J. Am Ceram Soc..83[92287-96(2000 urna Thermochemical Reactions and equilibria between Fluoromicas and silicate matrices A Petromimetic Perspective on Structural Ceramic Composites Todd T. King,* f Walter Grayeski, and Reid F. Cooper*, s Department of Materials Science and Engineering, University of wisconsin-Madison, Madison. Wisconsin 53706-1595 A petromimetie (geologicaF-analog) approach is applied to the interface. Both of these criteria emphasize the importance of design of alumina-fiber-reinforced glass-ceramic -matrix com engineering thermodynamically and mechanically functional inte posites that use a refractory, trioctahedral fluoromica fiber- faces or interphases between the fiber and the matrix phase. matrix interphase and feldspar matrixes. Studies of the solid- In considering these interfacial requirements, one realizes that state reaction couples between these silicate phases are the achievement of effective ceramic composite materials will be pursued to address the chemical tailorability of the interphase/ predicated on polyphase ceramic equilibria possessing high matrix interface from an engineering perspective. The mini- mechanical-behavior contrast among phases. Therefore, one ap- nization of alumina and silica activities within polyphase proach toward the fulfillment of these general design criteria is to feldspar-based matrixes via Mgo buffering is shown to be an onsider the wealth of data and understanding in the geologic ffective route toward a stable fluoromica interphase. An record: present right below our very feet is evidence of polyphase anorthite-2-vol%-alumina(CaAl Si2Os a-Al2O3)substrate, ceramic equilibria of phases-specifically silicates-with a wide hemically buffered with MgO, is shown to exhibit thermody variation of mechanical behavior. The combined disciplines of amie stability against fluorokinoshitalite(BaMg3lAlSiJO1oF2, igneous and metamorphic petrology and geochemistry indicate the up to temperatures potentially as high as 1460oC. The key to plethora of multicomponent silicate and oxide assemblages that he approach is the reduction of alumina activity via the exhibit high-temperature stability over geologic time scales. Thus, formation of MgAI O spinel. Similarly, the formation of one can consider the design of environmentally robust, elevated- forsterite(Mg, SiO4) stabilizes the mica in contact with matrix temperature ceramic composites from a petromimetic perspective compositions otherwise containing excess silica. The cationic Silicates possess a vast structural diversity, emanating from the interdiffusion between solid-solution feldspars and fluoromi- degree of Si-O--Si polymerization, that can provide the neces- as also is characterized. Coupled interdiffusion of k and sary mechanical contrast (i.e, intrinsic fracture toughness)to Si+ in exchange for Ba2+ and Al+ was observed between achieve debonding in fiber-reinforced ceramic composites as K-Ba solid-solution celsian and the barium-rich solid-solution articulated, for example, by the bimaterial interfacial crac end-member fuorokinoshitalite at 1300oC. A similar cationic deflection criterion developed by He and Hutchinson. Silicate xchange also is observed against the potassium-rich end- mechanical behavior ranges from extremely strong but brittle member fluorophlogopite(KMg3lAISi3JOJoF,), although in a framework silicates(fully polymerized, almost covalently bonded reverse direction, at temperatures of <1280C. The interfacial structures) to flexible sheet silicates(phyllosilicates )such as micas compositions identified via electron microprobe analysis spec- and clays(less-polymerized structures that incorporate notably fy one set of local equilibrium conditions from which global weak ionic or van der Waals bonding on specific crystallographic ceramic composite equilibrium can be achieved planes ). Thus, a refractory ceramic composite that uses micas as a thin, thermodynamically stable interphase to protect alumina fibers ically from cracks in an oxide/silicate matrix becomes a L. Introduction IBER-REINFORCED ceramic composites must fulfil two general as have a tetrahedral-cation-to-oxygen-anion(T: o)ratio of criteria to be effective structural materials for high-temperature 1: 2.5 and can be described by the structural formula aerobic environments. First, long- term thermochemical stability A0,.s-BY-[Alo-2Si4-2O1o(OH)2 (1) must be established between each of the ceramic component phases (i.e, fiber, matrix, and interphase) and the oxidizing where the Roman-numeral superscripts refer to the cation coordi environment. Second, useful ceramic composite systems must nation with O and/or OH anions, A represents a 12 exhibit significant toughness at elevated temperatures and rapid oordinated (i.e, large-diameter) alkali or alkaline-earth cation oading rates. Achievement of this second requirement is largely(known as the interlayer cation), and B represents a divalent or dependent on the debonding characteristics of the fiber/matrix trivalent octahedral (6-coordinated) cation. The T: O ratio require ment means that the relative amounts of al and si on the tetrahedral sites are affected by the specific occupancy of the octahedral and interlayer cation sites, for the above-described R. A. Condrate--contributing editor formula, with 100- anions, the sum of the tetrahedral Al and si ions must total 4. Cation-site occupancy defines the adjectives used to describe the micas. For example, the mineral muscovite which has the formula KAL,[Al SiJO,o(OH),, is described as a 189258 Received June 30. 1999 dioctahedral (i. e, only two of the three available octahedral sites ystems Program in Mechanics and Materials, through Grant No are occupied), trisilicic (i.e, three of the four tetrahedral sites contain Si), true(or flexible, i.e., the inter layer cation is an alkali) Member. American Ceramic Socie urrent ly with intel Corp, Santa car ight Center, Greenbelt,MD ca,the mineral kinoshitalite(BaMg [Al,Si,J0,o(OH),)is a trioctahedral, disilisic, brittle (i.e, the interlayer cation is an AUthor to whom correspondence should be addressed alkaline-earth element) mica. 287
Thermochemical Reactions and Equilibria between Fluoromicas and Silicate Matrices: A Petromimetic Perspective on Structural Ceramic Composites Todd T. King,* ,† Walter Grayeski,‡ and Reid F. Cooper* ,§ Department of Materials Science and Engineering, University of Wisconsin–Madison, Madison, Wisconsin 53706–1595 A petromimetic (geological–analog) approach is applied to the design of alumina-fiber-reinforced glass-ceramic-matrix composites that use a refractory, trioctahedral fluoromica fiber– matrix interphase and feldspar matrixes. Studies of the solidstate reaction couples between these silicate phases are pursued to address the chemical tailorability of the interphase/ matrix interface from an engineering perspective. The minimization of alumina and silica activities within polyphase, feldspar-based matrixes via MgO buffering is shown to be an effective route toward a stable fluoromica interphase. An anorthite–2-vol%-alumina (CaAl2Si2O8 1 a-Al2O3) substrate, chemically buffered with MgO, is shown to exhibit thermodynamic stability against fluorokinoshitalite (BaMg3[Al2Si2]O10F2), up to temperatures potentially as high as 1460°C. The key to the approach is the reduction of alumina activity via the formation of MgAl2O4 spinel. Similarly, the formation of forsterite (Mg2SiO4) stabilizes the mica in contact with matrix compositions otherwise containing excess silica. The cationic interdiffusion between solid-solution feldspars and fluoromicas also is characterized. Coupled interdiffusion of K1 and Si41 in exchange for Ba21 and Al31 was observed between K-Ba solid-solution celsian and the barium-rich solid-solution end-member fluorokinoshitalite at 1300°C. A similar cationic exchange also is observed against the potassium-rich endmember fluorophlogopite (KMg3[AlSi3]O10F2), although in a reverse direction, at temperatures of <1280°C. The interfacial compositions identified via electron microprobe analysis specify one set of local equilibrium conditions from which global ceramic composite equilibrium can be achieved. I. Introduction FIBER-REINFORCED ceramic composites must fulfil two general criteria to be effective structural materials for high-temperature aerobic environments. First, long-term thermochemical stability must be established between each of the ceramic component phases (i.e., fiber, matrix, and interphase) and the oxidizing environment. Second, useful ceramic composite systems must exhibit significant toughness at elevated temperatures and rapid loading rates. Achievement of this second requirement is largely dependent on the debonding characteristics of the fiber/matrix interface. Both of these criteria emphasize the importance of engineering thermodynamically and mechanically functional interfaces or interphases between the fiber and the matrix phase. In considering these interfacial requirements, one realizes that the achievement of effective ceramic composite materials will be predicated on polyphase ceramic equilibria possessing high mechanical-behavior contrast among phases. Therefore, one approach toward the fulfillment of these general design criteria is to consider the wealth of data and understanding in the geologic record: present right below our very feet is evidence of polyphase ceramic equilibria of phases—specifically silicates—with a wide variation of mechanical behavior. The combined disciplines of igneous and metamorphic petrology and geochemistry indicate the plethora of multicomponent silicate and oxide assemblages that exhibit high-temperature stability over geologic time scales. Thus, one can consider the design of environmentally robust, elevatedtemperature ceramic composites from a petromimetic perspective. Silicates possess a vast structural diversity, emanating from the degree of SiOOOSi polymerization, that can provide the necessary mechanical contrast (i.e., intrinsic fracture toughness) to achieve debonding in fiber-reinforced ceramic composites as articulated, for example, by the bimaterial interfacial crackdeflection criterion developed by He and Hutchinson.1 Silicate mechanical behavior ranges from extremely strong but brittle framework silicates (fully polymerized, almost covalently bonded structures) to flexible sheet silicates (phyllosilicates) such as micas and clays (less-polymerized structures that incorporate notably weak ionic or van der Waals bonding on specific crystallographic planes). Thus, a refractory ceramic composite that uses micas as a thin, thermodynamically stable interphase to protect alumina fibers mechanically from cracks in an oxide/silicate matrix becomes a very real possibility. Micas have a tetrahedral-cation-to-oxygen-anion (T:O) ratio of 1:2.5 and can be described by the structural formula A0.5–1 XII B2–3 VI @Al0 –2Si4 –2# IVO10~OH!2 (1) where the Roman-numeral superscripts refer to the cation coordination with O22 and/or OH2 anions, A represents a 12- coordinated (i.e., large-diameter) alkali or alkaline-earth cation (known as the interlayer cation), and B represents a divalent or trivalent octahedral (6-coordinated) cation. The T:O ratio requirement means that the relative amounts of Al and Si on the tetrahedral sites are affected by the specific occupancy of the octahedral and interlayer cation sites; for the above-described formula, with 10 O22 anions, the sum of the tetrahedral Al and Si ions must total 4. Cation-site occupancy defines the adjectives used to describe the micas. For example, the mineral muscovite, which has the formula KAl2[Al1Si3]O10(OH)2, is described as a dioctahedral (i.e., only two of the three available octahedral sites are occupied), trisilicic (i.e., three of the four tetrahedral sites contain Si), true (or flexible; i.e., the interlayer cation is an alkali) mica; the mineral kinoshitalite (BaMg3[Al2Si2]O10(OH)2) is a trioctahedral, disilisic, brittle (i.e., the interlayer cation is an alkaline-earth element) mica. R. A. Condrate—contributing editor Manuscript No. 189258. Received June 30, 1999; approved March 20, 2000. Research supported, in part, by the National Science Foundation, Division of Civil and Mechanical Systems Program in Mechanics and Materials, through Grant No. CMS-9414756. *Member, American Ceramic Society. † Currently with NASA/Goddard Space Flight Center, Greenbelt, MD. ‡ Currently with Intel Corp., Santa Clara, CA. § Author to whom correspondence should be addressed. J. Am. Ceram. Soc., 83 [9] 2287–96 (2000) 2287 journal
