Corrosion of Furnace Refractories by Molten Aluminum.

By understanding the causes of refractory corrosion and corundum growth, aluminum foundries can properly maintain their furnaces and reduce melt costs.

Refractories used in aluminum holding and melting furnaces are subjected to severe corrosion conditions. These refractories are degraded in service by the reaction with molten aluminum below the metal surface and by the reaction with the corundum that grows along the walls above the metal line.

This article discusses an experiment to determine which variables--temperature, atmosphere, alloy composition and fluoride salts--further promulgate the corrosion of aluminosilicate refractories when they are exposed to molten aluminum. This information can aid foundries in the selection of furnace refractories the operation of their furnace and the alloying procedures used on their molten aluminum.

Experimental Procedures

Two low-cement refractories with different alumina content were used (Table 1). Each refractory material (after being vibratory cast and dried according to the manufacturer specifications) was pre-fired at either 1500F (815C) or 2192F (1200C) for 5 hr. Refractory samples then were machined to cylindrical crucibles having an outer diameter and height of 45 mm with a central hole of 25 mm in diameter and 28 mm in depth.

Pure aluminum (Al), Al-5% magnesium (Mg), Al-5% Mg-5% zinc (Zn), and Al-5% Mg-7% silicon (Si) were the molten metals used in the experiment. These alloys then were cast in a graphite mold to obtain cylinders having a 20-mm diameter and 50-mm length. Discs of the alloys were cut to a thickness of 25 mm. The alloy then was placed in the refractory crucible. To determine the effect of cryolite on the corrosion resistance of these refractory materials, 0.5 weight % cryolite ([[Na.sub.3]Alf.sub.6]) was sprayed on the surface of the alloy in designated experiments. The crucible then was placed in the furnace.

Three types of heating procedures were used depending on the desirable atmosphere to be experimented. To determine the effect of oxygen (O), an electric resistance furnace was used where the lid of the furnace was placed on refractory bricks to allow the circulation of air from the outside atmosphere into the furnace chamber. This simulates the conditions that take place in zone 1 of the aluminum treatment furnaces as shown in Fig. 1. For the case where a limited supply of O was required, the same electric resistance furnace was utilized, but the lid tightly covered the top of the furnace, limiting the amount of O available during the melting process. This setup simulates the conditions taking place in zone 2 of the aluminum treatment furnaces (Fig. 1). For the case where 0 was to be present, a controlled-atmosphere electric furnace was used where the furnace chamber was purged with argon gas prior to heating, and then a constant argon (Ar) flow was maintained throughout the experiment. This setup simulates the conditions taking place in zone 3 of aluminum treatment furnaces (Fig. 1).

Each test lasted 4 days at a soaking temperature of either 1500F or 2192F for two different sets of experiments. This was done to determine the effect of melting temperature on the corrosion of the refractory samples. After each test, the crucible was sectioned using a diamond blade to reveal the extent of metal penetration into the refractory materials and any corundum formation.

Effects on Refractories

The results of the corrosion tests under different conditions for refractory material A with 60% alumina and refractory material B with 70% alumina are presented in Tables 2 and 3.

Temperature--Figure 2 shows that corrosion is more severe for the crucibles that were pre-fired at 2192F as illustrated by a deeper metal penetration into the refractory crucible in comparison to the prefiring temperature of 1500F. Previous work showed that at higher pre-firing temperatures, the non-wetting agents in the refractory materials lose their efficiency and promote the corrosion of refractory castables. Moreover, increasing the corrosion test temperature for a given pre-firing temperature should favor the reduction of oxides such as silica. Another reason for the reduced corrosion resistance of the refractories as the corrosion test temperature increases is that at higher temperatures, the viscosity and the surface tension of the molten metal is lower, which enhances the infiltration of the liquid metal into the refractory material.

