Journal of Materials Processing Technology 155–156 (2004) 1307–1312

High temperature oxide scale characteristics of low

carbon steel in hot rolling
Weihua Sun∗ , A.K. Tieu, Zhengyi Jiang, Cheng Lu
Faculty of Engineering, University of Wollongong, North-fields Avenue, Wollongong, NSW 2522, Australia

Abstract

In this paper, hot rolling tests of low-carbon steel were carried out on a 2-high Hille 100 experimental rolling mill at various speeds and
reductions. The rolling temperatures were between 1000 and 1030 ◦ C. Nitrogen protection was used to control the scale thickness when
the test samples were heated in the furnace and cooled in a cooling box. The rolling forces, rolling torques and scale thicknesses before and
after rolling were measured. Surface characteristics of the steel samples were analyzed with SEM and Atomic Force Microscope (AFM).
X-ray diffraction was performed to characterize the phase composition of the scale layers. The effects of scale thickness on rolling force
and torque were also investigated.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Hot rolling; Oxide scales; Surface roughness

1. Introduction range of 830–1150 ◦ C. Yu and Lenard [9] estimated the re-

sistance to deformation of the scale layer during hot rolling
A scale layer is always formed on the strip surface dur- of carbon steel strips.
ing hot rolling process on a hot strip mill. Primary scales However, little is known how the surface texture is trans-
formed on the surface of slabs in the reheating furnace ferred from one pass to the next. It has been suggested that
are removed by descaler. Secondary scale builds up again the texture is established by the work roll surface roughness
when the slab is waiting for further processing before each during the rolling process [10]. In the case of hot rolling, the
pass and before entering the finishing stands. Furthermore, actual surface texture is generated by a combination of the
tertiary oxidation scale layer develops after the slab is asperity contact and the lubricant contact between the strip
repeatedly descaled. Thus, the roll/strip interface in hot and the roll. It was assumed that the scale is compressed and
rolling always includes layers of scale. When the scale is begins to be elongated when it enters the roll bite [11]. When
subjected to rolling in the roll bite, it can act as a lubricant the oxide scale fractures and undertakes extreme pressure
if it is thick and ductile, or as abrasives in the three-body during rolling, the hot metal is extruded partially up into the
wear mechanism if it is hard and brittle. fine cracks and hence modifying the strip surface roughness.
Previous researches show that the oxide scale on the The objective of the present work is to examine the effect
steel surface contains three layers: hematite, magnetite and of hot rolling parameters on the characteristics of the oxide
wustite [1–3], even after any oxidation period longer than scale of low carbon steel. The morphology of the scale and
0.6 s [4]. Each scale element has different morphology and surface features of the strip before and after rolling was
mechanical properties [5,6]. Tiley et al. [7] carried out analyzed. Deformation behavior of the scale and its effects
hot compression tests on industrial reheating furnace scale on the rolling force and torques were also examined.
of mild steel, indicating considerable plastic deformation
when it was deformed at 30% strain with 0.1 and 1.0 s−1
strain rate over the temperature range of 650–850 ◦ C. 2. Experiment equipment, materials and procedures
Krzyzanowski and Beynon [8] conducted hot tension tests
after producing scales of 10–300 ␮m in thickness and also 2.1. Equipment
found noticeable plastic deformation in the temperature
Hot rolling experiments were carried out on a 2-high
Hille 100 experimental rolling mill with rolls of 225 mm di-
∗ Corresponding author. ameter and 254 mm barrel roll, which is made of high speed
E-mail address: [email protected] (W. Sun). steel (HSS) with a hardness of HRC55 and Ra of 0.4 ␮m in

0924-0136/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmatprotec.2004.04.167
1308 W. Sun et al. / Journal of Materials Processing Technology 155–156 (2004) 1307–1312

