1. Introduction
Continuous casting is the most economical mass production method around the world. The caster machine requires precise process control on every part beginning from tundish to the final slab torch cutter. One of the most sensitive parts is guide rolls that are positioned immediately below the downstream of the bottomless copper mold to the end of the metallurgical length. Guide rolls operate in high-temperature environments where they are in contact with the thinnest part of the solidified shell. In the meantime, the mold flux (coming from the copper mold) accompanied by air mist blow (to cool down the solid shell and proceed with solidification) result in the formation of an extremely acidic environment attacking the guide rolls at high temperature [1,2]. Thus, it is essential to reduce the corrosive wear damage to the guide rolls by understanding the corrosion happening on the surface of the guide rolls.
The continuous casting guide rolls are made of plain Carbon steel with wire cladding. Martensitic stainless steel having more than 13 wt.% Cr with added C to increase Martensitic start temperature (MS) is normally used in several passes over the guide roll. The purpose is to improve corrosion resistance, thermal fatigue, and wear resistance of the guide roll’s surface while the material is easily weldable [3]. Among all the possible welding processes, submerged arc welding (SAW) has been successively utilized on cylinder-shaped guide rolls. This process can be easily automated resulting in well-aligned beads [4-6].
From a microstructural point of view, most of the martensitic stainless steels pass through peritectic reaction during solidification meaning that the primary phase formed from the melt is δ-ferrite followed by γ-austenite during cooling. At low-temperature ranges, the austenite phase transforms into martensite. Although, δ-ferrite phase is beneficial to enhance the weldability of the material due to its ductility [7-9], it has poor corrosion resistance. The amount of retained δ-ferrite can be estimated by calculation of Cr and Ni equivalents according to the Schaeffler diagram [10-13].
Hadizadeh et al. observed the formation of fatigue-related beach marks called zebra lines happening on the surface of the failed rolls in the shape of parallel lines. They have stated that the interconnected δ-ferrite network at the cladding overlay in the overlap region (the boundary between two passes) is responsible for the formation of zebra lines at the surface of the roll. In addition, secondary sigma phase has been mentioned to form in the intermediate temperatures causing hardness increase and leading to fatigue progression lines [2]. However, they haven’t stated the relationship between the retained δ-ferrite amount and formation of cracks at the beads overlap area. In addition, the corrosive wear not only occurs at the overlap region but also inside each bead at a less severe rate.
Although, Hadizadeh et al. have stated that severe corrosion occurs at the boundary between two welding passes (e.g. welding overlays) so called zebra lines [2], a more in-depth analysis is necessary to find out why the reason for severe corrosion and the mechanism of crack growth. Thus in the current work, a sample of guide rolls after over 3000 casting heats (here, number of heats refers to the number of ladles carrying molten steel casting into the tundish of the continuous caster) was analyzed to find out the mechanism of corrosion. Furthermore, the effect of retained δ-ferrite on the surface roughness profile was studied in the overlap region of the wire cladding beads and inside the beads as well.
2. Experimental
The guide roll samples used in this paper were taken out from a continuous casting machine after over 3000 number of heats. The surface image of the roll is shown in Fig. 1. Detailed chemical composition of the overlayer wire clad is presented in Table 1. To observe the surface roughness profile of the damaged guide roll was observed with 3D optical microscopy using Olympus DSX500 (Olympus Corporations, Tokyo, Japan). Microstructural observations were carried out by polishing and etching the top and cross-section of the samples with Vilella’s etchant [14]. To characterize the crystal structure and grain size of the samples, electron backscattered diffraction (EBSD) was used by a scanning electron microscope (FE-SEM (JEOL JSM-7900F, Tokyo, Japan) equipped with energy dispersive spectroscopy (EDS). The EBSD data was then extracted by Oxford Aztec ver. 6.0 and analyzed by AztecCrystal ver. 2.1.
Table 1. Chemical composition of the wire-cladded material used for coating the guide roll
Fig. 1. The used guide roll being installed in the first segment of the continuous casting machine while working for more than 3000 number of molten steel heats.
