Microstructure and Phase Composition Changes in Polymer Fiber-Modified Thermacoat™ Lining Exposed to Contact with Liquid Steel

Abstract

Multicomponent clay and cenosphere linings protecting ceramic parts exposed to contact with liquid metal during continuous steel casting (CSC) are intended to diminish thermo-mechanical stresses at the beginning of this process. They are effective in their role, but due to their brittle nature, parts of them come off during transport or mounting. The admixture of polypropylene fibers into linings helps to alleviate problems with handling such parts, though the interaction of such a modified material with liquid steel should be re-assessed. The present experiment involved the preparation of a crucible with a Thermacoat™ (Vesuvius, Skawina, Poland) lining with the addition of Belmix™ (Belgian Fibers Group NV, Mouscron, Belgium) fibers and filling it with a drop of liquid steel. Next, the crucible was sectioned and the changes in its microstructure and phase composition were investigated with optical, scanning and transmission electron microscopy (OM/SEM/TEM) methods. This showed that the modified lining retained its non-wetting property against the steel of the non-modified material. The part with the lining, being in direct contact with the liquid steel, developed a highly porous layer filled with AlSiOx amorphous flakes with some larger blocky Al₂O₃ and SiO₂ particles. Right below this, a heat-affected zone (HAZ) consisting of fine γ-Al₂O₃ platelets immersed in amorphous silica was formed. Some of the voids with a size corresponding to polymer fiber diameter carried significant carbon deposits on their walls. The performed investigation indicated that the polymer fiber-modified linings were capable of withstanding at least short-term contact with liquid steel without instantaneous defragmentation, i.e., they retained the good high-temperature properties of the non-modified material.

1. Introduction

Multicomponent clay and cenosphere linings are routinely used in industry to protect parts of continuous steel casting (CSC) lines from thermo-mechanical shock developing during their first contact with molten metal [1,2]. Even if they are of the self-sacrificial type, these linings are capable of significantly extending the lifetime of glazed corundum lances through which gases are pumped into molten metal. Applied using a dip-coating technique [3], e.g., the immersion of substrates in their liquefied mixture, after drying, they are susceptible to crumbling and delamination during the handling and on-site installation of such protected parts. This problem could be alleviated by the admixture of polymer fibers into powders making up the lining materials, like in the case of sprayed fiber-reinforced concrete (FRC) [4,5], increasing their impact strength and resistance to crushing.

Experiments with one such lining material, known under the commercial name Thermacoat™, helped to establish that its modification with the addition of up to 1 wt. % of polymer Belmix™ fibers still allowed for their even distribution within this type of matrix [6]. These fibers are easily accessible and low-priced, which makes them a material of choice for anti-shrinkage additions to concrete bathroom floors, etc. Larger additions entail the formation of fiber agglomerates that promote crack growth under loading and decrease the material’s strength down to pre-modification level. Linings with randomly dispersed polymer fibers allowed for the achievement of reproducible ultimate flexural strength (UFS) at the level of 0.2 MPa. The latter parameter indicates the stress at failure in bending, which is of most interest in cases of these types of coatings and routinely applied in industries involved in the production of such parts. The pulling up of the treated part from the semi-liquid Thermacoat™ mixture automatically caused the vertical alignment of most of these relatively short fibers, which allowed for further improvement of the flexural strength of the lining by ~100% [7], i.e., in a similar way to in FRC materials [8,9]. Scanning electron microscopy observations of the pulled out fibers confirmed that at least some of the components of the Thermacoat™ mixture adhered to them, assisting in their anchoring. Aside from the presence of the fibers, the microstructure, and especially the porosity of both the modified and fiber-free materials, were practically the same. Heating runs using a differential scanning calorimeter (DSC) measuring the heat flow during continuous sample heating at 20 °C/min showed that the endothermic effect of polymer fiber melting at ~165 °C was dominated by that from water evaporation without affecting the sample’s overall shape. The response of a fiber-modified coating to contact with liquid steel could have more serious consequences as it takes place under an oxygen-bearing atmosphere, while the auto-ignition of such a polymer takes place at ~400 °C. In the worst-case scenario, the fast accumulation of combustion gases and a lack of enough escape channels could lead to catastrophic removal of the lining from the substrate. Therefore, a test covering Thermacoat™’s resistance to deterioration caused by contact with melted steel is much needed.