Table 1 cedure took about 23–25 min. The sample was quickly rolled
Chemical composition of the steel at 1000 ◦ C with a reduction from 7.5 to 41% and rolling
Element Chemical composition (wt.%) speeds between 0.12 and 0.72 m s−1 . After rolling, the sam-
C 0.18
ple were transferred into the cooling box which was con-
Si 0.18 nected to N2 flow and a flow rate of 20 l min−1 was selected.
Mn 0.95 This rolling test was carried out one sample after another.
P 0.026 The other series of tests were carried out in a similar proce-
S 0.027 dure, but the furnace was not connected with N2 and the steel
Cr 0.10
Ni 0.067
samples were manually tapped to remove the oxide scale
Cu 0.13 which was developed while soaking in the furnace. This is to
Mo 0.19 remove the surface primary scale before rolling and to sim-
Al–T 0.004 ulate the descaling effect. There were five or four samples
Ti <0.003 heated each time for 55–70 or 145–160 min, respectively, to
examine the effect of heating time on the behavior of the
secondary scale under different deformation conditions.
surface finishing. This rolling mill can operate at a rolling Samples for AFM, SEM and microscopic analysis were
speed up to 1 m s−1 , a load up to 150 t and a torque up to prepared from the rolled plates.
13000 kN m. It is equipped with two load cells to determine
the compensations for the mill modulus and bearing clear-
ance, and two torque gauges on the shaft to measure the 3. Results and discussion
individual roll torque. Two position transducers are used
to control the roll gap. Two DMC-450 radiation thickness 3.1. Topography of the oxide scale before and after rolling
gauges at entry and exit and a hydraulic AGC are available on
the Hille 100 rolling mill. A Pentium III computer was used As mentioned before, the steel surface is covered with a
in the experiment. Its maximum sampling rate is 250 k s−1 . layer of oxide scale during hot rolling. Thus, the topology of
In addition to the two optical pyrometers for controlling the rolled steel product surface here is actually the topology
the surface temperature, a Flir PM390 Thermo-camera by of oxide scale. 3D images of the steel surface after rolling
Thermoteknix Systems Limited was also used to measure can be seen from Fig. 1(a) and (c). The samples were heated
the temperature field of the strip surface. Nitrogen was fed for 60 and 65 min of which the soaking period is for 10
to the hearth of the reheating furnace in order to control and 15 min, respectively. After they were extracted from the
the scale thickness and the steel bars were cooled in a box reheating furnace, manual descaling operation was carried
connected with nitrogen gas straight after rolling. out mechanically on the samples at the entry of the roll bite.
A Nanoscope IIIA AFM from Digital Instruments and About 4–6 s later, the samples were rolled. Then the samples
Leico E440 SEM were used to analyze the topography of were picked up at the exit of roll bite and stored in a cooling
the strip surface. A Leico optical microscope and a Philips box, which was connected to N2 flow. The total time taken
PW1730 X-ray diffraction meter were also used characterize from descaling to the cooling box were 10.5–13 s.
the morphologies and kinetics of the oxide scale. Section analysis (Fig. 1(b) and (d)) was carried out along
the rolling direction with Digital Nanoscope software ver-
2.2. Materials sion 5.12b. As for the sample deformed at 25% at the rolling
speed of 0.12 m s−1 , the maximum vertical distance between
The test specimens were commercial low-carbon steel the peak and the valley, indicated with reversed triangles,
strip of dimension 10 mm×100 mm in thickness and width. was 1.912 ␮m over a horizontal distance of 6.45 ␮m (see
They were machined to 9.20 mm thickness, 100 mm width Fig. 1(b)). When the sample was deformed at 33.7% at the
and 450 mm length. The surfaces were carefully prepared rolling speed of 0.12 m s−1 , the value of the maximum ver-
to 0.5 to 0.7 ␮m in surface finishing, so that any influence tical distance was 1.391 ␮m over a horizontal distance of
of the original surface defects occurred from the industrial 6.04 ␮m. These values seem to be much smaller than those
manufacturing could be prevented. Table 1 shows the chem- in [12], which were obtained from simulated oxidizing tests
ical composition of the steel. on a Gleeble 3500 Thermo-Mechanical Simulator. As it will
be discussed later, the total thickness of the scale on the
2.3. Experiment procedure descaled steel surface after rolling is <26 ␮m. Deformation
effects could have an influence on the surface topology.
The temperature of the furnace was first raised to 1100 ◦ C A further examination of the surface morphology was
and the nitrogen gas was then connected. Five minutes later, conducted by scanning on the sample surface with SEM.
the test sample was put into the furnace hearth, waiting for Fig. 2 is the micrographs taken on the secondary oxide scale
the furnace temperature to reach 1100 ◦ C, the sample was surfaces that were non-deformed, micrograph (a), and after
soaked for 10 min to ensure uniform temperature. This pro- deformed 7.6% (Fig. 2(b)) and 25% (Fig. 2(c)), respectively,
 W. Sun et al. / Journal of Materials Processing Technology 155–156 (2004) 1307–1312 1309

Fig. 1. 3D image and section analysis of the rolled steel product surface: (a) and (b) after 25% deformation; (c) and (d) after 33.7% deformation. Rolling
speed was 0.12 m s−1 .