3. Results and discussion
3.1 Surface analysis
An image of the used guide roll is shown in Fig. 1. This guide roll has been used in the first section of the continuous casting machine for about over 3000 heats of molten steel. It can be seen that there are obvious parallel marks on the surface. Hadizadeh et al. have reported the zebra lines parallel to each other and stated that these parallel zebra lines correspond to the overlap region of two wire cladding rounds during clad welding process [2].
A closer look at the surface of the sample is shown in Fig. 2 at a selected zebra line. The 3D image clearly states that the zebra line possesses a deeper height compared with the other regions by average of 250 µm. In addition, out of the zebra line area, there exist different corrosion marks similar to grain boundary corrosion marks. Despite the martensite being the dominant phase at room temperature, another factor might affect the corrosion effects in these areas since the rolls operate at high temperatures (about 600℃) at the top segments in a continuous casting machine.
Fig. 2. Optical microscopy and 3D surface mapping of the used guide roll sample at (a), (b), (c) low magnification with higher magnification (d), (e) inside the zebra line, and (f), (g) outside the zebra line.
3.2 Microstructural characterization
Further analysis was conducted by grinding and polishing the cross-section and revealing the microstructure by etching. Fig. 3 shows the macrostructure of the cladding overlays positioning one zebra line in the center of the image. The higher magnification OM image of Fig. 3(b) depicts deep corrosion cracks (in blue arrows) grown into the cladding overlay. They appeared as deep corrosion marks on the surface like grain boundary corrosion marks (Fig. 2(f)). in addition, the overlap of the two different wire cladding bead overlays can be distinguished on the left side of Fig. 3(a). it appears that the overlap of the two beads does not correspond to the zebra line. In contrast, the heat-affected zone (HAZ) of the second bead belongs to the position of the zebra line which is contrary to the former literature [2].
Fig. 3. Macrostructure of wire cladding overlay on the used guide roll sample at the selected zebra line with higher magnification OM images at (b) at the cladding overlay and (c) at HAZ between two passes.
Due to the high-temperature working condition of the guide rolls, it is expected that at working temperature, is above the MS temperature of the wire cladded martensitic stainless steel. Therefore, the corrosion has taken place at the prior austenite grain boundary during operation [5].
SEM images were taken from the different distinct areas of the sample after polishing the ND and RD cross-sections presented in Fig. 4. Inside the zebra line, corrosion has occurred uniformly within the retained δ-ferrite phases (Fig. 4(a)-(b)). Out of the zebra line, the corrosion has intensified at a specific point and the crack has grown deep into the microstructure of the cladding overlay perpendicular to the surface (Fig. 4(c)-(d)). The corrosion product was formed around the crack.
Fig. 4. The SEM (a), (c) BSE images and (b), (d) SEI of the (a), (b) cross-section at the zebra line; (c), (d) cross-section out of zebra line at a corrosion crack mark.
EBSD was used to observe the microstructure of the cladding overlay and reconstruct the prior austenite grain. Fig. 5 shows a representative area of the top surface and RD cross-section. The grain boundary between two adjacent austenite grains can be seen in the reconstructed image (Fig. 5(b)). In EBSD Fig. 5(c), the only way to distinguish between martensite and δ-ferrite is by observing the orientation relationship and the strain induced in each phase by means of kernel average misorientation (KAM) map. In addition, the retained δ-ferrite has low levels of stains (blue color areas in Fig. 5(c)), while martensite contains high strain due to the displacive transformation nature of martensite (green color areas in Fig. 5(c)).
Fig. 5. EBSD (a) IPF map, (b) IPF reconstruct IPF map, and (c) KAM map of the cross-section of the top cladding overlay.
In order to construct the prior austenite grains, the well-known K-S orientation relationship was used to correlate the parent austenite with its corresponding child martensite phase according to the overlap of the two phases PFs [15,16]. The pole figures of martensite and reconstructed austenite are presented in Fig. 6 taken from a yellow dashed area in Fig. 5(b). By comparing both parent and daughter phases, it can be concluded that all of the martensite phases possess K-S (Fig. 6(h)); however, the retained δ-ferrite phases don’t obey either of them. Therefore, the reconstruction fails to obtain reliable austenite from the δ-ferrite phase.