Atmospheric conditions--Under the circulated air conditions, corundum formed from the reaction of aluminum with O at 2102F (1150C). However, for the test temperature of 1562F (850C) under the condition of circulated air, neither corundum formation nor metal penetration in the refractory was detected Tables 2 and 3). Corundum was detected to form under low partial pressure of O, but only in the presence of cryolite with Al-Mg and Al-Mg-Si alloys. Under the cover of Ar gas, corundum formation was not observed. However, the metal reacts with the refractory at high and low prefiring and test temperatures. In this condition, the only source of O to react with the aluminum alloy is from the oxides that are present in the refractory materials. Thus, the metal penetration in this condition is equivalent to the ability of the alloy to reduce the oxides such as [SiO.sub.2] in refractory materials.

Alloy composition--The presence of alloying elements also has a remarkable effect on the corrosion resistance of the refractories. The effect of the alloy composition was observed to change with the atmospheric condition. For instance, pure Al was seen to corrode both refractories at 2102F more easily than Al-Mg or Al-MgSi alloys in the presence of O, as shown by a higher penetration of the metal in the refractory. But at 1562F, neither pure Al nor the alloys reacted with any of the refractories--regardless of the prefiring temperature under the same conditions. However, under the cover of Ar gas and the absence of O, pure Al was observed to be less corrosive at 1562F than at 2102F (Fig.3). In these conditions, the increase in corrosion resistance only was observed with pure Al, which compares to the alloys for refractory material A that contain about 60% alumina.

The oxidation of pure Al is known to result in the formation of a thin protective layer of [Al.sub.2][O.sub.3] The incorporation of Mg into the alloy in the presence of O at a high temperature results in the formation of spinel and corundum with interconnected metallic channels. Although this composite contains metal channels, it acts as a barrier due to its greater thickness compared to the oxide that forms on pure Al. In addition, the formation of the spinel and corundum might block the pores of the refractories and thus reduce their infiltration by molten aluminum. However, Al-Mg alloys still corrode the refractory due to the supply of molten metal to the reaction front via the metal channels created within the corundum network. Corundum formation has been observed only when O and Mg are present and the temperature is high enough.

The role of Si and Zn in the corrosion of refractory materials, however, is not clear from the present work. While these two elements limited the extent of metal penetration in some experiments, they enhanced the corrosion in others. For Si, this may be explained by the fact that this element has two opposite effects on corrosion: it unfavorably reduces the metal viscosity but favorably decreases the tendency of the silica to be reduced since the metal is more concentrated to metallic Si. An addition of Zn was expected to increase the corrosion since it reduces the metal surface tension, however, reaction products such as [ZnAl.sub.2] [O.sub.4] may favorably interfere with the corrosion process.

Cryolite--Previous work has shown that cryolite can reduce the corrosion resistance of refractories in some cases and can improve the resistance in others. However, for the two refractories that were tested under all conditions at high temperatures, cryolite was observed to reduce the corrosion resistance of the refractories and favored the formation of corundum with the alloys in the presence of O.

In general, when liquid aluminum alloy is in contact with a refractory material containing silicates, a reaction product composed mainly of [Al.sub.2] [O.sub.3], forms. This product may act as a barrier against the penetration of the metal into the refractory. In most cases, however, the protective oxide layer that forms contains metal channels that keep feeding the metal to the reaction front and results in further degradation of the refractory. The presence of cryolite results in further corrosion as it accelerates the formation of the oxide product containing larger amounts of metal. It previously was reported that the directed oxidation of aluminum alloys in the presence of NaOH dopant occurs in the temperature range of 1562-2192F. In these conditions, the oxidation was observed to occur not only on the surface of the aluminum but also along the crucible wall. This suggests that the liquid aluminum can strongly wet the crucible wall because of the presence of the dopant, which promotes wetting.