2288 ional of the American Cei poat am bient incorporate OH ions in their structure, quartz-bearing (i.e silica-rich) phase assemblages (e. g, granites) silicate compounds at temperatures of the only micas that are found n silica-rich rocks are those which pressure, because of the loss of this incorporate octahedrally coordinated Fe+cations, such as biotite mperatures are too low to be of interest with the Fe2+ end-member being the mineral annite for high-temperature engineering applications. However, by sub-(KFe,[AISi,,o(OH))). Annite and iron-rich biotite are stable in ituting a nonpolar F anion for the polar OH in the trioctahe- quartz-bearing rock because a pure Fe2+ end-member orthopyrox dral micas, the structural stability is improved greatly. Many ene (i.e, FeSiO, does not exist; SiO, will not "attack"the synthetic trioctahedral fluoromicas remain stable at temperatures octahedral Fe2+ cation in the mica. However, enstatite(MgSiOj) f>1200C2,3 which is a Mg2+ end-member orthopyroxene, does exist, which The idea of applying refractory, synthetic fluoromicas as disallows phlogopite and silica to be joined within a shared potential interphases in real fiber-reinforced ceramic composite polyphase stability field ystems originated within the composites group from Corning, (One should note that, although thermodynamic stability of Inc, in the mid-1980s. Those researchers were able to successfully biotite is indicated with silica, making it perhaps of interest as an translate the tough behavior of fluoromicas to an actual ceramic interphase with quartz-based glass-ceramics, it is not possible to omposite system comprised of SiC fibers, a fluoromica inter- prepare a fluorodated, high-temperature polymorph of biotite. A phase, and an alumina-rich glass-ceramic matrix; , howe rystal-chemical phenomenon, articulated as a Fe-F avoidance questions regarding the oxidation resistance of the composite and principle, exists that prohibits the incorporation of large amounts the thermochemical compatibility of the fluoromica/alumina-rich of F anions into iron-bearing trioctahedral micas. matrix interface at high temperatures remained Cooper and Hall initiated the petromimetic approach by an These observations clearly indicate that a silicate assemblage that will be stable against a Mg-bearing trioctahedral(flu examination of the thermodynamics and kinetics of fluoromica oro)mica must be one in which the activities of Al o, and Sio stability when applied in an all-oxide ceramic composite syst Alzo, and asio, respectively)are minimized. At the same time, synthetic fluorophlogopite(KMg,[AISi3JO1,), which is a trioc- the results of Cooper and Hall indicated that the mica is stable tahedral fluoromica capable of withstanding temperatures up to against spinel and forsterite(Mg, SiO4). Relative to making a 1400C in a dry aerobic environment 2 Their research indicated glass-ceramic matrix for use with the Mg-trioctahedral fluoromica that this fluoromica was thermodynamically stable against alumin interphase, one then can ask the following question: does a glass-forming silicate composition exist(which generally requires to-1280C and that, because the reaction with alumina(described a very high silica activity) in which, too, both forsterite and spinel below thermodynamically) was characterized kinetically above this temperature, one also could identify an approach to use a are stable upon crystallization? The geologic record becomes usef fluorophlogopite interphase in an alumina-fiber composite above this temperature (a thin layer of MgAl, O, spinel is required, strate such stability. These rocks contain feldspars(specifically separating the fiber from the mica). We have pursued much work plagioclase, which is a coupled solid-solution of albite(Na [ Alsi,jos) and anorthite(Ca[Al Si2JOs), as well as orthoclase(K[alsi3jos)as mechanical functionality of the fluorophlogopite interphase, as the fully polymerized silicate phase(s) that can be formed easily into tecting alumina from propagating cracks ceramic with a composition that has appropriately low aAl-o, and Thus, the goal of the present work is to complete the petromi- io, values should work well as a composite matrix that is stable with metic approach through the design of a silicate glass-ceramic trioctahedral fluoromica. Of course, setting these activities at the matrix for a fluoromica-interphase/alumina-fiber composite. The appropriate level is easily done by ensuring that the crystallized ass-ceramic matrix contains both spinel and forsterite as condensed interest in glass-ceramics for this purpose is multifaceted but phases (Le aMgAlO, MgSiO,=I). This possibility is easly includes the ease of manufacturing a composite with a matrix that can be consolidated with relative ease in the glass form and then ranged through overall batching of the glass to be made for matrix made more refractory via devitrification, as well as the possibility pplication( the unity-activity requirement demands that only a small of achieving thermodynamic compatibility with the fluoromic amount of a condensed phase be present) Given that an end-member trioctahedral fluoromica represents, Establishing multiphase silicate equilibria with the fluoromica a five-component chemical system(four stoichio- requires an understanding of the nature of the solid-state reactivity metric oxides plus fluorine), the gibbs phase rule suggests that general equilibrium at an arbitrarily specified temperature and iments of Cooper and Hall, as well as of the geologic record of pressure would involve five phases. The petromimetic approac equilibria in mica-bearing rock, demonstrates that the reaction which uses a glass-ceramic composed of feldspar plus small (and, thus, destruction) of fluorophlogopite is driven primarily by amounts of forsterite and spinel, would appear to be"simplified the extraction of octahedral Mg cations to form stable aluminate or silicate compounds An additional consideration in this approach would be the When fluorophlogopite is placed in direct contact with alumina establishment of equilibrium between a specific refractory feldspar (Al,O3), the reaction(at standard state at >1280.C)is described and a specific trioctahedral fluoromica. Because they are more efractory, the alkaline-earth feldspars are of more interest as a composite matrix than are the alkali feldspars. However, whether KMg [AISijJO1oF2(s)+5Al203(s)-3Mg2SiO(s) the use of a brittle(alkaline-earth) fluoromica interphase instead of a true(alkali) fluoromica interphase is desirable for composite +=KAISi,O(s)+:MgAl,O(s)+=KF()+sif(g) application is not clear, based on their relative mechanical re- sponses at elevated temperature. We believe that a solid-solution +=, (g)(2) fluoromica may prove to be most useful mechanically; thus, the nature of the coupled solid-solution behavior between mica and A comparison of this result with that for the breakdown reaction involving a loss of fluorin Intrinsic t thermal eldspar becomes important stable at This experimental investigation, then, has two emphases geared 1470C, fluorophlogop gruently melts at 1400 C)indicates toward understanding the petromimetic approach to alumina-fibe that the formation of m spinel drives the reaction. Simi- reinforced glass-ceramic-matrix composites that use an active larly, the reaction of fluorophlogopite with hi phyllosilicate interphase. First, the thermochemical conditions compounds such as mu by)the specifying phase stability between fluoromica interphases and formation of magnesium silicates. The geologic record also is quite polyphase feldspar-based matrixes will be identified. Elevated- specific about this Phlogopite is never found in natural, temperature solid-state reactions will
Because natural micas incorporate OH2 ions in their structure, they decompose to other silicate compounds at temperatures of 500°–800°C at ambient pressure, because of the loss of this structural water. These temperatures are too low to be of interest for high-temperature engineering applications. However, by substituting a nonpolar F2 anion for the polar OH2 in the trioctahedral micas, the structural stability is improved greatly. Many synthetic trioctahedral fluoromicas remain stable at temperatures of .1200°C.2,3 The idea of applying refractory, synthetic fluoromicas as potential interphases in real fiber-reinforced ceramic composite systems originated within the composites group from Corning, Inc., in the mid-1980s. Those researchers were able to successfully translate the tough behavior of fluoromicas to an actual ceramic composite system comprised of SiC fibers, a fluoromica interphase, and an alumina-rich glass-ceramic matrix;4,5 however, questions regarding the oxidation resistance of the composite and the thermochemical compatibility of the fluoromica/alumina-rich matrix interface at high temperatures remained. Cooper and Hall6 initiated the petromimetic approach by an examination of the thermodynamics and kinetics of fluoromica stability when applied in an all-oxide ceramic composite system based on alumina fibers. The mica used in their study was synthetic fluorophlogopite (KMg3[AlSi3]O10F2), which is a trioctahedral fluoromica capable of withstanding temperatures up to 1400°C in a dry aerobic environment.2 Their research indicated that this fluoromica was thermodynamically stable against alumina to ;1280°C and that, because the reaction with alumina (described below thermodynamically) was characterized kinetically above this temperature, one also could identify an approach to use a fluorophlogopite interphase in an alumina-fiber composite above this temperature (a thin layer of MgAl2O4 spinel is required, separating the fiber from the mica). We have pursued much work demonstrating, at both ambient and elevated temperatures, the mechanical functionality of the fluorophlogopite interphase, as well as that of fluorokinoshitalite (BaMg3[Al2Si2]O10F2), in protecting alumina from propagating cracks.7,8 Thus, the goal of the present work is to complete the petromimetic approach through the design of a silicate glass-ceramic matrix for a fluoromica-interphase/alumina-fiber composite. The interest in glass-ceramics for this purpose is multifaceted but includes the ease of manufacturing a composite with a matrix that can be consolidated with relative ease in the glass form and then made more refractory via devitrification,9 as well as the possibility of achieving thermodynamic compatibility with the fluoromica interphase. Establishing multiphase silicate equilibria with the fluoromica requires an understanding of the nature of the solid-state reactivity of the phyllosilicate. An evaluation of the reaction-couple experiments of Cooper and Hall,6 as well as of the geologic record of equilibria in mica-bearing rock, demonstrates that the reaction (and, thus, destruction) of fluorophlogopite is driven primarily by the extraction of octahedral Mg21 cations to form stable aluminate or silicate compounds. When fluorophlogopite is placed in direct contact with alumina (Al2O3), the reaction (at standard state at .1280°C) is described by KMg3@AlSi3#O10F2~s! 1 1 2 Al2O3~s! 3 5 4 Mg2SiO4~s! 1 3 4 KAlSi2O6~s! 1 1 2 MgAl2O4~s! 1 1 4 KF~l ! 