Fig. 6. Pore figure (PF) of the parent fcc austenite at (a) {100}, (b) {110}, and (c) {111} with its daughter bcc martensite phase at (d) {100}, (e) {110}, and (f) {111} accompanied by the overlap of (g) {110}fcc and {100}bcc according to N-W and (h) {110}fcc and {111}bcc according to K-S.
To observe the damaged area more closely, a combined analysis of EBSD and EDS was performed on a selected crack in Fig. 7. The results of austenite grain reconstruction show that the crack has grown from a specific prior austenite grain boundary (Fig. 7(c)). Observation of the surface has revealed that the patterns of grain boundary corrosion marks (Fig. 2(f)) show does not correlate with actual width of prior austenite grain size. The prior austenite grain size is much smaller than the grain size that appeared because of corrosion marks. Thus, it can be inferred that the corrosion crack does not grow into every austenite grain boundary. Conclusively, another factor determines corrosion crack formation and growth. EDS mapping demonstrates the abundance of Cr, Mo, and oxygen at the corrosion crack zone.
Fig. 7. Combined EBSD (a) band contrast, (b) IPF map at Y-axis, (c) prior austenite grain reconstruction and EDS (d) C Ka, (e) Fe Ka, (f) Cr Ka, (g) O Ka, and (h) Mo La mapping on a corrosion crack growing at a specific prior austenite grain boundary. Note the austenite grain width marked by white arrows in (c).
Further analysis was performed at higher magnification at the selected crack as shown in Fig 8. It is apparent that the corrosion has proceeded at the boundary between retained δ-ferrite and martensite. The EDS mapping of important elements reveals the abundance of Cr in corrosion products (zero solutions) and the bcc retained δ-ferrite phases (Fig. 8(h)). The bcc martensite and retained δ-ferrite can be distinguished by the KAM map where martensite contains strain due to formation by massive transformation whereas δ-ferrite is a remnant of solidification process (Fig. 8). The abundance of oxygen in corrosion product only (not in δ-ferrite) shows that the corrosion initiates and proceeds only into the δ-α interface. In addition, a slight carbon enrichment at the δ-ferrite interface reveals carbide formation at the interface during its decomposition (Fig. 8(f)).
Fig. 8. SEM (a) BSE and (b) SE images of a corrosion crack proceeding into the cladding overlay accompanied by higher magnification combined EBSD (c) KAM, (d) IPF at Y-axis, (e) prior austenite reconstruction and EDS (f) C Ka, (g) Fe Ka, (h) Cr Ka, (i) O Ka, and (j) Mo La mappings at the area close to rack tip.
It is important to mention that the corrosion occurring at the zebra line follows the same mechanism as the cracked areas grown into the prior austenite grain boundaries. The corrosion marks can be witnessed in Fig. 4(a), (b). However, the difference that causes more severe corrosion at the zebra line arises from the difference in retained δ-ferrite content. In order to find out about the difference in retained δ-ferrite content of the HAZ (prone to cause zebra line corrosion marks on the surface of guide rolls) and matrix (prone to the formation of corrosion crack along the prior austenite grain boundary), the samples were etched by Vilella etchant to reveal the δ-ferrite within martensite phase [17]. Fig. 9 shows the microstructure of clad material revealing residual δ-ferrite at the HAZ (Fig. 9(b)) and inside the adjacent clad overlay (Fig. 9(c)). Several images were taken and analyzed by image analysis similar to Fig. 9(d) and 9(e).
Fig. 9. Cross-section etched (a) microstructure of the wire cladded at (b) HAZ and (c) inside the clad overlay and the corresponding (d), (e) image analysis.
Quantitative measurement of residual δ-ferrite amount at the HAZ and inside the clad overlay shows almost 10 times more at the HAZ region. The equilibrium phase diagram of the clad material obtained from ThermoCalc software shown in Fig. 10(b) reveals that the dominant phase right before melting is δ-ferrite which belongs to the HAZ area where heat input is not enough to fully melt the material, instead heating results in fully δ-ferrite phase which decomposes upon cooling [18]. Thus, a higher amount of δ-ferrite will remain compared to the clad overlay region (Fig. 10(a)).
