Materials Transactions, Vol. 44, No. 6 (2003) pp. 1106 to 1115
#2003 The Japan Institute of Metals
Fine-Grained Doubly Oriented Silicon Steel Sheets
and Mechanism of Cube Texture Development*
Toshiro Tomida, Shigeo Uenoya and Naoyuki Sano
Corporate Research & Development Laboratories, Sumitomo Metal Industries, Ltd., Amagasaki 660-0891, Japan
An attempt has been made to obtain strongly cube-textured sheet steels with fine grains employing oxide-separator-induced
decarburization. The steels of the chemical compositions around Fe–3%Si–1%Mn–0.05%C were hot-rolled, twice cold-rolled with in-between
annealing into 0.35 mm thickness and then subjected to the lamination annealing with separators containing SiO2 , which promoted
decarburization. During the lamination annealing in a ferrite and austenite two phase region of about 1350 K, the sheet materials were
decarburized down to about 0.001% of carbon concentration. At the same time, the cube texture {100}h001i remarkably evolved in the columnar
ferrite grains which grew inward from sheet surfaces. Cube-oriented nuclei emerged during the primary recrystallization prior to
decarburization, and they selectively grew in the columnar grains. After complete decarburization, the sheet materials consisted of ferrite grains
of about 0.4 mm diameter, more than 90% of which were well aligned with the cube orientation. The doubly oriented steel sheets thus obtained
showed a large magnetic induction of 1.87 T at 800 A/m, a small core loss of 1.2 W/kg at 1.5 T and 60 Hz and a low magnetostriction of
2:5 106 at 1.9 T in both the rolling and transverse directions. It is likely that the cube texture component arises from a near-{410}h001i
component present after intermediate annealing and preferentially grows by surface energy difference. The magnetization processes are also
discussed.
(Received February 28, 2003; Accepted April 30, 2003)
Keywords: cube texture, decarburization, phase transformation, doubly oriented silicon steel, core loss, magnetostriction
1.
Introduction
Cube texture {100}h001i develops relatively with ease
upon annealing after rolling in fcc metals such as cupper,
aluminum and Fe–Ni.1) However, it hardly evolves in bcc
Fe–Si by ordinary rolling and annealing processes. This
metallurgical nature has thrown obstacles on the basic as well
as practical studies on cube texture, which have historically
captured the interests of many researchers because of its
symmetrical simplicity and the beneficial influence on
magnetic properties. This article describes a process for the
development of cube texture in Fe–3%Si–1%Mn sheets, the
mechanism of texture development and the magnetic properties of thus obtained materials.
Since the easiest principal directions for Fe–Si to magnetize are h100i directions, the development of cube texture
dramatically improves magnetic properties in two perpendicular directions in the plane of sheet materials, the rolling
(RD) and transverse (TD) directions. Therefore, the investigations relevant to the cube texture in Fe–Si have been
mostly made in relation with the development of magnetic
materials suitable for electrical transformers and other
electro-magnetic operations. Such materials are so called
doubly oriented electrical steel sheets, which are typically
0.3 mm thick and contain about 3% of silicon. Assumus et
al.2) have first pointed out that a high temperature annealing
of very thin Fe–3%Si sheets leads to a discontinuous or
secondary growth of cube-oriented grains. This phenomenon
was later proven to be a surface-energy-controlled process
occurring when {100} planes had a lower surface energy than
the average primary matrix3,4) However, since this selective
driving force by surface energy difference decreases with
increasing sheet thickness,5) it has been often found that the
*This
Paper was Originally Published in Japanese in J. Japan Inst. Metals
66 (2002) 950–957.
cube-oriented growth is rather sluggish in thick materials and
the increase in thickness for example to 0.3 mm results in
either an incomplete secondary growth or an extremely large
grain size more than 20 mm.
Taguchi and Sakakura6,7) have later found a different
method based on alternate cold-rolling in two perpendicular
directions (cross rolling), which provides Fe–3%Si sheets
with a similar coarse-grained structure. In this method,
silicon steel sheets containing AlN as a grain growth inhibitor
are cross-rolled after hot-rolling, decarburized and then
annealed at a high temperature for secondary recrystallization to occur. Except for the elaborate rolling in cross
directions, this method is nearly identical to what is used for
the commercial production of well-known Goss-textured
{110}h001i electrical steels. The driving force for this grain
growth is due to grain boundary energy, and cube-oriented
secondaries grow in a {632}h9 15 5i matrix.8) No other
method proposed so far has provided as high a degree of cube
texture to Fe–Si sheets as the abovementioned two methods.
During the extension of the work due to Assumus et al., the
drawbacks in magnetic properties of doubly oriented steels
were found.9,10) Whereas the magnetic inductions at relatively high fields in easy magnetizing directions were enough
large, magnetic core loss and magnetostriction were found to
be unexpectedly high if compared to those of Goss-textured
Fe–3%Si sheets. The large core loss has been related to the
large grain size over 20 mm,10) which affects eddy current
loss in the manner suggested by Pry and Bean,11) and related
also to the movement of 90 magnetic domain walls.12) The
high magnetostriction has been ascribed to the latter by
which the domains in transverse directions to applied fields
are created (or annihilated).10) It has been pointed out by
Littmann9) that doubly oriented materials show a larger
dependence of core loss on grain size than Goss-textured
materials, suggesting that significantly better properties could
be obtained by a large reduction in grain size.