1 1 4 SiF4~ g! 1 1 4 AlF3~ g! (2) A comparison of this result with that for the intrinsic thermal breakdown reaction involving a loss of fluorine (metastable at 1470°C; fluorophlogopite congruently melts at 1400°C) indicates that the formation of MgAl2O4 spinel drives the reaction. Similarly, the reaction of fluorophlogopite with high-silica-activity compounds such as mullite result in (i.e., are driven by) the formation of magnesium silicates. The geologic record also is quite specific about this point. Phlogopite is never found in natural, quartz-bearing (i.e., silica-rich) phase assemblages (e.g., granites); the only micas that are found in silica-rich rocks are those which incorporate octahedrally coordinated Fe21 cations, such as biotite, with the Fe21 end-member being the mineral annite (KFe3[AlSi3]O10(OH)2). Annite and iron-rich biotite are stable in quartz-bearing rock because a pure Fe21 end-member orthopyroxene (i.e., FeSiO3) does not exist; SiO2 will not “attack” the octahedral Fe21 cation in the mica. However, enstatite (MgSiO3), which is a Mg21 end-member orthopyroxene, does exist, which disallows phlogopite and silica to be joined within a shared polyphase stability field. (One should note that, although thermodynamic stability of biotite is indicated with silica, making it perhaps of interest as an interphase with quartz-based glass-ceramics, it is not possible to prepare a fluorodated, high-temperature polymorph of biotite. A crystal-chemical phenomenon, articulated as a Fe–F avoidance principle,10 exists that prohibits the incorporation of large amounts of F2 anions into iron-bearing trioctahedral micas.) These observations clearly indicate that a silicate assemblage that will be stable against a Mg21-bearing trioctahedral (fluoro)mica must be one in which the activities of Al2O3 and SiO2 (aAl2O3 and aSiO2 , respectively) are minimized. At the same time, the results of Cooper and Hall6 indicated that the mica is stable against spinel and forsterite (Mg2SiO4). Relative to making a glass-ceramic matrix for use with the Mg-trioctahedral fluoromica interphase, one then can ask the following question: does a glass-forming silicate composition exist (which generally requires a very high silica activity) in which, too, both forsterite and spinel are stable upon crystallization? The geologic record becomes useful once again: aluminum-bearing, silica-undersaturated basalts demonstrate such stability. These rocks contain feldspars (specifically plagioclase, which is a coupled solid-solution of albite (Na[AlSi3]O8) and anorthite (Ca[Al2Si2]O8), as well as orthoclase (K[AlSi3]O8)) as the fully polymerized silicate phase(s) that can be formed easily into a glass via fusion and subsequent cooling. Thus, a feldspar glassceramic with a composition that has appropriately low aAl2O3 and aSiO2 values should work well as a composite matrix that is stable with a trioctahedral fluoromica. Of course, setting these activities at the appropriate level is easily done by ensuring that the crystallized glass-ceramic matrix contains both spinel and forsterite as condensed phases (i.e., aMgAl2O4 5 aMg2SiO4 5 1). This possibility is easily arranged through overall batching of the glass to be made for matrix application (the unity-activity requirement demands that only a small amount of a condensed phase be present). Given that an end-member trioctahedral fluoromica represents, at a minimum, a five-component chemical system (four stoichiometric oxides plus fluorine), the Gibbs phase rule suggests that general equilibrium at an arbitrarily specified temperature and pressure would involve five phases. The petromimetic approach, which uses a glass-ceramic composed of feldspar plus small amounts of forsterite and spinel, would appear to be “simplified” to a four-phase equilibrium. An additional consideration in this approach would be the establishment of equilibrium between a specific refractory feldspar and a specific trioctahedral fluoromica. Because they are more refractory, the alkaline-earth feldspars are of more interest as a composite matrix than are the alkali feldspars.11 However, whether the use of a brittle (alkaline-earth) fluoromica interphase instead of a true (alkali) fluoromica interphase is desirable for composite application is not clear, based on their relative mechanical responses at elevated temperature.8 We believe that a solid-solution fluoromica may prove to be most useful mechanically; thus, the nature of the coupled solid-solution behavior between mica and feldspar becomes important. This experimental investigation, then, has two emphases geared toward understanding the petromimetic approach to alumina-fiberreinforced glass-ceramic-matrix composites that use an active phyllosilicate interphase. First, the thermochemical conditions specifying phase stability between fluoromica interphases and polyphase feldspar-based matrixes will be identified. Elevatedtemperature solid-state reactions will be conducted between a 2288 Journal of the American Ceramic Society—King et al. Vol. 83, No. 9
tember 2000 Thermochemical Reactions and equilibria between Fluoromicas and Silicate Matrices two-phase mixture of anorthite and alumina(CaAl2Si2O8 analysis of the Mgo-buffered substrates revealed the presence of c-Al,O3, which is a well-studied, refractory, glass-ceramic matrix fine (s5 um)spinel particles comprising 2 vol% of the total phase) and fluorokinoshitalite(which is capable of withstanding anorthite matrix. temperatures up to -1460.C), to elucidate the specific thermo- chemical reactions occurring at the interface. These results will be ompared with the interfacial response observed between a MgO. ( Feldspar Preparation for Interdiffusion Study buffered, polyphase, anorthite-based matrix and fluorokinoshitalite a different glass-forming feldspar composition was used as the substrates. Thus, effective tactics for stability via Mgo buffering matrix substrate for the cationic interdiffusion studies. The specific will be established. Second. the cationic interdiffusion exhibited composition, based on the feldspar mineral celsian( BaAl2Si2O between(K, Ba)solid-solution feldspars and(K, Ba)solid-solution consists of a monoclinic(Ba, Sr, K) solid-solution phase, hereafter referenced as stabilized celsian (KoaBao sSroslo6AlL6Si24Os) fluoromicas will be examined in detail, to elucidate both the Sr+ cations are added to BaAl, Si,O% because glass compositions thermodynamic and kinetic responses at the interface. Character ization of the local equilibrium conditions that exist at these of 1Bao-1Al2O3 2SiO,(molar basis) crystallize into the high- interfaces may target specific matrix and interphase compositions emperature sian (hexane from which global thermodynamic equilibrium may be established BaAl2SiOs) 3, Hexacelsian is avoided in many glass-ceramic high-temperature, polyphase ceramic composites. In addition, a applications ily because of its large anisotropy in thermal thermodynamic model will be applied to characterize the under expansion coefficient, which causes substantial thermal-shock lying kinetic mechanisms problems, but also because of the existence of a low-temperature (300C) phase transformation from B-hexacelsian to metastable orthorhombic a-celsian. This B-c transformation involves a large Il. Experimental Approach volume decrease(0.3 vol%), which would pose serious difficul (1 Finoromica Preparation Methods ties for a hexacelsian matrix during the temperature cycling experienced in both fiber-composite fabrication and applicati The two fluoromica materials examined in this reaction study However the substitution of Sr+ cations for Ba+ cations in the were the end-member compositions of the trioctahedral fluoromica original glass composition facilitates the crystallization and stabil- solid-solution series [ Ba KL-yMg3[Al+Si3-JJOJoF2, where 0s ity of the desired monoclinic feldspar structure S x s l. fluorokinoshitalite, which is the barium-rich (x= 1) Reagent-grade oxide powders were mixed in the appropriate end-member, and fluorophlogopite, which is the potassium-rich proportions to achieve an overall stabilized-celsian composition (x=O)end-member, were prepared using analytical-reagent-grade such as that previously noted. Then, the mixed-oxide batch was xide, carbonate, and fluoride powders, mixed to produce stoichi- melted, quenched, and pulverized, following the same procedures ometric fluoromica compositions, plus 13.5 wt% additional fluo- as the above-described anorthite-based compositions. The final rine. The excess fluorine(added via the substitution of MgF, for article size of the stabilized-celsian glass powder was <10 um Mgo)compensates for the predictable halogen loss that occurs The glass powder precursor then was cold-pressed into pellets and during melting and homogenization. Consequently, the batch heat-treated at temperatures of 1240-1300C for 3-6h in ambient omposition is oxygen-poor; however, reaction with the melting environment supplies the oxygen required for mica formation air, to promote both particle sintering and glass crystallization. Powder XRD analysis of the dense substrates confirmed a fully Both batches were melted individually in a covered platinum crystalline, single-phase, stabilized-celsian matrix that possessed crucible at 1450C for 5 h in air. The resultant melt was poured clinic structure into a coherent patty, which crystallized to coarse-grained mica Then, the large mica crystals were pulverized, nergy mill, into a fine crystalline powder that had a (4) Reaction-Couple Experiments ze of 5 um. Reaction-couple specimens were produced for both the MgO- buffering and the (2) Feldspar Preparation for Magnesia-Buffering Study old-pressed, crystalline fluoromica pellet between the polished The starting polyphase, feldspar-based matrix composition used surfaces(I-um diamond grit) of two fully crystalline feldspa for the MgO-buffering study consisted of a two-phase mixture of pellets of interest. Then, the oxide sandwich was hot-pressed anorthite(CaAl2Si2O)and alumina(Al2O3) Oxide and carbonate uniaxially(l MPa) at 1200-1400C for 10-24 h in a flowing powders were mixed in the appropriate proportions to produce a starting bulk composition of anorthite that contained -3 wt% mens were heated at a rate of 200% C/h and furnace-cooled at a rate alumina. The mixed-oxide batch was melted in a platinum crucible of~400°Ch at 1650C for 10-12 h in air and then quenched by pouring The fully bonded specimens were carefully cross-sectioned directly into water. The resultant glass cullet was pulverized to a perpendicular to the fluoromica/feldspar interface with a low final particle size of <10 um. To produce the buffered anorthite speed diamond saw. Then, the surfaces of interest were ground and apositions, the previously described alumina-rich anorthite polished to a 1-um finish. After a thin(20 nm)coating of carbon der was combined with -5 wt% reagent-grade, crystal- was deposited onto the polished surface, the thermochemical O powder in a mortar and pestle. The two powders were response of the interface was characterized morphologically(using oined several times with acetone to ensure adequate mixing. microscopy(SEM)and chemically(using EDS and wavelength. powder precursors then were cold-pressed into pellet form and dispersive X-ray spectroscopy (WDS) rocessed at 1350@C for 6 h in air to sinter and fully crystallize the pellets into the desired polyphase, feldspar-based matrix substrate Although powder X-ray diffraction(XRD)analysis of the sintered Ill. Experimental Results: Description and Comment pellets confirmed the triclinic structure of fully crystalline anor- 1) Buffering with Magnesia: An Approach toward a thite, it was unable to resolve the presence of Al,O, in the Stable Fluoromica Interphase unbuffered substrates or spinel in the MgO-buffered substrates Figure I(a) is a BEl micrograph of the interfacial reaction because of their occurrence in minute concentrations. when the between the anorthite-2-vol%alumina aggregate and fluoroki- ombined techniques of backscattered electron (BED)and noshitalite processed at 1200 contrast in this imaging energy-dispersive X-ray spectroscopy(EDs) however technique is due to differences of the small(s5 um), finely dispersed particles of metric alu- atoms comprising the solid; bri ari hter cot directl were observed in the unbuffered represent holes or pores in the specimen. Thug es Micro- to the presence of heavier elements ural point-counting techniques ind final bulk compo- 98 vol% anorthite and fluorokinoshitalite appears brighter than the an
