o > 
 
 — .^o^'X;-"^?^-;/^ V^^V "v"^^*\/^' 'V-'^'^V' %''^^'li«- 
 
 0^ ^o^ *.V.V <\ 
 
 
 
 >" v' 
 
 
 4 o 
 
 * -i^ ■^ ♦ (Cu 88 A "<?P "^^ * ' 
 
 
 ^.^^•"'J- 
 
 •v ^ 
 
 
 -.'?^^''^<. 
 'h ^ 
 
 .V 
 
 
 ^^-n^ 
 
 
 •1 o 
 
 
 % ^^ ^ ' /^V/h!'' "^^^^ <^ »*f!Si^'. '^t^. A-^ ■ ♦: 
 
 
 .5.'>. 
 
 .^'% 
 
 
 
 
/^SttV. 
 
IC ^^^ 
 
 Bureau of Mines Information Circular/1981 
 
 High-Temperature Corrosion 
 Resistance of Ceramic Materials 
 
 By T. A. Clancy 
 
 ^^^ 
 
 UNITED STATES DEPARTMENT OF THE INTERIOR 
 
uJ 
 
 Information Circular 8843 
 
 High-Temperature Corrosion 
 Resistance of Ceramic Materials 
 
 By T. A. Clancy 
 
 UNITED STATES DEPARTMENT OF THE INTERIOR 
 James G. Watt, Secretary 
 
 BUREAU OF MINES 
 
 Research at the Tuscaloosa Research Center is carried out under a memorandum of agreement between the 
 Bureau of f^ines, U.S. Department of the Interior, and the tJniversity of Alabama. 
 
HIGH-TEMPERATURE CORROSION RESISTANCE OF CERAMIC MATERIALS 
 
 by 
 
 T. A. Clancy ' 
 
 ABSTRACT 
 
 Since many new mineral processing systems under development can Impose 
 severe corrosive conditions on construction materials, the Bureau of Mines 
 reviewed and assessed the literature on corrosion resistance of ceramic 
 materials. Corrosive conditions were categorized as: (1) acids and aqueous 
 solutions, (2) glass and slag/metal melts, and (3) gases. Information on the 
 corrosion resistance of ceramic materials, generally. Is disseminated by many 
 sources, each of which Is Interested In specific corrosive environments. 
 Based on this survey. It was noted that (1) the development of standardized 
 tests Is needed to study solution-type corrosion of ceramics, (2) slag/melt 
 corrosion data are the most comprehensive of the available data, and (3) bet- 
 ter definition of corrosion mechanisms in gaseous environments is needed. 
 
 INTRODUCTION 
 
 Part of the Bureau of Mines Mineral Resources Technology research program 
 is aimed at solving ceramic materials problems associated with mineral pro- 
 cesses that include corrosive environments. Operation at high temperature 
 in corrosive environments is often required in order to achieve energy effi- 
 ciency. In most cases research is done in small laboratory equipment that can 
 be discarded after one or more experiments. However, when the scale of the 
 research is Increased to provide reliable engineering and economic data. 
 Inadequate materials performance can become a major problem. Materials are 
 generally selected based on data for only one set of operating parameters or 
 without adequate knowledge of material behavior under actual operating condi- 
 tions. Changing parameters, as is often done when process scale-up is accom- 
 plished, results in new conditions where no data are available and material 
 Integrity is unknown. The Bureau's efforts in developing new extractive tech- 
 nology has shown that various ceramic materials are candidates for service in 
 high-temperature corrosive environments. Ceramic materials often have higher 
 corrosion resistance to certain chemicals than metallic compounds and have the 
 possibility of replacing alloys containing critical elements such as chromium. 
 
 ^Ceramic engineer, Tuscaloosa Research Center, Bureau of Mines, Tuscaloosa, 
 Ala. 
 
nickel, or cobalt. It is advisable, therefore that data regarding the corro- 
 sion behavior of ceramic materials be available. 
 
 As part of its ceramics research, the Bureau's Tuscaloosa Research Center 
 has reviewed literature sources to determine what information is available on 
 corrosion of ceramic materials. Much of the information found relates to the 
 use of refractory ceramics in more traditional areas such as steelmaking, and 
 petrochemicals. This literature review and technical assessment includes 
 information developed over many years in these traditional areas, as well as 
 current information being generated in the development of new minerals pro- 
 cessing technology. This report summarizes the findings of a review of 
 58 pertinent references, with 44 being cited in the text and the remainder 
 included for supplemental information, 
 
 BACKGROUND 
 
 Van Vlack (42)^ defines corrosion as the deterioration and loss of 
 material due to chemical attack. The simplest form of corrosion is by means 
 of chemical solution. Generalizations regarding chemical solution are — 
 
 1. Small molecular and ions dissolve most readily. 
 
 2. Solution occurs more readily when solvent and solute are structurally 
 similar. 
 
 one. 
 
 3. The presence of two solutes may produce greater solubility than only 
 
 The most common type of corrosion involves the electrochemical process of 
 metal oxidation. All metals differ in their tendency to release electrons and 
 form ions. The production of ions and electrons when a metal goes into solu- 
 tion builds up an electrode potential that depends on the nature of the metal 
 and the nature of the solution. The particular potential is measured by 
 determining the voltage difference between the metal and a standard hydrogen 
 electrode. Measurements have been made for a whole series of metals, commonly 
 called the electromotive series. 
 
 Electrochemical corrosion is less important in ceramic systems, because 
 the complex nature of ceramic systems makes it difficult to define the condi- 
 tions of electrode potentials and because oxide-type materials are also more 
 thermodynamically stable. Only in the case of glass tank refractories has 
 electrochemical corrosion been widely reported for ceramic systems (24). 
 
 Information on corrosion of ceramic materials is disseminated over 
 sources that are principally interested in specific corrosive environments or 
 operating conditions. Very few sources are available that compare corrosion 
 resistances of ceramic materials. Campbell (6^) compares resistances of 17 
 commercial refractories. Compared with most metals and alloys, ceramic 
 
 ^Underlined numbers in parentheses refer to the list of references at the end 
 of this report. 
 
materials are very corrosion-resistant, particularly when compared at high 
 temperatures. This is true largely because oxidized, or combined, state of 
 most metals, except the precious metals, is the thermodynamically stable 
 state. It is possible to determine from the relative free energies of a metal 
 and its oxide (or a material and its corroded state) whether or not the corro- 
 sion reaction is to proceed. A classic comparison is that both iron and alu- 
 minum will oxidize, but aluminum forms an adherent oxide layer that greatly 
 impedes further reaction. 
 
 Standardized corrosion testing of ceramic materials, other than slag 
 tests described later, has been established by the American Society for 
 Testing and Materials (ASTM) for only three types of ceramic materials: 
 
 1. ASTM C225-54, Resistance of Glass Containers to Chemical Attack, 
 
 2. ASTM C279-54, Chemical-Resistant Masonry Units. 
 
 3. ASTM C301-78C, Vitrified Clay Pipe. 
 
 The test for glass containers involves placing bottles containing: (a) 
 acid or (b) water in an autoclave at 121° C. The amount of attack is reported 
 as the amount of: (a) acid consumed or (b) acid required to titrate the water 
 solution. In cases where bottles are too small for testing, pieces of glass 
 are autoclaved in water. The tests for masonry (fired brick) and clay pipe 
 involve placing pieces of the material in acid (boiling temperature for the 
 bricks) for 48 hours. The amount of corrosion attack is expressed as a per- 
 centage of weight loss by the sample. 
 
 Corrosion of ceramics has generally been categorized as to the corrosive 
 environment of concern. Therefore, it is easiest to list three broad areas of 
 corrosion as — 
 
 1. Aqueous solutions. 
 
 2. Glass and slag melts. 
 
 3. Gas atmospheres. 
 
 However, there is often an overlapping of environments, as in coal gasi- 
 fiers and steelmaking processes that involve both slag/metal and gaseous con- 
 ditions. The effects of these three types of corrosion on ceramic materials 
 will be reviewed separately. 
 
 CORROSION BY AQUEOUS SOLUTIONS 
 
 Probably the simplest form of ceramic corrosion to understand is that by 
 acids or aqueous solutions. The bulk of such corrosion work has been done 
 with glass and enamels and, to a lesser extent, with acid-resistant brick or 
 tile. Morey (27) has reviewed the basic nature of chemical attack of glass by 
 water, acids, and alkaline solutions. He states that while water and acids 
 attack only constituents other than silica, alkaline solutions and 
 
hydrofluoric acid attack the silica. Water and other acids extract alkali 
 oxides leaving a hydrated porous layer at the surface, which is mostly silica 
 and through which further reactants must pass. Therefore, the rate of attack 
 in water and all acids except HF will decrease with time. Alkaline solutions, 
 in addition to attacking modifying oxides, also dissolve the high-silica 
 porous layer. The depths to which this layer can extend depends on the rela- 
 tive solution rates of silica and the other oxides in the alkaline solution. 
 When this condition becomes stabilized, the rate of attack tends to become 
 constant. As in the case of other chemical reactions, the rate of attack 
 Increases rapidly with temperatures. In general, the increase is of the order 
 of 2.5 times for each increase of 10° C. 
 
 Tests for glass corrosion have generally been of two types. In one case, 
 glass is immersed in water or a corrosive solution, and the corrosion rate is 
 measured in terms of the alkali constituents removed from the glass in a given 
 time. In the other case, a piece of glass is exposed to a corrosive reagent 
 for a definite time. The rate of reaction is measured in terms of the weight 
 lost per unit of area exposed. 
 
 Because many commercial applications employ enameled metals for contain- 
 ers or similar shapes, the corrosion resistance of enamels is important. 
 While glass shapes can be formed of high-silica (>90 pet Si02) and Pyrex-type^ 
 glasses that are resistant to water and acids, enamels contain more modifying 
 oxides to permit fusing on metal surfaces. Cook and Andrews (8^) evaluated the 
 corrosion resistance of eight varied enamels to different concentrations of 
 inorganic acids, alkalies, organic acids, and salt solutions at both room and 
 boiling temperatures. All the enamels were resistant at room temperature to 
 alkali and salt solutions. Acid-proof enamels were resistant to inorganic and 
 organic acids at room or boiling temperatures. All enamels showed low 
 resistance to boiling alkali solutions, and some were marked by boiling salt 
 solutions. 
 
 Priest (31) determined the corrosion resistance of glassed steel, which 
 is, (1) applied in greater thicknesses than enamels, (2) more resistant to 
 mechanical damage at thicknesses of 40 to 80 mils, and (3) fired at higher 
 temperatures than enamels. The glassed steel's corrosion resistance to hot 
 acids, hot water, steam, or alkali solutions at moderate temperatures was an 
 order of magnitude greater than that of enamels. He concluded that all acids 
 except HF and H3PO4 can be handled in glassed steel up to 150° C and often 
 higher. Resistance to alkalies is lower, but solutions to pH of 12 at 100° C 
 can be handled. Concentrated alkali may be used in glassed steel below 40° C. 
 Corrosion resistance to pure water is limited at temperatures above 95° C. 
 
 Nordberg (28) has rated a number of commercial glasses (Corning Glass 
 types) as to resistance to water acid solutions, and alkaline solutions. 
 These ratings are shown in table 1. 
 
 ■'Reference to trade names is made for identification only and does not imply 
 endorsement by the Bureau of Mines. 
 