Fine-Grained Doubly Oriented Silicon Steel Sheets and Mechanism of Cube Texture Development
In the previous report,13) the decarburization of silicon
steel sheets by SiO2 and the development of {100}h012i
texture have been described by one of the authors. As coldrolled Fe–3%Si–1.1%Mn–0.05%C sheet steels were laminated with separators containing SiO2 and then annealed
under a reduced pressure, decarburization occurred due to the
thermo-chemical reaction, 2C+SiO2 ! Si+2CO. At the
same time, {100}h012i texture developed to a high degree in
the columnar grains which grew inward from sheet
surfaces. While strongly textured, fully decarburized materials had a grain size of only about 0.6 mm. In this study,
therefore, an attempt has been made to obtain doubly oriented
silicon steel sheets with small grains by the oxide-separatorinduced decarburization. Prior to decarburization, two stages
of cold rolling were applied along with an in-between
annealing to the materials of which chemical compositions
were around Fe–3%Si–1%Mn–0.05%C. The influences of
cold rolling and intermediate annealing on the preferred
orientation of columnar grain growth and magnetic properties
of thus obtained materials were then investigated. It is shown
here that the preferred orientation lies in {100}h001i
orientation as an appropriate two-stage cold rolling is applied
prior to decarburization. This texture development leads to as
a high degree of cube texture as attained by secondary
recrystallization, a considerably finer grain size and better
magnetic properties. The mechanism of cube texture evolution and the magnetization processes in fine-grained doubly
oriented materials are discussed.
2.
Experimental Procedures
Ingots of vacuum melted steels of which compositions are
listed in Table 1 and similar to that used in the previous
study13) were processed to 3.0 mm thick plates by forging and
hot rolling. They were then pickled and cold-rolled to
intermediate thickness in the range between 0.6 and 1.5 mm.
After annealing for 30 s at temperatures ranging from 1073 to
1323 K, they were again cold-rolled to the final thickness of
0.35 mm. Rolling was all operated in a single direction; no
cross-rolling was applied.
The samples of 150 150 mm2 (sample A) or 180 500 mm2 (sample B) in dimensions cut from cold-rolled
sheets were laminated with oxide separators in the manner
described below. In the case of steel A, a sheet-form ceramic
material, which was about 0.5 mm in thickness and consisted
of about 3 mm diameter fibrous oxides (52 mass% SiO2 and
48 mass% Al2 O3 ), was used along with TiO2 powders
(Anatase type structure, 99% purity) as separators. The sheet
form material was prepared by a papermaking process. Prior
to the lamination, the sheet-form material was heat-treated in
the atmosphere at 1273 K for 1 h to remove an organic binder
that was necessary for the papermaking process. The TiO2
powder of an amount of 40 g/m2 was placed between two
pieces of the sheet-form material, and they were then together
Table 1
Chemical compositions of steels used (mass%).
C
Si
Mn
P
S
sol-Al
Steel A
0.049
2.97
1.0
0.0002
0.0005
0.0005
Steel B
0.063
2.80
1.3
0.0002
0.0009
0.0034
1107
inserted between the samples. This manner of lamination is
the same as that used in the previous study.13) The samples of
steel B were laminated with the sheet-form separators of
about 0.3 mm in thickness, which were directly made by the
papermaking process from the mixture of the above mentioned fibrous oxides, TiO2 powder and a small amount of
organic binder; the organic binder was not removed prior to
lamination in this case. The amounts of fibrous oxide, TiO2
powder and organic binder contained in this sheet-form
separator were about 40, 30 and 8 g/m2 , respectively.
The laminated samples were annealed in an = two phase
region of either 1348 or 1373 K under a vacuum of about
103 Pa as follows; and phases coexist at temperatures
ranging between 1000 and 1400 K in the present materials.13)
The samples of steel A were heated to 1348 K at a rate of 1 K/
min and kept for time periods up to 16 h. On the other hand,
the samples of steel B were first heated to 723 K at a rate of
1 K/min and kept for 5 h to remove the organic binder
contained in the separators. They were then heated to 1373 K
at a rate of 2 K/min and kept for 24 h. Some samples were
halfway cooled during the heating processes.
The samples at every stage of the processing described
above were examined in chemical composition, texture and
microstructures. Textures were measured by X-ray diffractometry with a cobalt or molybdenum target and EBSP
(Electron Back Scattering Pattern) coupled to a scanning
electron microscope. The measurements were made at
specimen surfaces and 20 to 30 mm below surfaces for Xray diffractometry and EBSP respectively, unless mentioned
otherwise. Microstructures were observed via optical microscopy and OIM (Orientation Imaging Microscopy) capability
of EBSP. The sample surfaces for optical microscopy were
polished and etched in Nital solution. Those for EBSP were
electro-polished after mechanical polishing.
The samples after decarburization, of which carbon
concentrations were reduced down to about 0.001%, were
examined in magnetic properties. The strips of 100 30 mm2 in dimensions were cut from the decarburized sheets
at various angles with respect to RD and stress-relieved at
1173 K for 2 h in a vacuum. The magnetic induction and core
loss of the strips were measured under alternating (sinusoidal) flux conditions at 50 or 60 Hz by a single strip tester
(SST) with horizontal double yokes made by TOEI Industry
Co., Japan. The magnetostriction of the same strips was
measured at 50 Hz by a laser displacement meter coupled to
the same type of SST (model TLVM made by TOEI Industry
Co., Japan).
3.
3.1
Experimental Results
Decarburization induced by oxide separators and
columnar grain growth of ferrite
The variation of chemical composition of steel A during
lamination annealing at 1348 K is shown as a function of
annealing time in Fig. 1. The decrease in carbon concentration is observed to occur proportionally to annealing time in
the early stages of annealing. Carbon concentration decreases
by about 50% for first 4 h, and little carbon remains after an
annealing for 16 h. At the same time, manganese concentration decreases, and silicon concentration slightly increases.
