You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/277911744

A Study of the Batch Annealing of Cold-Rolled HSLA Steels Containing


Niobium or Titanium

Article  in  Metallurgical and Materials Transactions A · August 2015


DOI: 10.1007/s11661-015-2949-6

CITATIONS READS

8 680

4 authors, including:

C. I. Garcia S.-H. Choi


University of Pittsburgh Sunchon National University
139 PUBLICATIONS   1,482 CITATIONS    108 PUBLICATIONS   2,268 CITATIONS   

SEE PROFILE SEE PROFILE

Anthony Deardo
University of Pittsburgh
189 PUBLICATIONS   2,332 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Study on mechanical and electrical reliability issues of flexible electrodes View project

Dual Phase Steel (590 & 980 MPa) View project

All content following this page was uploaded by Anthony Deardo on 06 July 2016.

The user has requested enhancement of the downloaded file.


A Study of the Batch Annealing of Cold-Rolled HSLA
Steels Containing Niobium or Titanium
CHAO FANG, C. ISAAC GARCIA, SHI-HOON CHOI, and ANTHONY J. DEARDO

The batch annealing behavior of two cold-rolled, microalloyed HSLA steels has been studied in
this program. One steel was microalloyed with niobium while the other with titanium. A suc-
cessfully batch annealed steel will exhibit minimum variation in properties along the length of
the coil, even though the inner and outer wraps experience faster heating and cooling rates and
lower soaking temperatures, i.e., the so-called ‘‘cold spot’’ areas, than the mid-length portion of
the coil, i.e., the so-called ‘‘hot spot’’ areas. The variation in strength and ductility is caused by
differences in the extent of annealing in the different areas. It has been known for 30 years that
titanium-bearing HSLA steels show more variability after batch annealing than do the niobium-
bearing steels. One of the goals of this study was to try to explain this observation. In this study,
the annealing kinetics of the surface and center layers of the cold-rolled sheet were compared.
The surface and center layers of the niobium steel and the surface layer of the titanium steel all
showed similar annealing kinetics, while the center layer of the titanium steel exhibited much
slower kinetics. Metallographic results indicate that the stored energy of the cold-rolled con-
dition, as revealed by grain center sub-grain boundary density, appeared to strongly influence
the annealing kinetics. The kinetics were followed by the Kernel Average Misorientation re-
construction of the microstructure at different stages on annealing. Possible pinning effects
caused by microalloy precipitates were also considered. Methods of improving uniformity and
increasing kinetics, involving optimizing both hot-rolled and cold-rolled microstructure, are
suggested.

DOI: 10.1007/s11661-015-2949-6
 The Minerals, Metals & Materials Society and ASM International 2015

I. INTRODUCTION annealing. The alloy design of the steel also plays a role in
the property variation. For example, it has been known
PRODUCT uniformity is an important attribute for since the early 1980s that titanium microalloyed steels
steel in a highly competitive market. In the production of were more difficult to recrystallize and had larger
annealed cold-rolled gages of microalloyed HSLA steels property variations than did niobium steels, even when
using the batch annealing process, minimum variation of annealed to the same cold spot strength and with
properties along the coil length is important. For cold- reasonably equal atom pct additions.[1] For example,
rolled and annealed steels, one of the principal sources of for steels annealed to the 350 MPa (50 KSi) strength
strength in partially recrystallized steels is the non- level, the Ti grade showed approximately twice the
recrystallized regions, where high dislocation densities strength variation along the coil length as did the Nb
survive through the anneal. The larger the amount of grade, even when annealed to the same cold spot
these regions of high retained strain or stored energy, the temperature. Interestingly, these observations on batch
higher will be the strength. The variation in heating rate annealing are opposite to those found in numerous
and annealing temperature between the ‘‘cold’’ spot (mid- studies of both the recovery and recrystallization of
coil length) and ‘‘hot’’ spot (coil ends) in large coils is austenite during hot rolling, and in the later line anneal-
largely responsible for the variation in strength and other ing studies of cold-rolled ferrite in ultra-low carbon or IF
mechanical properties along the coil length after batch steels. In the cases of hot rolling of austenite[2,3] and line
annealing of ferrite in IF steels,[4,5] it is now accepted that
Nb additions retard recovery and recrystallization more
CHAO FANG, is with the Global Foundries Inc., Malta, New
York 12020. C. ISAAC GARCIA, Professor, is with the Department strongly than do Ti additions. One of the goals of this
of Mechanical Engineering and Materials Science, University of study was to investigate the reasons for this behavior,
Pittsburgh, Pittsburgh, PA 15261. SHI-HOON CHOI, Professor, is especially the differences in annealing behavior between
with the School of Applied Materials Engineering, Sunchon National steels containing either Nb or Ti in a batch annealing
University, 413 Jungangno (315 Maegok) Suncheon, Jeonnam 540-
742, South Korea. ANTHONY J. DEARDO, is with the Basic Metals
simulation.[6]
Processing Research Institute (BAMPRI), Department of Mechanical It is well known that the kinetics of annealing are
Engineering and Materials Science, University of Pittsburgh, Pittsburgh, controlled by diffusivity, driving forces and retarding
PA 15261, and Finland Distinguished Professor with the Oulu forces. The driving force for recovery and recrystalliza-
University, Centre for Advanced Steels Research, Oulu, Finland. tion in annealing is the retained strain or stored energy
Contact e-mail: ajdeardo.pitt@gmail.com
Manuscript submitted July 2, 2014. of cold work, while the retarding force is caused by
Article published online May 19, 2015 solute drag and/or precipitate pinning of sub-grain and

