Effect of MgO on wetting and corrosion behaviour of corundum substrate by CaO–SiO 2 –MgO (

Abstract

Effect of MgO content on wetting and corrosion behaviour of corundum substrate by CaO–SiO₂–MgO (−15%Al₂O₃) molten slags was investigated at 1450°C via joint application of a sessile drop technique, SEM–EDS detection and FactSage thermodynamic software. All slags exhibit good wettability on the substrate, and the contact angle values are small. Increasing MgO content from 7% to 9% in CaO-SiO₂-MgO slags can increase the contact angle between the droplet and the substrate, while the substrate dissolution thickness and the slag penetration depth tend to decrease. Increasing MgO content from 8% to 15% in CaO-SiO₂-MgO-15% Al₂O₃ slags can also increase the contact angle, but significantly decrease the slag penetration depth. The indirect dissolution thickness increases instead due to the formation of a large amount of MgAl₂O₄ by the interface reaction. The wettability can give an indication of the slag penetration depth in the substrate but not the dissolution thickness of the substrate.

Introduction

Refractories are widely used in metallurgy, glass and other high-temperature industries. However, there are often complex interactions between molten slag and refractory during smelting process. Such interactions imply corrosion of refractory by molten slag. The interactions always speed up the refractory corrosion, and affect the quality of the smelting product. It is generally believed that the corrosion between molten slag and refractory mainly includes direct and indirect dissolution of refractory in molten slag and penetration of molten slag into refractory [1–7]. The corrosion is closely related to wetting phenomena [8, 9]. The wetting and corrosion between refractory and molten slags during smelting process is therefore an important issue in view of the costs of the refractories and contamination of the smelting product, which must be considered in the composition design for specific industrial vessel or equipment.

Alumina can be used as composite refractory or single refractory, which is widely serviced as refractory containers such as hot metal torpedo car or steel ladle in iron and steel making, refractory components such as tundish stopper and submerged entry nozzle (SEN) in continuous casting, and also as smelting furnace linings of aluminium and copper in nonferrous metallurgy. Many researches have been carried on the corrosion of alumina refractory by Al₂O₃ contained molten slags [1–5,7, 8,10–23]. The wettability between solid and liquid phases is usually characterized by contact angle, so the contact angle between alumina refractory and molten slag has also been studied [8,24–29]. In the past, most corrosion experiments were performed in the dipping, immersion, finger or crucible test, which could not link the refractory corrosion with the change of contact angle in one experiment. A refractory often has different corrosion behaviours in different slag systems. In order to further understand the wetting and corrosion of corundum substrate by molten slags under extreme conditions, in this work, CaO–SiO₂–MgO slags with and without Al₂O₃ and corundum substrates were employed as the research objects. Combining application of a sessile drop technique and SEM-EDS detection, the effect of MgO content on the wetting and corrosion behaviour of corundum substrate by the molten slags at 1450°C was investigated by measuring the contact angle of molten slag on the substrate, the dissolution thickness of the substrate in the molten slag, and the slag penetration depth in the substrate. The association of the contact angle with the corrosion was analysed. The corrosion mechanisms of the corundum substrate by molten slags with different compositions were also discussed.

Experimental

Sample preparation

Corundum sheets with a thickness of about 0.6 mm were purchased from Foshan Xinghongfei Electronic Technology Co., Ltd, China. Each corundum sheet was cut into several substrates with about 20 × 18 × 0.6 mm dimensions and surface roughness of Ra = 0.6–0.8 μm. The substrate composition was analysed by XRF, as shown in Table 1. XRD analysis was also conducted, the result (not be shown here) indicated that the component of the substrate was α-Al₂O₃.

Table 1.

Chemical composition of corundum substrate.

Component	Al₂O₃	SiO₂	CaO	Fe₂O₃	Other
Content/mass%	99.30	0.50	0.11	0.05	0.04

Five kinds of CaO–SiO₂–MgO (−15% Al₂O₃) slags were designed for comparison. The slag target compositions and basicities are shown in Table 2. Here binary, ternary, and quaternary basicities are listed for reference. It should be noted that the binary and ternary basicities are not affected by Al₂O₃. The denominator in the definition of quaternary basicity includes Al₂O₃, so the quaternary basicity decreases with the dissolution of corundum in the molten slags. The specific composition locations in phase diagram [30] are shown in Figure 1.

Figure 1.

Specific composition locations of Nos. 1–5 original slags in phase diagrams of CaO–SiO₂–MgO system (a) and CaO–SiO₂–MgO-15% Al₂O₃ system (b).

Table 2.

Composition of experimental slags.

Slag No.	Target composition (XRF results)/mass%					Basicity
Slag No.	CaO	SiO₂	MgO	Al₂O₃	Fe₂O₃	CaO/SiO₂	(CaO + MgO)/SiO₂	(CaO + MgO)/(SiO₂+ Al₂O₃)
1	50.0	43.0	7.0	–	–	1.16	1.33	1.33
1	(51.6)	(41.6)	(6.4)	(0.3)	(0.03)
2	49.2	42.4	8.4	–	–	1.16	1.36	1.36
2	(51.4)	(40.4)	(7.8)	(0.2)	(0.25)
3	48. 9	42. 1	9. 0	–	–	1.16	1.38	1.38
3	(50.7)	(40.4)	(8.4)	(0.3)	(0.32)
4	40.7	36.3	8.0	15.0	–	1.12	1.34	0.95
4	(42.5)	(34.7)	(7.5)	(15.2)	(0.16)
5	37.0	33.0	15.0	15.0	–	1.12	1.58	1.08
5	(39.0)	(30.9)	(13.3)	(16.4)	(0.40)

Original slags numbered from 1 to 3, were CaO–SiO₂–MgO system without Al₂O₃, and the composition points basically located in the same primary crystal zone (Figure 1(a)). Among the three slags, the binary basicities were the same, but MgO contents slightly increased in turn. Considering the experimental temperature was limited at 1450°C, the melting temperature of slag would be too high if the MgO content was too high, so too high MgO content was not permitted in CaO–SiO₂–MgO original slag. Original slags numbered from 4 to 5, were CaO–SiO₂–MgO-15% Al₂O₃ system, and the two composition points located in different primary crystal zones (Figure 1(b)). Among the two slags, the binary basicities were also the same, but MgO contents were 8% and 15%, respectively. The original slags with and without Al₂O₃ were used to compare the effect of MgO on the wetting and corrosion behaviour of the corundum substrate by the molten slags under extreme conditions. Since the other experimental conditions were the same, the differences in the wetting and corrosion behaviour of corundum substrate should be attributed to the different compositions of the original slags.