1108
T. Tomida, S. Uenoya and N. Sano
3.1
3
1
2.9
0.05
Mn
0.5
0.025
C
0
0
As-rolled
200
400
600
800
0
1000
Manganese Concentration (mass%)
Carbon and Silicon Concentration (mass%)
Si
Annealing Time, t /min
Fig. 1 Variation in chemical composition during annealing at 1348 K for
steel A.
The variations in carbon and silicon concentrations are
ascribed to the following thermo-chemical reaction between
the oxide separators and the specimens,13)
2½C þ hSiO2 i ¼ ½Si þ 2ðCOÞ;
ð1Þ
where h i, [ ] and ( ) denote pure solids, solute in solids and
gaseous states, respectively. The reaction consists of the
deoxidization of SiO2 , the dissolution of reduced silicon into
steels and the oxidation of carbon in steels to form CO gas. In
accordance with this reaction, the decarburization of 0.05%
carbon should lead to an increase of silicon concentration by
0.06%. Agreement between this estimation and the experimental observation on the silicon concentration variation is
clearly seen in Fig. 1. The decrease in manganese concentration is attributed to the vaporization of manganese, of which
equilibrium vapor pressure is rather high at elevated
temperatures.13)
With the original chemistry, the samples are composed of
and phases at the annealing temperature as mentioned
earlier. However, due to the above chemistry variation, the
phase transformation from to occurs. It is seen in Fig. 2
that columnar grains of phase develop inward from both
sheet surfaces induced by the phase transformation. Beyond
the growth fronts of columnar grains, the mixed structure
consisting of phase and pearlite is observed to exist, which
is the remnant of the and phases stable at the annealing
temperature [see Figs. 2(a) to (c)]. The mixed structure
disappears after 4 h of annealing, and then a more rapid
growth of grains follows since the pinning effect due to the
presence of = two phase structures disappears. Some
grains grow until they penetrate sheets in thickness directions, and the grain growth virtually stops. It is also seen in
Fig. 2 that the diameter of columnar grains gradually
increases maintaining the ratio between diameter and length
of columnar grains being from 1 to 2. These variations in
microstructures as well as chemical composition during
lamination annealing were unchanged by the variation in
Fig. 2 Optical micrographs of entire cross-sections of steel A after
annealing at 1348 K.
processing prior to the annealing, and they are in good
agreement with those observed in the previous study.13)
3.2
Influence of two-stage cold rolling on preferred
orientation
The influence of two-stage cold rolling with intermediate
annealing on the development of preferred orientation upon
decarburization was then studied. The samples cold-rolled
and annealed in various conditions were fully decarburized at
1348 K for 12 h, and the magnetic induction at 800 A/m (B8 )
in RD was measured. B8 is known as a measure of the degree
of h100i axis alignment in the direction of applied fields in
Fe–Si. When single crystals are magnetized in h100i
directions, a B8 nearly equal to saturation magnetization
(2.03 T for Fe–3%Si14)) is observed, while a B8 of about 70%
of the saturation is observed for polycrystalline materials
without texture. The contour map of measured B8 (ratios to
2.03 T) is shown in Fig. 3 as a function of thickness reduction
of cold rolling and temperature of intermediate annealing;
note that, given the reduction at the first rolling, the reduction
at the second rolling is determined since total reduction is
constant in this experiment. As the temperature of intermediate annealing is about 1000 K, B8 is only 70 to 75% of
the saturation magnetization and similar to those for
materials without texture. However, when the intermediate
Fine-Grained Doubly Oriented Silicon Steel Sheets and Mechanism of Cube Texture Development
sheet surface; the orientations of a few grains in Fig. 4 were
not possible to be determined because of poor surface
conditions. The average misorientation angle between the
{100} planes and the sheet surface was about 4 . The h100i
axes lying in the sheet plane are also well aligned with either
RD or TD so that the orientations of about 90% of grains lie
within 10 of the exact cube orientation; the orientations of
98% of grains lie within 15 of the cube orientation. The
grain size is about 0.4 mm in diameter, which has been
determined by EBSP including all the grain boundaries of
more than 1 in boundary angles. The observation by an
optical microscope gave a slightly larger grain size because
of poorer detectability on small angle boundaries.
Thickness Reduction at First Rolling
0.8
0.7
0.5
0.6
Temperature, T /K
1300
1200
0.9
1100
0.85
0.8
0.7
0.75
1000
0.4
0.5
0.6
0.7
1109
0.8
Thickness Reduction at Second Rolling
Fig. 3 Contour map of the ratio of B8 in the rolling direction to 2.03 T as a
function of thickness reduction of cold rolling and temperature of
intermediate annealing. The samples of steel A were decarburized at
1348 K for 12 h.
annealing temperature is raised above 1200 K, B8 more than
90% of the saturation can be obtained under appropriate
rolling conditions. The optimum thickness reductions upon
rolling to attain high B8 , i.e., a high degree of h100i axis
alignment with RD, are 70 to 75% for the first rolling and 50
to 60% for the second rolling.
Figure 4 displays the surface microstructure and texture of
the sample of which B8 in RD is more than 90% of the
saturation magnetization. The sample was cold-rolled at a
reduction of 75%, annealed at 1323 K, again cold-rolled at a
reduction of 53%, and then decarburized at 1348 K for 16 h.
A dramatically developed cube texture is observed. All of the
53 grains of which orientations have been determined by
EBSP have {100} planes lying within 12 with respect to the
3.3 Texture development process
The variation in texture throughout the processing in this
experiment was investigated using the conditions under
which cube texture markedly developed. In Fig. 5 is shown
the texture after hot rolling, which has been measured at a
depth of 0.1 mm below sheet surfaces. The presence of a
diffuse Goss texture component is seen, which is known to
emerge due to the shear deformation by the friction between
samples and roll surfaces during hot rolling. Near the center
layer of thickness, a {100}h011i texture component was
observed as a major component instead of the Goss texture
component. The deformation texture after the subsequent
cold rolling at a reduction of 75% involved {100}, {111} and
{211} components in the decreasing order of intensity. This
type of cold rolling texture is also known to appear in Fe–Si
and other low-alloyed steels.
