CORROSION OF POTLINING REFRACTORIES: A COMPARISON BETWEEN DIFFERENT CELL LINING DESIGNS USING A UNIFIED APPROACH R. Pelletier and C. Allaire CIREP-CRNF Dept. Of Eng. Physics & Materials Eng. Ecole Polytechnique (CRIQ campus) 8475 Christophe-Colomb Street Montreal, Quebec, Canada, H2M 2N9 ABSTRACT The corrosion chemistry of aluminosilicate potlining refractories has been studied from post-mortem analysis according to a unified approach that considers both metallic sodium and molten bath has the corrosive agents. The results show that the cell lining design has a great influence on the type of corrosion involved. Cells containing semi- graphitized cathode blocks are more prone, with time, to the attack of their refractory potlining by the action of the molten bath. Unlike this case, the use of amorphous blocks in such cells promotes the corrosion of their potlining refractories mostly by the action of the metallic sodium, irrespectively of the cell’s age. INTRODUCTION In the past, two corrosive agents have been identified as the main responsible for aluminosilicate corrosion in potlining: liquid sodium fluoride and metallic sodium. Two distinct theoretical approaches have been developed to interpret laboratories and post-mortem observations. However, each approach can only explain a fraction of the wide range of observations made in used potlinings. These two approaches have recently been unified, resulting in a better understanding of real life conditions. The theoretical strategy that has permit to reach such unification has been presented elsewhere [1] and is resumed in the following. UNIFIED APPROACH The following development was inspired from the work of Schøning et al. [2]. In the approach that proposes molten NaF as the corrosive agent, the reaction pattern is: NaF + [aluminosilicate refractory] Na3 AlF 6 + [Al-Si-Na-O] compounds (1) Basically, NaF attacks alumina to produce cryolite and Na2 O, which combines with the remaining refractory. In the other approach, the reaction pattern is: Na(g) + [atm] + [aluminosilicate refractory] [Al-Si-Na-O] compounds (2) The sodium reacts with the atmosphere (gaseous oxygen or carbon monoxide) and/or with the silica contained in the refractory to produce Na2 O, which reacts with the remaining refractory. The reaction with the atmosphere should be favored if the gaseous oxygen renewal in the lining is fast. The availability of gaseous oxygen depends on the lining permeability and on the gas tightness of the cell. In the unified approach, it was assumed that all the sodium is oxidized by the atmosphere. Under this assumption, the reaction pattern becomes: Na2 O + [aluminosilicate refractory] [Al-Si-Na-O compounds] (3) The reaction patterns presented in equations (1) and (3) are linked by the following basic reaction: 3Na2 O + 2AlF 3 6NaF + Al2 O3 (4) Combining reaction patterns (1) and (3) with the help of the reaction (4) results in the following unified corrosion pattern: Na2 O + NaF + AlF 3 + [aluminosilicate refractory] [Na-Al-F compounds] + [Al-Si-Na-O compounds] (5) A good mean of visualising the predictions resulting from this corrosion pattern is by drawing what could be called "corrosion maps". CORROSION MAPS Corrosion maps are designed to visualize the evolution of the mineral composition of refractories during the corrosion process. Fundamentally, they are projected views of coexistence diagrams along certain paths. Figure 1 shows the trigonal composition prism proposed by Rutlin and Grande [3], which summarizes the coexistence lines of the system Na6 F6-Si3/2 F6-Al2 F6 -Na6O3-Si3/2 O3-Al2 O3 . In this diagram, all the aluminosilicate