The five slags were prepared by analytical pure chemical reagent according to the proportion given in Table 2. No. 1 slag was mixed in a horizontal ball mill for 16 h. Nos. 2–5 slags were ground respectively, and mixed in an agate bowl, pre-melted in a graphite crucible of induction furnace, cooled on a big copper plate, and crushed in an iron can. The crushed slag powder was decarburized for 20 h in a muffle furnace at 1000°C under air. The mixed or pre-melted slag powder, whose compositions were analysed by XRF and the results are also given in Table 2, was pressed to form cylindrical slag samples with a diameter of 3 mm on a chip press. The cylindrical samples were sintered in a MoSi₂ rod furnace to have certain strength. Sintered cylinder samples were ground on sandpaper to about 5 mm height to keep the slag quality as consistent as possible. Corundum substrates and sintered cylindrical samples were cleaned with absolute ethanol in ultrasonic instrument, dried and placed in a dryer for reserve. Considering the possible deviation of XRF detection, the slag compositions in this work were based on the respective target compositions.

Experimental method

A cylindrical slag sample was placed on a corundum substrate, and formed a combination together with a corundum carrier. Then the combination was smoothly pushed into the constant temperature zone of a horizontal furnace of a high-temperature contact angle measuring instrument (Model OCA25-HTV1800, Dataphysics Instruments GmbH, Germany) along the inner wall of one end of the furnace mouth through a feeding ruler, and the sample level was adjusted if necessary. The experimental device is schematically shown in Figure 2. According to the specifications provided by the equipment supplier: contact angle measurement range: 0–180°, measurement resolution: ± 0.01°, accuracy ±0.1°.

Figure 2.

Schematic diagram of experimental device with sessile drop technique (1-CCD camera; 2-quartz window; 3-gas inlet; 4-sealing cover; 5-water outlet; 6-refractory; 7-corundum furnace-tube; 8-gas outlet; 9-light source;10-thermocouple outlet; 11-water inlet; 12-water cold flange; 13-B-type thermocouple; 14-corundum substrate; 15-sessile drop; 16-corundum carrier; 17-PC and monitor).

The horizontal furnace was set to heat at 5°C·min⁻¹ to the target temperature (1450°C). A CCD camera was turned on when the furnace temperature reached about 10°C–30°C below the melting point of the cylindrical slag. The camera was turned off and the video file was saved after holding temperature for a certain time. The sample was cooled in the furnace at a rate of 5°C·min⁻¹. The furnace temperature was measured by a B-type thermocouple below the corundum carrier and also corrected as the experimental temperature by another B-type thermocouple placed in the cylindrical slag position on the carrier before the experiment, and the fluctuation at the target temperature was within 2.5°C. High purity argon (99.999%) with a rate of 200 mL·min⁻¹ controlled by a mass flow meter (Sevenstar, model D07-7) was injected into the furnace tube during the whole experiment. The experimental conditions of the five slags are given in Table 3.

Table 3.

Experimental conditions, dissolution thickness and penetration depth results.

Slag No.	Run No.	OTS/μm	Slag mass/mg	Holding time/min	TSA/μm	SDT/μm	SPD/μm
1	1a	614	65.1	0	586	28	20
	1b	649	62.8	10	616	33	29
	1c	569	66.9	60	519	50	68
2	2a	649	91.6	10	619	30	25
2	2b	655	79.4	30	608	47	94
3	3a	626	90.7	10	597	29	23
4	4a	630	72.4	10	628	2	56
4	4b	644	97.5	30	641	3	76
5	5a	655	81.4	10	582	73	0

Note: The figure in run No. represents slag No.; OTS-original thickness (h₀) of substrate; TSA-thickness (h) of substrate after experiment; SDT-substrate dissolution thickness (Δh = h₀–h); SPD-slag penetration depth.

Detection method

Based on the obtained video, the moment corresponding to the same temperature (1365°C) was taken as the zero time; the SCA20 software built in the instrument was used to measure the contact angle after the cylindrical slag completely melted. The complete melting temperature of the cylindrical slag was determined by distinguishing a curved liquid surface formed after melting. The liquid surface of the slag was smooth and had no obvious bulge at the complete melting temperature. The contact angle at complete melting temperature was defined as the initial one. Then, the contact angle was measured every 1 min until the end of holding temperature, and the corresponding time and temperature were extracted from the image by the SCA20 software. In view of the slag–substrate interactions in this work, the measured contact angle should be referred to as the apparent contact angle [27], being lower than the real contact angle, which direct assessment was quite problematic.

After the experiment, the corundum substrate with solidified slag droplet was longitudinally cut, polished, and sprayed with gold. The microstructure and related compositions of the wetting centre were detected by a field emission scanning electron microscopy (SEM, Nova NovaSEM400, FEI Company, U.S.A.) and energy-dispersive X-ray spectrometer (EDS). Three main zones of A, B and C shown in Figure 3(a) were observed, and the substrate dissolution thickness and the slag penetration depth were determined. Before the experiment, the middle and the edge thicknesses of the original corundum substrate were observed randomly under an optical microscope, the maximum difference between them was found to be small and within ±5 μm. Therefore, the thickness of the zone C could be treated as the original thickness (h₀) of corundum substrate. The typical SEM image of zone C was shown in Figure 3(b). The zone A was used to determine the thickness (h) of the corroded corundum substrate based on the wetting centre, then h₀–h value was defined as the dissolution thickness (Δh) of the substrate. The zone B was used to determine the penetration depth of molten slag based on the wetting centre. It could be considered that the slag penetration in the wetting centre mainly referred to vertical penetration. For the experiments with Nos. 4 and 5 slags, the product layer was thick compared with Nos. 1–3 slags and the substrate thickness (h) after corrosion was measured below the product layer.

Figure 3.

Schematic of specific zones for SEM-EDS observation (a) and typical SEM image of zone C (b).