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 46A, AUGUST 2015—3635


grain boundaries. In this study, both the driving force II. EXPERIMENTAL PROCEDURE
and the retarding force for static recovery and recrys-
tallization have been investigated. The chemical compositions of the 350 Grade HSLA
The stored energy[7] in cold-rolled metals is associated steels (nominal YS 350 MPa, 50 KSi) used in this study
with various lattice defects generated during deforma- are listed in Table I. Commercial hot band and full-hard
tion and provides the initial driving force for recovery (cold rolled) samples were supplied from a thin slab
and recrystallization during the annealing process. The casting steel company. The cold reductions used were
total amount and distribution of the stored energy will approximately 58 pct for the Ti-bearing steel and 55 pct
dramatically affect the annealing behavior. In cold- for the Nb-bearing steel, respectively. The final thick-
rolled metals of high stacking fault energy such as nesses were approximately 1.25 mm for the Ti steel and
ferrite, the stored energy can be considered to be the 1.40 mm for the Nb steel. The difference in cold
sum of (i) the grain center dislocation density present in reduction was not considered important since they were
sub-grain boundaries after cold rolling, and (ii) the extra similar. Also, the differences in final gage were not
grain boundary energy resulting from the grain shape thought to be important during annealing since the
change that occurs during cold rolling. The average annealing temperatures were subcritical and the heating
dislocation density has been estimated in the past by and cooling rates were very slow.
both indirect methods, e.g., calorimetry, X-ray line Small samples were sectioned from these conditions
broadening, changes in strength,[8] and direct means, and reheated from room temperature to 943 K (670 C)
e.g., etch pits[9] and transmission TEM.[10] at a heating rate of about 16 K/h (16 C/h). All samples
Electron back-scattered diffraction (EBSD)[11] is a were put into the furnace at the same time and
relatively new analytical technique, which has evolved individual specimens were removed at every 20 K
into a powerful tool for the crystallographic analysis of (20 C) interval from 873 K to 933 K (600 C to
steels and other engineering materials using the scanning 660 C) and at different soaking times ranging from 2
electron microscopy (SEM). The automatic analysis of to 8 hours, then water quenched to room temperature.
these EBSD patterns yields the crystallographic orien- These quenched samples were cut along the rolling
tation, the phase present and the value indicating the direction into smaller pieces, about 10 mm in length and
quality of the diffraction pattern for every scan point. 5 mm in width, and thermo-mounted using bakelite for
From these data, the microstructure and texture of the optical metallography or copper powder for SEM/
scanned area can be reconstructed, as well as informa- EBSD metallography. Then, they were ground on
tion gained concerning the presence or absence of lattice 400 to 2400 grit silicon carbide abrasive paper and
perfection through elastic distortion of the Kikuchi polished with 0.05 lm alumina power on abrasive
bands and its resulting diffraction distortions.[12,13] cloths. The vibratory polishing machine and 0.05 lm
Taylor factors calculated from EBSD scanning data alumina power were used for the final polishing step.
were also used in early applications as a measurement of These samples could then be used for EBSD analysis. In
an orientation-dependent stored energy in cold-rolled addition, the samples were etched with 2 pct Nital to
steels.[14,15] However, the assumption of the upper observe the microstructures in the SEM. A TSL EBSD
bound or full constraint method (stress and strain rate system attached to a Philips XL30 FEG SEM was used
are assumed uniform in each grain) is not appropriate to for scanning the desired areas. The original raw data
simulate the heterogeneous distribution of stored energy were collected using the TSL OIM data collector
found in the deformed grains. software. The SEM was operated at 20 kV voltage and
In the current study, the EBSD data have been the beam spot size used was 6 or 5. The scanning step
analyzed to determine the summation of sub-grain size was about 0.2 lm.
boundary energy in the investigated steels. This has led For further analysis of small precipitates and to
to what is now referred to as Stored Energy Maps, obtain additional crystallographic data, transmission
which attempt to measure the effect of the elastic strain electron microscopy (TEM) utilizing a JEOL 200CX
in the matrix caused by dislocation density and solute and a JEOL 2100F was used. The TEM studies were
atoms.[16,17] To this grain-center stored energy was conducted at 200 kV, 118 lA, and a beam spot size of 2.
added the extra grain boundary energy resulting from Thin foil sample preparation involved sectioning the
the grain shape change associated with the cold rolling. samples, mechanical or chemical thinning, and elec-
The resulting sum was taken as the total stored energy. tropolishing. After thinning, the foils were then punched
The stored energies of the hot band, cold-rolled, and into 3 mm disks and electropolished until perforation
annealed conditions of the two steels were determined using a Fischione twinjet electropolisher with an elec-
using this approach. These results were then related to trolyte solution of Na2CrO3 (20 pct by weight) and
the observed annealing kinetics of the two steels. acetic acid (80 pct). Finally, the samples were cleaned