at the rolling speed of 0.12 m s−1 . On the non-deformed ent mechanical properties, the scale may display a different
oxide scale surface, there are occasionally irregular long behavior when deformed together with hot steel substrate
thin cracks, which are probably attributed by thermal-stress by compression and tension.
during cooling, see photo (a) in Fig. 2. After deformation, Fig. 3 illustrates the microstructures of the primary and
short and coarse cracks appear transversely to the rolling secondary oxide scales before and after deformation. The
direction (Fig. 2(b) and (c)). These SEM micrographs reveal samples for the test of primary oxide scale in Fig. 3 were
that visible plastic deformation the scale and the substrate heated one by one for about 23–25 min each in the furnace.
have occurred. This is similar and consistent to the wear Results of the rolling test show that the primary scale exhibits
behavior which was observed by Vergne et al. [13]. Fig. 2(b) its significant plastic deformation. Before deformation, the
and (c) also show that the plastically deformed areas exhibit thickness of the scale turns out to be 297 ␮m, which is much
the extruded “under-cells” where the surfaces are oxidized. larger than the values in [9,14]. The reason could be that the
heating time in [9,14] was much shorter than the present test.
3.2. Deformation behavior of the oxide scale and Cracks and large pores can be found in Fig. 3(a). After it was
morphological features of the secondary scale deformed 15.8% at rolling speed of 0.12 m s−1 , the primary
scale became 220 ␮m (Fig. 3(b)). At the deformation of
Morphologies of the primary oxide scale were discussed 33.7% when it was rolled at the speed of 0.48 m s−1 , the
in previous research [1–3,8,9]. However, due to their differ- thickness was 127.7 ␮m (Fig. 3(c)). This means that the scale

Fig. 2. Secondary oxide scale surface micrograph: (a) non-deformed; (b) and (c) deformed.
1310 W. Sun et al. / Journal of Materials Processing Technology 155–156 (2004) 1307–1312

Fig. 3. Microstructures of the primary and secondary oxide scales before and after deformation: (a) and (d) non-deformed primary and secondary scale;
(b) primary scale deformed 15.8% at rolling speed of 0.12 m s−1 ; (c) primary scale deformed 33.7% at rolling speed of 0.48 m s−1 ; (e) and (f) secondary
scale deformed 15.8% at rolling speed of 0.24 and 0.72 m s−1 , respectively.

deformed 25.9% when the sample was subjected a 15.8% components in the secondary layer, which are hematite, mag-
deformation and that it would deform 57% when the sample netite and wustite. Magnetite shows its strongest presence
deformed 33.7%. There is a significant trend that the scale at 2θ ◦ of 30.1, 35.5, 57.1 and 62.6◦ . However, hematite re-
thickness decreases quickly with the bulk deformation. The veals its significant presence at 33.1, 35.7, 49.5 and 54.2◦
rolling speeds also play a similar role on the thickness of even though it is shadowed by magnetite (Fig. 5).
the oxide scale (see Fig. 4). From Fig. 3(d), the interface between the scale and the
Photos in Fig. 3(d)–(f) are the microstructures for the sec- substrate became rougher. Under a reduction of 15.8%, the
ondary oxide scales before and after deformation. The sam- scale thickness became 24 and 24.2 ␮m, respectively, at
ples were heated for 145–160 min before it was descaled. rolling speeds of 0.24 and 0.72 m s−1 . Although a 15.8%
The initial secondary scale thickness was 29.5 ␮m in this ex- bulk deformation could reduce the thickness of the sec-
periment. X-ray diffraction results show that there are three ondary scale about 13.5%, the rolling speed does not seem

Fig. 4. Effects of reduction and rolling speed on the thickness of primary and secondary oxide scales. (a) Reductions from 7.6 to 41.3%; (b) rolling
speeds from 0.12 to 0.72 m s−1 .
 W. Sun et al. / Journal of Materials Processing Technology 155–156 (2004) 1307–1312 1311

Fig. 5. X-ray diffraction result of the secondary oxide scale.

to have an effect on the secondary scale thickness (see 3.3. Surface roughness
Fig. 4(b)).
On the other hand, the bulk reduction does not have re- After rolling, surface roughness was measured to examine
markable effects on the thickness of secondary scale (see the effects of rolling parameters on the surface roughness.
Fig. 4(a)). The samples were heated for 55–70 min totally. The results are shown in Fig. 6.
The approximate thickness of the secondary scales were Bulk deformation affects significantly the surface finish
10.9–11.4 ␮m after the samples were deformed 7.6, 25 of the sample surface. When the bulk deformation increases,
and 33.7% at rolling speed of 0.12 m s−1 . This could be the surface roughness decreases remarkably when oxide
the reason that the total scale thickness was so small that scale is thick (see Fig. 6(a)). For the steel surface with pri-
the difference in scale thickness, which was caused by the mary oxide scale, the surface roughness increases with re-
deformation, could be compensated by oxidation while the duction at first. According to the present test, after reduction
sample was delivered to the cooling box. However, the re- reached 15.8%, surface roughness decreases significantly
sults of the test show another interesting outcome that the with reduction. As for the descaled surfaces, the roughness
holding time of the samples will affect the secondary scale decreases very quickly when reduction increases from 7.6 to
thickness significantly. The magnitude of the scale thick- 25%. But when the bulk deformation increases further, the
ness value for 55 to 70 min heating time (Fig. 4(a)) is less roughness reduction slows down. However, when there is a
than half of the value of the scales which were produced thin scale layer on the steel surface, the ability of deforma-
in 145–160 min (Fig. 4(b)). This implies that the thermal tion to improve the surface roughness is impaired.
history of the steel in the furnace plays an important role On the other hand, the effect of rolling speeds on the
on the oxide scale thickness even after descaling. surface roughness is complex in this experiment. In the case

Fig. 6. Effects of reductions and rolling speeds on the surface roughness.