The aim of this work was to check, on a laboratory scale, the effect of pouring a small amount of melted steel into a crucible made of Thermacoat™ material modified with polymer fibers. Microstructure investigations were performed using light, scanning and transmission electron microscopy (LM/SEM/TEM) methods.

2. Experimental Procedure

A Thermacoat™ multicomponent powder mixture of clay, alumina, cenospheres, aluminum powder, fused silica, chamotte, elastomeric styrene, dextrin and acrylic emulsifying agents (rest) was obtained from Vesuvius, Skawina, Poland. After adding of 1 wt. % of polypropylene Belmix™ fibers (single strand: ~34 μm in diameter, length of 12 mm—Belgian Fibers Group NV, Mouscron, Belgium) it was diluted with water until a moldable state was achieved, which allowed us to make a shallow crucible out of it (Figure 1). Next, the crucible was placed in a dryer for 12 h. Afterwards, a small amount of low-carbon steel (~0.5 cm³) was poured into it. The same material, but with higher amount of water, was used for covering the surfaces of lances intended for tests in this work.

The parts of the crucible subjected to direct contact with the liquid steel were sectioned with a diamond saw and observed using LM with a Keyence VHX–7000 (Keyence Corp, Osaka, Japan) as well as a Thermo Fisher Scios II Dual Beam (Thermo Fisher Scientific, Eindhoven, The Netherlands) microscope, enabling imaging both with secondary and back-scattered electrons (SE/BSE). The latter one was also used to prepare a thin lamella for TEM investigations with Tecnai 200 kV FEG (FEI, Eindhoven, The Netherlands) and Themis 200 kV FEG (Thermo Fisher Scientific, Eindhoven, The Netherlands) microscopes. The microstructure was characterized based on observations in bright-field (BF) mode, while phase analysis was performed by acquiring selected-area (SA) electron diffraction patterns (EDP). The local chemical composition was determined using scanning-transmission mode and a high-angle annular dark field (STEM-HAADF) detector as well as maps of elements relying on an X-ray energy-dispersive spectroscopy (EDS) attachment.

3. Results

The fractured surfaces of the heat-resistant ceramic tube (darker contrast) with an outside lining of Thermacoat™ modified with polymer fibers are shown in Figure 2a. The close-up of the latter area shows that the standard mixing procedure of the Thermacoat™ components allowed us to obtain a relatively uniform distribution of polymer fibers (Figure 2b). However, the higher-magnification observations using the SEM method indicate that at the micro-scale, the fibers formed small colonies as they were pushed out to cenosphere-free areas (Figure 3). In the case of these observation methods, the brighter spots in the images were caused by irregular sharper points formed at the fiber ends cut by a diamond saw. Simultaneously, it is of importance that practically no tangles were observed, as only such defects could have led to the accumulation of larger amounts of gases in one place during their combustion.

The small measure of liquid steel poured into the Thermacoat™ crucible solidified in it in the form of a rounded drop. At the first site, darkening of the inside of the crucible was the only indication of their interaction. The section of the crucible observed with the LM method showed that its surface was covered with a darker layer varying in thickness (<0.1 mm), while below it, in the heat-affected zone (HAZ), the color of the Thermacoat™ changed to slightly whitish (down to 3–4 mm) (Figure 4). Otherwise, the microstructure was dominated by roughly spherical pores (<1 mm), representing remnants of the cenospheres (of size distribution up to 0.5 mm), like in other materials produced with their use [8]. The presence of polypropylene fibers was noted neither on the crucible near the surface areas, nor in the HAZ. The observations of the crucible cross-section with the SEM/SE detector helped to establish that the layer close to the melted metal was distinguished from that farther away by the greater presence of flat areas (see Figure 5a,b). The characteristic rounded shape of cenosphere shells of both smaller and larger diameter prevailed in both microstructures.