Table I. Nominal Compositions of the Steels Used in this Study (Weight Percent)

C Mn Si Cu Nb V Ti N
Ti-bearing 0.046 0.730 0.191 0.127 0.003 0.004 0.058 0.009
Nb-bearing 0.053 1.068 0.194 0.148 0.032 0.002 0.010 0.007

3636—VOLUME 46A, AUGUST 2015 METALLURGICAL AND MATERIALS TRANSACTIONS A


using a three step ethanol cleaning process to remove all boundary energy from the Read–Shockley[18] relation
residues from the specimens. can be expressed in Eq. [2]:
Since it is common for the microstructure to vary ( 0
h  0 i
through the thickness, each specimen was observed at cm hh 1  ln hh ; h0 < h
cðDhÞ ¼ ½2
five different depths. Figure 1 illustrates schematically cm ; h0  h
the five different layers or depths through the thickness
where measurements and observations were made. where h¢ = |Dh| for 0 £ |Dh| £ p and h¢ = 2p  |Dh|
Unless stated otherwise, the measurements and obser- for p £ |Dh| £ 2p. In this approach, Dh is the boundary
vations presented below were taken from the center or misorientation between two neighboring sub-grains, h*
mid-thickness region. Figure 1 also shows the orienta- is the misorientation limit for low-angle grain bound-
tion of the observed regions with respect to the original aries, which was chosen as 15 deg, and cm is the specific
hot band rolling direction. The microhardness values energy of high-angle grain boundaries which was taken
were measured using the Vickers test on a Leco M-400- to be 0.756 J/m2.
G Hardness Tester, employing a 500 g load and a The grain boundary contribution to stored energy was
15 seconds dwell time. The results were plotted as the estimated through the use of Eq. [3].
average value of three layers (surface, quarter point,
center) and with 10 measurements per layer. The Total high angle GB Energy
standard deviation of the mean value for any group of ½3
¼ SvðGBÞ  7:56  105 J=cm2
measurements was less than 4 pct.
To access the total stored energy of the cold-rolled
condition going into the anneal, two sets of measure- In this equation, the grain boundary area per unit
ments were made. The first was the sum of the stored volume, SV in cm2/cm3, was measured for the various
energy of the sub-grain boundaries found in the ferrite conditions using Eq. [4].[19]
grain centers. The second was the contribution from the
SvðGBÞ ¼ 0:429ðNL Þk þ 1:571ðNL Þ? ; ½4
ferrite grain boundaries which became elongated during
the cold rolling. These two were added to estimate the
where (NL)k and (NL)^ are the intercept numbers per
total stored energy of cold work available to drive the
unit length along the rolling and thickness directions,
recovery and recrystallization in the anneal.
respectively.
The sub-grain method[17] was used in this study to
However, it will be shown later that the energy stored
construct stored energy distribution maps. The sub-
in the grain boundaries is much lower than that stored in
grain structure in the grain centers of the deformed
the sub-grain boundaries. It is estimated that about
microstructure was used in the calculation of the
10 pct of the total stored energy resides in the high-angle
dislocation-related stored energy, since the plastic de-
ferrite grain boundaries, while the balance is related to
formation of polycrystalline steels normally shows the
the dislocations and sub-grain boundaries. Another
dislocations in a sub-grain boundary network, typical of
method of measuring and displaying the amount and
high stacking fault alloys. If the dislocation substructure
distribution of stored energy is by determining Kernel
can be simplified in the form of sub-grains of diameter,
Average Misorientation (KAM) maps.[20] This tech-
D, and boundary energy, c, then the stored energy, Si of
nique is based on the point-to-point lattice rotation
each lattice site, can be given by Eq. [1]:
within a grain or specimen. These data have a value for
c each pixel equal to the average misorientation of that
Si ¼ a ; ½1
D particular pixel with its neighbors
As the internal strain and lattice imperfection varies,
where a is the geometric constant ~3. If it is assumed the disparity is depicted by changes in value of the
that the boundary energy of a sub-grain is only a func- KAM. In the standard representation, the magnitude of
tion of the boundary misorientation (Dh) and not de- the KAM representing a near-perfect lattice region is
pendent on the boundary types (tilt and twist), the given a small value, whereas areas of larger elastic strain
are assigned larger values. In this study, the color blue is
assigned to very small KAM values (recrystallized
areas), while the color red to very large ones (high
stored energy). This color representation permits a
highly visible summary of the local elastic strain
N distribution in the specimen, including stored energy in
recrystallization studies and transformation strains in
D
phase transformation studies
Niobium and titanium are known to be very effective
R
Surface Quarter Center as microalloying additions in steels. They can influence
D the recovery and recrystallization processes in both hot
layer Point layer layer
T rolling of austenite and cold rolling and annealing of
ferrite. It is well known that they do so by solute drag
D
and/or pinning of sub-grain boundaries by nitride and
carbide precipitates, thereby influencing the final mi-
Fig. 1—Schematic illustration of areas examined in each specimen.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 46A, AUGUST 2015—3637