Results and discussion

Changes of slag drop shape and contact angle

Figure 4 shows morphological change process of the slag samples before and after melting, and some typical images of the slag samples at heating and holding temperature stages are also presented. It was found that the slag sample melted within a certain temperature range, and the molten slag gradually spread out on the corundum substrate and the contact area between them increased. It can be seen that the height of the slag droplet gradually decreased to a very low level during holding temperature, and some droplets were almost level with the substrate. Large spreading area and small contact angle of molten slag on the corundum substrate indicated that the five slags exhibited good wettability on the corundum substrate.

Figure 4.

Morphological change process of different slag samples on corundum substrate (The subgraph No. also represents run No. in Table 3. Stages: A-before melting; B-melting process; C-complete melting; D-heating; E-start of holding temperature; F-end of holding temperature).

Figure 5 shows the change of apparent contact angle between corundum substrate and Nos. 1–3 slags without Al₂O₃, respectively. The initial contact angles in runs 1a, 1b, 1c, 2a, 2b and 3a were 29°, 22°, 26°, 25°, 26° and 27°, respectively. Then, each contact angle value decreased sharply with time, and the decrease trend gradually slowed down after a period of time. During holding temperature, the contact angle value changed little. The final contact angle value fluctuated at around 6.5° for run 1b. Similarly, the final contact angle value at 5.9° for run 1c, at 12.0° for run 2a, at 12.3° for run 2b, and at 15° for run 3a. It can be seen that, for the experiments of the same slag with different holding times, the difference in the contact angle was less than 2° within the comparable time (temperature). Such a small difference indicated that the experimental reproducibility was good. However, the contact angle values of these slags were somewhat different from the relevant literature data [28,29], which might be related to different slag compositions and the mixed impurities such as iron oxide in the slag during crushing, pre-melting and pressing processes, as shown in Table 2. Under the condition that the temperature of 1450°C was held for 10 min, the final contact angle in run 3a was significantly larger than those in runs 2a and 1b. As can be seen in Figure 5, the contact angle values of Nos. 1–3 molten slags with the same binary basicity on the corundum substrate gradually increased with increase of MgO content within comparable time (temperature), indicating that increasing MgO content in the CaO–SiO₂–MgO slag without Al₂O₃ could reduce the slag wettability on the corundum substrate.

Figure 5.

Comparison of apparent contact angle between substrate and Nos. 1–3 slags without Al₂O₃.

Figure 6 shows the change of apparent contact angle between the corundum substrate and Nos. 4 and 5 slags with 15% Al₂O₃, respectively. Since the binary basicity and MgO content of No. 2 slag were close to those of No. 4 slag, respectively, the contact angle in run 2a was also added in Figure 6 for comparison. As can be seen from Figure 6, the contact angle values in runs 4a and 4b were similar within comparable time (temperature), and the initial contact angle values were about 30°; the final contact angle values decreased gradually to 14° at stage of holding temperature, which was slightly larger than that in run 2a within comparable time (temperature). No. 5 original slag had the highest MgO content in this work. The initial contact angle value in run 5a increased to 37° and the final contact angle value decreased to 17°, they were also all the largest. It can be seen that, with the increase of MgO content, the contact angle of CaO–SiO₂–MgO-15% Al₂O₃ slag on the corundum substrate increased within comparable time. However, the wettability of slags on the corundum substrate was still good. The contact angle in run 2a was smaller than that in run 4a at stage of holding temperature. It is inferred that adding Al₂O₃ into CaO–SiO₂–MgO slag could also increase the contact angle value of the slag on the corundum substrate under the conditions of similar binary basicity and MgO content.

Figure 6.

Comparison of apparent contact angle between substrate and Nos. 4 and 5 slags with 15% Al₂O₃.

Microscopic observation

Figures 7 and 8 are SEM images of zone A (see Figure 3(a)) at low magnification and zone B at high magnification, respectively. In Figure 8, the zone from the slag/substrate interface to the orange line was marked as the slag penetration zone, and its depth value was defined by an average distance. For the experiments with Nos. 4 and 5 slags, the zone from the slag/substrate interface to the red line was marked as the product layer; and the penetration depth of No. 4 slag was defined by an average distance from the red line to the orange line, but that of No. 5 slag was found to be zero. The measurement results of substrate dissolution thickness and slag penetration depth are also given in Table 3.

Figure 7.

SEM images of samples at low magnification (The subgraph No. also represents run No. in Table 3).

Figure 8.

SEM images and mapping of the samples at high magnification (The subgraph No. also represents run No. in Table 3).

Table 4 shows EDS analysis results of the corresponding spots in Figure 8. Since all elements in the slag and the substrate existed in the form of oxides, the distribution of oxygen element is not shown in the mapping of Figure 8. It can be seen from Figure 8 and Table 4 that the corundum not only directly, but also indirectly dissolved in the molten slag. The latter referred to the interface chemical reaction between the corundum substrate and the slag. New phases, such as MgAl₂O₄ (MgO·Al₂O₃ or MA, melting point: 2105°C) and CaAl₄O₇ (CaO·2Al₂O₃ or CA₂, melting point: 1762°C), could be formed via the interface reaction in this work. Obviously, the interface reaction also led to the increase of substrate dissolution thickness.

Table 4.

EDS results of corresponding spots in Figure 8.

Run No	spot	Element content/at%					Possible phase
Run No	spot	O	Mg	Al	Si	Ca	Possible phase
1a	1	58.78	5.40	4.85	15.13	15.84	Melilite
1a	2	59.83	0.12	3.09	18.11	18.85	Melilite
1a	3	58.65	4.37	7.18	13.72	16.08	Melilite
1a	4	58.41	0.35	32.47	0.58	8.20	CA₂
1a	5	59.95	0.17	39.80	–	0.08	Al₂O₃
1b	1	58.73	2.79	10.67	12.12	15.69	Melilite
1b	2	57.70	11.51	30.79	–	–	MA
1b	3	58.35	–	33.41	–	8.23	CA₂
1c	1	58.49	–	33.95	–	7.56	CA₂
2a	1	58.22	0.60	31.44	0.72	9.02	CA₂
2a	2	57.47	12.46	29.87	–	0.20	MA
2b	1	57.61	11.97	30.42	–	–	MA
2b	2	57.79	11.03	31.18	–	–	MA
2b	3	58.34	–	32.59	0.39	8.67	CA₂
3a	1	57.63	11.52	30.53	–	0.32	MA
3a	2	58.37	1.18	31.77	0.85	7.84	CA₂
4a	1	61.11	2.93	9.29	17.57	9.09	Anorthite
4a	2	58.71	2.65	10.63	12.10	15.92	Melilite
4a	3	58.29	–	33.17	–	8.54	CA₂
4a	4	57.69	11.54	30.76	–	–	MA
4a	5	59.88	–	39.52	–	0.60	Al₂O₃
4b	1	61.06	3.31	9.58	17.33	8.72	Anorthite
4b	2	57.64	11.65	30.58	–	0.13	MA
4b	3	58.42	–	32.49	0.61	8.48	CA₂
5a	1	57.33	13.36	29.31	–	–	MA
5a	2	61.33	0.79	13.54	15.89	8.46	Anorthite