1312 W. Sun et al. / Journal of Materials Processing Technology 155–156 (2004) 1307–1312

of 55–70 min of heating, the roughness of both the descaled References

and non-descaled increases with rolling speed. On the other
hand, the roughness decreases with increasing rolling speed [1] F. Matsuno, Blistering and hydraulic removal of scale films of rimmed
steel at high temperature, Trans. ISIJ. 20 (1980) 413–421.
when the sample heating time is 145–160 min. However,
[2] J.S. Sheasby, W.E. Boggs, E.T. Turkdogan, Scale growth on steels
heavy reduction has a significant influence on the surface at 1200 ◦ C: rationale of rate and morphology, Met. Sci. 18 (1984)
finish. 127–136.
[3] I. Iordanova, M. Surtchev, K.S. Forcey, V. Krastev, High-temperature
surface oxidation of low-carbon rimming steel, Surf. Interface Anal.
30 (2000) 158–160;
4. Conclusion M.A. El Baradie, A fuzzy logic model for machining data selection,
Int. J. Mach. Tools Manufact. 37 (1997) 1353–1372.
In this paper, hot rolling tests were carried out on a [4] Weihua Sun, A.K. Tieu, Zhengyi Jiang, Hongtao Zhu, Cheng Lu,
Hille 100 mill. Topology and SEM analysis indicate that Oxide scales growth of low carbon steel at high temperatures, in:
Proceedings of AMPT2003, Dublin, 8–11 July 2003 (under review).
cracks occurred in the secondary oxide scale when sub-
[5] D.T. Blazevic, Tertiary rolled-in scale: the hot strip mill problem of
jected to bulk deformation and new “under-cell” extruded. the 1990s, in: Proceedings of the 37th MWSP Conference, ISS-AIME
Morphological examination shows that the primary ox- XXXVII (1996) 33–38.
ide scale deforms when the bulk reduction increases or [6] M. Schutze, An approach to a global model of the mechanical
when the rolling speed increases at a certain reduction. behaviour of oxide scales, Mater. High Temp. 12 (1994) 237–247.
[7] J. Tiley, Y. Zhang, J.G. Lenard, Hot compression testing of mild
However, an increase of bulk deformation or increase of
steel industrial reheat furnace scale, Steel Res. 70 (1999) 437–440.
rolling speed has limited influence in reducing the sec- [8] M. Krzyanowski, J.H. Beynon, The tensile failure of mild steel oxide
ondary scale thickness. The secondary oxide scale is com- under hot rolling conditions, Steel Res. 70 (1999) 22–27.
posed of three components similar to those for the primary [9] Y. Yu, J.G. Lenard, Estimating the resistance to deformation of the
scale. The thickness is greatly influenced by the heating layer of scale during hot rolling of carbon steel strips, J. Mater.
Proces. Technol. 121 (2001) 60–68.
history. More importantly, the bulk deformation has a sig-
[10] H.R. Le, M.P.F. Sutcliffe, Analysis of surface roughness of cold-rolled
nificant effect on the surface finish of the hot rolled steel aluminium foil, Wear 244 (2000) 71–78.
products. [11] J.H. Beynon, Y.H. Li, M. Krzyzanowski, C.M. Sellars, Measur-
ing modeling and understanding friction in hot rolling of steel, in:
M. Pietrzyk et al. (Eds.), Proceedings of the Metal Forming 2000,
Balkema, Rotterdam, 2000, pp. 3–10.
Acknowledgements [12] W. Sun, A.K. Tieu, Z. Jiang, C. Lu, H. Zhu, Surface characteristics
of oxide scale in hot strip rolling, J. Mater. Proces. Technol. 140
This project is supported by an ARC-linkage grant. The (2003) 76–83.
first author would like to thank Australian Government [13] C. Vergne, C. Boher, C. Levaillant, R. Gras, Analysis of the friction
and wear behavior of hot work tool scale: application to the hot
for scholarship support (IPRS and UPA) to undertake this
rolling process, Wear 250 (2001) 322–333.
research. The authors are greatly thankful for the assis- [14] A. Shirizly, J.G. Lenard, The effect of scaling and emulsion delivery
tance from Dr. Hongtao Zhu while the rolling tests were on heat transfer during the hot rolling of steel strips, J. Mater. Proces.
carrying out. Technol. 101 (2000) 250–259.