crostructure and properties. There have been some

Microhardness (VHN)
300
interesting recent studies of the effects of niobium and/ Ti-based
or titanium additions in steels. Belyakov et al., studied Nb-based
250
the development of primary recrystallization in ferrite in
a steel containing 0.4 pct of volume of fine TiC 200
precipitations with an average particle size of
12 nm.[21] After sufficiently large cold strains, the 150
recrystallization developed readily upon annealing at CR 350 540 560 580 600 620 640 660
temperatures above 873 K (600 C). Not surprisingly, Quench Temperature (°C)
an increase in the cold strain as well as the annealing
temperature resulted in the acceleration of recrystalliza- Fig. 2—Comparison of average through thickness microhardness
tion kinetics. However, a certain amount of the cold values in specimens quenched from different annealing temperatures.
worked microstructure, about 15 pct of volume, re-
mained unrecrystallized even after annealing at a rather bearing steel, and it seems to reach a lower plateau at
high temperature of 973 K (700 C). The unrecrystal- around 903 K (630 C), while the hardness of Ti-bearing
lized portions were composed of grains with the h0 0 1i steel was still continuing to drop up to 933 K (660 C).
crystallographic direction parallel to the compression Therefore, the Nb steel had a faster rate of softening, i.e.,
axis. Both the low stored energies in these grains after recrystallization, than the Ti-bearing steel. The higher
cold deformation and the pinning of recrystallizing grain annealing kinetics exhibited by the Nb steel are important
boundaries by the dispersed carbides were discussed as since they would help minimize the difference in mi-
crucial factors that resulted in the incomplete recrystal- crostructure generated during the annealing process. The
lization. On the other hand, small amounts of niobium sluggish annealing kinetics shown by the Ti steel means
in steels have also been known to affect both the that the final microstructure would be more sensitive to
microstructure and properties. The topic of niobium in variations within the process, i.e., annealing near the hot
steels has been covered extensively in a review paper by spot versus cold spot.
DeArdo.[22] Recent research by Hutchinson et al., illus- This difference in annealing behavior could also be
trated the comparative effectiveness of solute Nb and observed by the microstructural changes observed in the
NbC particles in impeding grain boundary motion with SEM and shown in Figure 3, using secondary electron
a theoretical treatment.[23] The Nb can be present both diffraction imaging. This figure shows the microstructure
as solute in solution, where it is thought to exhibit a evolution at the centerline of the Nb-bearing steel from
strong solute drag effect, or as NbC precipitates, which Figures 3(a) through (e), while Figures 3(f) through (j)
are thought to be effective at pinning sub-grain and show the Ti-bearing steel. This same evolution is present-
grain boundaries. It was shown that, for a steel ed in Figure 4 based on the KAM approach. In Figure 4,
containing ~0.05 at pct Nb (~0.1 wt pct Nb), under only the center region of the Nb steel was shown because
typical recrystallization conditions, solute Nb is more the surface and center regions showed similar kinetics.
effective in the ferrite, whereas in the austenite, depend- Both the surface and center regions of the Ti steel were
ing on the exact recrystallization temperature, either used, since they exhibited such different kinetics. Clearly,
solute Nb or NbC precipitates may be more effective. the Nb-bearing steel recrystallized much faster than the
The goal of this current study was to determine the Ti-bearing steel, based on grain shape or aspect ratio, and
annealing kinetics of two cold-rolled microalloyed the KAM results, all of which corresponded well with the
HSLA steels during a batch annealing simulation, and hardness drops.[6]
to determine the operative driving and retarding forces. Kernel average misorientation maps shown in Fig-
The driving force was to be followed using the EBSD ure 4 calculated from EBSD data are used in this study
Stored Energy approach, while the retarding force was to evaluate the fraction of recrystallized area in each
to be evaluated by precipitate observations using thin condition. As mentioned above, in the KAM data
foil electron microscopy. A better understanding of representation, the red or orange colored regions denote
these forces would then be the basis for understanding high lattice distortion or high dislocation density, while
how to both improve property uniformity and shorten the blue colored regions denote a relatively undistorted
the required process time. lattice, i.e., full recrystallization. The fractions of re-
crystallized area were compared in Figure 5. The first
thing to note in the data of Figure 5 is that the kinetics
are faster in the surface regions than the center for each
III. RESULTS AND DISCUSSION
steel. The kinetic curves can be compared with regard to
The changes in microhardness (VHN-500 g) values, rates in the sequence from high to low: Nb(Surface),
measured at the mid-thickness location, from the cold Nb(Center), Ti(Surface), and Ti(Center), where the first
rolled to the annealed conditions up to 933 K (660 C) for three are similar, but the fourth is much slower. A
both steels are shown in Figure 2. A significant drop in comparison of the surface to center behavior revealed
hardness occurred between 853 K and 873 K (580 C and that the Nb-bearing steel had a more uniform annealing
600 C), meaning that significant softening by advanced behavior than did the Ti-bearing steel, since the surface
recovery or polygonization or the onset of recrystalliza- and center recrystallization kinetics in the Nb steel were
tion had started there for both steels. But the hardness of very similar. It should also be noted that the surface and
Nb-bearing steel decreased much faster than did the Ti- center curves indicated that annealing was nearly