TABLE 1. - Resistance of glasses to attack by acids and alkali 
 
 
 Weight loss, mg/cm^ 
 
 Designation 
 
 24 hr, 5 pet HCl, 
 100° C 
 
 6 hr, 5 pet NaOH, 
 100* C 
 
 96 Dercent silica. •••••••..••• 
 
 0.0004 
 .005 
 .01 
 .02 
 Leached 
 Disintegrated 
 .35 
 .01 
 
 0.9 
 
 Boroslllcate— chemical. ........ 
 
 1.4 
 
 Soda llnie~bulb. ............... 
 
 1.1 
 
 Lead elass— electrical. ........ 
 
 1.6 
 
 Boroslllcate-tungsten sealing. 
 Hleh lead 
 
 3.9 
 3.6 
 
 Aluminosilicate. .••••..••••••• 
 
 .35 
 
 Alkall-reslstant 
 
 .09 
 
 Rauch (32) evaluated the resistance of glass-ceramic materials, consid- 
 ered for use In gas turbine heat exchangers, to both sulfuric acid and sodium. 
 Acid resistance was evaluated by Immersing samples In concentrated H2SO4 at 
 300" C for 2 hours, followed by heat exposure for 2 hours at 320* C, relm- 
 merslon In hot acid, heat exposure for 2 hours at 650* C, relmmerslon In hot 
 acid, and heat exposure for 2 hours at 1,000* C. Therefore, each sample 
 received a total Immersion time of 6 hours In hot acid. Sodium resistance was 
 evaluated by placing samples In a 100 cm-' platinum crucible, covering the sam- 
 ple with dry sea salt powder, and placing the crucible In a furnace heated to 
 750* C. As the salt melted, more was added to maintain the liquid level above 
 the sample during the 2 hour test. The percentage of weight change was deter- 
 mined for each material after these exposures. The acid corrosion resulted In 
 weight losses for the glass-ceramic materials to various degrees, with a cor- 
 dlerlte base type showing the least loss. The salt corrosion resulted In 
 weight gains In all the materials except one. 
 
 Honeycomb shapes of the heat exchanger candidate materials were exposed 
 to 1.0 percent H2SO4 followed by three heat cycles at 300* C and were then 
 Immersed In boiling NaCl solution followed by heat soaking at 800* C for 3 
 weeks. Since the heat exchanger requires low expansion materials, change In 
 length of the shapes was determined after the acld-heatlng cycles and during 
 the boiling NaCl solution exposure. These results Indicated that a glass- 
 ceramic containing a mixture of petallte and cordlerlte materials retained the 
 lowest thermal expansion. 
 
 Cooke (9^) determined the chemical durability of eight lead boroslllcate 
 glasses containing 60 or 65 mole-percent SIO2. The compositions were chosen 
 so as to Include glasses lying both Inside and outside the region In which 
 llquld-llquld Immlsclblllty occurs. Durability measurements were carried out 
 on granular samples of both heat-treated and quenched glasses using solutions 
 of various pH values for 4 hours at 70° C. 
 
 It was found that, besides chemical composition, micros tructure had an 
 effect on durability. In phase-separated glasses, the durability Is deter- 
 mined by the location of the less durable phase. If It Is contained as 
 droplets within a stable place, durability will be high. In other cases, 
 durability Is low. The addition of LI2O to some glasses eliminated phase 
 
separation and increased durability. Data are also presented showing that 
 extraction of Si02 increases as pH increases, particularly around pH 12, and 
 that lead and silica extraction increase greatly as the B2O3 content of the 
 glass increases. 
 
 The thermodynamic stability and effect of pH of the solution on durabil- 
 ity of glasses are discussed. The very limited solubility of silica in water 
 is one of the main factors in the corrosion resistance of silicate glasses. 
 Silica reacts with water to form H2Si03, and the thermodynamic stability of 
 Si02 in water at 25° C is characterized by an activity versus pH plot. This 
 curve can be divided into three sections: one at pH 7 to 10, where solution 
 of H2Si03 occurs; one at pH 10 to 12, where solution is due to formation of 
 HSi03; and a third at pH 12 or over, where solution is by formation of 
 3103-2. 
 
 The effect of pH on solution of lithium silicate glasses is then illus- 
 trated. Unlike pure silica, in the case of lithium silicate glass, the activ- 
 ity of H2Si03 (glass) or the extent of ion exchange between the silicate and 
 water will depend on the activity of lithium ions in the solution. If lithium 
 ions are allowed to build up in the solution, almost a complete removal of 
 lithium silicate glass is possible even up to pH 14. Since the H2Si03 (glass) 
 becomes ionized and will dissolve above pH 9 or 10, lithium silicate glasses 
 are not durable in water. This is also the case with other alkali (potassium, 
 sodium) silicate glasses. 
 
 Similar aqueous reactions are presented for PbO and B2O3 in aqueous solu- 
 tions. Stability versus pH for PbO shows lead dissolves as Pb''"^ and PbOH in 
 the acidic range, and the Pb''"^ has much higher activity. While in the alka- 
 line range lead forms HPb02, and the activity of HPb02 is greater than that of 
 Pb (OH) 2 only at pH above 14. Thus PbO in a glass will increase alkaline 
 durability but decrease acid durability. Boron trioxide forms orthoboric acid 
 in aqueous solutions, which then ionizes in various ways depending on solution 
 concentration and pH. Stability diagrams for B2O3 indicate that the stability 
 is much less than that for Si02. Normal substitution of B2O3 for Si02 in 
 single-phase oxide glasses will cause the durability in water to deteriorate. 
 However, glasses containing boron can be produced having good durability by 
 controlling the phase-separated microstructures, as in the case of Pyrex-type 
 glasses. 
 
 While the general behavior of common glasses in corrosive solution envi- 
 ronments has been investigated fairly thoroughly, this is not true for other 
 ceramic products even though many are used in chemical construction applica- 
 tions. Very limited quantitative data on the behavior of such materials under 
 corrosive solution conditions are available. Robbins (33) has reviewed the 
 use of concretes, cements, acid-proof brick, porcelain, and stoneware under 
 corrosive solution conditions. All these materials are generally more resis- 
 tant to acid solutions than to basic ones. 
 
 Portland cement concretes are readily attacked by weak acids and even by 
 many organic compounds. Alumina concretes are more chemically resistant than 
 
Portland cement concretes and will resist solutions at a pH above 4 or 5. 
 Silicate cements are resistant to many acids, but not to alkaline conditions. 
 
 Porcelain and stoneware are vety similar in chemical resistance proper- 
 ties. Porcelain is usually made from higher purity material and can usually 
 withstand more severe conditions. These products withstand acids very well, 
 with the exception of hydrofluoric acid. At high temperatures, concentrated 
 phosphoric and sulfuric acids attack porcelain or stoneware. Steam will also 
 cause degradation of these materials. Robbins ( 33 ) presents a table comparing 
 the resistance of porcelain and stoneware to 14 different chemicals, but pro- 
 vides no information on the effect of increased temperatures. 
 
 Acid-proof brick or tile are characterized by high silica contents and 
 low porosities. They will resist most acids, except hydrofluoric, and are 
 reasonably resistant to bases. While the brick are impervious to solutions, 
 the mortars necessary to join them together are not and can provide a path for 
 penetration. 
 
 A review of literature on corrosion by solutions has indicated that 
 nearly all the data available are limited to glass materials. The corrosion 
 data available on other industrial ceramic products are very limited and pri- 
 marily based upon supplier information which is often not oriented towards the 
 specific conditions of interest. This lack of definitive data may partly be a 
 result of the wide variation in industrial products and applications involved. 
 Therefore, in order to properly compare the chemical resistance of these 
 ceramic materials, standardized descriptions of material classes are required. 
 Also, standardized testing, similar to that used for glass materials, is 
 required. 
 
 New mineral processing techniques being investigated such as hot acid or 
 base leaching, chloride leaching, and fused salt baths may require the use of 
 ceramic materials that have superior corrosion resistance. Therefore, engi- 
 neering data on the corrosion resistance of candidate materials to such corro- 
 dents are necessary. These data could be collected in a manner that would 
 provide rates of attack, as is done with corrosion attack data for metals. 
 
 CORROSION BY GLASS AND SLAG MELTS 
 
 The majority of work on corrosion of ceramic materials has been directed 
 toward corrosion by melts of glasses, slags, and metals. The basic principles 
 of dissolution of ceramic materials by such melts have been covered by Cooper 
 and Kingery (10). They have shown that the rate of dissolution can be con- 
 trolled by three types of reaction mechanisms. For stationary specimens, the 
 rate of dissolution can be controlled by molecular diffusion; that 
 is, diffusion in the absence of fluid flow. This mechanism is common for 
 ceramic systems, and the rate is dependent on diffusion of solute through a 
 stagnant boundary layer at the specimen surface. Also for a stationary spec- 
 imen, dissolution by free convection can be the controlling mechanism. This 
 convection is dependent on the rate of bringing fresh slag to the specimen. 
 In the case of forced convection, the dissolution is controlled by the rate of 
 transport in the solution and depends on the velocity of the forced 
 
convection. In the first case, the dissolution rate is proportional to the 
 square root of time, while in the other cases the rate is directly propor- 
 tional to time. Cooper and Kingery found that for dissolution of sapphire in 
 CaO-Al 203-5102 slag, the corrosion rate was dependent on transport in the 
 melt. 
 
 McGee (26) has reviewed slag tests for refractories and discusses three 
 types: (1) reaction test, (2) static slag test, and (3) dynamic slag test. 
 The first consists merely of heating a slag in contact with a refractory for a 
 given time and observing the amount of reaction. This test is good for cer- 
 tain applications, particularly as a screening method. The second type is 
 usually a cup or a pencil test. A cup test involves drilling a hole in a 
 brick, filling it with slag, heating it for a given time, sectioning the brick 
 after cooling, and observing the reaction. Pencil tests are conducted by 
 inserting a long, narrow specimen into a slag contained in an inert crucible. 
 ASTM procedure C-621 describes a method for glass tank refractories. Some 
 disadvantages do exist with static tests. If the quantity of slag is small, 
 its composition may change during testing. In the case of small crucibles, 
 convection current effects can be overlooked. Dynamic slag tests are intended 
 to reproduce conditions of an actual process. Typical dynamic slag tests are 
 ASTM procedures C-768 and C-622. The first is a drip slag test in which a 
 known weight of a slag is dripped onto a refractory specimen. In the second, 
 a refractory tank of the test specimens containing a known glass composition 
 is heated for 96 hours at a glass-melting temperature of 1,230° to 1,565° C, 
 then drained of its contents; the depth and volume of corrosion are then mea- 
 sured at the metal line. Such tests are useful when they simulate actual cor- 
 rosion. In the case of the drip slag test, the velocity of the slag is not 
 well defined, and the results may or may not be in the appropriate regime of 
 mass transport. It is obviously difficult to simulate intermittent processes 
 that have different slag attack mechanisms at different parts of the process. 
 McGee cautions that deliberately making a slag test too severe by producing 
 results in a shorter time may lead to the wrong results. He concluded that 
 it is usually better to increase the flow rate of the slag than to alter the 
 composition or increase its temperature. 
 
 The Bureau of Mines' Tuscaloosa Research Center has conducted various 
 dynamic slag corrosion studies using a rotary chamber lined with refractories. 
 Slag melts have included glass wool, steel slag, and coal ash-slag composi- 
 tions. The refractories are exposed to flowing slag at a desired temperature 
 for periods of 5 to 8 hours. The chamber is heated with an oxygen-propane 
 torch and slag material is fed continuously as the chamber rotates horizon- 
 tally. Eroded area for each refractory is determined at the conclusion of the 
 run. An application of this test is described by Pahlman (29). 
 