Fig. 10. (a) Image analysis result of δ-ferrite content inside the clad overlay and HAZ with (b) equilibrium pseudo-binary phase diagram of cladding material.
Thus, it can be concluded that the microstructure of wire cladded stainless steel on the guide roll consists of room temperature martensite and δ-ferrite bcc phases are schematically shown in Fig. 11. At intermediate temperature during operation, prior austenite grains are exposed to a corrosive environment and the corrosion is intensified at higher δ-ferrite region. Due to the ferritization that occurs at HAZ of cladding overlays (Fig. 11(a)), higher corrosion rates result in zebra line appearance on the surface of guide roll whereas the deep crack accompanied by corrosion can be witnessed in a general surface that follows a connected δ-ferrite network.
Fig. 11. Schematic illustration of (a) high-temperature ferritization at the HAZ region resulting in (b) higher retained δ-ferrite at room temperature.
4. Conclusions
After analysis of the used guide rolls installed in the continuous casting machine for over 3000 heats, the following conclusions were reached on the surface deterioration mechanism.
1. Surface corrosion can be divided into two distinct marks. First, grain boundary corrosion crack occurs because of the elongated network of retained δ-ferrite after dendritic solidification. Second, zebra lines coincide with the HAZ of the adjacent cladding beads.
2. Corrosion initiates from the δ-γ interface at high temperature and proceeds by the formation of (Cr, Mo) oxide products as a crack opening exposes the new surface.
3. Despite the findings of former literature [2], it appears that the zebra line corrosion marks are caused by high retained δ-ferrite content at the HAZ zone adjacent to the cladding overlay and not the overlay itself. In HAZ, ferritization results in approximately 10 times higher retained δ-ferrite content compared with other neighboring cladded areas.
Acknowledgement
This work was supported by the Technology Innovation Program (200147323, Development of roll coating process and coating powder for fabrication of giga grade steel plate) funded By the Ministry of Trade, Industry & Energy (MOTIE, Korea). This project aims to improve the life long of continuous casting guide rolls in the Steel industry.
References
- Nordenstrom H, Stenback-lund E. Technology and Market Study on Continuous Casting Rolls. Lulea University of Technology (2006).
- Hadizadeh B, Bahrami A, Eslami A, Abdian K, Younes Araghi M and Etezazi M., Eng Fail Anal., 108 (2020) 104346.
- Chudzicki J, Horn BD, Tsai C., 36th Mech. Work. Steel Process. Conf., Maryland, USA (1994) 67-77.
- Wen P, Cai Z, Feng Z and Wang G., Opt Laser Technol., 75 (2015) 207.
- Du Toit M, and Van Niekerk J., Welding in the World, 54 (2010) R342.
- Hemmati I, Ocelik V and De Hosson JTM, Surf Coatings Technol., 205(21-22) (2011) 5235.
- Brooks JA and Thompson AW, Int Mater Rev., 36(1) (1991) 16.
- Kotecki D.J., Welding research supplement, 65 (1986) 273-s.
- BROOKS J.A., THOMPSON A.W. and WILLIAMS J. C., Welding research supplement, (1984) 71-s.
- Hull F.C., Welding research supplement, 52 (1973) 193-s.
- Schaeffler S and Schaeffler AL. Met Prog, 106 (1974) 227.
- KOTECKI DJ and Siewert TA., Welding research supplement, 71 (1992) 171-s.
- Zargar T, Sadeghi F, Kim JW, Lee JS, Heo Y-U and Yim CH, Met Mater Int., 28 (2022) 2263.
- Niessen F, Villa M, Hald J, Somers MAJ, Mater. and Des., 116 (2017) 8.
- Abbasi M, Nelson TW, Sorensen CD and Wei L. Mater Charact., 66 (2012) 1.
- Cayron C, Artaud B and Briottet L., Mater Charact., 57(4-5) (2006) 386.
- Bermejo MAV, Weld J., 91(4) (2012) 113S.
- Westin E.M., "Microstructure and properties of welds in the lean duplex stainless steel LDX 2101", Royal institute of technology, Stockholm (2010).