From Figure 7, the solidified slag was basically composed of dark grey and light grey phases. From the mapping in Figure 8 and the EDS analysis in Table 4, for the experiments with Nos. 1–3 original slags, it can be seen that contents of Al₂O₃ and MgO were high in the dark grey phase; the light grey phase was rich in SiO₂ and CaO, with low Al₂O₃ content, and MgO content was also low, even zero. Al₂O₃ was detected in the solidified slag, indicating that the corundum substrate could automatically dissolve in the original slag without Al₂O₃ at high temperatures. In addition, it is found that the content of dissolved Al₂O₃ in the slag near the interface increased with the increase of holding time, indicating that the longer the holding time, the more Al₂O₃ dissolved in the slag, resulting in the increase of dissolution thickness of the substrate. For the experiments with Nos. 4 and 5 slags containing 15% Al₂O₃, the light grey phase in the slag was rich in Al₂O₃, CaO and SiO₂, while the dark MA phase was rich in MgO and Al₂O₃.

The molten slag penetrated in the corundum substrate mainly through pores, grain boundaries and channels containing low melting point continuous phases such as CaO–SiO₂. Through large pore, some large penetration zones could be formed in the substrate. For example, the slag penetration depth in run 2b reached 94 μm (see Table 3) at the holding time of 30 min, which was an unusual case due to large pore (Figure 8). From the mapping results (Figure 8), it can be seen that the Ca element had the largest distribution range due to high CaO content in the slag. The Ca distribution was the most uniform in the slag, and also the most pronounced in the penetration zone. Combining with other components such as SiO₂, CaO could react easily with the corundum substrate to produce calcium aluminosilicates with low melting point, resulting in Ca²⁺ penetration into the substrate. MgO content in the slag was relatively low, and MgO was also easy to react with the corundum to form MA phase at the slag/substrate interface, resulting in almost no Mg²⁺ penetration into the substrate. The coexistence distribution zone of CaO and SiO₂ in the substrate near the interface could be considered as a marker of penetration.

In runs 1a-3a, the original slags did not contain Al₂O₃. It can be seen from Figures 7 and 8 and Table 3 that, for run 1a, no holding time, and only at the heating and cooling stages, the substrate dissolution thickness and the slag penetration depth reached 28 and 20 μm, respectively. For runs 1a, 1b and 1c with the same No. 1 slag and different holding times, it can be inferred that the direct dissolution of corundum substrate and the slag penetration were faster at the early stage (including melting, heating up, etc.) and slower at the later stage; a thin but discontinuous new phase (MA) layer was found at the slag/corundum substrate interface. For runs 1b, 2a and 3a with the same binary basicity and holding time (10 min), the apparent contact angle values increased within comparable time (temperature) with the increase of MgO content in the original slag, and the amount of interface MA phase also tended to increase; consequently, the direct dissolution thickness and the penetration depth tended to decrease (Table 3). It is found that a small amount of CA₂ could also be generated at the interface (Table 4).

For runs 4a and 4b with the original slag containing 15% Al₂O₃, MA phase layer was also formed at the slag/corundum substrate interface, and was basically thick and dense; however there was still obvious slag penetration in the substrate. From Table 3, although the holding time (30 min) of run 4b was much longer than that (10 min) of run 4a, the slag penetration depth in the former was 20 μm more than that in the latter, indicating that the formation of MA phase and the slag penetration in the substrate mainly occurred at the early stages of heating up and holding time. For runs 4a and 2a with similar slag basicity and initial MgO content, the slag penetration depth in run 4a was significantly larger than that in run 2a (see Table 3) even though a large amount of MA phase was formed at the slag/substrate interface in the former. In addition, as shown in Table 3, the substrate dissolution thicknesses in runs 4a and 4b were all very small and almost the same, and much smaller than that in run 2a, indicating that the original slag with 15% Al₂O₃ had a significant inhibitory effect on the dissolution of corundum substrate. From another angle, the above results seemed to indicate that, adding Al₂O₃ into CaO–SiO₂–MgO slag could increase the apparent contact angle value of the slag on the corundum substrate and reduce the dissolution thickness of the substrate, but still increased the slag penetration.

Runs 4a and 5a with the original slag contained the same Al₂O₃ content of 15%. From Table 3 and Figures 7 and 8, runs 4a and 5a corresponded to two extreme cases of the substrate dissolution thickness and the slag penetration depth, respectively. As the MgO content in No. 4 original slag was 8%, the generation amount of MA phase was relatively small in run 4a; the slag penetration was still pronounced although the substrate direct dissolution was significantly inhibited. As the MgO content in No. 5 slag increased to 15%, a large amount of MA phase was generated in run 5a, which led to obvious indirect dissolution of the corundum substrate due to the interface chemical reaction; but the slag penetration depth decreased to zero, which should be attributed to the hindrance and inhibition effect of the newly generated MA phase with high melting point [31,32]. In addition, based on that the dissolution thickness of the substrate in No. 5 slag was larger than that in No. 4 slag, the real contact angle of No. 5 slag would also be greater than that of No. 4 slag. In other words, the change in the real contact angle here was consistent with that in the apparent contact angle as the MgO content increased. Therefore, for runs 4a and 5a with the same binary basicity and holding time (10 min), it can be seen from Table 3 and Figure 6 that the contact angle values increased within comparable time (temperature) with the increase of MgO content in the original slag from 8% to 15%, but the slag penetration depth decreased to zero and the indirect dissolution thickness of the substrate increased significantly due to the generation of a large amount of MA phase at the interface.