 One of the most severe corrosion environments exists in the basic oxygen 
 furnace (BOF) steelmaking process. Kappmeyer and Hubble ( 17 ) have presented a 
 review of the slag corrosion effects in this environment. Refractories used 
 in this process are exposed to temperatures up to 1,650° C in contact with 
 liquid steel and fluxing slags — chiefly lime, fluorspar, and mill scale — plus 
 an atmosphere created by the reaction of carbon from the metal with oxygen 
 injected through a lance. They present data indicating that the CO/CO2 ratio 
 
varies through the vessel and the basicity (CaO/Si02 ratio) varies as the tem- 
 perature increases during the heat. The refractories predominantly used in 
 the BOF vessel are carbon-containing MgO-based types, either pitch-bonded or 
 fired, pitch-impregnated ones. Slag erosion tests are conducted to compare 
 these refractories and to determine what refractory properties are most impor- 
 tant. Standard bricks compose the sidewalls of a laboratory furnace in which 
 oxygen is blown onto molten pig iron containing additions of slag formers. 
 Following the heating, the degree of erosion is measured and a slag erosion 
 value is determined. Testing has shown that low erosion levels for specimens 
 are consistent with higher MgO contents, higher hot strengths, higher bulk 
 densities, and higher residual carbon contents. It is generally believed that 
 the carbon inhibits deep slag penetration and thus lowers erosion levels. 
 Thus, a careful balance must be drawn to obtain the best slag erosion resist- 
 ance through maximum brick density and, at the same time, the maximum amount 
 of residual carbon within the minor pore volume. 
 
 One of the most complete discussions on corrosion of glass tank refrac- 
 tories is by Litvakovskii (24). Besides considering the standard effects of 
 diffusion and dissolution of refractories, he discusses the effects of elec- 
 trochemical corrosion. With regard to chemical reactions, it is noted that 
 dissolution may occur with or without preliminary chemical reactions occurring 
 within the refractory. Generally, the corrosion rate is not limited by a 
 chemical reaction, but by the rate of diffusion between the reactants. One 
 factor greatly affecting this diffusion rate is the thickness of the melt con- 
 tact layer with the refractory. The thicker this contact layer, the slower is 
 the corrosion. The rate of formation and duration of maintenance of the con- 
 tact layer are the most important factors for the corrosion resistance of 
 materials. This is analogous to situations with metals where strength and 
 durability of the oxide film on the metal are the most important conditions of 
 resistance to corrosion. 
 
 Litvakovskii mentions several cases that substantiate the electrochemical 
 effect on corrosion. Investigations have shown that a certain difference of 
 potentials exists at high temperatures between the glass mass and the refrac- 
 tory oxides. These potentials are due to the contact between electrolytes of 
 different concentration and different temperatures, as well as to the differ- 
 ent oxidizing and reducing characteristics of the contacting substances. 
 These potentials can establish anodic and cathodic conditions between a 
 refractory and a glass melt and, as in metals, if the refractory is anodic, it 
 is the material that will corrode. This electrochemical effect is used to 
 explain the case of melting sheet glass in which the refractory is cathodic 
 and only dissolves at the melt line where anodic portions can form. Con- 
 versely, in melting borosilicate glass, the whole refractory surface is anodic 
 and dissolves intensively, but more uniformly. Data are given for electrode 
 potentials based on measurements made using a Pt-oxide-sheet glass-Pt cell at 
 1,100° C. A curve is given showing that, with an increase in potential dif- 
 ference between glass and refractory, corrosion is intensified. The conclu- 
 sion is made that a refractory and glass should be matched so that the 
 potential difference between them should be positive and have a minimum value. 
 
10 
 
 Kennedy and Peoppel (18) exposed about 50 commercial refractories to a 
 variety of acidic coal ash slags similar to slags expected in coal gasifiers. 
 Such slags consist mainly of Si02, CaO, Fe203, AI2O3, MgO, and alkalies. The 
 eutectic temperatures for these slags are low, and the prospect of finding a 
 refractory that will last for years is dim. It may, in fact, be necessary to 
 utilize water cooling as has been done with blast furnace linings. Therefore, 
 slag corrosion tests were run both with and without water cooling. The three 
 most predominant components of the slags used were Si02, CaO, and AI2O3, cre- 
 ating base/acid ratios of between 0.5 and 0.8. The refractory samples were 
 arranged around the circumference of the furnace shell to form a ring, into 
 which up to 75 kg of slag was charged. While the furnace was not rotated, 
 some rotary action was induced by the three burner flames, which were fired 
 tangential to the furnace circumference. 
 
 In the first three runs, no water cooling was used and the furnace tem- 
 perature was held at 1,450° to 1,500* C for 200 hours. The most corrosion- 
 resistant sample was a fused-cast chrome-spinel refractory, which showed 
 almost no penetration. A fused-cast alumina and a sintered chrome-alumina 
 refractory also showed good resistance. 
 
 Subsequent runs were made using water cooling to produce thermal gradi- 
 ients through the walls of the furnace. Water temperatures were monitored so 
 that heat flux could be calculated. The fourth test run included: (1) 
 nitride bonded SiC, (2) silicate bonded SiC, (3) oxynitride bonded SiC, and 
 (4) a silicon oxynitride product. The temperature was held at 1,410° to 1,450° 
 for 500 hours. All the silicon carbide bricks showed little attack, contrary 
 to a previous run without water cooling. The silicon oxynitride product was 
 much less resistant than the silicon carbide types. 
 
 The fifth test run included four direct-bonded basic brick of a nominal 
 60 pet MgO content, with the temperature held at about 1,500° C for 500 hours. 
 The four brick exhibited similar resistance to corrosion. The sixth test run 
 evaluated three fused-cast and one sintered product. The fused-cast products 
 showed better resistance, but developed thermal shock cracks. The seventh 
 test run exposed alumina -chrome and alumina-silica products. A chemically 
 bonded alumina-silica type corroded badly, but a chemically bonded alumina- 
 chrome type did not. 
 
 In conclusion, a fused cast chrome-spinel refractory exhibited superior 
 corrosion resistance, while alumina, alumina -chrome, and magnesia-chrome types 
 formed spinel reaction layers at the slag interface. Increased chroraia con- 
 tent resulted in increased resistance. Silicon carbide products showed good 
 corrosion resistance when water cooled, but gaseous corrosion problems 
 remained unresolved. 
 
 Kim (19) investigated the attack of a 95 pet MgO fired refractory in con- 
 tact with Al203-containing slags. A master slag of composition 46 CaO ' 30 
 Si02 * 17 FeO * 7 MnO was used as a reference. Cylindrical rods 4-1/4 inches 
 long by 3/4-inch diameter were cut from the refractory and suspended into the 
 slag, contained in a platinum crucible at a furnace temperature of 1,600° C. 
 The slags used could be divided into four categories; that is, the master 
 
11 
 
 slag, the master slag plus MgO, the master slag plus AI2O3, and the master 
 slag plus MgO plus AI2O3. Two slag immersion times were used, 30 seconds 
 and 5 minutes. Following immersion, the samples were raised from the furnace 
 and sectioned for petrographic examination. 
 
 The results of the slag tests may be summarized as follows. The AI2O3- 
 containing slags produced the greatest amount of attack. No specimen survived 
 for 5 minutes in slags containing 10 pet or more AI2O3. For Al203-containing 
 slags, the addition of MgO in small amounts did not reduce attack. However, 
 with the addition of 13 pet and 15 pet MgO to these slags, the attack was 
 reduced. A specimen immersed in a low-basicity slag (C/S = 1,0) necked to 
 failure in 5 minutes, while a specimen immersed in the master slag (C/S = 1.5) 
 survived 100 minutes before failing. Both this low basicity slag and slags 
 with high contents of AI2O3, demonstrated strong tendencies to penetrate and 
 climb above the slag line in the specimens. The mechanisms of slag attack on 
 the magnesite by Al203-containing slags is by disruption of the silicate 
 matrix rather than by dissolution of MgO grains. Increased AI2O3 contents 
 resulted in greater attack because of the formation of highly fluid Ca-Al sil- 
 icates. The low-basicity slag was also detrimental to the magnesite refrac- 
 tory, but the higher viscosities of the reaction products and their localized 
 nature cause less disruption than do the Al203-containing slags. 
 
 In another paper, Kim ( 20 ) investigated the formation of a dense MgO 
 layer on carbon-containing magneslte-slag interfaces. There have been three 
 mechanisms suggested for the improved slag resistance by carbon-containing 
 magnesites: (1) a decreased slag penetration resulting from an increase of 
 wetting angle between the brick and slag; (2) the generation of internal brick 
 gases and pressure buildup retarding slag penetration; and (3) the carbon 
 reduction of MgO and subsequent oxidation of magnesium at the hot face to pro- 
 duce a dense impervious layer of MgO. Slags of C/S ratios of 1.0, 1.5, and 
 2.0, and containing the FeO, MnO, AI2O3, and MgO components, were melted in 
 crucibles cut from a 95 pet MgO, pitch-impregnated brick. The crucibles con- 
 tained 50 grams of the slag and were held at 1,600° C for 2 hours in an air 
 atmosphere. After this exposure, the crucibles were sectioned for petro- 
 graphic examination. Also, microprobe analysis scans were made across the 
 slag-refractory interface. 
 
 The following is a summary of the test results. A dense MgO layer does 
 form on the refractory hot face and is effective in reducing slag penetration. 
 This dense layer is not able to form, however, in iron oxide-free slags. The 
 slag should have a high CaO/Si02 ratio or should contain MgO if the layer is 
 to survive. The thickness of the layer observed increases as the CaO/Si02 
 ratio and MgO content of the slag increase. 
 
 Alumina was used as in the slag as a tracer, and the silicates behind the 
 dense MgO were found to be native to the brick since they contain no detect- 
 able A1203. A mechanism of intermittent destruction and reconstruction of the 
 dense MgO layer is suggested based on slag penetration, magnesium vapor oxida- 
 tion, MgO dissolution, and rebuilding of the layer at the resultant limit of 
 renewed slag penetration. 
 
 BBMWWaglWW BBiHBH 
 
12 
 
 Binns (4^) reviewed the use of special ceramics in handling molten iron 
 and steel. The nature of the steelmaking environment is described, and those 
 ceramic materials that are widely used in steelmaking are discussed. The 
 properties such as wettability, atmospheric stability, and mineralogy, which 
 make these materials useful for containing molten iron and steel, are 
 described. The factors that tend to impair the performance of, or lead to the 
 deterioration of, these materials are listed as — 
 
 1. Reactions with constituents of molten steel. 
 
 2. Reaction with oxidation products of steel and with steelmaking slags. 
 
 3. Attack by hot reducing or oxidizing gases at atmospheric or low 
 pressures. 
 
 4. Damage by steady state or transient thermal stresses. 
 
 5. Excessive heat conduction, leading to freezing. 
 
 The physical and chemical properties of commercially available oxides and 
 several mixed oxides are presented. The materials included were AI2O3, BeO, 
 MgO, CaO, Zr02, Cr203, Ce02, Th02, Hf02, MgO*Al203, Zr02'Si02, and CaO«Zr02. 
 These materials are compared in terms of stability in steelmaking conditions 
 and in contact with molten iron and steel. At present, MgO and Zr02 offer 
 the best combinations of properties, although the zirconates may offer some 
 potential. 
 