two-phase mixture of anorthite and alumina (CaAl2Si2O8 1 a-Al2O3, which is a well-studied, refractory, glass-ceramic matrix phase) and fluorokinoshitalite (which is capable of withstanding temperatures up to ;1460°C2 ), to elucidate the specific thermochemical reactions occurring at the interface. These results will be compared with the interfacial response observed between a MgObuffered, polyphase, anorthite-based matrix and fluorokinoshitalite substrates. Thus, effective tactics for stability via MgO buffering will be established. Second, the cationic interdiffusion exhibited between (K,Ba) solid-solution feldspars and (K,Ba) solid-solution fluoromicas will be examined in detail, to elucidate both the thermodynamic and kinetic responses at the interface. Characterization of the local equilibrium conditions that exist at these interfaces may target specific matrix and interphase compositions from which global thermodynamic equilibrium may be established in high-temperature, polyphase ceramic composites. In addition, a thermodynamic model will be applied to characterize the underlying kinetic mechanisms. II. Experimental Approach (1) Fluoromica Preparation Methods The two fluoromica materials examined in this reaction study were the end-member compositions of the trioctahedral fluoromica solid-solution series [BaxK12x]Mg3[Al11xSi32x]O10F2, where 0 # x # 1. Fluorokinoshitalite, which is the barium-rich (x 5 1) end-member, and fluorophlogopite, which is the potassium-rich (x 5 0) end-member, were prepared using analytical-reagent-grade oxide, carbonate, and fluoride powders, mixed to produce stoichiometric fluoromica compositions, plus 13.5 wt% additional fluorine. The excess fluorine (added via the substitution of MgF2 for MgO) compensates for the predictable halogen loss that occurs during melting and homogenization.12 Consequently, the batch composition is oxygen-poor; however, reaction with the melting environment supplies the oxygen required for mica formation. Both batches were melted individually in a covered platinum crucible at 1450°C for 5 h in air. The resultant melt was poured into a coherent patty, which crystallized to coarse-grained mica during cooling. Then, the large mica crystals were pulverized, using a fluid-energy mill, into a fine crystalline powder that had a mean particle size of ;5 mm. (2) Feldspar Preparation for Magnesia-Buffering Study The starting polyphase, feldspar-based matrix composition used for the MgO-buffering study consisted of a two-phase mixture of anorthite (CaAl2Si2O8) and alumina (Al2O3). Oxide and carbonate powders were mixed in the appropriate proportions to produce a starting bulk composition of anorthite that contained ;3 wt% alumina. The mixed-oxide batch was melted in a platinum crucible at 1650°C for 10–12 h in air and then quenched by pouring directly into water. The resultant glass cullet was pulverized to a final particle size of ,10 mm. To produce the buffered anorthite compositions, the previously described alumina-rich anorthite glass powder was combined with ;5 wt% reagent-grade, crystalline MgO powder in a mortar and pestle. The two powders were recombined several times with acetone to ensure adequate mixing. Both the unbuffered glass and MgO-buffered glass/crystalline powder precursors then were cold-pressed into pellet form and processed at 1350°C for 6 h in air to sinter and fully crystallize the pellets into the desired polyphase, feldspar-based matrix substrate. Although powder X-ray diffraction (XRD) analysis of the sintered pellets confirmed the triclinic structure of fully crystalline anorthite, it was unable to resolve the presence of Al2O3 in the unbuffered substrates or spinel in the MgO-buffered substrates, because of their occurrence in minute concentrations. When the combined techniques of backscattered electron imaging (BEI) and energy-dispersive X-ray spectroscopy (EDS) were used, however, small (#5 mm), finely dispersed particles of stoichiometric alumina were observed in the unbuffered anorthite substrates. Microstructural point-counting techniques indicated a final bulk composition of ;98 vol% anorthite and ;2 vol% alumina. Similar analysis of the MgO-buffered substrates revealed the presence of fine (#5 mm) spinel particles comprising ;2 vol% of the total anorthite matrix. (3) Feldspar Preparation for Interdiffusion Study A different glass-forming feldspar composition was used as the matrix substrate for the cationic interdiffusion studies. The specific composition, based on the feldspar mineral celsian (BaAl2Si2O8), consists of a monoclinic (Ba,Sr,K) solid-solution phase, hereafter referenced as stabilized celsian (K0.4[Ba0.5Sr0.5]0.6Al1.6Si2.4O8). Sr21 cations are added to BaAl2Si2O8 because glass compositions of 1BaOz1Al2O3z2SiO2 (molar basis) crystallize into the hightemperature, hexagonal polymorph of celsian (hexacelsian, b-BaAl2Si2O8).13,14 Hexacelsian is avoided in many glass-ceramic applications, primarily because of its large anisotropy in thermal expansion coefficient, which causes substantial thermal-shock problems, but also because of the existence of a low-temperature (;300°C) phase transformation from b-hexacelsian to metastable, orthorhombic a-celsian. This b–a transformation involves a large volume decrease (;0.3 vol%), which would pose serious difficulties for a hexacelsian matrix during the temperature cycling experienced in both fiber–composite fabrication and application.13 However, the substitution of Sr21 cations for Ba21 cations in the original glass composition facilitates the crystallization and stability of the desired monoclinic feldspar structure.15 Reagent-grade oxide powders were mixed in the appropriate proportions to achieve an overall stabilized-celsian composition, such as that previously noted. Then, the mixed-oxide batch was melted, quenched, and pulverized, following the same procedures as the above-described anorthite-based compositions. The final particle size of the stabilized-celsian glass powder was ,10 mm. The glass powder precursor then was cold-pressed into pellets and heat-treated at temperatures of 1240°–1300°C for 3–6 h in ambient air, to promote both particle sintering and glass crystallization. Powder XRD analysis of the dense substrates confirmed a fully crystalline, single-phase, stabilized-celsian matrix that possessed the monoclinic structure. (4) Reaction-Couple Experiments Reaction-couple specimens were produced for both the MgObuffering and the interdiffusion studies by sandwiching one cold-pressed, crystalline fluoromica pellet between the polished surfaces (1-mm diamond grit) of two fully crystalline feldspar pellets of interest. Then, the oxide sandwich was hot-pressed uniaxially (;1 MPa) at 1200°–1400°C for 10–24 h in a flowing dry-argon environment (flow rate of ;10 cm3 /min). The specimens were heated at a rate of 200°C/h and furnace-cooled at a rate of ;400°C/h. The fully bonded specimens were carefully cross-sectioned perpendicular to the fluoromica/feldspar interface with a lowspeed diamond saw. Then, the surfaces of interest were ground and polished to a 1-mm finish. After a thin (;20 nm) coating of carbon was deposited onto the polished surface, the thermochemical response of the interface was characterized morphologically (using secondary electron imaging (SEI) and BEI in scanning electron microscopy (SEM)) and chemically (using EDS and wavelengthdispersive X-ray spectroscopy (WDS)). III. Experimental Results: Description and Comment (1) Buffering with Magnesia: An Approach toward a Stable Fluoromica Interphase Figure 1(a) is a BEI micrograph of the interfacial reaction between the anorthite–2-vol%-alumina aggregate and fluorokinoshitalite processed at 1200°C. The contrast in this imaging technique is due to differences in the atomic number (Z) of the atoms comprising the solid; brighter contrast corresponds directly to the presence of heavier elements. Features that appear black represent holes or pores in the specimen. Thus, in this image, fluorokinoshitalite appears brighter than the anorthite–2 vol% September 2000 Thermochemical Reactions and Equilibria between Fluoromicas and Silicate Matrices 2289
yournal of the American Ceramic Sociery-King et al. (a) than either fluorokinoshitalite or anorthite-2 vol%o alumina sub- sequent WDS analysis of this high-Z layer revealed a composition that corresponded to stoichiometric celsian(i.e, BaAl,Si,Os). At fluorokinoshitalite 1286 C. the reaction intensified and resulted in a wider reaction zone (-60 Hm) that consisted of stoichiometric celsian, spinel, and anorthite(Fig. 1(b). The presence of spinel confirms that the reaction at the interface is driven by the combination of MgO from fluorokinoshitalite and excess Al2O, from the anorthite pellet, as shown in the following two-step reaction BaMg.[Al SigJO1oF2+5O2(g)- BaAl2Si2O8 anorthite +2 vol% alumina 3Mgo+ f, (g) MgO+Al2O3→MgAl2O4 4) Fluorokinoshitalite is not in thermodynamic equilibrium with anorthite-2 vol% alumina at 1200%C In an attempt to minimize the destruction of the fluoromica phase via the depletion of Mg cations, a MgO-buffered, (b)2 anorthite-2-vol%alumina pellet (now alumina +2 vol% spinel) fluorokinoshitalite was reacted against fluorokinoshitalite at 1200%C. The interfacial illustrated Fig. I(c). Although the fluorokinoshitalite/MgO-buffered anorthite interface was still slightly reactive(governed by eqs. (3)and (4), the reaction was significantly reduced (compare Fig. I(a)), as illustrated by the now-thinner (-2 Hm thick), discontinuous celsian layer. The arrows highlight regions where interfacial celsian formation oc- curred. The generation of smaller quantities of product phases (i.e celsian) clearly establishes the effectiveness of Mgo buffering in omoting stable trioctahedral fluoromica/anorthite +2 vol% umina interfaces by arresting the loss of Mgt cations from the phyllosilicate. Complete minimization of the interfacial reaction in this case could be accomplished by increasing the amount of Mgo anorthite+2 vol% alumina additions to the anorthite-2-vol%-alumina matrix and/or increas- ing the high-temperature annealing time of the buffered feldspar 48 HIRONS let, to ensure that all free alumina has reacted to spinel. These results also are applicable to silica-rich (i.e, anorthite + minor hase-pure silica) matrix compositions: for Mgo buffering in this case, the silica is combined with the mgo to form forsterite thereby substantially diminishing the silica activity in the matrix. Combining the ideas, it is important to note that the stoichiometric Mgo-buffered anorthite MgAl, O, is in equilibrium with forsterite: thus, a two-phase pinel forsterite buffer within the feldspar matrix should stabilize the Mg+-trioctahedral mica completely (2) Cationic Interdiffusion: Equilibrium between Multicomponent Feldspar and Fluoromica Figure 2 shows a BEl image of a stabilized-celsian/fluoroki- noshitalite interface processed at 1300C. Although the reaction couple split along the interface during the final stages of polishing, fluorkkinoshitalite as evidenced by the -20-pm-wide interfacial crack, the separated bstrates still illustrate