Discussion

Under this experimental condition, the corrosion of corundum substrate included direct and indirect dissolution of corundum in the molten slag and penetration of molten slag into corundum. The direct dissolution reaction of the substrate can be expressed as

$${\rm A}{\rm l}_ 2{\rm O}_ 3\lpar {\rm s} \rpar \to \lpar {{\rm A}{\rm l}_ 2{\rm O}_ 3} \rpar $$

A l_{2} O_{3} (s) \to (A l_{2} O_{3})

(1)

where (Al₂O₃) denotes alumina dissolved in molten slag, later (MgO) and (CaO) also have similar meanings. It is generally considered that the rate-limiting step in the direct dissolution process is the diffusion of (Al₂O₃) in the boundary layer [14,20–22]. The following mass transfer equation can be used for the direct dissolution process:

$$J = \beta\lpar {C_i-C_0} \rpar $$

J = β (C_{i} - C_{0})

(2)

where J is the mass transfer flux; β is the mass transfer coefficient of (Al₂O₃); C_i and C₀ are (Al₂O₃) concentrations at the slag/corundum interface and in the bulk of slag droplet, respectively. It can be seen from Equation (2) that, the greater the values of (C_i–C₀) and β, the greater the dissolution rate of corundum in molten slag. For Nos. 1–3 original slags without Al₂O₃, C₀ in the initial stage was zero, and the value of (C_i–C₀) was the maximum. As the wetting process proceeded, the corundum dissolved into the slag, then (Al₂O₃) content in the slag bulk became higher and higher, and (C_i–C₀) decreased accordingly; furthermore, with the increase of (Al₂O₃) content, the quaternary basicity of the slag decreased and the viscosity would increase even under heterogeneous condition [30,33], resulting in the decrease of β value. The above reasons led to a decrease of diffusion rate, which was consistent with the phenomenon observed in the experiments that the corundum substrate dissolved more at the early stage of holding temperature and less at the later stage. Using FactSage thermodynamic software [30], it can be calculated that Al₂O₃ solubility prior to the generation of new phase is 18.5 mass% in No. 1 original slag at 1450°C, and those in Nos. 2 and 3 slags are 18.3 and 18.2 mass%, respectively. So it can be considered the automatic dissolution trend of corundum substrate in the molten slag was very large. Under the condition of the same binary basicity, the Al₂O₃ solubility decreased slightly with the increase of MgO content in Nos. 1–3 original slags, which also explained the result on the slight decrease of substrate dissolution thicknesses (see Table 3) at holding time of 10 min. For Nos. 4 and 5 original slags with 15% Al₂O₃, the initial value of (C_i–C₀) was very small, so the direct dissolution trend of corundum substrate in the two molten slags was not obvious and the direct dissolution thickness (see Table 3) could be basically ignored. It can be concluded that the driving force for the corundum to direct dissolution into the slag decreases with the increase of Al₂O₃ content in the original slag. In addition, it is seen from Table 4 that high-melting-point phases such as MA and CA₂ were generated at the slag/substrate interface (see Figure 8), they were believed to inhibit the direct dissolution of the substrate to a certain extent [2,34].

Here, the formation of MA and CA₂ might be related to two reaction mechanisms: Firstly, (MgO) and (CaO) reacted with solid corundum substrate, as given in the following Reactions (3) and (4), respectively. Secondly, (MgO) and (CaO) also reacted with (Al₂O₃) contained in the original slag, as given in Reactions (5) and (6), respectively. Reactions (3)–(6) at 1450°C can be analysed by FactSage thermodynamic software [30].

$$\lpar {{\rm MgO}} \rpar {\rm}+{\rm A}{\rm l}_ 2{\rm O}_ 3\lpar {\rm s} \rpar {\rm}={\rm MgA}{\rm l}_ 2{\rm O}_ 4\lpar {\rm s} \rpar \; \; \; \; \; \; \Delta G_1^\theta = -68002{\rm J\times mo}{\rm l}^{-1}$$

(MgO) + A l_{2} O_{3} (s) = MgA l_{2} O_{4} (s) Δ G_{1}^{θ} = - 68002 J \times mo l^{- 1}

(3)

$${\rm 1/2}\lpar {{\rm CaO}} \rpar {\rm}+{\rm A}{\rm l}_ 2{\rm O}_ 3\lpar {\rm s} \rpar {\rm}={\rm 1/2CaA}{\rm l}_ 4{\rm O}_ 7\lpar {\rm s} \rpar {\rm \; }\; \Delta G_2^\theta = -50896{\rm J\times mo}{\rm l}^{-1}$$

1 / 2 (CaO) + A l_{2} O_{3} (s) = 1 / 2 CaA l_{4} O_{7} (s) Δ G_{2}^{θ} = - 50896 J \times mo l^{- 1}

(4)

$$\lpar {{\rm MgO}} \rpar + \lpar {{\rm A}{\rm l}_ 2{\rm O}_ 3} \rpar {\rm}={\rm MgA}{\rm l}_ 2{\rm O}_ 4\lpar {{\rm s}{\rm .s} .} \rpar \; \; \; \; \; \Delta G_3^\theta = -98210{\rm J\times mo}{\rm l}^{-1}$$

(MgO) + (A l_{2} O_{3}) = MgA l_{2} O_{4} (s . s .) Δ G_{3}^{θ} = - 98210 J \times mo l^{- 1}

(5)

$${\rm 1/2}\lpar {{\rm CaO}} \rpar + \lpar {{\rm A}{\rm l}_ 2{\rm O}_ 3} \rpar {\rm}={\rm 1/2CaA}{\rm l}_ 4{\rm O}_ 7\lpar {\rm s} \rpar \; \; \Delta G_4^\theta = -81106{\rm J\times mo}{\rm l}^{-1}$$

1 / 2 (CaO) + (A l_{2} O_{3}) = 1 / 2 CaA l_{4} O_{7} (s) Δ G_{4}^{θ} = - 81106 J \times mo l^{- 1}

(6)

where ΔG_i^θ (The subscript i takes values of 1–4) is the standard Gibbs free energy change of corresponding Reactions (3)–(6). For (MgO), (CaO) and (Al₂O₃), their respective pure liquid substances are taken as the standard states. It is well known that a certain amount of MgO or Al₂O₃ can dissolve in the MA(s) to form a spinel solid solution (MA(s.s.)) phase [34,35], MA in MA(s.s.) takes pure stoichiometric solid substance as the standard state. MA(s) and MA(s.s.) can be formed in Reactions (3) or (5). Here, Reactions (3) and (5) were used as an example of forming MA(s) and MA(s.s.), respectively. The ΔG_i^θ values are less than zero, indicating that the above reactions can proceed under the standard state. Further, the activities of the corresponding components in each original slag can be calculated from FactSage thermodynamic software and the Gibbs free energy (ΔG_i) values of each reaction are obtained by the isothermal equation under the non-standard state, as shown in Table 5.