After the intermediate annealing at 1323 K, the texture
shown in Fig. 6 appeared near sheet surfaces. As seen in Fig.
6(a), although weak, the texture has a component in which
h100i axes lie in RD and {100} planes are rotated from the
rolling plane by about 15 about RD. This orientation lies
near {410}h001i as indicated in Fig. 6(b) which is the plots of
the crystal orientation distribution densities in the range
between cube and Goss orientations. The maximum of
RD
Goss
Cube
1.5
0.5
1.5
1.5
1
1
TD
2
2
2.2
1 2.2
0.5
Fig. 4 Crystallographic orientation analyses by EBSP for a fully decarburized sample of steel A by annealing at 1348 K for 16 h.
(a) Stereogram of {100} poles of grains and (b) an OIM image for the
analyzed area with 3-D vector representation of orientation of individual
grains. The vectors represent the measured orientations of h100i axes of
individual grains.
1
Fig. 5 {100} pole figure for steel A measured after hot rolling by X-ray
diffractometry.
1110
T. Tomida, S. Uenoya and N. Sano
100
<100>
<111>
<211>
<210>
<110>
(a)
at 1348 K
(b)
Axis Density
10
1
0.1
0.01
700
Asrolled
900
0
1100
200
400
Annealing Time, t /min
Temperature, T/K
Fig. 7 Variation in axis density measured by X-ray diffractometry for steel
A during (a) heating and (b) annealing at 1348 K. The density is displayed
as multiples of random distribution.
Fig. 6 (a) {100} pole figure and (b) crystal orientation distribution by
EBSP for steel A after intermediate annealing at 1323 K. The distribution
from {100}h001i to {110}h001i is displayed as multiples of random
distribution.
distribution density is located between {410}h001i and
{310}h001i. In contrast, the distribution density at cube
orientation is small in this stage of the processing. In addition
to the near {410}h001i texture component, a {111}h011i
component was also observed.
Figure 7 shows the texture present after the final cold
rolling at a reduction of 53% and those after lamination
annealing. The deformation texture after the final cold rolling
is similar to that observed after the first rolling. The
difference is that the {111} component after the final rolling
is stronger than that after the first rolling and the {100}
component behaves vice versa. During the subsequent
heating to about 900 K, recovery and then primary recrystallization of phase occur so that the cold rolling texture is
weakened. However, the {111} and {100} components again
increase when temperature is raised to above 1200 K at which
decarburization as well as columnar grain growth of phase
begins. Then the {100} component only continues to increase
and becomes the strongest single component during decarburization at 1348 K. After 4 h of annealing, at which the
growth fronts of columnar grains migrating from both sheet
surfaces impinge at the thickness center (see Fig. 2), the
intensity of the {100} component becomes more than 20
times the random intensity. The {100} component further
evolves during the subsequent growth of grains at the
expense of all other components.
In Fig. 8 is displayed the {100} pole figure for the sample
which has just undergone primary recrystallization by
heating up to 923 K after final cold rolling. The sample was
little decarburized and the size of recrystallized grains was
about 10 mm. A weak but distinct cube texture component is
observed to be present. This means that the nuclei for the
development of cube texture appear during the primary
recrystallization prior to decarburization.
The evolution processes of cube texture are again shown in
Figs. 9 to 11 as the variation in orientation distribution in
close proximity to {100}h001i. The analyzed stages of
texture development are the one in which samples have just
undergone primary recrystallization after final cold rolling
(the state just mentioned above), the beginning stage of
columnar grain growth of phase in which the columnar
RD
1
Goss
Cube
2.3
0.7
0.7
0.7
1
2.3
1
1.7
1.7
TD
1
1
0.35
0.7
0.7
2.3
1.7
Fig. 8 {100} pole figure by EBSP for steel A heated to 923 K after final
cold rolling.
Fine-Grained Doubly Oriented Silicon Steel Sheets and Mechanism of Cube Texture Development
ND
φ
1
φ
0.8
RD
TD
0.6
up to 923 K
(9.5)
0.4
0.2
soaked
at 1348 K
(250)
Cube
Goss
Normalized Distribution Function (arb. unit)
Normalized Distribution Function (arb. unit)
Cube
up to 1348 K
(45)
1111
{100}<011>
ND
1
0.8
ϕ1
RD ϕ1
TD
0.6
up to 923 K
(9.5)
0.4
up to 1348 K
(45)
0.2
soaked
at 1348 K
(250)
0
0
0
0
15
30
45
φ
Rotation about RD,
Fig. 9 Normalized orientation distribution functions for steel A in the
range from {100}h001i to {110}h001i (rotation about RD) at the three
stages of cube texture evolution. The samples were (dotted) heated up to
923 K, (dashed) heated up to 1348 K, and (solid) annealed for 16 h at
1348 K. The numbers within parentheses are the distribution densities at
{100}h001i as multiples of random distribution.