3638—VOLUME 46A, AUGUST 2015 METALLURGICAL AND MATERIALS TRANSACTIONS A


Fig. 3—Microstructure evolution of Nb-bearing steel: (a) cold rolled, (b) 873 K (600 C), (c) 893 K (620 C), (d) 913 K (640 C), (e) 933 K
(660 C); and Ti-bearing steel: (f) cold rolled, (g) 873 K (600 C), (h) 893 K (620 C), (i) 913 K (640 C), (j) 933 K (660 C). Centerline observa-
tions.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 46A, AUGUST 2015—3639


Fig. 4—Kernel average misorientation maps of Nb-bearing steel at center layer: (a) 873 K (600 C), (b) 893 K (620 C), (c) 913 K (640 C), (d)
933 K (660 C); and Ti-bearing steel at surface layer: (e) 873 K (600 C), (f) 893 K (620 C), (g) 913 K (640 C), (h) 933 K (660 C); and
Ti-bearing steel at center layer: (i) 873 K (600 C), (j) 893 K (620 C), (k) 913 K (640 C), (l) 933 K (660 C).

completed throughout the Nb steel at 898 K (620 C), property level variation frequently observed between the
whereas the annealing of the Ti steel was still evolving two steels in batch annealing.
up to 948 K (670 C), especially at the mid-thickness. The grain boundary misorientation distributions for
These different features from surface to center and in the both steels were also measured using EBSD. The
time scale are very likely the cause of the difference in fraction of low-angle grain boundaries (<15 deg) is high

3640—VOLUME 46A, AUGUST 2015 METALLURGICAL AND MATERIALS TRANSACTIONS A


100 Stored energy distribution maps and changing trends
calculated from the EBSD scanning data from the
90
centerline location using the grain center sub-grain
Fraconof recrystallized area (%) 80 method are shown in Figures 7 and 8. Since the starting
hot band grain sizes are different, it was decided that the
70
total stored energy should be used, as it comprised
60 contributions from both the grain boundaries and grain
center sub-grain boundaries. The total stored energy in
50
the surface and centerline regions of both hot rolled and
40 cold rolled, including contributions from both grain
Ti-bearing Surface center sub-grain boundaries and ferrite grain bound-
30 aries, is presented in Figures 9 and 10, and summarized
Ti-bearing Center
20 in Tables II, III, and IV. Figures 7, 8, and 10 were
Nb-bearing Surface reconstructured from data taken at 10009 with a step
10 Nb-bearing Center size of 0.2 lm.
0
The data of Table II, calculated based on the average
CR 580 600 620 640 660 670 670-1 670-2 670-4 670-6 grain size and assuming no variation in ferrite grain size
Annealing Stages or Temperature (°C) through the thickness in the hot-rolled condition, reveal
that the GB contribution is about 10 pct of the grain
Fig. 5—Comparison of the fraction of recrystallized area during an- center sub-grain contribution for both steels, and
nealing based on KAM measurements. furthermore comprises about 10.3 pct (Ti steel) to
11.4 pct (Nb steel) of the total stored energy. The stored
energy in the Nb steel is 87 pct higher than for the Ti
0.45 steel at the surface and 34 pct higher in the centerline
0.4 regions. Clearly, the hot band of the Nb steel has a much
larger stored energy than the Ti steel.
0.35 Table III, for the cold-rolled conditions, shows that
0.3
the total stored energy for the Nb steel is 11.4 pct higher
than for the Ti steel in both the surface and centerline
Fraction