the entire thermochemical response at 1300%C. One can observe that the contrast in the mica darkens ithin -100 um of the interface, whereas that in the feldspar brighter in the first -50 um beyond the interface. In general, the result indicates an increase in the concentration of high-z elements Fig. 1. B El micrographs illustrating the morphology of reaction interfaces into the celsian and their loss from the mica. The only exceptions to this trend are large, more brightly contrasted fluorokinoshitalite 2 vol% alumina/fluorokinoshitalite reaction at 1200C (the primary grains that are oriented with their basal planes parallel to the reaction-induced feature consists of a single, continuous layer of stoichiometric reaction-couple interface, clearly demonstrating the impact of the sian, 6 um thick),(b) anisotropic fluoromica crystal structure on the interfacial response. on at 1286.C(the reaction in Fig. 1(a) intensifies, producing product These observed compositional trends across the interface are metric celsian, anorthite, and spinel), and (c) Mgo-buftered indicative of an interdiffusional process between the feldspar and ite/filuorokinoshitalite reaction at 1200C(notice the les-severe, discon- the fluoromica. Electron microprobe analysis of a bonded section nature of the celsian layer, highlighted by the arrows, in comparison he interface without Mgo additions(see Fig. 1(a))) of the same sample identifies the diffusing ionic species as K Si, Ba, and Al, which are displayed in a WDS concentra- tion profile across the interface(see Fig. 3(a)). Although the diffusional anisotropy of the fluoromica results in considerable alumina. The brightest feature in Fig. 1(a) is a continuous, variation in the X-ray signals (particularly in the 5 um section "6-um-thick interfacial layer that corresponds to a phase com- ent to the original interface on the fluorokinoshitalite side, the posed of a relatively higher concentration of heavy- elements than-expected K+ concentration and higher-than-expected
alumina. The brightest feature in Fig. 1(a) is a continuous, ;6-mm-thick interfacial layer that corresponds to a phase composed of a relatively higher concentration of heavy-Z elements than either fluorokinoshitalite or anorthite–2 vol% alumina. Subsequent WDS analysis of this high-Z layer revealed a composition that corresponded to stoichiometric celsian (i.e., BaAl2Si2O8). At 1286°C, the reaction intensified and resulted in a wider reaction zone (;60 mm) that consisted of stoichiometric celsian, spinel, and anorthite (Fig. 1(b)). The presence of spinel confirms that the reaction at the interface is driven by the combination of MgO from fluorokinoshitalite and excess Al2O3 from the anorthite pellet, as shown in the following two-step reaction: BaMg3@Al2Si2#O10F2 1 1 2 O2~ g! 3 BaAl2Si2O8 1 3MgO 1 F2~ g! (3) MgO 1 Al2O3 3 MgAl2O4 (4) Fluorokinoshitalite is not in thermodynamic equilibrium with anorthite–2 vol% alumina at 1200°C. In an attempt to minimize the destruction of the fluoromica phase via the depletion of Mg21 cations, a MgO-buffered, anorthite–2-vol%-alumina pellet (now alumina 1 2 vol% spinel) was reacted against fluorokinoshitalite at 1200°C. The interfacial response is illustrated as a BEI image in Fig. 1(c). Although the fluorokinoshitalite/MgO-buffered anorthite interface was still slightly reactive (governed by Eqs. (3) and (4)), the reaction was significantly reduced (compare Fig. 1(a)), as illustrated by the now-thinner (;2 mm thick), discontinuous celsian layer. The arrows highlight regions where interfacial celsian formation occurred. The generation of smaller quantities of product phases (i.e., celsian) clearly establishes the effectiveness of MgO buffering in promoting stable trioctahedral fluoromica/anorthite 1 2 vol% alumina interfaces by arresting the loss of Mg21 cations from the phyllosilicate. Complete minimization of the interfacial reaction in this case could be accomplished by increasing the amount of MgO additions to the anorthite–2-vol%-alumina matrix and/or increasing the high-temperature annealing time of the buffered feldspar pellet, to ensure that all free alumina has reacted to spinel. These results also are applicable to silica-rich (i.e., anorthite 1 minor phase-pure silica) matrix compositions: for MgO buffering in this case, the silica is combined with the MgO to form forsterite, thereby substantially diminishing the silica activity in the matrix. Combining the ideas, it is important to note that the stoichiometric MgAl2O4 is in equilibrium with forsterite:16 thus, a two-phase spinel 1 forsterite buffer within the feldspar matrix should stabilize the Mg21-trioctahedral mica completely. (2) Cationic Interdiffusion: Equilibrium between Multicomponent Feldspar and Fluoromica Figure 2 shows a BEI image of a stabilized-celsian/fluorokinoshitalite interface processed at 1300°C. Although the reaction couple split along the interface during the final stages of polishing, as evidenced by the ;20-mm-wide interfacial crack, the separated substrates still illustrate the entire thermochemical response at 1300°C. One can observe that the contrast in the mica darkens within ;100 mm of the interface, whereas that in the feldspar is brighter in the first ;50 mm beyond the interface. In general, the result indicates an increase in the concentration of high-Z elements into the celsian and their loss from the mica. The only exceptions to this trend are large, more brightly contrasted fluorokinoshitalite grains that are oriented with their basal planes parallel to the reaction-couple interface, clearly demonstrating the impact of the anisotropic fluoromica crystal structure on the interfacial response. These observed compositional trends across the interface are indicative of an interdiffusional process between the feldspar and the fluoromica. Electron microprobe analysis of a bonded section of the same sample identifies the diffusing ionic species as K1, Si41, Ba21, and Al31, which are displayed in a WDS concentration profile across the interface (see Fig. 3(a)). Although the diffusional anisotropy of the fluoromica results in considerable variation in the X-ray signals (particularly in the 5 mm section adjacent to the original interface on the fluorokinoshitalite side; the lower-than-expected K1 concentration and higher-than-expected Fig. 1. B EI micrographs illustrating the morphology of reaction interfaces between fluorokinoshitalite and anorthite-based glass-ceramics ((a) anorthite 1 2 vol% alumina/fluorokinoshitalite reaction at 1200°C (the primary reaction-induced feature consists of a single, continuous layer of stoichiometric celsian, 6 mm thick), (b) anorthite 1 2 vol% alumina/fluorokinoshitalite reaction at 1286°C (the reaction in Fig. 1(a) intensifies, producing product phases of stoichiometric celsian, anorthite, and spinel), and (c) MgO-buffered anorthite/fluorokinoshitalite reaction at 1200°C (notice the less-severe, discontinuous nature of the celsian layer, highlighted by the arrows, in comparison with the interface without MgO additions (see Fig. 1(a))). 2290 Journal of the American Ceramic Society—King et al. Vol. 83, No. 9
September 2000 Thermochemical Reactions and equilibria between Fluc The interfacial compositions of the stabilized celsian and fluorokinoshitalite establish the thermochemical conditions for one of perhaps several two-phase equilibria. Affected by the use of pecific starting compositions and the diffusion dynamics, many fluorokinoshitalite equilibrium tie lines could exist between a Ba-K solid-solution feldspar and a Ba-K solid-solution mica. Tables I and ll display the experimentally measured interfacial compositions of all atomic species in the system at 1300C (all compositional analysis was conducted at interfacial regions that showed minimal effects of mica anisotropy). Experimentally measured bulk compositions interface separation also are presented for comparison. Although the response observe at 1200 C was much less dramatic(interdiffusion of K+and Si+ for Ba- and Al extended away from the original interface only 40 um into the mica and 25 um into the feldspar), the nterfacial compositions, shown in Tables Ill and Iv, are quite similar to those measured at 1300C. The apparent insensitivity of the interfacial compositions to the processing temperature suggests stabilized celsian that thermochemical equilibrium can be established over a wide temperature range with a narrow range of compositions, despite the non-unique compositional conditions that generally describe a two-phase solid-solution equilibrium. 24,25 Knowledge of these thermochemical conditions introduces the opportunity for the design of globally stable fluoromica interphase/silicate matrix interfaces Fig. 2. BEI micrograph of a stabilized celsian/fluorokinoshitalite reaction K(+Sr)-stabilized celsian and fluorokinoshitalite are not stable interface processed at 1300.C. The fluorokinoshitalite substrate separated against each other at 1400C. Fine particles of stoichiometric from the stabilized celsian substrate on final polishing. Gradual contrast celsian(BaAl, Si,O,) and spinel(MgAl, 04) were the only addi- hanges in both phases( the brighter phase is celsian, and the less-bright tional phases observed along the interface. In addition, large(2 placing Ba+. Fluorokinoshitalite grains oriented with basal planes um)bubbles were produced along the exterior edges of the rallel to the interface show exceptionally bright contrast, in comparison fluoromica substrate, indicating the volatilization of gaseous flu- to grains oriented perpendicular to the interface, indicating that these orin(F2). The formation of celsian and spinel phases, as well as former grains do not participate easily in the ion exchange gas bubbles, strongly suggests that interfacial reaction occurred via the combination of MgO from the fluoromica with slight alumina excesses from the feldspar-based matrix, as articulated by eqs. 3) Ba2+ concentration correspond to a fluoromica grain that was and(4). These results again suggest that Mgo buffering may be aligned parallel to the interface), the coupled nature of this process highly effective in achieving thermodynamic equilibrium between C g represented clearly by the simultaneous diffusion of K and alumina-rich stabilized celsian and fluorokinoshitalite. even at cations into the fluorol ations into the feldspar. Charge neutrality of the interdiffusing species is The thermochemical response between stabilized celsian and maintained, des p a concentration gradientof the Al+ species. Interdiffusion Figure 4 shows micrographs of such an interface that has been extended out as far as 100 um away from the interface in the processed at 1300%C. The gradual change in contrast within the mica and -50 um in the feldspar at this temperature, which is fluorophlogopite, visible in the bEl image in Fig. 4(a), again consistent with the information available from BEl (Fig. 2) indicates compositional gradients across the interface, however, in Furthermore, the coupled cationic exchange between these two this case, high-Z elements increase in concentration toward the materials is isomorphous; i.e., both phases are stable against each interface. EDS linescans(Fig. 5)confirm the interdiffusion of the other, but their compositions change. As illustrated in Fig 3(b), the K+ species from the mica and the Ba2+ species from the feldspar mpositional profiles of both F and Mg- cations remain (the coupled counterparts, Si" exchange for Al, although not constant throughout the fluoromica at levels consistent with bulk thown in this figure also were observed). Similar to the reaction fluorokinoshitalite compositions, but then decrease abruptly to between stabilized celsian and fluorokinoshitalite, uphill diffusio zero at the original stabilized -celsian/fluorokinoshitalite interface observed again, except this time for the Si species demonstrating the phase stability of the fluoromica On closer examination of the 1300%C reaction, fine particles A sharp change in Sr+ concentration at the interface also is (<5 um) of stoichiometric spinel (MgAl,O4)and leucite evident in Fig. 3(b), which suggests that the mobility of St (K[AISizJO,)were observed along the interface via WDS analysis cations is much slower than that of Ba- cations in the mica(i.e, Figure 4(b) shows a SEl image of the reaction, where the dark D5r+< DBa+, where D is the intrinsic(self-)diffusion coeffi- regions correspond to spinel. The presence of these two specific cient; the subscript represents the ionic species, and the superscript product phases confirms that the stabilized celsian contains finite represents the phase). Therefore, the cationic exchange between amounts of excess alumina; thus, the interfacial reaction is driven the Ba2+ species from the fluoromica and the Sr+ species from by the depletion and subsequent combination of Mg2+ cations the stabilized celsian is not a dominant diffusional mechanism and from the mica to form spinel, as described in Eq. (2). At reaction the interfacial is addressed almost exclusively via the temperatures of <1280C, the alumina-bearing stabilized-celsian/ upled interdiffusion of K+ and Si+ for Ba2+ and AP+. This fluorophlogopite interface remained stable-no breakdown of the particular exchange vector has been reported to occur in the mica because of reaction with the alumina was observed, which is celsian-adularia(KAISi,Os)feldspar series, -20 as well as in the consistent with the equilibrium articulated in Eq (2) atural kinoshitalite-phlogopite mica series.