refractories lie on the Si3/2 O3 -Al2 O3 line, which is the silica-alumina line. Si 3/2 O 3 Na2 Si2 O 5 Na2 SiO3 NaAlSi 3 O8 NaAlSiO4 Na4 SiO4 Al 6Si 2 O 13 NaAl 11 O17 Al 2O 3 NaAlO2 Na6O3 Si3/2F6 Na2 SiF6 Al 2F6 Na3 AlF6 Na5 Al3F14 Na 6F 6 Figure 1: Coexistence lines in the quaternary reciprocal system Na2 O-SiO 2 -Al2 O3 NaF-SiF 4 -AlF 3 at subsolidus temperature proposed by Rutlin and Grande [3]. During the corrosion process, at least those described by the corrosion pattern (5), the mineral composition of a refractory shifts from a point on this line (initial composition) to another located in the Na6 O3-Na6 F6-Al2 F6 plane (ultimate composition). The exact location of the latter point depends on the proportion of Na2 O, NaF and AlF 3 in the mixture corroding the refractory. These two points delimit a line corresponding to a corrosion path. Each path is different because it passes through a certain number of domains at specific proportions of refractory and corrosive agents. Note that in coexistence diagrams such as the one shown in figure 1, the domains are bounded by four triangular planes that are each delimited by three coexistence lines. Because all the paths that converge to the same ultimate composition form a plane, they can be represented on a single corrosion map. Therefore, a map shows the behavior of all the aluminosilicate refractories in a specific corrosion condition. With only a few maps, it is possible to predict the corrosion path of any aluminosilicate refractory in all the conditions taken into account by the corrosion pattern (equation (5)). To draw these maps, all the coexistence domains have to be expressed as mathematical functions. Reference [1] presents an example of how each boundary on the corrosion map is calculated, starting from the following equation as the reactive side of the corrosion pattern: Z Y Y X (100− X) ⋅ Na2 O + ⋅ NaF + ⋅ AlF3 + ⋅ SiO2 + ⋅ Al2 O3 ⇒ MNa2O MNaF MAlF3 ⋅ R B MSiO2 MAl2O3 (6) In this equation, X, Y and Z are the number of grams of each chemical, the Mi's are their molecular weights and RB is the NaF/AlF 3 mass ratio of the initial fluorides melt. This equation remains the same for all the domains. For the products side, we have to go systematically through all the existing triangular planes inside the trigonal prism of the coexistence diagram shown in Figure 1. The three compounds delimiting each plane are used as reaction products to complete the reaction. One more equation is needed to solve the system. It is the following: R Na = 2⋅ Z M Na 2 O 2 ⋅Z Y + M Na O M NaF 2 (7) The parameter RNa is defined as the molar fraction of the sodium introduced as oxide, with respect to all sources of sodium (Na2 O and NaF). Many combinations of parameters can be used as axes when drawing corrosion maps. A good choice is the weight percent of corrosive agents, i.e. Na3 AlF 6 + NaF + Na2 O, introduced (Y-axis) as a function of the silica content of the refractory (X- axis). This choice for the Y-axis is particularly useful in an engineering point of view. Since the quantity of corrosive agents under carbon blocks increases as the potlining ages, the Y-axis can be seen as time. For a similar reason, it can also be seen as the vertical position into the refractory lining. Depending on the corrosion conditions involved, the number of domains visible and their shape vary significantly. Two important parameters control the number and the shape of domains: RNa and the bath ratio, RB. The parameter RNa has a much more dramatic effect on the corrosion maps. Figure 2 shows the corrosion maps obtained with RNa = 0 or 1, at RB = 3.2 (in wt.). On such