The corrosion of the refractory sample by the molten metal and by the newly formed corundum is operated by two separate mechanisms. Whenever corundum was formed, three conditions were observed: the refractory was resistant to both corundum and molten metal attack (Fig. 4a); the refractory was corroded by the corundum, but was resistant to the molten metal (Fig. 4b); and the refractory was corroded by both corundum and the molten metal (Fig. 4c). In the case where corrosion took place by both corundum and molten metal attack, there always is a distinct separation between the two corroded areas in the refractory. This area corresponds to the metal line. Thus, corrosion due to molten metal and the formation of corundum takes place independently, but may aid each other when the extent of the corrosion is severe enough.

Based on the results, the corrosion of the refractory lining at and below the metal line in aluminum treatment furnaces mainly takes place by the reduction of the oxides (mainly [SiO.sub.2]) under the action of the molten metal. The molten metal is brought above the metal line in such furnaces by the corundum that forms.

The conditions promoting each of the above mechanisms include: favoring the reduction of silica when partial pressure of oxygen decreases, metal temperature increases, pre-firing temperature increases, when cryolite is present; and enhanced directed oxidation of aluminum when partial pressure of O increases, and Mg, Si and cryolite are present.

When comparing the two tested materials, it can be observed from Tables 2 and 3 that material B, which contains 70% alumina, has a higher resistance to corrosion by pure aluminum at the higher pre-firing and test temperatures. At a pre-firing temperature of 1500F and a test temperature of 1562F, its resistance to corrosion by pure aluminum (under Ar) is even less than that of material A, which contains about 60% alumina. However, at these two pre-firing temperatures and atmospheres, material B showed more resistance to Al-Mg-Zn and Al-Mg-Si alloys at the test temperature of 1562F.

Conclusions

The following conclusions can be drawn from these experiments:

* The corrosion of aluminosilcate refractories at or below the bellyband area in aluminum holding and melting furnaces occurs during the reduction of the oxides (mainly [SiO.sub.2]) by the molten metal;

* Such corrosion is favored at higher operating temperatures, higher prefiring temperatures (for castables or other unshaped refractories) and a lower supply of O;

* While Mg can increase the extent of corrosion at low temperatures below the bellyband area in aluminum treatment furnaces, its presence reduces the corrosion of refractory materials at high temperatures near the metal line;

* The presence of Si and Zn in the alloy does not have a clear effect on the reduction of the refractory oxides. However, Si in the presence of Mg promotes the corundum growth at the metal line.