--- Both the The influence of the anisotropic fluoromica structure on the feldspars (BaI-KJAl os) and the micas interdiffusion process again is evidenced clearly along the stabi- (Ba,KI- Mg3[All+ Si3_,o(OH))exhibit complete isomor- lized-celsian/fluorophlogopite interface. The bicontrasted grain phous solid solution over their entire compositional ranges (i.e highlighted by the arrow in Fig. 4(a)illustrates the ease of where 0 s x s 1). Therefore, it is not surprising that diffusion along fluoromica basal planes than in th heterogeneous solid-solution equilibrium exists between stabilized direction. The brighter contrast entering from the A schematic representation of this layer edges of the fluoromica represent areas where the Ba+ Fig. 3(a) cation has been successfully exchanged for the k cation. The dark
Ba21 concentration correspond to a fluoromica grain that was aligned parallel to the interface), the coupled nature of this process is represented clearly by the simultaneous diffusion of K1 and Si41 cations into the fluoromica and of Ba21 and Al31 cations into the feldspar. Charge neutrality of the interdiffusing species is maintained, despite the required “uphill diffusion” (i.e., diffusion up a concentration gradient) of the Al31 species. Interdiffusion extended out as far as ;100 mm away from the interface in the mica and ;50 mm in the feldspar at this temperature, which is consistent with the information available from BEI (Fig. 2). Furthermore, the coupled cationic exchange between these two materials is isomorphous; i.e., both phases are stable against each other, but their compositions change. As illustrated in Fig. 3(b), the compositional profiles of both F2 and Mg21 cations remain constant throughout the fluoromica at levels consistent with bulk fluorokinoshitalite compositions, but then decrease abruptly to zero at the original stabilized-celsian/fluorokinoshitalite interface, demonstrating the phase stability of the fluoromica. A sharp change in Sr21 concentration at the interface also is evident in Fig. 3(b), which suggests that the mobility of Sr21 cations is much slower than that of Ba21 cations in the mica (i.e., DSr21 mica ,, DBa21 mica , where D is the intrinsic (self-)diffusion coefficient; the subscript represents the ionic species, and the superscript represents the phase). Therefore, the cationic exchange between the Ba21 species from the fluoromica and the Sr21 species from the stabilized celsian is not a dominant diffusional mechanism and the interfacial response is addressed almost exclusively via the coupled interdiffusion of K1 and Si41 for Ba21 and Al31. This particular exchange vector has been reported to occur in the celsian–adularia (KAlSi3O8) feldspar series,17–20 as well as in the natural kinoshitalite–phlogopite mica series.21–23 Both the feldspars ([Ba12xKx]Al22xSi21xO8) and the micas ([BaxK12x]Mg3[Al11xSi32x]O10(OH)2) exhibit complete isomorphous solid solution over their entire compositional ranges (i.e., where 0 # x # 1). Therefore, it is not surprising that a heterogeneous solid-solution equilibrium exists between stabilized celsian and fluorokinoshitalite. A schematic representation of this diffusion process is shown in Fig. 3(a). The interfacial compositions of the stabilized celsian and fluorokinoshitalite establish the thermochemical conditions for one of perhaps several two-phase equilibria. Affected by the use of specific starting compositions and the diffusion dynamics, many equilibrium tie lines could exist between a Ba-K solid-solution feldspar and a Ba-K solid-solution mica. Tables I and II display the experimentally measured interfacial compositions of all atomic species in the system at 1300°C (all compositional analysis was conducted at interfacial regions that showed minimal effects of mica anisotropy). Experimentally measured bulk compositions also are presented for comparison. Although the response observed at 1200°C was much less dramatic (interdiffusion of K1 and Si41 for Ba21 and Al31 extended away from the original interface only ;40 mm into the mica and ;25 mm into the feldspar), the interfacial compositions, shown in Tables III and IV, are quite similar to those measured at 1300°C. The apparent insensitivity of the interfacial compositions to the processing temperature suggests that thermochemical equilibrium can be established over a wide temperature range with a narrow range of compositions, despite the non-unique compositional conditions that generally describe a two-phase solid-solution equilibrium.24,25 Knowledge of these thermochemical conditions introduces the opportunity for the design of globally stable fluoromica interphase/silicate matrix interfaces. K(6Sr)-stabilized celsian and fluorokinoshitalite are not stable against each other at 1400°C. Fine particles of stoichiometric celsian (BaAl2Si2O8) and spinel (MgAl2O4) were the only additional phases observed along the interface. In addition, large (;2 mm) bubbles were produced along the exterior edges of the fluoromica substrate, indicating the volatilization of gaseous fluorine (F2). The formation of celsian and spinel phases, as well as gas bubbles, strongly suggests that interfacial reaction occurred via the combination of MgO from the fluoromica with slight alumina excesses from the feldspar-based matrix, as articulated by Eqs. (3) and (4). These results again suggest that MgO buffering may be highly effective in achieving thermodynamic equilibrium between alumina-rich stabilized celsian and fluorokinoshitalite, even at 1400°C. The thermochemical response between stabilized celsian and fluorophlogopite also was explored at elevated temperatures. Figure 4 shows micrographs of such an interface that has been processed at 1300°C. The gradual change in contrast within the fluorophlogopite, visible in the BEI image in Fig. 4(a), again indicates compositional gradients across the interface; however, in this case, high-Z elements increase in concentration toward the interface. EDS linescans (Fig. 5) confirm the interdiffusion of the K1 species from the mica and the Ba21 species from the feldspar (the coupled counterparts, Si41 exchange for Al31, although not shown in this figure, also were observed). Similar to the reaction between stabilized celsian and fluorokinoshitalite, uphill diffusion is observed again, except this time for the Si41 species. On closer examination of the 1300°C reaction, fine particles (,5 mm) of stoichiometric spinel (MgAl2O4) and leucite (K[AlSi2]O6) were observed along the interface via WDS analysis. Figure 4(b) shows a SEI image of the reaction, where the dark regions correspond to spinel. The presence of these two specific product phases confirms that the stabilized celsian contains finite amounts of excess alumina; thus, the interfacial reaction is driven by the depletion and subsequent combination of Mg21 cations from the mica to form spinel, as described in Eq. (2). At reaction temperatures of ,1280°C, the alumina-bearing stabilized-celsian/ fluorophlogopite interface remained stable—no breakdown of the mica because of reaction with the alumina was observed, which is consistent with the equilibrium articulated in Eq. (2). The influence of the anisotropic fluoromica structure on the interdiffusion process again is evidenced clearly along the stabilized-celsian/fluorophlogopite interface. The bicontrasted grain highlighted by the arrow in Fig. 4(a) illustrates the relative ease of diffusion along fluoromica basal planes than in the perpendicular direction. The brighter contrast entering from the exposed interlayer edges of the fluoromica grain represent areas where the Ba21 cation has been successfully exchanged for the K1 cation. The dark Fig. 2. BEI micrograph of a stabilized celsian/fluorokinoshitalite reaction interface processed at 1300°C. The fluorokinoshitalite substrate separated from the stabilized celsian substrate on final polishing. Gradual contrast changes in both phases (the brighter phase is celsian, and the less-bright phase is mica) near the interface indicates an interdiffusion process of K1 replacing Ba21. Fluorokinoshitalite grains oriented with basal planes parallel to the interface show exceptionally bright contrast, in comparison to grains oriented perpendicular to the interface, indicating that these former grains do not participate easily in the ion exchange. September 2000 Thermochemical Reactions and Equilibria between Fluoromicas and Silicate Matrices 2291
Journal of the American Ceramic Society-King et al ol.83.No.9 40- WDS Linescan across Stabilized-Celslan Fluorokinoshitalite interface Conditions: 1300C, 18.5 h, argon Original a 010203040.5060708090100110 Distance(u JKt Stabilized-Celsian Fluorokinoshitalit 20" WDS Linescan across (b) Stabilized-Celsian Fluorokinoshitallte Interface Conditions: 1300C. 18. 5 h, argon ☆F 夂ⅹ TmT 60708090100110 Distance(um Stabilized-Celsia Fluorokinoshitalite WDS linescans ac terface processed at 1300oC In Fig. 3(a), th feldspaison to that for Ba2+; the exchan nica. In Fig 3(b), the strontium profile suggests that the diffusion of Sr+ is extremel is not a significant energy-dissipative process in establishing equilibrium across the Ag and F species are evidenced as sentially immobile
Fig. 3. WDS linescans across a stabilized celsian/fluorokinoshitalite interface processed at 1300°C. In Fig. 3(a), the interdiffusing species are K1 and Si41 from the feldspar, in exchange for Ba21 and Al31 from the fluoromica. In Fig. 3(b), the strontium profile suggests that the diffusion of Sr21 is extremely slow, in comparison to that for Ba21; the exchange of Sr21 for Ba21 is not a significant energy-dissipative process in establishing equilibrium across the interface. Mg21 and F2 species are evidenced as being essentially immobile. 2292 Journal of the American Ceramic Society—King et al. Vol. 83, No. 9