maps, the thin lines represent equilibria involving only cryolite as fluoride, while the thick lines correspond to equilibria involving only villiaumite (NaF) as fluoride. Therefore, when a domain is surrounded by only one type of line, the assemblage contains only one fluoride, either cryolite or villiaumite. Consequently, by varying RNa from 0 to 1, a very wide range of conditions can be covered with the proposed unified corrosion pattern. This shows that the two previously antagonist approaches constitute in fact the limiting cases of a more general corrosion pattern. DRY CELLS VERSUS WET CELLS It was previously shown [1] that the parameter RNa is useful to differentiate between two major types of cells, i.e., wet cells or dry cells, with respect to the unified approach predictions. Wet cells are those in which the potlining refractories are ultimately corroded mostly by the action of molten bath. In such a case, the presence of both cryolite (Na3 AlF 6 ) and villiaumite (NaF) in the corrosion products is predicted. Unlike this case, the potlining refractories in dry cells are mostly corroded by the action of metallic sodium, such that ultimately cryolite is not predicted to coexist inside the corrosion products. The maximum value of RNa leading to the presence of coexisting cryolite in the corrosion products, irrespectively of the corrosive agents content is: RNa** = 3/(2RB + 3) (8) Thus, cells having their RNa value higher or lower than RNa** can be defined as dry or wet cells, respectively. POST-MORTEM ANALYSIS Based on the above-mentioned unified approach, corrosion maps were obtained corrosion agents [wt%] 100 R Na = 0.0 R B = 3.2 80 NAS2 60 NAS 6 40 β-A 20 A A3S 2 S 0 0 20 40 60 80 SiO2 in the refractory [wt%] 100 corrosion agents [wt%] (a) 100 RN a = 1.0 RB = 3.2 80 N 2S N3 S2 60 NS NA 40 NS2 20 β -A A 0 0 NAS2 A3 S2 NAS6 20 40 60 80 SiO2 in the refractory [wt%] S 100 (b) Figure 2: Corrosion maps obtained for (a) RNa = 0 and (b) RNa = 1, at RB = 3.2 (in wt.). from corroded refractories taken from the potlining of two AP-30 cells. Both cells were using the same 130 mm thick insulating refractory material underneath a 130 mm thick protective refractory and a bedding mix. The original characteristics of these materials are given in Table I. Although the protective refractories were different in both cells, they were involving similar compositions, as shown in Table I. The major difference between the two cells was the type of cathode blocks they were using. Cell “A” was lined with amorphous cathode blocks while cell “B” had semi- graphitized cathode blocks. The properties of these blocks are given in table I. Table I: Characteristics of the cells potlining materials (from the manufacturers) Materials Cell Chemical composition [wt. %] SiO2 Al2 O3 Fe2 O3 MgO CaO Na2 O K2 O TiO2 Al2 O3 +TiO2 Bulk density [g/cm3 ] Porosity [%] Amorphous SemiBedding blocks graphitized mix blocks A B A and B Protective refractory Insulation A B A and B 60 - 64 ---------------------29 - 33 1.97 – 2.08 74 8.2 6.5 1.2 5.0 0.4 1.5 0.7 ---0.57 20 - 25 b 76 b ---------------------------1.54 ---------------------------1.63 0.01 99.5 0.01 0.01 0.04 0.39 ------1.15 60 - 65 26 - 32 5-7 0.5 – 1.0 0.5 – 1.0 ------1-2 27 - 34 1.84 – 1.92 16 - 18 a 26 a ---- ---- a : Total porosity b: Apparent porosity The refractory core samples were taken in both cells on the central longitudinal axis. They were then submitted to semi-quantitative XRD and XRF analyses. Cells “A” and “B” were autopsied after 1361 and 1710 days, respectively. Both cells were operating with a bath ratio (RB) of about 1.1 (in wt.). To obtain the corrosion