               Composition of the Refractory Materials Tested
Compounds                Material Material
                            A        B
[Al.sub.2][O.sub.3]       59.6%    70.0%
[SiO.sub.2]               34.0%    23.1%
[Fe.sub.2][O.sub.3]        0.8%     0.8%
CaO                        3.8%     2.8%
MgO                        0.2%     0.3%
[TiO.sub.2]                1.3%     1.7%
[Na.sub.2O] + [K.sub.2O]   0.3%     0.2%
Non-Wetting Agents         yes      yes
                     Results of the Corrosion Tests on
                  Refractory Material A with 60% Alumina
Pre-Firing Temperature 2192F (1200C)                          1500F (815C)
Test Temperature       2102F (1150C)             1562F (850C) 1562F (850C)
Alloy                    Corundum       Metal       Metal        Metal
                          Formed     Penetration Penetration  Penetration
Circulated air
Al                          No          5 mm         0 mm         0 mm
Al-5% Mg                    No          0 mm         0 mm         0 mm
Al-5% Mg-7% Si             Yes [*]     1-2 mm        0 mm         0 mm
                           Yes [**]    6-7 mm        0 mm         0 mm
Enclosed air
Al                          No         4-5 mm        0 mm         0 mm
                            No          5 mm         0 mm         0 mm
Al-5% Mg                   Yes [*]      0 mm         0 mm         0 mm
                           Yes [*]     3-4 mm        0 mm         0 mm
Al-5%Mg-5%Zn                No         0.5 mm        0 mm         0 mm
Al-5%Mg-7%Si                No         0.5 mm        0 mm         0 mm
                           Yes [**]     3 mm         0 mm         0 mm
Argon gas
Al                          No         5-7 mm       1-2 mm       0.5 mm
Al-5%Mg                     No         1-2 mm        2 mm         1 mm
Al-5%Mg-5%Zn            Not tested   Not tested     2-3 mm        2 mm
Al-5%Mg-7%Si                No         4-5 mm       3-4 mm        3 mm
(*.)No contamination of crucible
(**.)Crucible contaminated with corundum
0.5 wt.% [Na.sub.3][AIF.sub.6] added
                     Results of the Corrosion Tests on
                  Refractory Material B with 70% Alumina
Pre-Firing Temperature 2192F (1200C)                          1500F (815C)
Test Temperature       2102F (1150C)             1562F (850C) 1562F (850C)
Alloy                    Corundum       Metal       Metal        Metal
                          Formed     Penetration Penetration  Penetration
Circulated air
Al                          No         0.5 mm        0 mm         0 mm
Al-5%Mg                    Yes [*]      0 mm         0 mm         0 mm
Al-5%Mg-7%Si               Yes [**]     3 mm         0 mm         0 mm
                           Yes [*]     4-5 mm        0 mm         0 mm
Enclosed air
Al                          No         2-3 mm        0 mm         0 mm
                            No          5 mm         0 mm         0 mm
Al-5%Mg                     No         1-2 mm        0 mm         0 mm
                           Yes [*]     3-4 mm        0 mm         0 mm
Al-5%Mg-5%Zn                No          0 mm         0 mm         0 mm
Al-5%Mg-7%Si                No         0.5 mm        0 mm         0 mm
                           Yes [*]      3 mm         0 mm         0 mm
Argon gas
Al                          No          3-4mm       1-2 mm       1-2 mm
Al-5%Mg                     No          1 mm         2 mm         1 mm
Al-5%Mg-5%Zn            Not tested   Not tested      1 mm        0.5 mm
Al-5%Mg-7%Si                No         3-4 mm       1-2 mm       0.5 mm
(*.)No contamination of crucible
(**.)Crucible contaminated with corundum
0.5 wt.% [Na.sub.3] [AIF.sub.6] added

Techniques for the Prevention of Corundum

There are several ways to deal with corundum formation. First, the use of low silica content firebrick has been among the earliest recommendations because the less silica that is available for conversion, the better. The advent of low cement castables and hard refractories incorporating non-wetting agents to reduce molten aluminum penetration also have lessened the propensity for corundum formation.

Refractory composition and refractory surface treatment also appear to play an important role in minimizing corundum formation. A study showed in testing that corundum did not develop on commercial phosphate bonded 70% and 82% alumina, zircon or spinel refractories. In addition, a phosphoric acid treatment or a spinel mortar coating prevented corundum formation on 70% and 90% aluminum refractories exposed to recycled aluminum scrap melting. The phosphoric acid treatment most likely closes off some of the refractory's porosity, rendering it less permeable to molten aluminum. In another study, tar bonded magnesite brick, mullite and zircon brick, and mortar have been found to provide excellent resistance to corundum conversion.

In another study in which four types of castable alumina refractories for aluminum melting were studied, it was determined that refractory composition is the most important factor in preventing corundum formation assuming melting furnaces are properly managed and operated.

Modern furnace operating practices are successfully managed by avoiding unnecessary melt superheat, direct flame impingement and negative furnace pressures, which reduces corundum formation reaction rates by limiting the oxygen available. Doors, sills and flue ducts also should be properly maintained. Periodic cleaning and prudent use of wall cleaning furnace fluxes can reduce the corundum formation problems and ease cleanup before it becomes unmanageable. However, these cleanings should be careful, thorough and avoid gouging of the base furnace refractory brick. In addition, fluxes should be used sparingly as their corrosive nature in wall cleaning almost always affects the base refractory.

--Adapted from 'Mechanism of Corundum Formation and Prevention Techniques" by David V. Neff Metaullics Systems, and Raymond G. Teller, BP Research, as presented at the 1989 AFS Molten Aluminum Processing Conference.