 Properties of nonoxide materials are presented in a manner similar to 
 that for the oxides. The materials included were carbides, borides, nitrides, 
 silicides, and sulfides. The only nonoxides considered to be truly commer- 
 cially available for large-scale usage are silicon carbide, silicon nitride, 
 and zirconium diboride. The usefulness of SiC and Si3N4 is strongly dependent 
 on the particular bonding phase employed as well as the fabrication tech- 
 niques. The ZrB2 material usually exists as a mixture with SiC or graphite, 
 which tends to increase oxidation and thermal shock resistance. 
 
 Bates (3-) reviews the requirements for materials to be used in MHD gener- 
 ators. The corrosion problems in these units are great and are due primarily 
 to coal ash slag and potassium seed liquid, which provides a current flow. 
 Temperatures will range from 600° to 1,800° C, and the added problem of ero- 
 sion compounds the problem. The slag and seed can react with MHD refractories 
 by direct chemical reaction with vapor or liquid, penetration of vapor or liq- 
 uid into pores or cracks, or diffusion of elements from the slag or seed. 
 Corrosion by liquid K2SO4 seed appears to be more severe than that by vapor 
 phase K2SO4. Porosity, as has been mentioned by others, has a profound effect 
 on the corrosion rate. Data are presented indicating that at near zero 
 porosity, corrosion rates for MgO and AI2O3 are about 1 cm/year. The loga- 
 rithm of the corrosion rate increases linearly with porosity. Impurities are 
 known to enhance corrosion rates. The attack on commercial oxide refrac- 
 tories, which occurs at grain boundary-located impurity phases, is greater 
 than for high-purity materials. 
 
13 
 
 A description of two coal slags is given. One type (Montana) is basic, 
 while another (Pennsylvania) is acidic. This wide difference in acidity will 
 undoubtedly require the use of different refractories. Seed material can also 
 be absorbed into the slag (up to about 30 pet), which can alter the corrosive- 
 ness of the slag. Also, iron can readily diffuse from the slag into alumina 
 or magnesia refractories. The consequence of this penetration under MHD con- 
 ditions has not yet been determined. Although emphasis on MHD refractories 
 has been placed on AI2O3 and MgO, other oxides such as Zr02, BeO, and Th02 
 have shown good resistance to potassium seed slags. 
 
 Zborowska (44) investigated the corrosion resistance of dense strontium 
 zirconate (SrZr03) material in contact with AI2O3, CaO, MgO, Si02, and K2CO3, 
 as well as coal ash. The zirconate material is a candidate for use in MHD 
 generators and would be exposed to coal ash slags. Strontium zirconate has a 
 melting point of about 2,900° C and has good corrosion resistance against 
 alkalies. 
 
 Synthesis of SrZr03 was carried out by prefiring pellets of SrC03 and 
 Zr02 powders, grinding the product, adding a polyvinyl alcohol binder and die 
 pressing crucibles. The corrosion tests were carried out by placing approx- 
 imately 1 gram of the substances in crucibles and exposing the crucibles to 
 1,700" C in a gas furnace for 2 hours. After cooling, the crucibles were 
 examined microscopically and by X-ray microanalysis. 
 
 No trace of corrosion effects was seen for crucibles containing AI2O3, 
 CaO, or MgO. Considerable corrosion occurred when Si02 was used. A reaction 
 layer was formed that was rich in zirconium and silicon. The coal ash corro- 
 sion produced a similar effect, but produced a reaction layer of only about 
 one-half the thickness of that by Si02 corrosion. 
 
 Pahlman (29) evaluated the stability of alumina-base refractories in 
 Western lignite ash slag environments. The refractories tested were (1) high- 
 alumina types (50 to 99 pet AI2O3), (2) fire clays, (3) high-alumina castables 
 (93 to 95 pet AI2O3), (4) regular castables (60 to 70 pet AI2O3), and (5) a 
 high-alumina plastic type. Both static and dynamic slag tests were conducted. 
 The lignite ashes used for slag testing were a high-alkali type and a low- 
 alkali type. Besides the chemical components of Si02, CaO, AI2O3, MgO, Fe203, 
 and SO3, the high-alkali type contained 9.3 pet alkali (Na20 + K2O), and the 
 low-alkali type contained only 1.4 pet alkali. 
 
 Static slag tests were run by immersing bars of each refractory in the 
 ash and heating at 1,300° C for 4 hours in an alumina crucible. The reduction 
 in cross-sectional area was determined for each refractory. The degree of 
 corrosion attack by these static tests could be related closely to the phase 
 composition present. Alumina refractories composed of mostly mullite, some 
 ar-Al203, and no cristobalite performed better than refractories of similar 
 alumina content but containing (>-Al203, mullite, and cristobalite. The 
 presence of cristobalite, as a result of sintering that did not permit suffi- 
 cient mullite formation, resulted in excessive expansion due to formation of 
 alkali silicates. Tests conducted in which cristobaltte-containing samples 
 were exposed only to alkali vapors indicated that vapor-phase reaction is 
 
14 
 
 sufficient to cause corrosion effects. Chemically bonded high-alumina samples 
 did not perform as well as did sintered ones. 
 
 Dynamic slag tests were run by placing cut refractory specimens around 
 the perimeter of a steel drum. The drum was rotated at about 6 rpm, and the 
 interior faces of the specimens were heated with a propane-oxygen torch at 
 1,400° C. Ash pellets were fed into the exit end of the drum once temperature 
 equilibrium had been reached. Ash pellets (100 grams each) were fed into the 
 drums every 10 minutes for the duration of an 8-hour run. At the end of the 
 run, the slag pool was poured out of the drum, and the average area change for 
 each specimen was determined. In general, the dynamic corrosion resistance 
 increased with increasing AI2O3 content. However, one f used-cast type of only 
 92 pet AI2O3 performed as well as the 99 pet AI2O3 types. Bloating of speci- 
 mens in the dynamic tests did not occur, as had been seen in the static tests. 
 This is likely due to alkali vapors not being able to penetrate the rotating 
 slag layer and the containment of cristobalite-alkali reactions at the slag 
 interface. The dynamic tests indicated that tar-impregnated refractories per- 
 formed much better than conventional sintered ones of similar chemistry and 
 mineral phase composition. However, it was not clear whether chemical bonding 
 enhanced corrosion resistance of these alumina refractories. In general, cor- 
 rosion resistance for monolithic refractories was lower than for sintered ones 
 of similar chemistry. It is recommended that, for use in alkali slag and 
 alkali vapor environments, refractories of 90 to 99 pet AI2O3 be used. 
 
 One refractory application area that is exposed to considerable alkali 
 slag attack is the regenerator checkers of a glass tank furnace. These brick 
 are exposed to the action of alkaline dust carried over from the raw batch 
 and, while checker temperatures are not as high as tank temperatures, checkers 
 may reach 1,370" C. Basic refractories have largely replaced fireclay and 
 high-alumina refractories, especially in the hotter section of the checkers. 
 Many of the studies of effects of alkali on these refactories have been post- 
 service examinations of checkers. Van Dreser and Cook (41) examined three 
 types of used checker refractories — (1) periclase or magnesite, (2) forster- 
 ite, and (3) chrome-magnesite — after 2 years of service. The conclusions 
 reached were that (1) the periclase and magnesite materials exhibited some 
 growth and recrystallization at the hot face, but were still serviceable; (2) 
 grain growth and volume changes in the spinel grains of the forsterite refrac- 
 tory caused a general cracking and friable condition; and (3) the chrome- 
 magnesite exhibited the most carryover penetration, resulting in numerous 
 voids and disintegration. 
 
 Steinhoff (38) developed a testing method to determine the degree of cor- 
 rosion of refractories by alkalies. A cylindrical test sample with a hole 
 drilled in the axis is exposed to corrosion by alkali vapors evolved in the 
 hole from a mix of fine fireclay plus Na2C03, Na2S04, Na2S, Na2B407, or mixes 
 of these salts. Magnesite brick showed a minimum of penetration, and no reac- 
 tion melts were observed. Fireclay refractories are easily covered in the 
 alkali vapors with a glaze that acts as a protecting layer against penetration 
 by alkalies. Sillimanite-type refractories scale under the influence of 
 
15 
 
 penetration by alkali vapors. For checker brick with a service temperature of 
 1,200" C, ordinary firebrick may be satisfactory, but for higher temperatures, 
 magnesite is necessary. 
 
 Schulte (36) describes the corrosion of magnesite refractories in check- 
 ers operating in the range 700° to 1,000* C. Chemical changes in composition 
 of brick zones are noticeable, and CaO and MgO are markedly removed by diffu- 
 sion, or possibly by vaporization. The cross section of the brick is 
 decreased, but usually no loosening of the structure or disintegration is 
 observed. Experiments are described for reproduction of corrosion of magne- 
 site by sulfates and SO3, but no clear corrosion effects were observed by the 
 action of a gas atmosphere containing SO3 or in a Na2S04 melt. This is 
 explained by the short reaction time and absence of periodic formation and 
 decomposition reactions that occur in actual checker conditions. 
 
 Touhig (40) has evaluated various ceramic materials as components for 
 lithium-sulfide batteries. The materials would be in contact with molten 
 lithium alloys at temperatures of between 375* and 425* C. Free energies of 
 reaction for several technical ceramics indicate that AI2O3 and Zr02 are 
 unstable with lithium, but no reaction is anticipated with AIN, BeO, MgO, 
 Si3N4, Y2O3, and BN. Impurities contained in these materials may, however, 
 cause some reaction to occur. 
 
 Screening tests on the above materials, consisting of exposure to molten 
 lithium at 400* C, indicated that the most promising candidate materials were 
 Y2O3, MgO, and BeO. The first two oxides were slightly discolored, suffered 
 some loss in weight, and remained nonconductive. The BeO specimen exhibited 
 no evidence of any reaction. A similar study in a molten mixture of 
 LiAL-LiCl-KCl produced less degradation but produced comparable qualitative 
 compatibilities. Overall, the BeO and Y2O3 materials appear to be the most 
 promising for use in the lithium-sulfur battery. 
 
 Singh ( 37 ) evaluated the resistance of Si3N4 and a 50 pet Si3N4 + 50 pet 
 AI2O3 material to molten lithium and molten sodium. Both ceramics are 
 expected to be compatible as a result of the positive free energy for the 
 reaction: 
 
 1/4 Si3N4 + 3 (Na or Li) = (Na3N or Li3N) + 3/4 Si. 
 
 Tests for sodium compatibility were performed at 550* C in a closed type 316 
 stainless steel capsule under argon. Similar tests for molten lithium were 
 conducted at 400* C in a closed style 304 stainless steel crucible under 
 helium. Exposures were for 300 hours. 
 
 No appreciable weight change was noted in the samples exposed. X-ray 
 diffraction patterns obtained before and after the tests were identical, and 
 no reaction products were detected. The samples exposed sodium showed evi- 
 dence of attack at the grain boundaries. After exposure to lithium, all sam- 
 ples exhibited an extensive crack network. Crack kinetics tests were run by 
 inspecting samples after exposure for both 1 and 10 hours. Samples exposed 
 for 10 hours appeared the same as those after 300 hours. Materials exposed 
 
16 
 
 only 1 hour showed cracking, but retained their integrity to a greater extent 
 than those exposed for 10 hours. 
 