0.25 regions. However, the total stored energy of the cold


band for each location agrees well with the observed
0.2
recrystallization kinetics as presented in Figure 5, and
0.15 therefore can help explain the observed variation in
annealing behavior at different locations in the two
0.1
steels. Higher stored energy in the starting cold-rolled
0.05 state supplies higher initial driving force for recrystal-
lization which led to a higher rate of recrystallization.
0 Therefore, based on the data of Table III, the Nb steel
0 5 10 15 20 25 30 35 40 45 50 55 60 65
Misorientation (Degree)
surface would be expected to have the highest recrys-
tallization kinetics, followed by the Nb steel centerline,
Ti steel surface, and then Ti steel centerline. This is what
Cold Rolled Partially Recrystallized
is observed, in fact. The data of Table IV illustrate the
Nearly Fully Recrystallized increase in total stored energy when the hot band
undergoes cold rolling. Again, the increase in total
Fig. 6—Misorientation distribution changing trend during annealing. stored energy resulting from cold rolling correlates well
Centerline area-Ti steel. with the variation of annealing kinetics with steel and
location, i.e., Nb steel surface the highest and Ti steel
center the lowest.
in the fully hard cold-rolled samples, since the very high Figure 10 reveals that there is a certain amount of
dislocation density resides mainly in the sub-grain, i.e., deformed areas and grains that have rather little stored
low-angle boundaries. As annealing progresses and energy, even after cold rolling. A similar observation
recrystallization commences, these low-angle boundaries was also made in the Belyakov et al. study discussed
are replaced by high-angle grain boundaries. Therefore, earlier.[21] There can be two interpretations of this
as annealing progresses, the amount or fraction of low- observation. First, that there can be a change in grain
angle boundaries will diminish, while the high-angle shape in cold rolling with little accumulation of stored
boundaries (>15 deg) increases, as presented in Figure 6 energy, and second, that the dislocations generated
for the centerline region of the Ti steel. The transition during cold rolling were very rapidly incorporated into
from low-angle to high-angle boundaries is another the sub-grain boundaries, i.e., either dynamic or rapid
indication of the consumption of the stored energy as static recovery. A similar phenomenon has been ob-
recrystallization progresses. Both steels exhibited similar served before in the cold deformation of iron single
trends, but at different annealing stages. crystals.[24]

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 46A, AUGUST 2015—3641


(a) (e)

Min: 0.503, Max: 17.956 Min: 0.487, Max: 17.978


(b) (f)

Min: 0.319, Max: 18.527 Min: 0.418, Max: 18.202


(c) (g)

Min: 0.303, Max: 16.304 Min: 0.475, Max: 18.016


(d) (h)

Min: 0.309, Max: 13.473 Min: 0.338, Max: 18.111

Fig. 7—Stored energy distribution maps of Nb-bearing steel at center layer: (a) 873 K (600 C), (b) 893 K (620 C), (c) 913 K (640 C), (d)
933 K (660 C); and Ti-bearing steel at center layer: (e) 873 K (600 C), (f) 893 K (620 C), (g) 913 K (640 C), (h) 933 K (660 C).

3642—VOLUME 46A, AUGUST 2015 METALLURGICAL AND MATERIALS TRANSACTIONS A


Ti-bearing Fr Nb-bearing Fr Ti-bearing SE Nb-bearing SE

100.00 6.00

Fraction of Recrystallized Area (%)


5.68

Average Stored Energy (J/cm3)


90.00
5.15 5.12 5.00
80.00 4.88
70.00 4.00
60.00 3.73

50.00 3.08 3.00

40.00
2.00
30.00
1.64
20.00 1.27
0.92 1.00
10.00 0.71

0.00 0.00
Cold Rolled 600°C 620°C 640°C 660°C
Stage

Fig. 8—The change in average grain center stored energy (SE) and the fraction of recrystallized area (Fr) during annealing.

(a) (b) (c) (d)

Min: 0.453, Max: 16.694 Min: 0.328, Max: 16.366 Min: 0.282, Max: 16.126 Min: 0.265, Max: 16.483

Fig. 9—Stored energy distribution maps for hot band samples of (a) Nb steel at surface layer, (b) Nb steel at center layer, (c) Ti steel at surface
layer, (d) Ti steel at center layer.

(a) (b) (c) (d)

Min: 0.724, Max: 18.059 Min: 0.810, Max: 18.141 Min: 0.384, Max: 18.515 Min: 0.377, Max: 18.274

Fig. 10—Stored energy distribution maps for cold-rolled samples of (a) Nb steel at surface layer, (b) Nb steel at center layer, (c) Ti steel at sur-
face layer, (d) Ti steel at center layer.