Table 5.

Activities of components in the original slag and ΔG_i values of each reaction at 1450°C.

Slag No	(MgO) activity	(CaO) activity	(Al₂O₃) activity	ΔG₁/J·mol⁻¹	ΔG₂/J·mol⁻¹	ΔG₃/J·mol⁻¹	ΔG₄/J·mol⁻¹
1	1. 50 × 10⁻²	2. 10 × 10⁻³	–	−7841	−6733	–	–
2	1. 73 × 10⁻²	2. 09 × 10⁻³	–	−9885	−6733	–	–
3	1. 79 × 10⁻²	2. 04 × 10⁻³	–	−10364	−6525	–	–
4	2. 08 × 10⁻²	3. 65 × 10⁻³	1. 00 × 10⁻²	−12524	−10593	23237	106271
5	7. 77 × 10⁻²	6. 24 × 10⁻³	8. 41 × 10⁻³	−31403	−14534	−1391	23754

Checking the composition of spinel phase in Table 5, it is found that a small amount of Al₂O₃ dissolved in the MA formed in this work, indicating that the activity of MA in spinel solid solution phase should be less than 1, which was beneficial to the formation of MA from thermodynamics. It can be seen from Table 5 that the ΔG₁ and ΔG₂ values related to the five slags were all less than zero even if the activities of MA and CA₂ were taken as the maximum value of 1, indicating that MA and CA₂ could be generated by Reactions (3) and (4) under the non-standard state, respectively. Moreover, MgO activity in the five slags increased with the increase of MgO content, and the ΔG₁ value became more negative, indicating that the generation of MA became easier and easier, which was consistent with the phenomenon observed in the experiments. For Nos. 1–3 slags, the distribution of MA formed at the interface was discontinuous. This was because MgO content in these slags was low; with the dissolution of Al₂O₃ into the slags, MgO content in the slags further decreased, resulting in only a small amount of MA formed, which was also consistent with the research results of Sarkar [1], Lee [7] and Zhang et al. [36].

For the reactions in the same slag, the ΔG₁ value was less than the ΔG₂ value, indicating that MA was generated preferentially. From the mapping (Figure 8) and EDS analysis (Table 4) after the experiment, it is found that the amount of MA was more than that of CA₂. The CA₂ layer formed at the interface was also discontinuous. In the corundum substrate, Al₂O₃ could form CA₂ with CaO in the permeated slag. For No. 4 slag with 15% Al₂O₃, compared with Nos. 1–3 original slags without Al₂O₃, the amount of the MA phase from Reaction (3) significantly increased and accumulated on the surface of the substrate, which significantly reduced the direct dissolution of the substrate. However, No. 4 slag still showed a certain vertical penetration [37–39], perhaps due to its lowest melting point, a large surface tension [40], and a small contact angle on the substrate. It is speculated that the penetration of No. 4 slag occurred prior to the formation of a large amount of MA phase, and the increase of surface tension of the original slag was considered to be the main factor for the increase of slag penetration depth.

Compared with No. 4 slag, MgO content in No. 5 slag increased to 15%, which promoted indirect dissolution of the substrate (ΔG₁ value of Reaction (3) with No. 5 slag was the most negative), resulting in a significant increase in the amount of MA phase formation, subsequent with the increases in the slag viscosity and the contact angle between the slag and the substrate. And CaO concentration in No. 5 slag relatively decreased, consequently almost no slag penetration into the substrate was observed. However, the indirect dissolution of the substrate significantly increased the dissolution thickness. Since the ΔG₁ value was far less than the ΔG₂ value, the generation of MA in run 5a was much easier than that of CA₂, and therefore the CA₂ phase was not even found.

It can be seen from Table 4 that the ΔG₃ and ΔG₄ values related to No. 4 slag were both more than zero, indicating that Reactions (5) and (6) could not occur in No. 4 original slag. In fact, for runs 4a and 4b, no MA and CA₂ phase were found in the slag phase, which was consistent with the theoretical calculation results. It is found that the slag penetration depth was large, but the dissolution thickness of the substrate was very small. This was because the melting point of No. 4 slag was low, it is speculated that the vertical penetration of the slag started and reached a certain degree once slag melting. After the substrate slightly dissolved, Al₂O₃ content near the interface increased, while the activity of MA decreased due to the dissolution of a certain amount of Al₂O₃ in the spinel phase, and the generation of MA by Reaction (5) became easy. After a large amount of MA was formed, the slag penetration would be inhibited.

Reaction (5) related to No. 5 original slag could generate spinel solid solution phase at 1450°C in run 5a due to ΔG₃ = −1391 J·mol⁻¹ < 0, in which the activity of MA was 0.56 from the calculation of FactSage software. It is seen that Reactions (3) and (5) related to No. 5 slag could occur, and Reaction (3) was the main one. The ΔG₄ value was more than zero, indicating that Reaction (6) could not occur in No. 5 original slag.