 Robinson (34) detailed the effect of alkali attack on carbon refrac- 
 tories. It is noted that the mechanism of alkali attack on carbon is differ- 
 ent than that on alumina-silica refractories. Compounds of potassium were 
 used to evaluate attack, although sodium compounds will produce similar reac- 
 tions. It is possible for the carbon to react with gaseous, liquid, or solid 
 alkali. A variety of carbon materials were tested. These were categorized as 
 to graphite content, ash content, and apparent porosity. 
 
 The first Investigation concerned the effect of alkali on carbon oxida- 
 tion. Carbon refractories were impregnated with KOH, and weight loss (TGA) 
 runs were made. As the KOH content of the refractory was increased, the 
 amount of weight loss increased and the initiation temperature was lowered. 
 Dilatometric determinations on alkali-impregnated specimens showed that alkali 
 increases the expansion that occurs in argon or C02. 
 
 The corrosive reaction of carbon was investigated by exposing a carbon 
 rod above a charge of molten alkali in a fireclay crucible for 6 hours at 
 1,315** C. To offset the effects of air leaks, a flowing gas atmosphere was 
 used. It was found that a CO2 atmosphere resulted in significant alkali 
 attack, while argon and CO atmospheres produced no attack. 
 
 Since alkali vapors do cause softening, but not the cracking that has 
 been associated with alkali attack, it was desired to determine if molten 
 alkali would produce carbon cracking. Test samples were cut from carbon 
 refractories and a cavity drilled in each. The cavity was packed with a 
 selected alkali and the specimens were heated at various temperatures in a 
 tube furnace with controlled gas atmospheres. Both KOH and K2CO3 were used to 
 produce molten alkali. Molten KOH penetrates the pores of the carbon and 
 reacts with carbon to form K2CO3 in an argon atmosphere. The attendant expan- 
 sion is sufficient to crack the refractory when saturation by liquid is com- 
 plete. Melts of K2CO3 behave similarly, except that temperatures above 900° C 
 are necessary because K2CO3 melts and reacts with carbon above 900° C. Atmo- 
 spheres of CO2, CO, and air will prevent or retard absorption of the melt by 
 the refractory. Also, a slow rate of heating favors reaction of the hydroxide 
 external to the refractory, while rapid heating rates encourage absorption. 
 The reaction of the alkali with alumina-silica ash material produces a high 
 viscosity melt that does not penetrate the refractory. 
 
 Information on slag-melt corrosion of ceramic materials is the most com- 
 prehensive for any corrosive environment. This is probably a result of the 
 large-scale use of ceramics and the amount of ceramic experience in the slag- 
 melt area. Therefore, the greatest volume of data is available in this area, 
 and anyone requiring data on specific processes has many sources of 
 information. 
 
 This summary of slag and melt corrosive attack on ceramic materials has 
 shown that testing is normally formulated to simulate actual processing prac- 
 tice as closely as possible. Of course, any investigation evaluates 
 
17 
 
 candidate materials under conditions that may not fully represent actual prac- 
 tice. Considering the complex nature of most of the slag or melt processes, 
 full simulation is difficult. In this type of testing, complete definition of 
 test variables such as melt chemistry, refractory properties, melt viscosity 
 and feed rate, atmosphere conditions, and other pertinent variables must be 
 defined. Only by defining these conditions as completely as possible can 
 proper simulation be accomplished. Then, for any specific slag-melt system 
 being evaluated, meaningful data will be obtained regardless of the process of 
 interest. 
 
 CORROSION BY GAS ATMOSPHERES 
 
 Corrosion of ceramic materials by gases should be considered as similar 
 to corrosion by liquids in that the mechanisms of diffusion, reaction, and 
 dissolution are still pertinent. However, because of the nature of gaseous 
 flow, boundary layers are very thin and corrosion rates may be very high. 
 Also, in gaseous processes, replenishment of the corroding medium is usually 
 rapid so that a condition analogous to a saturated corrosive slag is not 
 likely to occur. 
 
 The problems of gaseous corrosion of ceramic materials have, in the past, 
 been limited primarily to processes in the petrochemical industry and the iron 
 and steel industry. However, the recent demand for energy generation from 
 coal and fossil fuels has created interest in processes in which high- 
 temperature, gaseous environments are prevalent. Bakker and Crowley (1) have 
 produced an excellent review of processes in which gaseous corrosion problems 
 are common. They review petrochemical, blast furnace, and coke oven processes 
 that create gaseous corrosion problems and describe refractory experience and 
 selection in those areas. 
 
 Petrochemical processes cited are (1) catalytic cracker units, (2) cata- 
 lytic reformers, and (3) secondary reformers. In catalytic cracker units, 
 operating temperatures are generally 540° to 760* C and conditions vary from 
 oxidizing to reducing throughout the complete unit. Refractory linings in the 
 reactor section are typically filled with carbon. Problems in the units are 
 mainly confined to the transfer lines and cyclones where erosive wear is a 
 real possibility. These areas are typically lined with a high-alumina, 
 erosion-resistant cas table. In catalytic reformers, temperatures are only 
 around 540° C, but operating pressures may reach 300 psi. The atmosphere is 
 rich in hydrogen, which can reduce silica at temperatures above 980° C. 
 However, silica-free refractories are not necessary, and concretes containing 
 50 to 70 pet AI2O3 perform well. Conditions in secondary reformers most 
 closely approximate coal gasification conditions. The gas composition is rich 
 in hydrogen, carbon monoxide, and steam. Operating temperatures are about 
 1,095° C, and pressures are in the 200 to 300 psi range. Because of silica 
 reduction problems, U.S. practice prefers the use of essentially silica-free, 
 95 pet alumina concretes. 
 
 Historically, blast furnaces have been lined with fireclay refractories. 
 Gaseous corrosion failure mechanisms in blast furnaces are a result of either 
 carbon monoxide disintegration or alkali bursting. Carbon monoxide 
 
 HBB 
 
18 
 
 disintegration occurs around 500° C and is the result of catalytic reduction 
 of CO to carbon in the pores of a refractory in the presence of free iron or 
 Iron oxide. It is pointed out that only iron or iron oxides act as catalysts 
 and iron silicates So not. Thus, carbon monoxide disintegration resistance 
 can be built in by high-firing the refractory to insure formation of iron sil- 
 icate^. Alkali bursting is caused by the vaporization of alkali compounds in 
 the lower section of the furnace. Alkali vapors then ^JJ^^^IJP^f %?J^^^^^^^ 
 reoxidize and react with silicates in the refractory. The alkali ^ili^^^es 
 have higher specific volumes, and the refractory is disrupted. Most alumina- 
 silica refractories are subject to alkali attack and, in some cases, carbon- 
 containing refractories are used instead. 
 
 Bakker and Crowley (1) also discuss the corrosive effects of S^^^^Jhat 
 are found in coal gasificltion units. The reduction of silica or silicates by 
 hydrogen is a possibility. This reduction is lessened in the P^^f ^^^^^^^^^^ 
 stelm and so may not be a real problem in some gasifiers. Corrosion by steam 
 leaching may also be a problem in gasifiers. It appears that steam leaching 
 irmore^pronounced in refractories that have high contents of 81-^^ P^-^-> 
 such as normal fireclay bricks. Carbon disintegration may be a P^^J^^"^ ^^ 
 calcium aluminate concretes are able to form a CaCO, phase. Corrosion by 
 alkali vapors will depend on the conditions of gas atmosphere and types of 
 alkali compounds in the coal ash. 
 
 Crowley (11-12) has documented the reduction of silica in refractories by 
 hydrogen. His— tTIdies showed that weight loss, due to the formation of gase- 
 ous sfo, increased with hydrogen partial pressure, gas 'l^^ /f^> ^^'f^H^' 
 ture. In the presence of steam, this weight loss was retarded. The results 
 indicated that this weight loss will occur at temperatures above 925 C. 
 Refractories having silica contents ranging from 8 to 95 pet were evaluated. 
 R:co^endations fof use in reducing conditions - encountered in reformer 
 units are that phosphate bonded 85 pet alumina brick, 99 pet alumina brick, 
 and calcium aluminate concretes low in silica be selected. 
 
 Wistrom (43) investigated the properties of a dense, 94 pet ^IzO 3 cast- 
 able after exp^ure to a 47.5 steam - 52.5 CO (vol-pet) atmosphere at 465 psia 
 and 199° C for 6, 12, and 18 days. Within 6 days, the ^^i^i"V}^'"^^^^^,0„. 
 cement phase decomposed to calcium formate, Ca (HC02)2 and boehmite, AlO OH). 
 Other than the amount of calcium formate formed, there was no apparent differ 
 enee in the decomposition of the eastable or of ^^^^^/^^f ^^/J^^'^f ^^e 
 cements. All of the CaO in the eastable and a majority of the CaO in the 
 cemen (excepting one) was converted to calcium formate, ^he weight gain, 
 decrease in porosity, and general increase in modulus of rupture can be 
 explained by'th^ relation of the cement with the steam-CO atmosphere to form 
 caLium formate and boehmite. Whether the eastable had ^een prefired or only 
 dried prior to exposure had no major effect on these chemical y^^ctlons. 
 These data indicate that the bonding provided by the boehmite in castables 
 exposed as described can replace that which is lost by the decomposition of 
 the original cement phases. However, the boehmite -^11/-^™^°-/^/^^^^^^' 
 temperaLres. Basically all the reaction of the eastable is confined to the 
 cement phases, and the alumina aggregate is essentially unreactive. 
 
19 
 
 Ballard (2) investigated the effect of steam-CO atmospheres on refractory 
 bond phase in — (1) a dense-fired alumina brick, (2) a phosphate-bonded ramming 
 mix, and (3) a calcium aluminate cement castable. These materials were 
 exposed to mixtures of steam and CO and to 100 pet of both steam and CO at 
 385, 750, and 1,015 psia at 260° and 950° C. After exposure, the bulk den- 
 sity, apparent porosity and modulus of rupture (MOR) of the materials were 
 measured. The phosphate- and cement-bonded samples were also analyzed by 
 DTA/TGA and X-ray diffraction (XRD). The dense alumina samples were also 
 examined by Auger electron spectroscopy. 
 
 The dense alumina material exhibited a reduction in MOR that was a func- 
 tion of steam pressure and coincided with segregation of calcium at grain 
 boundaries. The phosphate-bonded material showed good corrosion resistance to 
 the atmospheres, except for the case of saturated steam. The strength loss in 
 this material is attributed to attack of the AIPO4 bond phase, although no 
 reduction in the quantity of AIPO4 was detectable by XRD. The calcium alumi- 
 nate cement bond phases displayed more chemical reactivity than either of the 
 two other types. Reactions occurred between the cement phases and steam-CO 
 atmospheres below 300° C. Calcium formate and especially boehmite provide 
 additional bonding, since samples containing these phases showed an increase 
 in MOR. Above 900° C, no evidence was found for reactions of the cement bond 
 phases with the test atmospheres. 
 
 Sadler (35) has conducted a continuing program directed toward selection 
 of suitable refractory liner materials for nonslagging high- and low-Btu coal 
 gasifiers. Fourteen commercial refractories have been tested before and after 
 exposure to a variety of atmospheres ranging from 760° to 1,100° C, pressures 
 from 400 to 1,000 psi, and gas velocities from 0.04 to 1.4 ft/min for periods 
 up to 1,000 hours. Atmospheres include typical high- and low-Btu coal gasi- 
 fier atmospheres as well as steam, H2, CO, and CO2. An exposure of refrac- 
 tories to high-Btu gases containing sodium and potassium hydroxide vapors as 
 well as an exposure to alkali-impregnated refractories to high-Btu gas at 
 980° C were conducted. The major results can be summarized as follows: 
 
 1. Steam is the most corrosive gaseous component of the coal gasifier 
 environment toward the refractories tested. 
 
 2. Silica loss, due to volatilization, from intermediate-alumina con- 
 crete containing high-purity cement is negligible in high-Btu coal gasifier 
 environments below 1,000° C. 
 