Since the lattices of the microalloyed precipitates they form on grain or sub-grain boundaries, they can
(NaCl type) do not fit well with either the austenite hinder the motion of these defects leading to the
(FCC) or ferrite (BCC) lattices, the particles must suppression of recovery and recrystallization. Palmiere
form heterogeneously on pre-existing crystalline de- et al. showed that the local pinning force exerted by
fects such as grain or sub-grain boundaries, interphase particles on a flexible sub-grain boundary can be
interfaces or even individual dislocations.[22] When expressed in Eq. [3]:

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 46A, AUGUST 2015—3643


Table II. Comparison of the Stored Energy (J/cm3) in the Hot Band Before Cold Rolling

Ti Steel Nb Steel
Steel/Location
S. E. Region GB Matrix Total GB Matrix Total
Surface 0.280 0.750 1.03 0.493 1.40 1.893
Centerline 0.242 0.651 0.893 0.323 0.87 1.193

Table III. Comparison of Stored Energy (J/cm3) with Location in Cold Band Before Anneal

Ti Steel Nb Steel
Steel/Location
S. E. Region GB Matrix Total GB Matrix Total
Surface 0.470 5.52 5.99 0.582 6.09 6.672
Centerline 0.470 5.15 5.62 0.582 5.68 6.262

Table IV. Total Change in Stored Energy (J/cm3) Hot Rolled to Cold Rolled

Steel/Location CR Total HB Total Difference


Ti surface 5.99 1.03 4.96
Ti center 5.62 0.893 4.729
Nb surface 6.672 1.893 5.479
Nb center 6.262 1.193 5.069

IV. CONCLUSIONS
FP ¼ 3cfl=ð2pr2 Þ; ½3
1. The Batch Annealing Simulation used in this study
where c is the surface energy between the particle and can be useful in programs to reduce property vari-
the matrix, f is the local volume fraction of particles, l is ability along the length of annealed coils and to un-
the sub-grain diameter, and r is the particle radius.[25] derstand how to accelerate the process.
Therefore, for an array of particles to exert a large 2. The KAM and stored energy maps are valuable
pinning force on a boundary, the ratio of f/r2 must be methods of studying the formation and elimination
high. In this investigation, several types of particles were of retained strain or stored energy during cold roll-
observed in annealed specimens, TiN, Ti4C2S2 and TiC ing and annealing. They are also helpful in follow-
in the titanium steel and Fe3C in the niobium steel. ing the progress of softening during annealing.
These particles were identified by selected area diffrac- 3. The two steels used in this study exhibited different
tion and EDS.[6] Although there were fine particles recrystallization kinetics principally because of dif-
observed in both steels, very few exhibited a f/r2 ratio ferent levels of stored energy in the cold-rolled con-
high enough to cause possible pinning effects that would dition. At similar cold reductions, the difference in
be expected to retard the observed annealing kinetics. stored energy was caused by different hot band
The role of particle pinning was, therefore, disregarded. stored energy and also possibly by variations in
It should be mentioned that the small amount of TiC work hardening during the cold rolling.
precipitates found in the titanium steel exhibited the 4. The Ti steel would be expected to have a larger var-
Kurdjumov–Sachs orientation relationship, which iation in properties along the coil because of the
means the particles precipitated in the prior austen- large difference between hot spot and cold spot an-
ite.[6,22] The absence of fine TiC or TiN with large f/r2 in nealing behavior. The large difference observed be-
the Ti steel during annealing might be caused by tween surface and center regions observed in these
annealing well below the nose of the TTT curve for experiments for the Ti steel is a predictor of varia-
precipitation of TiC in ferrite,[26] or by particle coars- tion in properties along the coil length. The smaller
ening during the very slow heating rate and long difference for the Nb steel means more uniform
annealing times at the subcritical temperatures.[27] Jang properties.
et al., have shown that the TiC particle size can 5. The hot band microstructure plays an important
approach 3000 lm in radius at 928 K (700 C) in steels role in the behavior during cold rolling and anneal-
that are hyper-stoichiometric with respect to Ti/C ratio ing. Increasing the stored energy of the hot band
and held 1 hour at temperature.[27] This might be using higher cooling rates on the runout table, and
another reason why particles fine enough to retard lowering the coiling temperature, or using larger
recovery and recrystallization were not observed in the cold reductions, would be expected to increase the
Ti steel, which is also hyper-stoichiometric. hot band stored energy, shorten the annealing time,

