superior magnetic properties by a sharp Gosstexture developed by secondary recrystallization. RGO is produced by a two stage cold rolling process with intermediate annealing and subsequent primary recrystallization. By ODF analysis a high Goss intensity after primary reerystallization was empirically proved to be advantageous for the development of a sharp final Gosstexture in the following secondary recrystallization. This, in turn, requires a better understanding of how to achieve the beneficial primary Goss-texture formed by the basic processes of nucleation and growth. In the-present study the early state of recrystallization was investigated. Only the EBSP-method (lectron_.back_ scattering ]]attern), recently developed by Dingley, provides a sufficient submieron spatial resolution to measure the orientations of the early nuclei together with their cold worked and recovered environment.

1. INTRODUCTION Except for "in situ"-and for "dynamical" recrystallization, the recrystallization of cold worked steels leads to textures being completely different from the preeeeding cold worked texture/1-3/. Mainly two different models aim to describe the development of this recrystallization texture, called "oriented nucleation" and "selective growth". The model of "oriented nucleation" is supported by results where, after the onset of recrystallization, nuclei with orientations being typical for the recrystallization texture were found very dose to so called "transition bands" or "shear bands"/1, 2/. On the other hand experiments on deformed Fe3%Si single crystals being artificially nucleated at one end demonstrated the mechanism of growth selection between the variously oriented nuclei and for the fastest growing nucleus an orientation relationship to the deformed matrix characterized by a 27* < 110> rotation.
In the following an experiment is presented which resulted in proving the selective growth mechanism for primary recrystallization in grain oriented silicon steel. Starting from cold worked sheet an early state of reerystallization was prepared by quenching after short time heating. Single grain orientations were measured in the giving a Gauss-type scattering to each measuring point, a continuous "nucleation texture" eodd be calculated, which was compared to the deformation texture and to that texture having developed at completion of reerystallization.
2. EXPERIMENTAL 2.1. Specimen preparation The material used for the present investigation contained the following amounts of alloying elements in weight percent: Si: 3,16% / C: 0,029% /Mn:0,061% / S: 0,02% / P: 0,008% / N: 0,002% / + traces. The hot rolled strip of 2 mm thickness is "hot strip annealed" in a dry gas atmosphere at 10300C for 150 s and then waterquenched and pickled. This is followed by a first cold rolling to the intermediate thickness of 0.65 mm, by a recrystallization annealing at 9800C for 180 s and by second cold rolling to the final thickness of 0.26 ram. There are three types of materials on which texture investigations were carried out: a) Material "coldroll"is the starting material obtained by the above procedure. b) Material ')ullrec" is obtained from "coldroll"by heat treatment at 720"C in saltbath for 180 s which leads to complete recrystallizafion. c) Material "nuc" is obtained from "coldroll" by heat treatment at 720"C but for the much shorter time of about 2 to 4 s in order to generate nuclei.
On the transverse sections (defined by sheet plane normal and roiling direction) of the annealed specimens a Nital etching was performed in order to evaluate the recrystallized portion in the SEM. For further investigations specimens having only few nuclei were selected, whereas those without any nuclei or with too many recrystallized grains already being in contact with each other were disposed of. Transverse sections were polished to a very fine grid of 0.5 p,m and additionally sputtered with neutral argon atoms in order to avoid any structure on the surface. In this way accurately fiat surfaces as required for EBSP-analysis were obtained. For materials "coldroll" and 'zllrec", also fiat specimens for X-ray texture measurements were cut out of the sheets.
At specimen "nuc" local single orientations were determined in the SEM, applying the EBSP-teclmique recently developed by Venables and Dingley /7, 8/. 203 orientation were measured for individual isolated nuclei distributed over the cold worked matrix. From the so obtained individual orientations a continuous ODF was calculated by superpositioning of the Gaussian scattering functions one for each measured orientation with a half width of 10 providing even and odd coefficients for true ODF calculation. All ODF's are plotted in ql constant sections through the Euler angle space {tPl, ,tP2}.