 3. High-alumina concretes made from tabular alumina and high-purity cal- 
 cium aluminate cements lose strength when exposed to steam atmospheres at 
 1,000 psi and 760° C or above. 
 
 4. Brick containing 45 to 60 pet alumina show adequate resistance to 
 coal gasifier environments, and some phosphate-bonded concretes are unaffected 
 while other gradually lose strength. 
 
 5. Silicon carbide refractories are aggressively corroded (converted to 
 silica) by atmospheres containing high partial pressures of steam. 
 
20 
 
 6. The presence of saturated alkali vapors at 980° C seem to have no 
 effect on cold crushing strength of alumina refractories, while all refrac- 
 tories suffer deterioration when exposed to molten NaOH and KOH at 980° C. 
 
 Swaroop (39) has reviewed the feasibility of using ceramic coatings on 
 metal structures in a coal gasifier environment. Tables are given which 
 indicate the relative amounts of NH3, CO2, CO, H2, CH4, H2O, and H2S to be 
 expected in a coal gasifier operating in the temperature range 800° to 
 1,200° C. The most corrosive gas is expected to be H2S. Therefore, the 
 sulfidation and oxidation of potential ceramic coatings are discussed. Five 
 types of coatings that conceivably could be used to coat metallic structures: 
 (1) intermetallic compounds that form compact oxide layers, (2) intermetallic 
 compounds that form glassy oxide layers, (3) alloy coatings that form compact 
 oxide layers (4) noble metals and alloys, and (5) stable oxides that provide a 
 barrier to the corrosive environment. The only one of interest in the report 
 is the fifth type. 
 
 The protective capability of an oxide layer may be estimated by the 
 Pilling-Bedworth (P-B) ratio, which relates the specific volume of the oxide 
 formed to the specific volume of the corresponding metal. If this ratio is 
 below 1.0, an unprotective layer is formed, and if it is over 1.0, a protec- 
 tive layer is formed. However, for large values, the layer develops internal 
 stresses and will scale. The ratios presented for potential coating materials 
 indicate that carbides and nitrides of Ti, Si, and Zr should have the highest 
 resistance to scaling. 
 
 Besides this P-B ratio for compounds, one should also consider the 
 stability of the compounds in coal-gas environments. Free energy of reaction 
 values are given at 1,026° C for a number of ceramic compounds in coal-gas 
 mixture. It is noted that most of the coating compounds will resist sulfida- 
 tion. Also, the listed compounds (except for SiC) form an oxide layer that 
 is protective from further oxidation, Silicon carbide will not oxidize at 
 1,026° C at the oxygen pressure of a coal gasifier. Also, oxides such as 
 AI2O3, Zr02, and Cr203 will not undergo sulfidation. 
 
 Four methods are noted as suitable for applying these ceramic coatings: 
 (1) chemical vapor deposition, (2) vitreous enameling, (3) plasma spraying, 
 and (4) physical vapor deposition (sputtering, vacuum evaporation, etc.). 
 Conclusions are that ceramic compounds of the type AI2O3, Zr02, or Cr203; SiC, 
 HfC, or ZrC; Si3N4, TiN, or HfN; and TiB2, ZrB2, or HfB2 have the desirable 
 properties that could offer corrosion resistance to coal-gas conditions. It 
 would be best if these compounds could be graded with suitable alloys to match 
 the base metal. Also, plasma spray techniques are probably best for applica- 
 tion since vapor deposition techniques are not well-enough developed for these 
 materials. 
 
 Kraner (21) reviewed the effects of various atmospheres and gaseous con- 
 stituents on the life of steelplant refractories. The fluxing ability of 
 alkali compounds and accessory materials such as lead compounds, zinc com- 
 pounds, sulfides, chlorides, and fluorides is discussed. Many of these 
 
21 
 
 compounds have significant vapor pressures at steelmaking temperatures and 
 will migrate throughout a furnace and react with refractories. 
 
 The influence of oxygen content in these atmospheres is illustrated by 
 the role played by iron oxides. Hematite, Fe203, will lose oxygen and form 
 Fe304. The Fe304 is relatively refractory and will not drastically affect 
 refractory performance. However, if reducing conditions are allowed to exist 
 (as by formation of CO), the Fe304 will be reduced to FeO, which is a strong 
 flux for alumina-silica refractories. 
 
 Kraner also discusses the CO disintegration problem in blast furnace 
 walls and the alkali bursting problem. Although the alkali contents of ores 
 and coke are not high, after long periods of operation the refractory may con- 
 tain up to 20 pet alkalies. The formation of minerals such as nephelite, 
 Na20*Al203*2Si02, results and such minerals are not refractory. 
 
 The use of basic refractories in open hearths is discussed. Chrome- 
 containing bricks are subject to expansion effects because of the oxidation- 
 reduction of iron oxide. Therefore, chrome-containing bricks are not used in 
 contact with liquid metal, but are used in roofs. Magnesia refractories are 
 preferred for areas in contact with liquid metal. 
 
 The effects of oxidation-reduction reactions in carbon-containing MgO BOF 
 refractories have been reported by Leonard (23) , Pickering (30) , and 
 Carniglia (7^). These papers show that the reaction: 
 
 MgO + C = Mg + CO 
 
 can be predicted by comparing the free energies of formation at temperature 
 for both MgO and CO. At steelmaking temperatures of about 1,600° C, the CO 
 equilibrium pressure is about 1.0 atm and the magnesium pressure at equi- 
 librium would be about 0.01 atm. Therefore, any carbon in the refractory not 
 oxidized would react with MgO until the magnesium pressure reached 0.01 atm. 
 Removal of magnesium vapor by flowing gases would cause the reduction reaction 
 to continue. 
 
 As Mg vapor diffuses out of the brick, oxygen diffuses in and a dense 
 layer of MgO is produced about one-fourth to one-half inch into the brick. 
 This type of oxidation-reduction reaction continues as the BOF undergoes heat- 
 ing cycles. The reaction rate increases with temperature and decreases with 
 increasing Mg or CO pressure at constant temperature. The dense MgO layer 
 dissolves quickly in low lime-silica slags, but is quite impervious to high- 
 silica slags. 
 
 The effect of vanadium attack on refractories has been reported by Jones 
 and Hardy (16), Lehmann (22), and Gottardi and Bonetti (14). These studies 
 indicate that vanadium in oil ash used in furnaces is very corrosive to 
 refractories because of the formation of low melting vanadates (melting point 
 about 1,050° C). The first reference states that fuels containing vanadium 
 are very destructive to fireclay refractories, and the second states that 
 
22 
 
 magnesite refractories show less resistance to vanadium attack than do fire- 
 clay refractories. 
 
 A recent paper by Havranek (15) compares cup slag testing to a new method 
 of evaluating alkali attack. The refractories tested were blast furnace types 
 and were first tested In a standard slag cup test (DIN 51069) under controlled 
 CO/CO2 atmospheres at 1,000" C, 1,200° C, and 1,400° C for 5-hour periods. 
 The degree of attack and amount of cup expansion were determined. A total of 
 103 cup tests were run at various temperatures. The results obtained were 
 quite variable and difficult to reproduce, and It was concluded that this type 
 of test was unsuitable for selection of alkall-reslstant refractories. 
 
 Therefore, a more precise method based on the change of a physical param- 
 eter of the refractory was desired. A method based on the hot modulus of rup- 
 ture after alkali attack was developed. Six, 25- by 25- by 150-mm test bars 
 are cut from each refractory. The hot modulus of rupture at a selected tem- 
 perature Is then determined on three bars. Bulk density and porosity are 
 determined on the other tjiree bars. These bars are then dried and placed in a 
 silicon carbide box measuring 250- by 170- by 250-mm after the bottom has been 
 covered with 50 to 60 mm of fine coke. The test bars are then covered with 
 500 g of potassium carbonate, and the lid is cemented on the box with refrac- 
 tory cement. The box and contents are then held at the selected test tempera- 
 ture for 5 hours. 
 
 The test bars are examined visually after the test, after which soluble 
 alkali salts are removed by boiling until a neutral pH is obtained. The hot 
 modulus of rupture of these bars at 1,250° C is then determined. After cool- 
 ing the bars, bulk density and porosity are determined. Refractories that had 
 good resistance to alkali at 1,250° C were (1) brick with low porosity and 
 excess silica glass, (2) alumina or silicon carbide brick with a mullite bond, 
 and (3) silicon carbide brick with an oxynitride bond. After exposure to 
 alkali at 1,400° C, only the latter two types showed good resistance. These 
 results appear consistent with experience in the use of blast furnace 
 refractories. 
 
 Brooks (_5) investigated the corrosive effects of fuel oil combustion 
 products on silicon nitride and silicon carbide. The silicon materials were 
 exposed in a burner duct to combustion gas and in crucibles to molten sodium 
 sulfate and vanadium pentoxlde. The gas environment tests were run at 900°, 
 1,000°, and 1,400° C, while the melt tests were run at 900° C only. 
 
 The test results Indicated that fuel oils containing 50 ppm sodium pro- 
 duced a deposition of Na2S04, which reacted with the silicon materials to form 
 Na2Si205. This reactant, which is molten above 780° C, severely attacked both 
 self -bonded SIC and reaction-bonded Si3N4. Corrosion by vanadium was less 
 severe and was only noted when sodium salts were present so that the protec- 
 tive SIO2 layer was destroyed. Localized corrosion could be related to areas 
 of low O2 pressure, which would result in the formation of SIO. Hot-pressed 
 Si3N4 demonstrated better corrosion resistance. Instead of exhibiting deep 
 penetration of deposits, the reaction with the hot-pressed material was 
 
23 
 
 limited to a narrow layer. The corrosion Is dependent on the sodium content 
 of the fuel, and this fact may affect the use of such materials In gas 
 turbines. 
 
 Information on engineering properties of ceramic materials complied by 
 Lynch (25) Includes corrosion or oxidation resistance on most of the common 
 oxides, carbides, nitrides, and slllcldes. The Information Is quite general 
 or Is limited to single data points, except In a few cases. The data on these 
 groups of materials are summarized below. 
 
 Oxides 
 
 Aluminum oxide Is chemically one of the most stable refractory compounds 
 and up to the 1,700° - 1,815° C range Is resistant to all gases, except moist 
 fluorine. It may be used In either oxidizing or reducing atmosphere as high 
 as 1,925° C for short periods. 
 
 Beryllium oxide Is stable In all dry atmospheres, but undergoes consider- 
 able weight loss above 1,095° C In even low concentrations of water vapor. 
 The weight loss Is attributed to formation and volatilization of Be (OH) 2. 
 
 Magnesium oxide Is unaffected by neutral or oxidizing atmosphere even 
 above 1,980° C, but Is easily reduced. The powder dissolves slowly In water 
 unless highly sintered, but dense material Is stable In presence of water 
 vapor. 
 
 Calcium oxide reacts readily with water at room temperature and also 
 hydrates In air. It will react with CO2 and SO2 on heating. 
 
 Zirconium oxide Is stable In oxidizing and moderately reducing atmos- 
 pheres, but will decompose at 2,200° C In contact with carbon, nitrogen, or 
 hydrogen. 
 