3644—VOLUME 46A, AUGUST 2015 METALLURGICAL AND MATERIALS TRANSACTIONS A


and increase property uniformity along the coil 8. R. Abbaschian and R. Reed-Hill: Physical Metallurgy Principles,
length. 4th ed., CT, Cengage Learning, Stamford, 2009.
9. W.H. Robinson: Techniques for the Direct Observation of Structure
and Imperfection, Techniques in Metals Research, Vol. 2, Part 1,
Interscience Publishers, New York, 1968, pp 291–340.
10. P. Hirsch, A. Howie, R.B. Nicholson, D. Pashley, and M.J.
Whelan: Electron Microscopy of Thin Crystals, Butterworths,
ACKNOWLEDGMENTS London, 1967, pp. 415–33.
11. F.J. Humphreys: J. Mater. Sci., 2001, vol. 36, pp. 3833–54.
The authors would like to thank the steel companies 12. S.-H. Choi: Mater. Sci. Forum, 2002, vols. 408–412, pp. 469–74.
that form part of the BAMPRI-Thin Slab Casting and 13. J. Wu: PhD Thesis, University of Pittsburgh, 2005.
Rolling Consortium for providing the steel samples 14. S.H. Choi, F. Barlat, and J.H. Chung: Scripta Mater., 2001,
and for their financial support for this program. In ad- vol. 45, pp. 1155–62.
dition, the authors would like to thank the Basic Me- 15. N. Rajmohan, Y. Hayakawa, J.A. Szpunar, and J.H. Root: Acta
Mater., 1997, vol. 45, pp. 2485–94.
tals Processing Research Institute (BAMPRI) and the 16. M.L. Taheri, H. Weiland, and A.D. Rollett: Metall. Mater. Trans.
Mechanical Engineering and Materials Science Depart- A, 2006, vol. 37A, pp. 19–25.
ment, University of Pittsburgh for providing the fa- 17. S.H. Choi and Y.S. Jin: Mater. Sci. Eng. A, 2004, vol. A371,
cilities to conduct this work. Special thanks are pp. 149–59.
18. W.T. Read and W. Shockley: Phys. Rev., 1950, vol. 78, p. 275.
extended to Dr. Kengun Cho, Dr. Xiaojun Liang, and 19. E.E. Underwood: in Quantitative Microscopy, R.T. DeHoff and
Professor Mingjian Hua as well as to Hoe-Seok Yang, F.N. Rhines, eds., McGraw-Hill, New York, 1968, p. 77.
Long Li, Albert Stewart, Cole Van Ormer and Dr. 20. L.N. Brewer, D.P. Field, and C.C. Merriman: in Electron
Ken Goldman, for their generous and constant help. Backscatter Diffraction in Materials Science, A.J. Schwartz, M.
Kumar, B.L. Adams, and D. Field, eds., Springer, New York,
2009.
21. A. Belyakov, F.G. Wei, K. Tsuzaki, K. Kimura, and Y. Mishima:
REFERENCES Mater. Sci. Eng. A, 2007, vol. 471, pp. 50–56.
22. A.J. DeArdo: Int. Mater. Rev., 2003, vol. 48 (6), pp. 371–402.
1. S.R. Goodman and H.S.L.A. Steels: Technology & Applications, 23. C.R. Hutchinson, H.S. Zurob, C.W. Sinclair, and Y.J.M. Brechet:
ASM, Materials Park, 1984, pp. 239–252. Scripta Mater., 2008, vol. 59, pp. 635–37.
2. L.J. Cuddy: Plastic Deformation of Metals, Academic Press, New 24. H. Hu: in Recovery and Recrystallization of Metals, L. Himmel,
York, 1975, pp. 129–140. ed., Interscience, New York, 1963, pp. 311–78.
3. A.J. DeArdo: Microalloying ‘95, The Iron and Steel Society, 25. E.J. Palmiere, C.I. Garcia, and A.J. DeArdo: Metall. Mater.
Warrendale, 1995, pp. 15–35. Trans. A, 1996, vol. 27A, pp. 951–60.
4. D.O. Wilshynsky-Dressler, D.K. Matlock, and G. Krauss: Physical 26. S. Freeman: The Effect of Second Phase Particles on the Me-
Metallurgy of IF Steels, The ISI Japan, Tokyo, 1994, pp. 13–33. chanical Properties of Steel, The Iron and Steel Institute, London,
5. H. Takechi: Microalloying ‘95, The Iron and Steel Society, War- 1971, pp. 152–56.
rendale, 1995, pp. 71–82. 27. J.H. Jang, H.L. Chang, N.H. Heung, H.K.D.H. Bhadeshia,
6. C. Fang: Ph. D. Thesis, University of Pittsburgh, 2011. and W.S. Dong: Mater. Sci. Technol., 2013, vol. 29, pp. 1074–
7. A.L. Titchener: Prog. Met. Phys., 1958, vol. 7, pp. 247–338. 79.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 46A, AUGUST 2015—3645

View publication stats

You might also like