map associated to an analyzed sample, its RNa value has to be determined. To achieve this, the mixture of oxides in the corrosion products has to be considered as a single non- stoechiometric compound. In such a case, the products side of reaction (5) becomes: ⇒ A A ⋅ NaF + ⋅ AlF 3 + B ⋅ NaAl a Si b O (3 a + 4 b +1 ) 2 M NaF M AlF3 ⋅ R F (9) In equation (9), a and b are respectively the Al/Na and Si/Na molar ratios of the oxides found in the analyzed sample. The parameter RF is the NaF/AlF 3 mass ratio of the fluorides also found in the corroded sample. Using the same technique than the one used to build the corrosion maps, it is possible to obtain a mathematical function relating the chemical composition of the corroded sample (X*) with the corrosion conditions. This function is the following: X* = where: 100 ⋅ (2 b ⋅ H) (10) M Al O 2 3 ⋅ { a ⋅ H + ( R − R ) } + 2 b ⋅ H B F M SiO2 R R H = B Na ⋅ (2 R F + 3) + 3 ⋅ (R B − R F ) (1 − R Na ) (11) Reorganizing equation (10) allows to obtain the sodium ratio, R* Na, necessary so that X* = X°, where X° is the uncorroded refractory composition. The expression for R* Na is the following: (12) * M Al2 O3 * ⋅ (3a + 1) − 100 − X ⋅ 6b (R B − R F ) ⋅ X ⋅ M SiO2 * R Na = * M Al2O3 * X ⋅ ⋅ [ ( R − R ) − a ⋅ R ⋅ ( 2 R + 3 ) ] + 100 − X ⋅ 2 b ⋅ R ⋅ ( 2 R + 3 ) B F F B F B M SiO 2 ( ( ) ) Equation (12) was used to calculate the RNa* value of each corroded sample. This allowed to build a corrosion map specific to each sample. Then, each sample was located on its corresponding map. The initial (uncorroded) composition of a sample (Xo ) was taken as its X-coordinate, while its Y-coordinate, identified as Y*, was calculated using the following equation: Y* = where: Z= [Z + Y ⋅ (1 +1 RB )] [Z + Y ⋅ (1 +1 RB ) +100] M Na 2 O 2 ⋅ M NaF ⋅ R Na ⋅Y (1 − R Na ) (13) (14) and: MAlF3 ⋅ R B ⋅ X* ⋅ (2 ⋅ R F + 3) MSiO ⋅ b 2 Y= * 2 ⋅ R B ⋅ R Na ⋅ (2 ⋅ R F + 3) + 6 ⋅ (R B − R F ) * (1− R Na ) (15) RESULTS AND DISCUSSION The designation and location of the analyzed samples taken from the two autopsied cells are given in Table II. Their XRD and XRF analyses are presented in Tables II and III, respectively. The corrosion maps of the samples taken from cells "A" and "B" are shown on Figure 3 and 4, respectively. As shown in Table II, there is a good correlation between the predicted minerals in the corroded samples and those identified by XRD. However, in some samples, the presence of compounds such as Na-Al-Si-O, Na-Ca-Al-Si-O and Na-K-Al-Si-O were detected by XRD instead of pure Nepheline (NaAlSiO 4 ). As expected, nor Albite (NaAlSi3 O8 ) or sodium disilicate (NaSi2 O5 ) were detected by XRD since these minerals are most often present as amorphous phases in corroded refractories. An important result concerns the values of RNa and RF obtained from the corroded samples of both cells. As shown in Table II, the RNa value for cell “A” lies between 0.85 and 0.90, while the RF value is infinite in all cases. This suggests that cell "A" can be classified as a dry cell. Since RB for both cells is 1.1 (in wt.), their corresponding RNa** value is 0.58 (see equation (8)). The latter is in fact much lower than the three above RNa values. Unlike in cell “A”, the RNa values obtained for samples B-1 and B-2 (0.25 and 0.30, respectively) in cell B are much lower than RNa** . The RF values for these two samples are also much lower than in cell “A”, i.e., 1.9 and 1.5 (in wt.), respectively. These results are typical of a wet cell which alloys more melt percolation through the cathode blocks during operation. As shown in Table I, the porosity of the semi- graphitized blocks used in cell “B” is about 1.5 times higher than that of the amorphous blocks used in cell “A”. An interesting point is that sample B-3 (cell “B”) shows RNa and RF values more typical of a dry cell, i.e., 0.89 and 213, respectively. Since the sample B-3 was taken closer to the cathode blocks than samples B-1 and B-2, this result suggests that even a wet cell first behave like a dry cell during early operation. This would be consistent with the fact that after a cell start-up, the molten bath can percolate through the cathode blocks only after the latter have been infiltrated by sodium [4]. Thus the refractories underneath these blocks should theoretically first be corroded by the action of Na2 O before the percolating melt reaches them. Table II: Comparison between the unified approach and the observations. Cell Sample A A-1 A A B B A-2 A-3 B-1 B-2 Position a [mm] 114 244 293 114 141 RF RNa* Prediction Analysis, XRD b ∞ 0.90 Na2 Si2 O5 + Albite Nepheline Glass ∞ ∞ 1.9 1.5 0.88 0.85 0.25 0.30 NaF Na-Al-Si-O (56) + Nepheline (25) Others Glass Nepheline (42) + Na-Al-Si-O (2) NaF (39) Na2 SiO3 ---- Na2 SiO3 (13) Others Na2 Si2 O5 Nepheline Glass Nepheline (49) + Na-Al-Si-O (4) Na2 SiO3 Na2 SiO3 (14) NaF Albite Nepheline NaF (34) Glass Na-Ca-Al-Si-O (13) Cryolite Cryolite (15) NaF ------------Albite Nepheline NaF (14) Chiolite (27) Fluorite (12) Cristobalite (11) Others Glass Na-Ca-Al-Si-O (6) + Na-K-Al-Si-O (70) Cryolite (6) NaF Na2 Si2 O5 Nepheline Cryolite B B-3 228 213 0.89 Corundum ------Albite + Na2 Si2 O5 Nepheline NaF ---- Corundum (4) Cristobalite (5) Others Glass Na-K-Al-Si-O (41) NaF (35) Others a: Position of the sample. The distances are given with respect to the bottom of the pot shell. b: Values between parentheses are weight percentages of the crystalline phases estimated using a semi-quantitative method. Table III: Chemical composition of the corroded samples (XRF). Chemical composition [wt. %] SiO2 Al2 O3 Na2 O TiO2 CaO Fe2 O3 F A-1 A-2 A-3 B-1 B-2 B-3 36.6 5.9 19.5 0.4 3.1 33 1.0 43.9 18.4 30.9 1.0 1.4 2.7 6.6 41.1 22.8 29.5 0.6 1.6 2.3 4.3 16.5 24.7 27.3 0.3 4.7 1.1 31.4 35.6 27.1 19.9 0.5 2.6 3.1 12.2 43.8 24.0 26.3 1.1 1.8 1.2 4.2 Based on the above overall results, potlining refractories in a dry cell tend to be converted into a mixture of Nepheline and sodium silicate compounds (Na2 SiO 3 and Na2 Si2 O5 ), which coexist with sodium fluoride (see Table II and Figure 3, samples A-1 to A-3). In a wet cell, the corrosion of the refractories tends to lead principally to the formation of Nepheline and Albite, coexisting with cryolite in presence or not of sodium fluoride (see Table II and Figure 4, samples B-1 and B2). For a wet cell, the possible transition from an initial dry to a final wet operating conditions, on service, may lead to a mixture of the two above phase systems, as shown in Table II and Figure 4 for sample B-3 (presence of Nepheline, Albite, Na2 Si2 O5 and NaF). Based on the ternary Na2 O-Al2 O3-SiO 2 phase diagram, the above phase system comprising Nepheline, Na2 SiO 3 and Na2 Si2 O5 , which has been associated to the corroded refractories from cell “A” (dry cell), should involve low melting point eutectic compounds. In such condition, refractory bricks with high alumina:silica ratio are required [5]. The purpose of such bricks is to prevent the formation of an oxide melt during the operation of the cell. However, in wet cells (such as in cell B), the presence of a melt in the potlining refractories cannot be prevented since an important amount of molten bath percolates through the cathode blocks during operation. In such a case, decreasing the refractories alumina:silica