Normalized Distribution Function (arb. unit)
Cube
{110}<110>
ND
θ
1
0.8
RD θ
TD
0.6
0.4
up to 923 K
(9.5)
0.2
soaked
at 1348 K
(250)
up to 1348 K
(45)
0
0
15
30
Rotation about TD,
45
θ
Fig. 10 Normalized orientation distribution functions for steel A in the
range from {100}h001i to {110}h110i (rotation about TD) at the three
stages of cube texture evolution. The samples used and manner of
presentation of data are the same as those in Fig. 9.
grains are developed to the depth of 60 mm (the sample heated
up to 1348 K, see Fig. 2(a)), and the stage in which samples
are fully decarburized. It is seen in Fig. 9 that the cube texture
component that emerges during primary recrystallization is
largely diffuse and widely spreads to Goss orientation. Then
this diffuseness greatly and continuously lessens in the later
stages in which the columnar grains grow due to decarburization. In Fig. 10 that displays the distributions in the range
15
Rotation about ND,
30
45
ϕ1
Fig. 11 Normalized orientation distribution functions in the range from
{100}h001i to {100}h011i (rotation about ND) for steel A at the three
stages of cube texture evolution. The samples used and manner of
presentation of data are the same as those in Fig. 9.
between {100}h001i and {110}h011i (a crystal rotation about
TD), a similar behavior of the variation in diffuseness of cube
texture is observed, although the diffuseness after primary
recrystallization is somewhat smaller than that seen in Fig. 9.
However a different behavior of the variation in diffuseness
of texture is observed in the distributions in the range
between {100}h001i and {100}h011i [a crystal rotation about
the sheet normal (ND)] as shown in Fig. 11; note that {100}
planes are maintained to be parallel to the sheet plane by the
crystal rotation in this range. In the early stages of columnar
grain growth, the diffuseness of cube texture in this range
does not so lessen as that seen in Figs. 9 and 10. The density
in the orientations far from cube orientation is unchanged or
in part increases, as long as the {100} planes of crystals are
parallel to the sheet plane. In the later stages, however, those
misoriented components are depleted so that the cube texture
is entirely sharpened.
3.4 Magnetic properties
The magnetic properties of fully decarburized strips of
steel B are presented in Fig. 12 as a function of the angle of
applied fields to RD. The conditions of cold rolling and
intermediate annealing for these samples are the same as
those described in the section 3.3. The chemical composition
after decarburization was Fe–2.9%Si–0.7%Mn–0.0007%C,
and the resulting grain structure was similar to that shown in
Fig. 4. The strips showed large magnetic induction and small
core loss in RD and TD. In these directions B8 are about
1.87 T, and the 50 and 60 Hz losses at an induction of 1.5 T,
W15=50 and W15=60 , are about 0.9 and 1.2 W/kg respectively.
These losses in easy directions are notably smaller than those
for the previously reported doubly oriented Fe–3%Si strips
with grains of 25 mm in diameter.9,10) The W15=60 for 0.3 mm
thick doubly oriented sheet materials, which were slightly
thinner than the present materials and showed a comparable
1112
T. Tomida, S. Uenoya and N. Sano
3.5
2.5
1.9
RD
Bm=1.9T
3
1.7
W 15/50
2
1.5
1.5
B8
1
BP
1.3
0.5
0
30
60
90
Angle to RD, φ
0
0
Magnetostriction, λ /10-6
2.5
Core Loss, W/Wkg -1
Magnetic Induction, B /T
W 15/60
25 to RD
Bm=1.7T
45 to RD
Bm=1.5T
0
0
65 to RD
Bm=1.7T
Fig. 12 Magnetic induction and core loss of decarburized samples as a
function of angle to the rolling direction. The 0.35 mm thick samples of
steel B were decarburized at 1373 K for 20 h. Bp is the calculated induction
based on the model described in the text.
B8 of 1.86 T, was reported to be 1.74 W/kg;10) thinner
materials generally exhibit smaller core losses. It is also seen
in Fig. 12 that the induction and core loss vary symmetrically
with respect to the directions of applied fields, which is of
course due to the cube texture present.
Magnetostriction of the abovementioned samples is shown
in Fig. 13. The magnetostriction in easy magnetizing
directions as well as in the direction of 45 to RD is positive
regardless of magnetic induction. However, in the directions
of 25 and 65 to RD, the magnetostriction becomes negative
below an induction of 1.5 T. It is of particular interest that the
magnetostriction in easy directions is considerably smaller
than that reported for the doubly oriented Fe–3%Si strips
with large grains. Although the magnetostriction for our
material is only 2:5 106 at 1.9 T, Foster and co-workers10)
reported that the 0.3 mm thick Fe–3%Si doubly oriented
strips showed a DC magnetostriction of 22:1 106 at 1.5 T.
4.
TD
Bm=1.9T
0
-2
-1
0
1
2
Magnetic induction, B/ T
Fig. 13 Magnetostriction loops for the same samples as shown in Fig. 12.
Growth in Radial Directions
Low Surface Energy
{100} Grain
Sheet Surface
R
α/γ Two Phase
Structure
t
Discussion
4.1 Driving force of cube texture evolution
The preferred orientation of columnar grain growth lay in
{100}h001i as appropriate two-stage cold rolling was applied
prior to the oxide-separator-induced decarburization. A cubeoriented component was observed to emerge during primary
recrystallization after final cold rolling, and they selectively
grew during columnar grain growth. Here, let us first discuss
the mechanism of the selective growth.
In the previous study13) using the same oxide-separatorinduced decarburization, the preferred orientation of columnar grain growth was observed to lie in {100}h012i as a
single-stage cold rolling was applied before decarburization.