eptember 2000 ns and equilibria between Fl as and Silicate Matrices 2293 Table I. Bulk and Interfacial Compositions of K-Sr Stabilized Celsian and Fluorokinoshitalite reacted at 1300oc Stabilized celsian Fluorokinoshitalite lon species Bulk Interface Interface Bulk 13.04±0.80 22.86±3.41 19.95±1.06 27.22±0.4 5.07±0.56 2.07±0.4 8.06±0.75 5.94±0.12 0.71±0.1 15.00±0.44 13.50±0.2 13.57±008 17.88±1.33 61±0.50 10.90±0.29 962±0.22 59±1.14 31.94±1.4 22±0.32 Total 100.07±0.69 9964±0.15 9955±0.41 9961±0.14 Table l. Stoichiometry of Bulk and Interfacial allows the driving force for chemical diffusion to be reduced to the Compositions of K-Sr Stabilized Celsian and adients in concentration only Fluorokinoshitalite reacted at 1300oc One can describe the motion of an individual ionic species i using the Fick/Einstein relationship(Fick's first law). In this relationship, a Ba,K.Sr,Mg,XALSi)O,(Ba, K,Sr )Mg/AI si)o oF. flux is driven by the gradient in electrochemical potential Bulk do a 0.31 A=、SD9%D(+ZF 000 0.04 0.02 where j, is the flux of the ith species, c the concentration, D the 0 2.85 defg intrinsic(self-)diffusion coefficient, m the electrochemical poten- 1.91 tial(which consists of the chemical-potential (u)and electrical potential(c) components), Z the valence, s the distance, F the Faradays constant, R the universal gas constant, and T the temperature. Thus, for the diffusing species in the stabilized- elsian/fluorokinoshitalite reaction couple displayed in Fig. 3(a), the fluxes in both phases can be described as K-rich region indicates that Ba cations have not yet reached the center section of the grain in appreciable quantities. Not surprisingly, 2+DBa+ dnBi+ the more-spacious interlayer planes allow cationic interdiffusion to JBa2+ proceed with relative ease, compared with diffusion through the more densely packed and strongly bonded tetrahedral and octahedral sheets CK+DK+ dnk of the phyllosilicate; interdiffusing cations, particularly K+ and Ba2+ 丿B (6b) cross the stabilized-celsian/fluoromica interface will be slowed bstantially when encountering phyllosilicate grains oriented parallel Dot-d (or almost parallel)to the interface IV. Discussio CAl+ DAP+ dmap+ JAIS Characterization of the Multicomponent Interdiffusion The experimental results of the reaction olid-solution CS++DSH+ amsH K-Ba feldspar and solid-solution K-Ba tri Js clearly demonstrate coupled cationic interdi exchange of K+ and Si+ cations for Ba 1+cations. A flux equation for the 02- species is included because the limited Therefore, a diffusion model that addresses quaternary systems is database available for ionic diffusion in silicate necessary to describe the interdiffusive response at the interface that the intrinsic mobility of the o- igorously. Because of the complexities that are involved in at of the tetrahedral cation s anion may perhaps be faster describing the related motion of four cationic species, however continuity conditions should be considered for this anal some general assumptions can be made to develop a tractable first, which can be used to greatly simplify the problem, description of the phenomenon. The first assumption is that the is charge neutrality, that is reaction-couple substrates (i.e, the reactants)act as infinite sources of the diffusing species. This assumption is justified easily ∑:=002/m2+jk+-2ip-=-3Ap+-4/s(7) considering the minute diffusion distances relative to the overall One then can substitute Eqs. (6)into Eq.(7). Simplification depth( thickness)of the reactants. The second assumption is that cal equilibrium is maintained at all times, despite the significant becomes involved when one realizes that concentration gradients that evolve during the reaction. While a CBa2+DBa2+, CK+DK+>>co2-Do2->>CAJ+ DAF+, CsH+Ds:+(8) standard assumption for diffusion analyses, this assumption in- cludes the fact that, on initiation and continuation of the reaction That is, the products of concentration and mobility of the interlayer he new (reacted) interfacial com of both mica and cations and, perhaps, of oxygen, far outstrip those of the remain fixed. The third assumption is that tetrahedral cations Al+ and Si+. If such is the case, then or lution behavior adequately de the diffusion process, cannot, in this interdiffusion problem, sustain an electrochemical
K1-rich region indicates that Ba21 cations have not yet reached the center section of the grain in appreciable quantities. Not surprisingly, the more-spacious interlayer planes allow cationic interdiffusion to proceed with relative ease, compared with diffusion through the more densely packed and strongly bonded tetrahedral and octahedral sheets of the phyllosilicate; interdiffusing cations, particularly K1 and Ba21, across the stabilized-celsian/fluoromica interface will be slowed substantially when encountering phyllosilicate grains oriented parallel (or almost parallel) to the interface. IV. Discussion: Characterization of the Multicomponent Interdiffusion The experimental results of the reaction between solid-solution K-Ba feldspar and solid-solution K-Ba trioctahedral fluoromica clearly demonstrate coupled cationic interdiffusion involving the exchange of K1 and Si41 cations for Ba21 and Al31 cations. Therefore, a diffusion model that addresses quaternary systems is necessary to describe the interdiffusive response at the interface rigorously. Because of the complexities that are involved in describing the related motion of four cationic species, however, some general assumptions can be made to develop a tractable description of the phenomenon. The first assumption is that the reaction-couple substrates (i.e., the reactants) act as infinite sources of the diffusing species. This assumption is justified easily, considering the minute diffusion distances relative to the overall depth (thickness) of the reactants. The second assumption is that local equilibrium is maintained at all times, despite the significant concentration gradients that evolve during the reaction. While a standard assumption for diffusion analyses, this assumption includes the fact that, on initiation and continuation of the reaction, the new (reacted) interfacial compositions of both mica and feldspar remain fixed. The third general assumption is that ideal-solution behavior adequately describes the diffusion process, a point easily justified in that, given the lack of any transitionmetal cation species, the point-defect concentrations, by whose motion chemical diffusion occurs, are very small; this assumption allows the driving force for chemical diffusion to be reduced to the gradients in concentration only. One can describe the motion of an individual ionic species i using the Fick/Einstein relationship (Fick’s first law). In this relationship, a flux is driven by the gradient in electrochemical potential: ji 5 2 ciDi RT dhi dj 5 2 ciDi RT S dmi dj 1 ZiF df dj D (5) where ji is the flux of the ith species, c the concentration, D the intrinsic (self-)diffusion coefficient, h the electrochemical potential (which consists of the chemical-potential (m) and electricalpotential (f) components), Z the valence, j the distance, F the Faraday’s constant, R the universal gas constant, and T the temperature. Thus, for the diffusing species in the stabilizedcelsian/fluorokinoshitalite reaction couple displayed in Fig. 3(a), the fluxes in both phases can be described as jBa21 5 2 cBa21DBa21 RT dhBa21 dj (6a) jK1 5 2 cK1DK1 RT dhK1 dj (6b) jO22 5 2 cO22DO22 RT dhO22 dj (6c) jAl31 5 2 cAl31DAl31 RT dhAl31 dj (6d) jSi41 5 2 cSi41DSi41 RT dhSi41 dj (6e) A flux equation for the O22 species is included because the limited database available for ionic diffusion in silicate minerals suggests that the intrinsic mobility of the O22 anion may perhaps be faster than that of the tetrahedral cations.26 Two continuity conditions should be considered for this analysis. The first, which can be used to greatly simplify the problem, is charge neutrality; that is, O zi ji 5 0 or 2jBa21 1 jK1 2 2jO22 5 23jAl31 2 4jSi41 (7) One then can substitute Eqs. (6) into Eq. (7). Simplification becomes involved when one realizes that cBa21DBa21 , cK1DK1..cO22DO22..cAl31DAl31 , cSi41DSi41 (8) That is, the products of concentration and mobility of the interlayer cations and, perhaps, of oxygen,26 far outstrip those of the tetrahedral cations Al31 and Si41. If such is the case, then one cannot, in this interdiffusion problem, sustain an electrochemicalpotential gradient of the Ba21, K1, or O22 species. Therefore, the Gibbs energy of the interdiffusion reaction will be dissipated by the diffusion of the tetrahedral cations (see Schmalzried27); i.e., no Table I. Bulk and Interfacial Compositions of K-Sr Stabilized Celsian and Fluorokinoshitalite Reacted at 1300°C Ion species Composition (wt%) Stabilized celsian Fluorokinoshitalite Bulk Interface Interface Bulk Ba21 13.04 6 0.80 22.86 6 3.41 19.95 6 1.06 27.22 6 0.49 K1 5.07 6 0.56 2.22 6 0.75 2.07 6 0.49 0.08 6 0.08 Sr21 8.06 6 0.75 5.94 6 0.12 0.71 6 0.15 0.26 6 0.09 Mg21 0.19 6 0.02 0.36 6 0.24 15.00 6 0.44 13.50 6 0.21 Al31 13.25 6 0.06 13.57 6 0.08 9.48 6 0.43 10.29 6 0.13 Si41 20.95 6 0.22 17.88 6 1.33 12.61 6 0.50 10.90 6 0.29 O22 39.62 6 0.22 36.98 6 1.29 34.59 6 1.14 31.94 6 1.48 F2 0.00 0.00 7.22 6 0.32 6.36 6 0.25 Total 100.07 6 0.69 99.64 6 0.15 99.55 6 0.41 99.61 6 0.14 Table II. Stoichiometry of Bulk and Interfacial Compositions of K-Sr Stabilized Celsian and Fluorokinoshitalite Reacted at 1300°C Composition Celsian, (BaaKbSrcMgd)(AleSif )O8 Mica, (BaaKbSrc)Mgd(AleSif )O10Fg Bulk Interface Interface Bulk a 0.31 0.57 0.67 0.99 b 0.42 0.20 0.25 0.02 c 0.30 0.23 0.04 0.02 d 0.03 0.05 2.85 2.21 e 1.59 1.74 1.62 1.91 f 2.41 2.20 2.08 1.95 g 1.75 1.68 September 2000 Thermochemical Reactions and Equilibria between Fluoromicas and Silicate Matrices 2293
Table Ill. Bulk and Interfacial Compositions of K-Sr Stabilized Celsian and Fluorokinoshitalite Reacted at 1200%C Composition(wt%) Stabilized celsian Fluorokinoshitalite lon species Bulk Interface erface 13.52±0.10 25.42±0.10 20.78±0.58 27.29 0.46 5.13±0.10 1.31±0.10 1.59±0.16 721±0.10 5.32±0.10 0.57±0.1 十一十 ±0.10 1442±0.18 0.10 ±0. 21.45±0.10 7.64±0.10 13.11±0.13 1102±0.20 33.27±0.71 32.11±0. 5.63±0.12 5.25±0.08 Total 10091±0.10 100.97±0.10 9973±0.0 100.23±0.55 Table IV. Stoichiometry of Bulk and Interfacial 3(a), the concentrations of these two tetrahedral cations vary with Compositions of K-Sr Stabilized Celsian and the distance s away from the interface(which is E=0).Thus, one Fluorokinoshitalite reacted at 1200oC realizes that D'eld Dmca(where the superscripts are associated with feldspar and mica phases, respectively ) Furthermore, neither Ba,ksSr.MguKAlSi)Os (Ba,ksSreMgAAl SHOnoFs tend nor Dmica has a unique value; instead, each is a function of distance, time, and temperature. The diffusion profile, of Si",for example, in each phase is characterized by the error-function form a 0.31 that is the solution to Ficks second law. In this problem, however the continuity condition () must be active everywhere 0.04 2.85 including at the interface. The significant consequence of this fact defg 2.75 is that the specific two-phase(local) equilibrium established at the 2.01 2.41 nterface is a function of the initial compositions of the reactants 1.38 and the relative values of Feld and Dmica at E=0. Thus, writing Eq (9)specifically for the Si t flux at the interface(using the eq dXs样+ energy of reaction is dissipated by the motions of other species(an (12) application of the"steady-state approximation"frequently used in species would disallow the accumulation of an electrical approach has been fully articulated by Jostand evertheless, the motions of Al and Si cations are coupled for this mica-feldspar interdiffusion problem, the result is by a molar -flux constraint, which is the second continuity condi- substitution of AP+and Si cations between feldspar and me tion. This constraint means that there must be a one-to (13) that is, maintenance of the mica and feldspar structures requires that the tetrahedral cation fluxes be coupled. Mathematically, this where the subscript"int" denotes a value at the interface and the constraint is articulated as subscript "bulk" denotes a value associated with a reactant. In application to the diffusion data presented in Tables I and Il, for JAI the reaction of stabilized celsian with fluorokinoshitalite at One can substitute the Fick-Einstein flux equations for the Al+ 1300 C, Xfel int =0.169, xs 4, bulk=0.185, xs int=0.108,and and Si+ species from Eqs. (6)into Eq (9)and, realizing that the 0. 