In the actual process, with continuous dissolution of the substrate in the molten slag, the Al₂O₃ content increased, while the contents of other components decreased correspondingly, so the thermodynamic analyses of the above reactions might change. But anyway, the substrate was first wetted after the slag sample melting, and then the substrate was dissolved directly and indirectly in the molten slag. In general, under the experimental conditions, Nos. 1–3 slags did not contain Al₂O₃, and the direct dissolution trend of the substrate was large; when the MgO content in each original slag increased slightly, the direct dissolution thickness and penetration depth decreased slightly; however, with the direct dissolution of the substrate, the increase of Al₂O₃ content in the molten slag could partially offset the increase of MgO content in the original slag, resulting in a very small degree of indirect dissolution of the corundum substrate due to the chemical reaction. Nos. 4 and 5 slags contained 15% Al₂O₃, which could inhibit the direct dissolution of the substrate to a large extent; when the MgO content in the original slag increased from 8% to 15%, the interface chemical reaction with the substrate was promoted, the indirect dissolution of the substrate was intensified, and the penetration of the slag into the substrate was prevented due to the massive generation of the new phase with high melting point. It should be pointed out that the direct dissolution was usually undesirable while the indirect might be beneficial if the boundary layer led to a lower rate of corrosion.

Conclusions

In this work, the contact angles of CaO–SiO₂–MgO (−15% Al₂O₃) molten slags on the corundum substrate, the substrate dissolution thicknesses in the molten slags, and the slag penetration depths in the substrate were measured by the sessile drop technique combined with the SEM–EDS detection, and effect of MgO content on the wetting and corrosion behaviour of substrate by the molten slags at 1450°C was investigated. The main conclusions are as follows:

CaO–SiO₂–MgO (−15% Al₂O₃) slags show good wettability to corundum substrate, and the contact angles are small. When the temperature is held at 1450°C for 10 min, the final contact angle values of Nos. 1–5 slags fluctuate at around 6.5°, 12°, 15°, 14° and 17°, respectively.

In CaO–SiO₂–MgO slags without Al₂O₃, the trend of automatic dissolution of the corundum substrate is large. Under the condition that the slag basicity and the holding time remain unchanged, increasing the MgO content from 7% to 9% in the original slag can increase the contact angle between the slag droplet and the substrate, while the substrate dissolution thickness and the slag penetration depth tend to decrease.

In CaO–SiO₂–MgO-15% Al₂O₃ slags, when the MgO content is as low as 8%, the dissolution thickness of corundum substrate is very small, but the slag penetration depth is large; when the slag basicity and Al₂O₃ content remain unchanged, increasing the MgO content to 15% in the original slag can increase the contact angle, but significantly reduce the slag penetration depth; due to the formation of a large amount of MA phase in the interface reaction, the indirect dissolution corrosion of the substrate significantly increases instead.

The wettability (or contact angle) can give an indication of the slag penetration depth in the substrate but not the dissolution thickness of the substrate.

Footnotes

Disclosure statement

No potential conflict of interest was reported by the author(s).

References

Sarkar

, Sohn

HY.

Interactions of alumina-based and magnesia-based refractories with iron melts and slags: a review. Metall Mater Trans B. 2018;49:1860–1882. doi:10.1007/s11663-018-1300-1

Sandhage

, Yurek

GJ.

Direct and indirect dissolution of sapphire in calcia-magnesia-alumina-silica melts: dissolution kinetics. J Am Ceram Soc. 1990;73:3633–3642. doi:10.1111/j.1151-2916.1990.tb04269.x

Sandhage

, Yurek

GJ.

Indirect dissolution of sapphire into calcia-magnesia-alumina-silica melts: electron microprobe analysis of the dissolution process. J Am Ceram Soc. 1990;73:3643–3649. doi:10.1111/j.1151-2916.1990.tb04270.x

Sandhage

, Yurek

GJ.

Indirect dissolution of (Al,Cr)₂O₃ in CaO-MgO-Al₂O₃-SiO₂ (CMAS) melts. J Am Ceram Soc. 1991;74:1941–1954. doi:10.1111/j.1151-2916.1991.tb07813.x

Bates

JL.

Heterogeneous dissolution of refractory oxides in molten calcium-aluminum silicate. J Am Ceram Soc. 1987;70:C-C-55–C-C-57. doi:10.1111/j.1151-2916.1987.tb04965.x

Jansson

, Brabie

, Jönsson

Corrosion mechanism and kinetic behaviour of MgO-C refractory material in contact with CaO-Al₂O₃-SiO₂-MgO slag. Scand J Metall. 2005;34:283–292. doi:10.1111/j.1600-0692.2005.00748.x

Lee

, Zhang

Melt corrosion of oxide and oxide-carbon refractories. Int Mater Rev. 1999;44:77–104. doi:10.1179/095066099101528234

Cheng

, Zhao

et al.Wetting, cladding and corrosion properties of alkaline slag on dense corundum refractories. Ceram Int. 2022;48:5795–5804. doi:10.1016/j.ceramint.2021.11.127

Yoon

, Lee

Wetting, spreading and penetration phenomena of slags on MgAl₂O₄ spinel refractories. ISIJ Int. 2017;57:1327–1333. doi:10.2355/isijinternational.ISIJINT-2016-578

10.

Tang

, Wu

, Wang

et al.Comparative evaluation investigation of slag corrosion on Al₂O₃ and MgO-Al₂O₃ refractories via experiments and thermodynamic simulations. Ceram Int. 2017;43:16502–16511. doi:10.1016/j.ceramint.2017.09.034.

11.

Song

, Liu

, Lv

et al.Corrosion behavior of Al₂O₃ substrate by SiO₂-MgO-FeO-CaO-Al₂O₃ slag. J Mater Res Technol. 2020;9:314–321. doi:10.1016/j.jmrt.2019.10.060

12.

Kumar

, Khanna

, Ikram-ul-Hag

et al.Corrosion behavior of Al₂O₃-C refractories with casting mould meniscus slags at 1550°C. Steel Res Int. 2016;87:46–56. doi:10.1002/srin.2014005902

13.

Vázquez

, Pena

, De Aza

et al.Corrosion mechanism of polycrystalline corundum and calcium hexaluminate by calcium silicate slags. J Eur Ceram Soc. 2009;29:1347–1360. doi:10.1016/j.jeurceramsoc.2008.08.031

14.

Cho

, Fan

Diffusional dissolution of alumina in various steelmaking slags. ISIJ Int. 2004;44:229–234. doi:10.2355/isijinternational.44.229

15.

Tripathi

, Malfliet

, Blanpain

et al.Dissolution behavior and phase evolution during aluminum oxide dissolution in BOF slag. Metall Mater Trans B. 2019;50:1782–1790. doi:10.1007/s11663-019-01590-4

16.