The lamellae for the TEM observations of the near-surface areas of the crucible were cut with a focused ion beam (FIB) from the place of its contact with the liquid metal at the side and bottom of the drop, which was practically the only flat place on the material. Their observation revealed that on the crucible surface, a highly porous layer of varying thickness was formed (Figure 6). It was characterized by the presence of only a few rounded voids, of which the larger ones resembled a collapsed cenosphere. Otherwise, the layer was filled with flattened, interconnected, fine, jagged particles with the occasional presence of larger blocky ones. Imaging using a BSE detector sensitive to Z (atomic number) indicated that some fine spherical particles of higher average atomic mass, and therefore, differentiating themselves with bright contrast, showed a tendency to segregate at the bottom of this layer (Figure 6c). Right below this layer was spread a relatively homogenous material with some irregularly shaped medium-sized pores characteristic of the dried Thermacoat™ microstructure [6].

The STEM-HAADF images of a section of the crucible from the area closest to the molten steel, backed by EDS maps presenting the distribution of chemical elements, helped to establish that the fine jagged particles were mostly in the amorphous AlSiO_x sialon phase (Figure 7). The larger blocky ones were either amorphous silica (SiO₂) or crystalline alumina (Al₂O₃). The very small rounded particles turned out to be iron oxides. Similar observations performed on the parts of the crucible located below its upper layer showed that the Thermacoat™ material was completely fused together, forming an amorphous matrix with dispersed plate-like fine precipitates (Figure 8). The accompanying EDS maps allowed us to identify them as silica and alumina, respectively. Some of the larger voids in this area carried on their walls significantly more of a carbon deposit than that which usually accumulates on thin foil surfaces due to hydrocarbon contamination (Figure 7, map of carbon distribution). Therefore, this could be attributed to remnants of the burnt-out polypropylene fibers, assuming that the empty cavern partly collapsed in the softened matrix (the present voids were in the micrometer range, while the average diameter of the Belmix™ fibers was ~34 μm).

The TEM/BF images of the part of the crucible located several micrometers below the steel drop confirmed the observations performed with the STEM/HAADF method, that its exposure to high temperature transformed it into a dense amorphous matrix filled with plate-like precipitates (Figure 9). A similar “needle-like” microstructure was observed previously in a heat-treated calcium–magnesium–aluminum–silicate (CMAS) coating subjected to high-temperature treatment (5 h/1250 °C), which the authors tentatively (due to a lack of TEM/electron diffraction data) attributed to one of the alumina forms [10]. The matrix of their material was the first to liquefy during heating, which should also have taken place in the present case.

The SAEDP acquired from the larger area (Figure 9b) produced numerous spots arranged along the radii that are characteristic of the γ-Al₂O₃ phase (i.e., 111/200/131/400/133 planes, respectively), while the amorphous matrix was represented by diffused intensity in the background. The electron diffraction obtained from an individual platelet (Figure 9c) showed a series of spots that are also characteristic of this phase, i.e., with the [001]γ-Al₂O₃ zone axis.

4. Discussion

The laboratory-scale test involving the placement of a drop of liquid steel over the Thermacoat™ modified with polypropylene fibers was only a kind of first approximation of working conditions in the vat filled with molten steel. Ceramic lances coated with such linings are used to blow oxygen into molten metal, which is connected not only with exposure to high temperature, but also to shear forces caused by the disturbed flow of the liquid steel. However, the sacrificial role of the linings means that they would effectively serve their purpose as long as they avoid self-delamination within first second of such exposure. A problem in this case might be an explosive combustion of polypropylene fibers, e.g., the reaction of such composites to high temperature. From that point of view, placing even a small amount of molten steel over a Thermacoat™ substrate should give a much-needed answer. And indeed, the experiment performed within this work showed that most of the material remains in its original form, even though exposure to temperature in excess of 1500 °C resulted in partial evaporation of some of its components, leaving behind a heavily porous upper layer and HAZ below it. Most of the upper layer consisted of flake-like amorphous sialon particles with only occasional larger silicon or alumina particles, while the HAZ was formed of an amorphous silica matrix filled with numerous plate-like γ-Al₂O₃ crystallites. The difference in the phase composition of both areas was a result of a strong temperature gradient due to the low heat conductivity of the Thermacoat™. The small size of the steel drop and, consequently, the small amount heat transferred into the substrate material was the reason for the dominance of the HAZ over the upper layer. In case of the on-site test on the steel mill, the degradation may start in the same way, but it should immediately progress through the evaporation of the silica matrix from the deeper buried areas. This is because the evaporation of silica already starts at temperatures above 800 °C. The γ-Al₂O₃ crystallites could eventually turn to α-Al₂O₃ [11], but not to AlSiOx particles, i.e., the growth of the upper layer just by consuming the HAZ would not be possible.