This selective growth was related to the surface free energy
difference between {100} and other planes of phase. It is
clear that if the surface free energy of {100} planes is the
lowest in all surface orientations, the {100}-oriented particles or islands in = two phase structures tend to
Growth in Axial Directions due
to Decarburization
Fig. 14 Schematic representation of columnar grain growth.
preferentially grow at the expense of phase during
decarburization-induced phase transformation until sheet
surfaces are entirely covered with phase. Moreover, if the
radial growth of columnar grains shown in Fig. 14 is possible
to occur, the {100} texture may further develop by the
surface energy difference. The alignment of h012i directions
along RD has been then ascribed to the recrystallization of
cold-rolled structures, which takes place prior to decarburization. It was postulated that a {100}h012i-oriented component might appear during the recrystallization and acted as
nuclei for the selective growth in later processes.13) The
results obtained in the present experiment, the {100}h001ioriented component observed in the matrix after primary
Fine-Grained Doubly Oriented Silicon Steel Sheets and Mechanism of Cube Texture Development
recrystallization as well as the development of cube texture,
are consistent with the above texture mechanism based on
surface energy difference. The sluggish sharpening of the
cube texture component with respect to the rotation about ND
shown in Fig. 11 is also reasonable if surface free energy is
the main selective driving force.
Taking the effects of surface free energy into consideration, the driving force D for the radial growth of columnar
grains shown in Fig. 14 may be written without the term for
pinning effects as
D ¼ M½b ð1=Rav 1=RÞ þ s =t;
ð2Þ
where M is the mobility of grain boundaries, and b and s
are grain boundary energy and the difference in surface
energy respectively. R is the radius of the grain under
consideration, Rav is the average grain radius, and t is the
thickness of decarburized layers or the length of columnar
grains. Since the dominant texture component due to the
primary recrystallization after final cold rolling has been a
{111} component, the s under consideration should be the
free energy difference between {100} and {111} planes.
Mills et al.15) have reported for Fe–3%Si that the s can be
at most about 180 mJ/m2 under an atmosphere of a low
oxygen partial pressure at an elevated temperature, which is
about 10% of the surface energy itself; Walter and Dunn16)
reported a smaller surface energy difference of 70 mJ/m2
between {110} and {100} planes when the {110} planes had
the lowest surface energy. The b for Fe–3%Si has been
reported to be about 470 mJ/m2 .17) Therefore, let s and b
be 180 mJ/m2 and 470 mJ/m2 respectively. Assuming that Rav
is equal to the layer thickness t as observed in this
experiment, the ratio r of the magnitude of driving force by
surface energy [the second term in eq. (2)] to that by grain
boundary energy [the first term in eq. (2)] is then described as
r ¼ 0:38=j1 Rav =Rj:
ð3Þ
This ratio is larger than 1.0 as 0:72Rav < R < 1:62Rav and it
becomes infinite as R becomes Rav . It means that the selective
driving force by surface free energy on grain growth could be
larger than the driving force by grain boundary energy in a
relatively wide range of grain size. In other words, the {100}
grains completely surrounded by {111} grains can grow
unless the grain size is smaller than 72% of the average grain
size. In contrast, the {111} grains totally surrounded by
{100} grains cannot grow unless the grain size is larger than
162% of the average. Although this estimation is rather rough
and perhaps gives the maximal possible contribution of
surface energy, it is suggestive of a high plausibility of the
texture mechanism by surface energy difference.
It is also evident from this estimation that the radial growth
is, to a large extent, driven by grain boundary energy. As
noted in the previous report,13) the influence of grain
boundary energy may manifest itself in the later stages of
columnar grain growth where the numbers of {100}-oriented
grains considerably increases and the surface-energy-related
mechanism is no longer effective. The likely relevant
phenomenon is seen in Fig. 11 in which the cube texture
component sharpens with respect to the crystal rotation about
ND; i.e., the grains that have {100} planes parallel to the
sheet plane but are largely misoriented from cube orientation
1113
are gradually depleted. As {100}-oriented columnar grains
impinge and grow further at the expense of neighboring
{100}-oriented grains, such grain growth will proceed in the
manner that high angle boundaries are preferentially depleted
since they are in general of high energy. Since the {100}oriented grains that are largely misoriented from the true
cube orientation are mostly surrounded by abundantly
existing well-oriented cube grains, they are frequently
surrounded by relatively high angle boundaries. It is therefore
probable that such misoriented {100} grains are depleted in
the later stages of decarburization to reduce grain boundary
energy.
4.2 Emergence of cube texture component
The relation between the initial orientations of crystals and
the orientations of recrystallized grains after cold rolling and
annealing has been extensively studied for Fe–3%Si. Walter
and Hibbard18) reported that a cube texture component
appeared as single crystals of {210}h001i orientation were
cold-rolled and recrystallized upon annealing. Detert19) and
Aspden20) also suggested that the presence of the texture
components with orientations from {100}h001i to
{210}h001i before cold rolling was an important factor for
the cube texture component to emerge during subsequent
primary recrystallization. As seen in Fig. 6, a component with
the orientation near {410}h001i, which lies between
{100}h001i and {210}h001i and is away only by 10 from
{210}h001i, was observed after the intermediate annealing in
this experiment. The occurrence of this component was then
followed by the emergence of a cube texture component upon
primary recrystallization after final cold reduction. Therefore
the cube texture component is likely to arise from the near{410}h001i component present after the intermediate annealing.
Also the near-{410}h001i component could be related to
the diffuse Goss texture present after hot rolling (see Fig. 5).
Varlakov and Romashov21) reported that cold rolling and
annealing of polycrystalline sheets with a well-developed
Goss texture resulted in the appearance of a {210}h001i
component in a primary recrystallized matrix. Harase et al.22)
examined the textures after cold rolling and annealing using
single crystals with the Goss orientation. Their results
indicated that the {210}h001i component emerged during
the subsequent grain growth at a relatively high temperature
after primary recrystallization. The near {410}h001i component in the present experiment, of which orientation is close
to {210}h001i as aforementioned, therefore may arise from
the diffuse Goss texture component present after hot rolling.