 Titanium oxide may be reduced and Is restricted to oxidizing 
 environments. 
 
 Chromium oxide Is resistant to reduction by carbon and hydrogen. 
 
 Calcium and strontium zlrconates are stable In air at high temperatures, 
 but will react with silica or alumina. 
 
 Carbides 
 
 Silicon carbide Is the most stable of the carbides up to about 1,650° C. 
 It Is stable below 980° C and above 1,150° C owing to a protective silica 
 coating. Between these temperatures, the oxidation Is a function of the phys- 
 ical state of this silica. The rate of oxidation above 1,150° C Is affected 
 by a decrease In viscosity of the silica due to Impurities, formation of vola- 
 tile SIO In vacuum or reducing conditions, and the mass velocity of the 
 oxygen-bearing environment. 
 
24 
 
 Appreciable oxidation of titanium, hafnium, and zirconium carbides occurs 
 as low as 430° C. Increased oxidation rates of HfC and ZrC coincide with 
 phase changes of the oxide coatings. These carbides are reportedly attacked 
 by nitrogen at high temperatures. However, samples of TiC and ZrC were non- 
 reactive with hydrogen at 2,200° C. 
 
 Chromium carbides are not significantly affected in air up to 1,270° C, 
 although carbi-des of tungsten and molybdenum are less stable. Oxidation 
 occurs with these materials between 540° and 815° C, and oxidation becomes 
 catastrophic at slightly higher temperatures. Carbides of uranium, thorium, 
 and plutonium react with water and moist air to form hydrogen, hydrocarbons, 
 and hydrated oxides. 
 
 Nitrides 
 
 Nitrides are poor in oxidation resistance except for BN and Si^N^ up to 
 1,095° C. Most nitrides. are stable in reducing atmospheres and, of course, 
 nitrogen. Controlled oxidation of Si3N4 above 1,700° C produces an 
 oxynitride. Less weight change in Si3N4 occurs in the higher density, hot- 
 pressed material. Silicon nitride is stable in neutral or reducing atmos- 
 pheres up to 1,870° C, It is attacked very slowly in boiling water and is 
 slightly soluble in water at 1,500 psi and 315° C. 
 
 Boron nitride is only slightly oxidized in air at 980° C and is insoluble 
 in water, but hydrolyzes in boiling water and aqueous solutions. It is stable 
 in inert atmospheres, but reacts with carbon above 1,980° C. In hydrogen, it 
 can be used above 1,980° C and will resist chlorine up to 700° C, 
 
 Silicides 
 
 Oxidation of the silicides decreases as the silicon content increases. 
 Tantalum silicide is relatively stable in air up to 1,650° C. Chromium sili- 
 
 cide oxidizes in air, and the brittle oxide layer offers little protection. 
 
 Molybdenum disilicide resists oxidation up to 1,700° C. Above 1,755° C and an 
 
 oxygen pressure of 10 torr, oxidation increases owing to formation of gaseous 
 
 SiO. Tungsten disilicide oxidizes to form WO3 and Si02, Above 1,040° C, WO3 
 volatilizes, and a stable Si02 coating forms. 
 
 The review on gaseous corrosion of ceramic materials indicates that this 
 area has been investigated heavily in recent years. These recent investiga- 
 tions have dealt with more complex and more severe gas conditions than were 
 encountered previously. As in the case of slags and melts, the intent has 
 been to simulate process conditions, and this can be difficult to accomplish. 
 While actual simulation of O2, CO2, CO, H2, and steam conditions has been con- 
 ducted extensively, testing has been directed towards oxide ceramics and only 
 limited data are available on carbides, nitrides, silicides, or borides. 
 Also, the effects of environments of nitrous oxides, sulfurous oxides, vola- 
 tile metal oxides, halogen gases, etc., are not well known. Of course, as in 
 the case of slag-melt corrosion, the specific conditions of the process of 
 interest will dictate the corrosion information that is necessary. As pro- 
 cesses lead to environments containing various other gaseous components, the 
 
25 
 
 corrosion effects of such components must be determined. A major aim of any 
 gaseous corrosion testing should be to determine the attack mechanisms predom- 
 inating so that judicious selection of materials may be made. Testing to 
 determine mechanisms requires that standardized and accepted test procedures 
 be developed. 
 
 SUMMARY 
 
 The information available on the corrosion behavior of ceramic materials 
 is disseminated by many sources, each of which are principally interested in 
 specific corrosive environments. In the case of corrosion by acid as well as 
 basic solutions, it was noted that quantitative corrosion information is 
 available only for glass materials. Information for other classes of ceramic 
 products is only qualitative and generally is supplied by the manufacturers. 
 Standardization of both material classifications and test procedures would be 
 required to better describe the corrosion resistance of ceramic materials. 
 Ideally, it should be possible to quantitatively define the rates of corrosion 
 attack by specific reagents and conditions on different classes of ceramic 
 products. One advantage in solution corrosion testing is that exposure condi- 
 tions are probably the simplest to define of the three types being considered, 
 since solution concentration and temperature are the prime variables. 
 
 The area of slag-melt corrosion contains the most comprehensive informa- 
 tion of the three general areas of corrosion. This type of corrosive attack 
 is difficult to fully simulate at the test level because of the many variables 
 that must be defined. Complete definitions of the refractories being tested 
 and the slag or melt material are necessary to obtain meaningful data. The 
 aim of most slag-melt corrosion investigations is usually to determine what 
 particular refractory will best contain a specific slag or melt. This type of 
 evaluation serves well to assess new developments in refractory materials. 
 The data collected in regard to existing mineral or energy conversion pro- 
 cesses should aid in selection of suitable materials for new processes. 
 
 The area of gaseous corrosion has received a great deal of study in 
 recent years. Systems such as coal gasification, gas turbines, and MHD have 
 led to much of this interest. Simulated testing in O2, CO, CO2, H2, and steam 
 environments has resulted in a large amount of data on gas-refractory reac- 
 tions, but primarily for oxide ceramics and not for carbides, nitrides, sili- 
 cides, etc. The mechanisms of many of these reactions are not well known, and 
 standardized tests are required to establish the corrosion behavior in various 
 gaseous environments. As future energy conversion processes lead to corrosive 
 gas environments such as nitrous oxides, sulfurous oxides, and volatile metal 
 vapors, the effects of such environments on ceramic structural materials 
 will need to be determined. 
 
 An effort directed toward establishing the relative importance of the 
 materials performance problems associated with evolving metallurgical pro- 
 cesses, critical materials substitution, and energy conversion processes is 
 needed to ascertain what areas demand the majority of investigative work. 
 Once this importance is established, the specific problems of corrosion attack 
 on ceramic materials can be evaluated and means taken to improve their 
 
26 
 
 performance. The information in this review indicates that data are available 
 for a limited number of corrosive environments, but quantitative data are 
 needed in certain areas in which ceramic materials will potentially be exposed 
 to new, untested service conditions. 
 
 Research at the Bureau's Tuscaloosa Research Center has been initiated to 
 evaluate the resistance of commercial ceramic materials to hydrochloric acid 
 solutions. Work also is underway to evaluate the relative resistance of com- 
 mercial refractories with regard to slag attack in the production of glass 
 wool from mineral waste materials such as slate wastes and porphyry copper 
 tailings. Future research is being directed toward developing quantitative 
 data on the resistance of ceramic materials to acids such as H2SO4, HNO3, 
 H3PO4, and HF; bases such as NaOH and NH4OH; and various gaseous environments 
 encountered in the roasting and/or reduction of low-grade, complex refractory 
 ores. 
 
27 
 
 REFERENCES 
 
 1. Bakker, W. T. , and M, S. Crowley. Requirements for Ceramics in Coal 
 
 Gasification Processes. Proc. ERDA/NSF Workshop on Ceramics for Energy 
 Applications, Columbus, Ohio, November 1975, pp. 153-200. 
 
 2. Ballard, W. V., and D. E. Day. Stability of the Refractory-Bond Phases 
 
 In High Alumina Refractories in Steam-CO Atmospheres. Am. Ceram. Soc. 
 Bull., V. 57, No. 7, 1978, pp. 660-666. 
 
 3. Bates, J. L. Behavior/Requirements for MHD Support Systems. Proc. 
 
 ERDA/NSF Workshop on Ceramics for Energy Applications, Columbus, Ohio, 
 November 1975, pp. 254-270. 
 
 A. Binns, D. B. The Use of Special Ceramics in Handling Molten Iron and 
 Steel. J. Brit. Ceram. Soc, v. 77, No. 1, 1978, pp. 1-13. 
 
 5. Brooks, S., and D. B. Meadowcroft. The Corrosion of Silicon Based 
 
 Ceramics in a Residual Fuel Oil Fired Environment. Proc. Brit. Ceram. 
 Soc, No. 26, July 1978, pp. 237-250. 
 
 6. Campbell, I. E. High Temperature Materials and Technology. John Wiley 
 
 and Sons, Inc., New York, 1967, 1,022 pp. 
 
 7. Carniglia, S. C. Limitations on Internal Oxidation-Reduction Reactions 
 
 in BOF Refractories. Am. Ceram. Soc. Bull., v. 52, No. 6, 1973, 
 pp. 160-165. 
 
 8. Cook, R. L., and A, I. Andrews. Chemical Durability of Porcelain Enam- 
 
 els. J. Am. Ceram. Soc, v. 28, No. 9, 1945, pp. 229-256. 
 
 9. Cooke, D., and A. Paul. Chemical Durability of Glass Enamels Containing 
 
 B2O3, Part I: Lead Borosilicates. J. Brit. Ceram. Soc, v. 77, No. 4, 
 1978, pp. 104-116. 
 
 10. Cooper, A. R. , and W. D. Kingery. Dissolution in Ceramic Systems: I. 
 
 Molecular Diffusion, Natural Convection and Forced Convection Studies 
 of Sapphire Dissolution in Calcium Aluminate Silicate. J. Am. Ceram. 
 Soc, V. 47, No. 1, 1964, pp. 37-43. 
 
 11. Crowley, M. S. Hydrogen-Silica Reactions in Refractories. Part I. Am. 
 
 Ceram. Soc. Bull., v. 48, No. 7, 1967, pp. 679-682. 
 
 12. . Hydrogen-Silica Reactions in Refractories. Part II. Am. Ceram. 
 
 Soc. Bull., V. 49, No. 5, 1970, pp. 527-530. 
 
 13. Cutler, I. B. Corrosion Behavior of Ceramics. Proc. ERDA/NSF Workshop 
 
 on Ceramics for Energy Applications, Columbus, Ohio, November 1975, 
 pp. 65-82. 
 
28 
 
 14. Gottardl, V., and G. Bonetti. Corrosion of Magnesite Brick in Glass Fur- 
 
 naces. Silicates Ind. , v. 29, No. 4, 1964, pp. 129-131. 
 
 15. Havranek, P. H. Alkali Attack on Blast Furnace Refractories. J. Brit. 
 
 Ceram. Soc, v. 77, No. 3, 1978, pp. 92-97. 
 
 16. Jones, M. C. K. , and R. L. Hardy. Petroleum Ash Components and Their 
 
 Effect on Refractories. Ind. and Eng, Chem. , v. 44, 1952, 
 pp. 2615-2619. 
 
 17. Kappmeyer, K, K. , and D. H. Hubble. Pitch-Bearing MgO-CaO Refractories 
 
 for the BOP Process. Ch. in High Temperature Oxides, ed. by A. M. 
 Alper, Academic Press, Inc., New York, 1970, 358 pp. 
 