The amorphous/crystalline form of the am-SiO₂/γ-Al₂O₃ composite documented in the HAZ in this paper was a prevailing microstructure of the investigated samples. However, it should be noted that even if an advanced TEM sample preparation technique like FIB allows us to obtain lamellae from areas pin-pointed during SEM observation, their picking-up is impossible in cases of more corrugated surfaces. The am-SiO₂/γ-Al₂O₃ is also brittle, as demonstrated by the cracks in Figure 9a, but at high temperature, even under the small load of the steel drop, it formed practically the only flat areas on the material. Other Thermacoat™ constituents like chamotte melt at 1800 °C, a temperature which was not reached during the present experiment. Consequently, its particles contributed to the development of a rough surface, and because of that, it was omitted from the investigation. However, taking into account the fact that the formation of the am-SiO₂/γ-Al₂O₃ composite was promoted by the presence of silica in the original Thermacoat™ mixture (through the direct addition of 10 wt. % and indirectly through its presence in cenospheres [12,13] at a similar level), the portion of this composite should also be significant, i.e., representative of a major component of the heat-treated material. The participation of cenospheres in the formation of this composite is upheld by the presence of fine platelets of γ-Al₂O₃ as well as rounded Fe-rich particles, which most probably also came from them.

Polypropylene fibers were absent in the upper layer of the crucible as their melting and auto-ignition temperatures are 165 °C and ~400 °C, respectively. However, their combustion, as they were buried deep in the HAZ, might have been retarded by the lack of oxygen. The presence of larger carbon deposits on the walls of larger voids (of sizes corresponding roughly with the fiber diameter) were indeed documented, but they may have also been formed by the agglomeration of other combustible Thermacoat™ components, like starch and similar components [6]. There are some question marks in the case of real working conditions, during which liquid steel may compress the upper layer, cutting off the release of the products of combustion, but laboratory-scale tests are rarely able to equal tests-by-work. Even though the process of the removal of polypropylene fibers from the Thermacoat™ under contact with liquid steel could be only presumed, the most important thing is that it was neutral as concerns the shape preservation of the crucible and the linings applied to the parts already used in CSC production lines.

5. Summary

The modification of a multicomponent Thermacoat™ with the addition of up to 1 wt. % of Belmix™ polypropylene fibers, allowing us to minimize its damage during pre-installation handling, turned out neutral as concerns its response to laboratory-level contact with liquid steel. The performed experiment indicated that the main inorganic constituents of Thermacoat™, i.e., fused silica and cenospheres, contributed to the formation of the am-SiO₂/γ-Al₂O₃ composite. The latter, even in its self-sacrificial role, was the only component capable of decreasing the thermo-mechanical shock exerted on other vital parts of CSC production lines.

The introduction of polypropylene fiber-modified linings to industry practice will have to be preceded by tests-by-work, which interfere with the currently accepted mill procedures and are therefore generally opposed by industrial plants. Additionally, even a small rise in material costs is never welcomed. However, the positive outcomes of these experiments are that not only would the proposed linings bring benefits in the form of a lower number of factory rejects, but they would also cut down on the amount of hard-to-recycle ceramic waste.