The EBSP-method
Here in a modified SEM (JEOL 840) a stationary electron beam hits the highly tilted (70*) specimen sketching the back,scattered electron image on a transparent phosphor screen. This back_scattering electron image consists of sets of crossing lines due to the chatmeling contrast/7/. This image is detectable through a lead glass window by a low light TV-camera. After contrast improvement the backscattered image is displayed on a TV-monitor. The orientation of specific local areas can now be determined by positioning a cursor on each of two different zone axes of the diffraction pattern on the monitor in order to calculate the orientation by a microcomputer/8/. Prior to orientation determination the investigation area was selected applying a special backscattering electron detector with a low aperture being highly sensitive for orientation contrast while the SEM is operating in the scanning mode. This orientation contrast image is rather weak due to a long working distance at a 70* tilted specimen stage. Therefore a 20 times integration and storing of the backscattering electron image at an image analyzer was used to significantly improve the contrast. After switching the SEM to spot mode the electron beam could be moved to different places inside the stored image, a stably operating SEM provided. Being controlled by the image analyzer computer the beam positioning was discrete. At a model material the spatial resolution could be proved to be better than 0,3 Im by crossing a grain boundary. Thus the local textures of very small nuclei and even of subgrain areas in a deformed matrix could be investigated 19/but will be reported elsewhere/10/. Because of this high spatial resolution sometimes it was not easy to distinguish between cold worked and recrystallized matrix from stationary pattern on the EBSP monitor. Therefore an infinitesimal beam movement was performed by shifting the cursor around the aimed position. A changing of the monitor image pattern indicates deformed matrix whereas stationary pattern ensures a recrystallized nucleus to be observed.  Fig. 3 presents the ODF calculated from the single orientation data of 203 isolated nuclei being distributed over the specimens cross section. This nucleation texture, although otherwise very weak, has a maximum very close to {111} < 112 >, i.e. to the maximtma of the rolling texture located on the gamma-fibre (Fig. 1). The nucleation texture as a whole is completely different from the texture in Fig. 2 obtained after completion of primary recrystallization.

c) Environment of the nuclei
The early stage nuclei without exception wcrc situated along the grain boundaries. At no other possible nucleation site such as transition bands/12/or shear bands/2/ rccrystallization nuclei wcrc actually observed. (Also no "matching planes" reported to occur after the onset of secondary rccrystallization /13/ could bc found in the present investigation.) Between the nucleus' orientation and that of its direct environment always large angle orientation changes wcrc found, but no pronounced orientation relationships wcrc detectable. By scanning along the rolling and normal direction orientation changes of +_ 12 inside the deformed matrix grains could be observed.

DISCUSSION
The intensity maximum of the gamma-fibre of the nucleation texture (Fig.3) is very close to the major cold work texture (Fig.l) component and thus is assumed to bc formed by "in situ" rccrystallization. This is plausible also since the deformed matrix regions in ga_mma-fibrc orientations (i.e. near < 111 > ND) possess the highest amount of stored energy. Since all orientations contributing to the nucleation tcxturc arc measured on rccrystallizcd grains isolated in the cold worked matrix and of less than 1 p.m diameter, growth selection processes at that early state can bc neglected. The < 111> ND oriented nuclei can thus bc interpreted as being evolved by oriented nucleation in retaining their former orientation. The other nuclei arc almost randomly distributed and some weak maxima arc probably duc to fluctuations resulting from limited statistics. In particular the weak maxima near cubc and Goss must bc decreased in their levels because of their high multiplicity duc to their symmetry/14/. Thus the nucleation texture can bc interpreted as a largely random texture resulting from heterogeneous nucleation at the grain boundaries superimposed by an oriented nucleation duc to "in situ" rccrystallisation of orientations near {111} < 112 >. Since the final rccrystallization texture (Fig. 2) is very different from the nucleation texture, it cannot be formed by oriented nucleation. Thus a selective growth mechanism must bc responsible for the development of the rccrystalliztion texture. During this selection process the < 111> I I ND components being the maxima of the nucleation texture decrease while those with < 100> I I RD increase2 During this process apparently Goss oriented nuclei arc preferred to grow into {111} < 112 > regions which is in agreement with the well known growth relationship 27-rotation around < 110 >/4/. Thcrcforc during rccrystallization the nuclei of the Goss-componcnt may grow stronger the higher the component { 111} < 112 > in the deformed state. coldroll $* 7-1.