ratio should prevent the penetration of the fluoride melt since it favors the Albite formation which increases the melt viscosity [2,6]. In intermediate cell operating conditions where RNa is closed to RNa** , the used of successive protective layers of low and high alumina refractories maybe required to prevent both fluoride melt penetration and oxide melt formation [7]. corrosion agents [wt%] 100 RN a = 0.90 RB = 1.1 RF= inf. 80 corrosion agents [wt%] N2 S N3 S2 60 NS NA NS2 40 20 β-A NAS2 A3S2 A NAS6 0 0 100 20 40 R Na = 0.88 R B = 1.1 R F= inf. 80 60 80 S 100 A-2 N2 S N3 S2 60 NS NA NS 2 40 β-A 20 NAS2 A3S2 A NAS6 0 0 corrosion agents [wt%] A-1 100 20 40 RN a = 0.85 RB = 1.1 RF = inf. 80 60 80 100 A-3 N2 S N 3S2 NS NA 60 S NS2 40 β-A 20 A 0 0 NAS2 A3S2 NAS6 20 40 60 80 SiO2 in the refractory [wt%] S 100 Figure 3: Computed corrosion maps of samples A-1, A-2 and A-3 from cell “A”. corrosion agents [wt%] 100 RNa = 0.25 RB = 1.1 RF = 1.9 80 B-1 NAS2 NAS6 60 β-A 40 20 A3S2 A S 0 corrosion agents [wt%] 0 20 100 R Na = 0.30 R B = 1.1 R F= 1.5 80 60 80 100 B-2 NAS 2 60 NAS6 β-A 40 20 A3S 2 A S 0 0 corrosion agents [wt%] 40 100 20 40 RNa = 0.89 RB = 1.1 RF= 213 80 60 80 100 B-3 N 2S N3 S2 60 NS NA NS 2 40 β-A 20 A 0 0 NAS2 A3S2 NAS6 20 40 60 80 SiO 2 in the refractory [wt%] S 100 Figure 4: Computed corrosion maps of samples B-1, B-2 and B-3 from cell “B”. CONCLUSIONS This work shows that the aluminum reduction cell lining design has a great influence on the type of corrosion imposed to the refractories during service. Cells using semi- graphitized cathode blocks were shown to be more prone to the attack of their refractory lining by the action of the molten bath. This is the case of wet cells, which should benefit of the use of low alumina refractories (A/S ≅ 0.33) to prevent the penetration of such fluoride melt during operation. Unlike this case, the use of amorphous blocks in such cells was shown to promote the corrosion of their potlining refractories mostly by the action of the metallic sodium. This is the case of dry cell, which should benefit of the use of high alumina refractories (A/S ≅ 0.85) to prevent the formation of an oxide melt on service. ACKNOWLEDGEMENTS The authors are very grateful to Norsk-Hydro and Alcoa for their financial support during the realization of this work. Special acknowledgements are addressed by the authors to Dr. O.-J. Siljan from Norsk-Hydro, and to Dr. A. Tabereaux and Mr. P. Clery from Alcoa, for their technical support. REFERENCES 1. R. Pelletier, C. Allaire, O.-J. Siljan and A. Tabereaux, “Corrosion of Potlining Refractories: A unified Approach”, JOM , manuscript No. 00-121, to be published in August 2001. 2. C. Schøning, T. Grande and O.-J. Siljan, "Cahode Refractory Materials for Aluminium Reduction Cells", Light Metals 1999, 231-238. 3. J. Rutlin and T. Grande, "Phase Equilibria in Subsystems of the Quaternary Reciprocal System Na2 O-SiO 2 -Al2 O3 -NaF-SiF 4 -AlF 3 ", J. Am. Ceram. Soc., Vol. 82 N° 9, 2538-44, 1999. 4. E.W. Dewing, “The Reaction of Sodium with Non Graphitic Carbon: Reactions Occuring in the Lining of Alumina Reduction Cells”, Transaction of the Metallurgical Society of AIME, 227, p. 1328, 1963 5. C. Allaire, "Refractory lining for alumina electrolytic cells", J. Am. Ceram. Soc., 75, (8), 2308-11, 1992. 6. T. Grande and J Rutlin, “Viscosity of Oxyfluoride Melts Relevant to the Deterioration of Refractory Linings in Aluminium Reduction Cells”, Light Metals 1999, 431-436. 7. F. Brunk, "Corrosion and Behaviour of Fireclay Bricks of Varying Chemical Composition Used in the Bottom Lining of Reduction Cells", Light Metals 1994, 477-482.