 18. Kennedy, C. R. , and R. B. Peoppel, Corrosion Resistance of Refractories 
 
 Exposed to Molten-Acidic Coal-Ash Slags. Interceram. , v. 27, No. 3, 
 1978, pp. 221-226. 
 
 19. Kim, S., P. S. Nicholson, and W. Lu. Attack of Magnesite Refractories 
 
 by Alumina-Containing Slags, Am. Ceram. Soc, Bull., v. 57, No. 7, 
 1978, pp. 652-656. 
 
 20. . BOF Slag-Induced Formation and Destruction of Dense MgO Layers in 
 
 Carbon-Containing Magnesites. Am. Ceram. Soc. Bull., v. 57, No. 7, 
 1978, pp. 649-656. 
 
 21. Kraner, H. M. Refractories Service Conditions in the Blast Furnace. J. 
 
 Am. Ceram. Soc, v. 25, No. 11, 1942, pp. 311-320. 
 
 22. Lehmann, H. , S, Kienow, and P, Artelt, (The Corrosion Behavior of Basic 
 
 Refractory Bricks Against V2O5,) Torrind-Ztg, Keram. Rundschau, v. 
 88, Nos. 7, 8, and 10, 1964, pp, 153-159, 225-231; abs, in Chem. Abs., 
 V. 61, 1964. 
 
 23. Leonard, R. J., and R. H. Herron. Significance of Oxidation-Reduction 
 
 Reactions Within BOF Refractories. Am. Ceram. Soc, Bull,, v. 55, No. 
 1, 1972, pp. 1-6. 
 
 24. Litvakovskii, A. A. Fused Cast Refractories, Israel Program for 
 
 Scientific Translations, Jerusalem, 1961, 273 pp, 
 
 25. Lynch, J, F,, C, G, Ruderer, and W, H. Duckworth. Engineering Properties 
 
 of Selected Ceramic Materials, Battelle Memorial Inst,, Columbus, 
 Ohio, 1966, 
 
 26. McGee, T, D, Laboratory Slag Tests To Simulate Refractory Corrosion, 
 
 Interceram., v. 27, No. 4, 1978, pp. 416-419. 
 
 27. Morey, G, W. The Properties of Glass, ACS Monograph Series, Reinhold 
 
 Pub. Corp., New York, 2d ed. , 1954, 591 pp. 
 
M Mi ii Mn n Mii'n i w i niir ' "v^wwiiimmwii ii i iiH[ 
 
 29 
 
 28. Nordberg, M. E. Chemical Durability. Corning Glass Works Monograph; 
 
 available upon request from T. A. Clancy, P. 0. Box L, University, 
 Ala. 35486 
 
 29. Pahlman, J. E., C. F. Anderson, and S. E. Khalafalla. Stability of 
 
 Alumina-Base Refractories in Western Lignite Ash Slag Environments. 
 BuMines RI 8334, 1979, 16 pp. 
 
 30. Pickering, G. D. , and J. D. Batchelor. Carbon-MgO Reactions in BOF 
 
 Refractories. Am. Ceram. Soc. Bull., v. 50, No. 7, 1971, pp 611-614. 
 
 31. Priest, D. K. Corrosion Concepts for Glassed Steel. Am. Ceram. Soc. 
 
 Bull., v. 39, No. 10, 1960, pp. 507-510. 
 
 32. Rauch, H. W. Improved Ceramic Heat Exchanger Material. General Electric 
 
 Co., NASA Contract NAS 3 - 19698, November 1977, 30 pp. 
 
 33. Robbins, M. D. For Moderate Temperatures and Severe Conditions, Use Non- 
 
 metallic Inorganics. Chem. Eng., v. 65, No. 18, 1958, pp. 123-134. 
 
 34. Robinson, G. C, P. Schroth, and W. D. Brown. Alkali Attack of Carbon 
 
 Refractories. Am. Ceram. Soc. Bull., v. 58, No. 7, 1979, pp. 668-675. 
 
 35. Sadler, L. Y. , III, N. S. Raymon, K. H. Ivey, and H. Heystek. Evaluation 
 
 of Refractory Liner Materials for Use in Nonslagging, High Btu Coal- 
 Gasification Reactors. Am. Ceram. Soc. Bull., v. 58, No. 7, 1979, 
 pp. 705-714. 
 
 36. Schulte, K. (Corrosion of Magnesite Checkerwork Bricks by Alkali Sul- 
 
 fates amd SO3.) Glastech. Ber. , v. 38, No. 11, 1965, pp. 457-461; abs. 
 in Chem. Abs, v. 64, 1966. 
 
 37. Singh, R. N. , and W. D. Tuohig. Compatibility of Si3N4 and Si3N4-Al203 
 
 With Liquid Na and Li. J. Am. Ceram. Soc, v. 58, Nos. 1-2, 1975, 
 pp. 70-71. 
 
 38. Steinhoff, E. (The Reasons for the Destruction of Checkerwork Brick in 
 
 Glass Furnaces.) Glastech. Ber., v. 28, 1955, pp. 265-272; abs. in 
 Chem. Abs., v. 49, 1955. 
 
 39. Swaroop, R. Ceramic Coatings for Components Exposed to Coal-Gas Envi- 
 
 ronments, A Review. Argonne Nat. Lab., Rept. ANL-76-124, December 
 1976, 263 pp. 
 
 40. Tuohig, W. D., and J. E. Battles. Ceramics in Lithium Metal Sulfide Bat- 
 
 teries. Proc. ERDA/NSF Workshop on Ceramics for Energy Applications, 
 Columbus, Ohio, November 1975, pp. 204-230. 
 
 41. Van Dreser, M. L., and R. H. Cook. Deterioration of Basic Refractories 
 
 in a Glass Regenerator. Am. Ceram. Soc. Bull., v. 40, No. 2, 1961, 
 pp. 68-73. 
 
30 
 
 42. Van Vlack, L. H. Elements of Materials Science. Addison-Wesley Pub. 
 
 Co., Reading, Mass., 1964, 445 pp. 
 
 43. Wistrom, K. W. , and D. E. Day. Chemical Reactivity of Calcium Aluminate 
 
 Cement Bond Phases in a Steam-CO Atmosphere at 199° C. Am. Ceram. Soc. 
 Bull., V. 57, No. 7, 1978, pp. 680-684. 
 
 44. Zborowska, M. , M. Grylicki, and J. Zborowski. The Corrosion Resistance 
 
 of Strontium Zirconate Ceramics to Coal Ash Constituents and Potassium 
 Seed of MHD Generators. Ceramurgia Internat., v. 5, No. 1, 1979, 
 pp. 28-30. 
 
31 
 
 BIBLIOGRAPHY 
 
 1. Adams, P. B. , and D. L. Evans, Chemical Durability of Borate Glasses. 
 
 Ch. in Borate Glasses, ed. by L. D. Pye, V. D. Frechette, and N, J, 
 Kreidl, Plenum Press, New York, 1978, pp. 525-536. 
 
 2. Alcock, S., and D. R. F. Spencer. Application and Performance of Basic 
 
 Refractories in Secondary Steelmaking Ladles. J. Brit. Ceram. Soc. , 
 V. 77, No. 2, 1978, pp. 45-57. 
 
 3. Bose, S., and T. D. McGee. Corrosion of MgO Single Crystals by BOF 
 
 Slags. Am. Ceram. Soc. Bull., v. 57, No. 7, 1978, pp. 674-684. 
 
 4. Clark, C. B. Reaction of Fused Cast Alumina Refractories With Metals and 
 
 Slags. J. Metals, No. 9, 1966, pp. 1047-1052. 
 
 5. Crowley, M. S. Refractory Problems in Coal Gasification Reactors. Am. 
 
 Ceram. Soc. Bull., v. 54, No. 12, 1975, pp. 1072-1075. 
 
 6. Crowley, M. S., and J. F. Wygant. Critical Ceramic Applications in Fos- 
 
 sil Fuel Processing. Am. Ceram. Soc. Bull., v. 53, No. 11, 1973, 
 pp. 828-832. 
 
 7. Dial, R. E. Refractories for Coal Gasification and Liquefaction. Am. 
 
 Ceram. Soc. Bull., v. 54, No. 7, 1975, pp. 640-643. 
 
 8. Gac, F. D. Stability of Refractory Castables to Steam - N2 and Steam-CO 
 
 Atmospheres. Los Alamos Scientific Lab., Rept. LA-6508-T, September 
 1976, p. 72. 
 
 9. Gilbert, S. V. Rotary Slag Testing of Basic Refractories. Am. Ceram. 
 
 Soc. Bull., V. 55, No. 7, 1976, pp. 645-648. 
 
 10. Howe, R. A., T. D. McGee, and C. E. Osterholtz. Attack of a Tar-Bonded 
 
 Periclase Refractory by a Dolomitic BOF Slag. Am. Ceram. Soc. Bull., 
 V. 55, No. 2, 1976, pp. 205-209. 
 
 11. Miller, M. A., and W. L. Shott. Corrosion of Refractories by Blast Fur- 
 
 nace Slag. Am. Ceram. Soc. Bull., v. 47, No. 7, 1968, pp. 648-653. 
 
 12. Nelson, R. T. Destructive Effect of Alkali on Carbon Brick. Am. Ceram. 
 
 Soc. Bull., V. 35, No. 5, 1956, pp. 188-190. 
 
 13. Snajdr, E. A., B, Klaudinyi, J. A. Middleton, and D. S. Yang. Evaluation 
 
 of Slag and Refractory Compatibility for Low-Btu Gasification of Coal 
 for Electric Power Generation. Rept. to U.S. Dept. Energy under Con- 
 tract EX-76-C-01-1545, December 1977. 
 
 14. Trojer, F. Deterioration of Cement Rotary Kiln Linings by Alkali 
 
 Sulfides and Sulfates. Am. Ceram. Soc. Bull., v. 47, No. 7, 1968, 
 pp. 630-634. 
 
 * U.S. QOVERNMEMT PRIUriNG OFFICE : 1981 - 349-265 
 


 ^^..'b'^ 
 
 
 o > 
 
 v-^'\/* V'^'^V V''^^*'y'' V*^''*/ 
 
 
 
 O > 
 
 
 
 >^ .. 
 
 
 
 
 '- '^'-o^ !^^i^': ^o\^' :m^^^ '-n^o^ »-.^^»'- ^r.^^ :m^^\ ^^j..^ 
 
 
 k"^' 
 
 
 •^0^ 
 
 \ '•-^P'/ . ^^'"-"^ ^ '-^^^^ J"^-^. '^J ^^"\ '^^'' ^^'^-- 
 
 
 ^'- -^Ao^ » 
 
 \/ %^^%oo \;^V V».o^ "-■ •' 
 
 
 
 
 '^o^ °^^^iB»-. '^n\> ■'jm^>,.\ '-^^.^ 
 

 
 
 
 bV" 
 
 
 ' ^>' 
 
 <i^°^ 
 
 
 
 
 
 : %.** •• 
 
 
 .^'% 
 
 0" .i.5j^'. ^o ^-j^" .c:,;^. >„ c° 
 
 ^<^_ •.;-.' .^^ 
 
 'fi.V^ -iS^M!^^^ ^^Mc^ 
 
 'bV 
 
 •^o^ 
 
 
 
 
 
 ."^V. 
 
 
 
 .^'% 
 
 aO' V'-r^\^^^ *^o^ *.v. 
 
 
 
 
 'bv^ 
 
 
 
 , "^' A" '<}>