Chen

, He

, Wang

Dissolution behavior of Al₂O₃ into tundish slag for high-al steel. J Mater Res Technol. 2020;9:11311–11318. doi:10.1016/j.jmrt.2020.07.107

17.

Monaghan

, Chen

Dissolution behavior of alumina micro-particles in CaO-SiO₂-Al₂O₃ liquid oxide. J Non Cryst Solids. 2004;347:254–261. doi:10.1016/j.jnoncrysol.2004.09.011

18.

Valdez

, Prapakorn

, Cramb

et al.Dissolution of alumina particles in CaO-Al₂O₃-SiO₂-MgO slags. Ironmak Steelmak. 2002;29:47–52. doi:10.1179/030192302225001965

19.

Bui1

, Ha

, Chung

et al.Effect of alumina content and solid phase in molten flux on dissolution of alumina. Met Mater Int. 2005;11:319–326. doi:10.1007/BF03027336

20.

Choi

, Lee

, Kim

JS.

Dissolution rate of Al₂O₃ into molten CaO-SiO₂-Al₂O₃ slags. ISIJ Int. 2002;42:852–860. doi:10.2355/isijinternational.42.852

21.

Taira

, Nakashima

, Mori

Kinetic behavior of dissolution of sintered alumina into CaO-SiO₂-Al₂O₃ slags. ISIJ Int. 1993;33:116–123. doi:10.2355/isijinternational.33.116

22.

Liu

, Zou

, Guo

et al.Phase field simulation study of the dissolution behavior of Al₂O₃ into CaO-Al₂O₃-SiO₂ slags. Comput Mater Sci. 2016;119:9–18. doi:10.1016/j.commatsci.2016.03.034

23.

Wang

, Xue

, Zhang

et al.Thermodynamic corrosion behavior of Al₂O₃, ZrO₂ and MgO refractories in contact with high basicity refining slag. Ceram Int. 2019;45:20664–20673. doi:10.1016/j.ceramint.2019.07.049

24.

Kim

, Lee

, Chung

Dissolutive wetting and spreading phenomena between Al₂O₃ substrate and CaO-Al₂O₃ liquid slags. Metall Mater Trans B. 2016;47:1209–1216. doi:10.1007/s11663-016-0587-z

25.

Monaghan

, Zhang

, Longbottom

RJ.

Wettability of CaO-SiO₂-MgO-Al₂O₃ liquid slag on solid Al₂O₃. Metall Mater Trans B. 2021;52:4023–4032. doi:10.1007/s11663-021-02317-0

26.

Shen

, Fujii

, Nogi

Wettability of some refractory materials by molten SiO₂-MnO-TiO₂-FeO_x slag. Mater Chem Phys. 2009;114:681–686. doi:10.1016/j.matchemphys.2008.10.021

27.

, Lv

, Xiang

et al.Wetting behavior of Al₂O₃ substrate by calcium ferrite series melts. ISIJ Int. 2015;55:483–490. doi:10.2355/isijinternational.55.483

28.

Seo

, Kim

, Paik

YH.

Wetting characteristics of CaO−SiO₂−Al₂O₃ ternary slag on refractory oxides, Al₂O₃, SiO₂ and TiO₂. Met Mater Int. 2001;7:479–483. doi:10.1007/BF03027090

29.

Choi

, Lee

HG.

Wetting of solid Al₂O₃ with molten CaO-Al₂O₃-SiO₂. ISIJ Int. 2003;43:1348–1355. doi:10.2355/isijinternational.43.1348

30.

CRCT-Thermfact Inc. and GTT Technologies . https://www.factsage.com/. Accessed 10 Jan 2023.

31.

Wang

, Glaser

, Du

SC.

Improvement of resistance of MgO-based refractory to slag penetration by in situ spinel formation. Metall Mater Trans B. 2015;46:749–757. doi:10.1007/s11663-014-0277-7

32.

Wang

, Du

SC.

Improvement of resistance to slag penetration in magnesia-based refractory with colloidal alumina addition II by means of a proper particle size distribution of MgO. Metall Mater Trans B. 2016;47:1858–1865. doi:10.1007/s11663-016-0649-2

33.

Allibert

, Gaye

, Geiseler

et al. Slag Atlas, 2nd Ed, Verein Deutscher Eisenhüttenleute (VDEh), Verlag Stahleisen GmbH, D-Düsseldorf, 1995: pp. 372,381.

34.

Fujii

, Nagasaka

, Hino

Activities of the constituents in spinel solid solution and free energies of formation of MgO, MgO Al₂O₃. ISIJ Int. 2000;40:1059–1066. doi:10.2355/isijinternational.40.1059

35.

Park

, Todoroki

Control of MgO·Al₂O₃ spinel inclusions in stainless steels. ISIJ Int. 2010;50:1333–1346. doi:10.2355/isijinternational.50.1333

36.

Zhang

, Rezaie

, Sarpoolaky

et al.Alumina dissolution into silicate slag. J Am Ceram Soc. 2000;83:897–903. doi:10.1111/j.1151-2916.2000.tb01291.x

37.

Wanibe

, Tsuchida

, Fujisawa

et al.Fundamental study on the infiltration of slags into refractories with the slagging reaction. Trans ISIJ. 1983;23:322–330. doi:10.2355/isijinternational1966.23.322

38.

Wanibe

, Tsuchida

, Fujisawa

et al.Infiltration of corrosive liquids into capillary tubes. Trans ISIJ. 1983;23:331–339. doi:10.2355/isijinternational1966.23.331

39.

Riaz

, Mills

, Bain

Experimental examination of slag/refractory interface. Ironmak Steelmak. 2002;29:107–113. doi:10.1179/030192302225003477

40.

Allibert

, Gaye

, Geiseler

et al. Slag Atlas, 2nd Ed, Verein Deutscher Eisenhüttenleute (VDEh), Verlag Stahleisen GmbH, D-Düsseldorf, 1995: pp. 440,447,448.

Effect of MgO on wetting and corrosion behaviour of corundum substrate by CaO–SiO 2 –MgO (–Al 2 O 3 ) slags

Abstract

Introduction

Experimental

Sample preparation

Experimental method

Detection method

Results and discussion

Changes of slag drop shape and contact angle

Microscopic observation

Discussion

Conclusions

Footnotes

Disclosure statement

References