In addition, since the optimal temperature of intermediate
annealing for cube texture to evolve was relatively high
around 1300 K, a grain growth after primary recrystallization
might be associated with the appearance of the near{410}h001i component. Hence the observed variation in
texture during the process prior to decarburization can be
phenomenologically understood based on the known relation
between annealing textures and initial crystal orientations
before cold reduction in Fe–Si, although the present materials
contain much carbon and manganese.
1114
T. Tomida, S. Uenoya and N. Sano
4.3
Magnetization processes in fine-grained doubly
oriented steels
Using the crystallographic data by EBSP shown in Fig. 4,
the magnetic induction B8 can be calculated. Becker and
Döring23) have proposed that the domain distribution in Fe–
Si sheets around the knee of magnetization curve is described
as magnetic domains are distributed only among the three
easy directions of magnetization closest to the applied field
such that the net magnetization is parallel to the applied field.
The knee of magnetization curve occurs around an applied
field of 800 A/m, i.e., around B8 . The magnetization Bp and
the fractions of domain Pj in the three easy directions for a
single crystal are then described as
Bp ¼ Bs =ð1 þ 2 þ 3 Þ
Pj ¼ j =ð1 þ 2 þ 3 Þ ðj = 1, 2 and 3Þ;
which are called Kaya’s law.24) Here 1 , 2 and 3 are the
directional cosines of the three easy directions closest to the
applied field with respect to the applied field and Bs is the
saturation magnetization. We may modify these equations for
polycrystalline materials as follows,
X
B p ¼ Bs
i =ð1;i þ 2;i þ 3;i Þ;
ð4Þ
Pj ¼
X
i
i j,i =ð1;i þ 2;i þ 2;i Þ;
ð5Þ
i
where i denotes the i-th grain and i is the volume fractions
of the i-th grain. There is an ambiguity in the definition of Pj
in eq. (5) since the directions of easy axes vary depending on
grains. Therefore we redefine j’s (1, 2 and 3) to denote the
easy directions closest to the applied field as well as closest to
the fixed axes of the material, RD, TD and ND respectively.
Putting the crystallographic orientation and fraction data for
53 grains in Fig. 4 into eq. (4), the magnetization Bp shown in
Fig. 12 has been obtained; we assume here that Bs is equal to
that for Fe–3%Si (2.03 T). The calculation is reasonably in
good agreement with the measurement of B8 , although the
observed B8 is somewhat larger than that calculated. A
similar tendency has been reported for {110}h001i oriented
Fe–3%Si. Therefore it is obvious that the domain distribution
around an applied field of 800 A/m is well described by the
abovementioned state proposed by Becker and Döring and
the texture observed at the sheet surface shown in Fig. 4 well
represents the texture of the entire material.
Apart from magnetic induction, the notable magnetic
properties of doubly oriented materials found in the present
experiment are small core loss and low magnetostriction in
easy directions, which are considerably smaller than those
reported for the doubly oriented Fe–3%Si sheets with large
grains more than 20 mm. Except texture, the factors that need
to be considered in respect of those properties are the spacing
between main magnetic domain walls and the presence of
magnetic domains in the direction perpendicular to applied
fields. Core loss increases with increasing the domain wall
spacing as suggested by Pry and Bean,11) since the velocity of
domain wall motion upon magnetization is increased,
resulting in a large increase in eddy current loss. Since the
domain wall spacing generally decreases with decreasing
grain size, reduced domain wall spacing may be the primary
reason for the decreased core loss observed in this experiment; the grain size of the present materials is smaller than
that of the previous materials by a factor of about 60.
Magnetostriction is, on the other hand, directly related to
the above second factor, the presence of transverse domains.
As cube-textured Fe–Si is magnetized in easy directions,
magnetostriction is observed if the magnetic domains in the
directions perpendicular to applied fields are created or
annihilated upon magnetization. Therefore the observed
small magnetostriction in RD and TD should be ascribed to
a domain structure for which the volume fraction of
transverse domains is rather small.
Based on the calculation described above and the data on
magnetostriction, the volume fraction of transverse domains
in our materials may be deduced. Using eq. (5) and the data
by EBSP shown in Fig. 4, the volume percentages of domains
present at the knee of magnetization for our material have
been obtained as 89.5% in RD, 6.4% in TD and 4.1% in ND,
when magnetized in RD. Thus the volume percentage of
transverse domains at an induction around B8 is about 10.5%.
On the other hand, during the magnetization up to 1.9 T
(around B8 ), our materials have showed a magnetostriction of
about 2:5 106 in RD. This value of strain is about 7% of
3100 =2, the strain yielded by the domain transformation
from the [010] or [100] to the [001] in a single crystal; 100 is
23:7 106 for Fe–3%Si.25) Therefore, if it is assumed that
entire samples are homogeneously strained with the volumeweighted average of strains and the imperfection of cube
texture is ignored, the observed magnetostriction is explained
as the annihilation of transverse domains of about 7% in
volume fraction. Since the volume fraction of transverse
domains around B8 is 10.5%, that in demagnetized states is
then estimated to be about 17.5 (7+10.5)%. This value is
roughly one fourth of the volume fraction of transverse
domains in the cube-textured material exhibiting a magnetostriction of 22:1 106 . The low-core-loss property shown
in Fig. 12 can also be related to this domain structure, in
which magnetization would occur mostly by 180 domain
wall motion.
5.
Conclusions
The texture development in silicon steel sheets by the
oxide-separator-induced decarburization combined with twostage cold rolling and the magnetic properties of thus
obtained doubly oriented sheet materials have been studied.
The study has led to the following conclusions.
(1) By lamination annealing with separators consisting of
SiO2 , TiO2 and Al2 O3 at about 1350 K, 0.35 mm thick
sheet steels with chemical compositions around Fe–
3%Si–1%Mn–0.05%C can be fully decarburized down
to 0.001% of carbon concentration. Associated with
decarburization, columnar grains of phase grow
inward from sheet surfaces.
(2) Cube texture dramatically evolves in the columnar
grains as appropriate two-stage cold rolling with
intermediate annealing is applied to the materials prior
to decarburization. The proper two-stage cold rolling
process for cube texture to develop is the one consisting
of a cold rolling at a reduction around 65%, a
Fine-Grained Doubly Oriented Silicon Steel Sheets and Mechanism of Cube Texture Development
(3)
(4)
(5)
(6)
(7)
subsequent annealing at 1250 K for 30 s, and a final cold
rolling at a reduction around 55%.
Texture components near {110}h001i and {410}h001i
exist below sheet surfaces after hot rolling and intermediate annealing of two-stage cold rolling, respectively. Then a cube texture component emerges upon
primary recrystallization after final cold rolling, presumably originating from the near-{410}h001i component.
The cube texture component develops during decarburization resulting in a highly oriented grain structure of
about 0.4 mm grain size, in which more than 90% grains
are closely aligned with cube orientation. In the early
stages of the cube texture evolution, the cube texture
component develops in the manner that texture components with {100} planes away from the sheet plane are
mainly depleted. Then, in the later stages of the
evolution, the cube texture evolves at the expense of
all the other texture components.
The main selective driving force for cube texture
evolution appears to be the orientation dependence of
surface free energy, while the influence of grain
boundary energy is also suggested by the behavior of
texture evolution observed in the later stages.
The doubly oriented sheet materials thus obtained show
a large induction of 1.87 T at 800 A/m, a small core loss
of 1.2 W/kg at 1.5 T and 60 Hz, and a low magnetostriction of 2:5 106 at 1.9 T in the rolling and
transverse directions. The observed small core loss
appears to be due to small grain size by which magnetic
domains are refined, while the low magnetostriction is
related to a reduced volume fraction of transverse
domains.
The calculated induction based on the observed grain
structure and the method proposed by Becker and
Döring is in good agreement with the observation of
induction at 800 A/m. The volume fraction of transverse
domains present in a demagnetized state, which has
been deduced based on the above calculation and the
observation of magnetostriction, is about 17.5%.
1115
Acknowledgments
The authors gratefully acknowledge instructive discussions with Dr. Taiji Nishizawa (Professor Emeritus of
Tohoku University), Dr. Shigeharu Hinotani, and Dr.
Yasuhiro Maehara with R&D Laboratories of Sumitomo
Metal Industries Ltd. They also wish to thank Dr. Sumio
Kobayashi (the former director of the R&D Laboratories) and
Mr. Yasuyuki Tozaki (the director of the R&D Laboratories)
for their inevitable supervision throughout the research.
REFERENCES
1) G. Gottstein and K. Lücke (ed.): Proc. 5th Int. Conf. on Textures of
Materials, (Springer-Verlag, 1978).
2) F. Assmus, K. Detert and G. Ibe: Z. Metallk. 48 (1957) 344–349.
3) J. L. Walter: Acta Metall. 7 (1959) 424–426.
4) J. L. Walter: J. Appl. Phys. 36 (1965) 1213–1220.
5) K. Foster, J. J. Kramer and G. W. Wiener: Trans. Metall. AIME 227
(1963) 185–188.
6) S. Taguchi and A. Sakakura: Acta Metall. 14 (1966) 405.
7) Kinzokubutsuri 7 (1961) 221–223.
8) J. Harase and R. Shimizu: J. Japan. Inst. Metals 53 (1989) 745–752.
9) M. F. Littmann: J. Appl. Phys. 38 (1967) 1104–1108.
10) K. Foster, J. Seidel and J. W. Shilling: Magn. Magn. Mater. 2 (1973)
971–975.
11) R. H. Pry and C. P. Bean: J. Appl. Phys. 29 (1958) 532–533.
12) G. W. Wiener, K. Foster and D. S. Shull: J. Appl. Phys. 38 (1967)
1102–1103.
13) T. Tomida: Mater. Trans. 44 (2003) 1096–1105.
14) R. M. Bozorth: Ferromagnetism, (D. Van Nostrand: New York, 1951)
77.
15) B. Mills, M. McLean and E. D. Hondros: Philos. Mag. 27 (1973) 361–
368.
16) J. L. Walter and C. G. Dunn: Trans. Metall. AIME 215 (1959) 465–471.
17) E. D. Hondros and L. E. Stuart: Philos. Mag. 17 (1968) 711–727.
18) J. L. Walter and W. R. Hibbard: Trans. Metall. AIME 212 (1958) 731.
19) K. Detert: Arch. Eisenhüttenwesen 35 (1964) 75–83.
20) R. G. Aspden: Trans. Metall. Soc. AIME 227 (1963) 905–909.
21) V. P. Varlakov and V. M. Romashov: Fiz. Metal Metalloved. 13 (1962)
671–675.
22) J. Harase, Y. Ushigami and N. Takahashi: Mater. Sci. Forum 157–162
(1994) 899–904.
23) R. Becker and W. Döring: see J. W. Shilling and G. L. Houze, Jr., IEEE
Trans. Magn. MAG-10 (1974) 195–223.
24) S. Kaya: see S. Chikazumi, Physics of Ferromagnetism Vol. II,
(Syokabo Tokyo, 1984) 232–233.
25) A. Hubert: see J. W. Shilling and G. L. Houze, Jr., IEEE Trans. Magn.
MAG-10 (1974) 195–223.