101; these values, when placed in Eq.(13), giv lectrical potential is mitigated by the rapid motion of the alkali A simple comparison by inspection of the and alkaline-earth species, solve for the binary chemical (inter ldif- gradients of Si composition at 5=0 in Fig 3(a)(see Eq(12)) of the simple molar-flu supports a relative difference of this magnitude. Application of Eq onstraint, results in a simple Nernst-Planck form, i.e These differences in interdiffusion coefficients for the tetrahedral DAB+D species are consistent with the limited database for similar inter- D=xu-Dau+Xa D s diffusion reactions in metamorphic rock assemblages where X is the mole fraction, The above-given argument, given the US Summary and Comment simplifies to one that can be analyzed using the typically applied form of Ficks first law, that is, for each phase The results of this investigation illustrate two different aspec of the petromimetic approach to the successful engineering of functional fluoromica-interphase/silicate-matrix interfaces for ap- (I1) plication in an alumina-fiber ceramic composite. The first aspect demonstrates the necessity of chemical buffering-i.e, the need with a similar expression for jAi+ for a polyphase matrix that includes both spinel and forsterite--to Because of the differences in the structures of the two phases in stabilize the fluoromica interphase structurally against degradation the reaction couple, the intrinsic diffusion coefficients for the Al due to alumina- or silica-excess feldspar-matrix stoichiometries. The and Si*+ cations will be different in each phase. Furthermore, as is MgO additions decrease the activity of the excess species by com- readily apparent from the diffusion profiles presented in, e.g., Fig. bining to form either spinel or forsterite. Thus, the thermodynamic
energy of reaction is dissipated by the motions of other species (an application of the “steady-state approximation” frequently used in chemical kinetics28). Another ramification of Eq. (8) is that one would expect the diffusive motions of the Al31 and Si41 species to be decoupled electrically, i.e., the rapid motions of the other ionic species would disallow the accumulation of an electrical potential. Nevertheless, the motions of Al31 and Si41 cations are coupled by a molar-flux constraint, which is the second continuity condition. This constraint means that there must be a one-to-one substitution of Al31 and Si41 cations between feldspar and mica; that is, maintenance of the mica and feldspar structures requires that the tetrahedral cation fluxes be coupled. Mathematically, this constraint is articulated as jAl31 5 2jSi41 (9) One can substitute the Fick–Einstein flux equations for the Al31 and Si41 species from Eqs. (6) into Eq. (9) and, realizing that the electrical potential is mitigated by the rapid motion of the alkali and alkaline-earth species, solve for the binary chemical (inter)diffusion coefficient D˜ , which, because of the simple molar-flux constraint, results in a simple Nernst–Planck form, i.e., D˜ 5 DAl31DSi41 XAl31DAl31 1 XSi41DSi41 (10) where X is the mole fraction. The above-given argument, given the assumptions noted, is rigorous; thus, the interdiffusion problem simplifies to one that can be analyzed using the typically applied form of Fick’s first law; that is, for each phase, jSi41 5 2D˜ dXSi41 dj (11) with a similar expression for jAl31. Because of the differences in the structures of the two phases in the reaction couple, the intrinsic diffusion coefficients for the Al31 and Si41 cations will be different in each phase. Furthermore, as is readily apparent from the diffusion profiles presented in, e.g., Fig. 3(a), the concentrations of these two tetrahedral cations vary with the distance j away from the interface (which is j 5 0). Thus, one realizes that D˜ feld Þ D˜ mica (where the superscripts are associated with feldspar and mica phases, respectively). Furthermore, neither D˜ feld nor D˜ mica has a unique value; instead, each is a function of distance, time, and temperature. The diffusion profile, of Si41, for example, in each phase is characterized by the error-function form that is the solution to Fick’s second law. In this problem, however, the continuity condition (Eq. (9)) must be active everywhere, including at the interface. The significant consequence of this fact is that the specific two-phase (local) equilibrium established at the interface is a function of the initial compositions of the reactants and the relative values of D˜ feld and D˜ mica at j 5 0. Thus, writing Eq. (9) specifically for the Si41 flux at the interface (using the Eq. (11) form for flux), D˜ feld dXSi41 feld dj 5 D˜ mica dXSi41 mica dj (12) one can use the derivative of the error-function form of XSi41(j,t) to solve for the relative values of D˜ feld and D˜ mica. Such an approach has been fully articulated by Jost29 and Swenson et al.;30 for this mica–feldspar interdiffusion problem, the result is XSi41,int feld 2 XSi41,bulk feld XSi41,bulk mica 2 XSi41,int mica 5 S D˜ mica D˜ feld D 1/ 2 (13) where the subscript “int” denotes a value at the interface and the subscript “bulk” denotes a value associated with a reactant. In application to the diffusion data presented in Tables I and II, for the reaction of stabilized celsian with fluorokinoshitalite at 1300°C, XSi41,int feld 5 0.169, XSi41,bulk feld 5 0.185, XSi41,int mica 5 0.108, and XSi41,bulk mica 5 0.101; these values, when placed in Eq. (13), give D˜ mica ' 5D˜ feld. A simple comparison by inspection of the gradients of Si41 composition at j 5 0 in Fig. 3(a) (see Eq. (12)) supports a relative difference of this magnitude. Application of Eq. (13) to the data at 1200°C (Tables III and IV) gives D˜ mica ' 2D˜ feld. These differences in interdiffusion coefficients for the tetrahedral species are consistent with the limited database for similar interdiffusion reactions in metamorphic rock assemblages.31 V. Summary and Comment The results of this investigation illustrate two different aspects of the petromimetic approach to the successful engineering of functional fluoromica-interphase/silicate-matrix interfaces for application in an alumina-fiber ceramic composite. The first aspect demonstrates the necessity of chemical buffering—i.e., the need for a polyphase matrix that includes both spinel and forsterite—to stabilize the fluoromica interphase structurally against degradation due to alumina- or silica-excess feldspar-matrix stoichiometries. The MgO additions decrease the activity of the excess species by combining to form either spinel or forsterite. Thus, the thermodynamic Table III. Bulk and Interfacial Compositions of K-Sr Stabilized Celsian and Fluorokinoshitalite Reacted at 1200°C Ion species Composition (wt%) Stabilized celsian Fluorokinoshitalite Bulk Interface Interface Bulk Ba21 13.52 6 0.10 25.42 6 0.10 20.78 6 0.58 27.29 6 0.46 K1 5.13 6 0.10 1.31 6 0.10 1.59 6 0.16 0.05 6 0.05 Sr21 7.21 6 0.10 5.32 6 0.10 0.57 6 0.13 0.13 6 0.11 Mg21 0.00 0.06 6 0.10 14.42 6 0.18 13.45 6 0.15 Al31 13.34 6 0.10 14.23 6 0.10 9.90 6 0.17 10.91 6 0.38 Si41 21.45 6 0.10 17.64 6 0.10 13.11 6 0.13 11.02 6 0.20 O22 40.58 6 0.10 37.00 6 0.10 33.27 6 0.71 32.11 6 0.19 F2 0.00 0.00 5.63 6 0.12 5.25 6 0.08 Total 100.91 6 0.10 100.97 6 0.10 99.73 6 0.09 100.23 6 0.55 Table IV. Stoichiometry of Bulk and Interfacial Compositions of K-Sr Stabilized Celsian and Fluorokinoshitalite Reacted at 1200°C Composition Celsian, (BaaKbSrcMgd)(AleSif )O8 Mica, (BaaKbSrc)Mgd(AleSif )O10Fg Bulk Interface Interface Bulk a 0.31 0.64 0.73 0.99 b 0.41 0.12 0.19 0.00 c 0.26 0.21 0.04 0.00 d 0.00 0.01 2.85 2.75 e 1.55 1.83 1.77 2.01 f 2.41 2.17 2.25 1.95 g 1.42 1.38 2294 Journal of the American Ceramic Society—King et al. Vol. 83, No. 9
September 2000 quilibria between Fluoromicas and Silicate Matrices stabilized h celsian fluorophlogopite Distance (um ig. 5. EDS linescan across the stabilized-celsian/fluorophlogopite inter- face reacted at 1300.C for 10 h. The annotation of the plot notes the schematic representation of the interdiffusion process between the solid- solution feldspar and the solid-solution trioctahedral mica. case of te,isl460°or 1388°C. ely. These same chemical-buffering technique should be equally applicable to alumina-rich stabilized-celsian matrices. Preliminary studies also indicate that Mgo additions to barium-stuffedcordierite-matrix compositions can sufficiently reduce the alumina or silica activities to provide thermochemical abilized fluorophlogopite feldspar mineral, the interlocking doublering structure of cordi. rite is quite similar to that of framework silicates, which suggests elian an ideal matrix-phase mechanical response. This aspect is one The second approach toward stable interfaces uses the chemical flexibility offered between phases exhibiting a solid-solution equilibrium. The interdiffusional attributes of K-Ba solid-solution celsian against both end-members of the fluorokinoshitalite- fluorophlogopite K-Ba solid-solution series, involving the cationic exchange of K and Si for Ba- and Al, provides a wide compositional and thermal stability range. The compositions characterizing the interfacial response between stabilized celsian and fluoromica establish the exact thermochemical conditions necessary for maintaining global equilibrium between the inter- phase and matrix, up to temperatures that are limited only by the fluoromica melting temperature Both of these approaches offer substantial latitude in designing functional interfaces between fluoromica interphases and alumi 10 MICRONS nosilicate feldspar matrices. This design flexibility can be directly attributed to the chemical and structural complexity of the fluoromi- cas:the manipulation of multivariable, polyphase systems, although tedious to characterize, may offer the best approach toward property- Fig. 4. Interfacial morphology of a stabilized-celsian/fluorophlogopite tailorable interfaces in engineering ceramic composites. reaction couple processed at 1300.C. Figure 4(a) is a BEl micrograph illustrating the interfacial morphology. Within the fluorophlogopite, one easily can see the incorporation of heavier-Z elements near th spectroscopy reveals the incorporation of Ba- into the mica(see Fig. 5). The arrow in the figure indicates the incomplete interdiffusion in one mica ed and grain whose basal plane is approximately parallel to the reaction interface Figure 4(b) is an SEl micrograph identify ing the interfacial reaction Drs. Kenneth Chyung and Steven Dawes( Corming, Inc, Corning, NY). The authors products. As confirmed via WDS, points"B" and"C in the figure are truly grateful for the help of Doug Waelchli(University of wisconsin-Madison) ouple specimens. The authors also thank Dr. Y. Austin orrespond to spinel and leucite, respectively, these phases indicate the manv A and"E in the figure correspond to unreacted stabilized celsian and fluorophlogopite, respectively, whereas point"D"in the figure indicates a appreciative of Mark Paquette for his patience and perseverance in the frustrating task cul rain into which ion exchange(K+ and Si++for Ba2+and Al+)has of preparing laminate specimens for optical and electron microscopy. Namaste References need to form more product phases is minimized, effectively protectin IM. Y. He and J. W. Hutchinson,"Crack Deflection at an Interface Between the fluoromica interphase from a catastrophic loss of Mgo 2H. R Shell and K. H. Ivey, "Fluorine Micas, "Bull. U.S. Bur. Mines, 647(1969) Following this methodology, thermodynamic equilibrium can R. F. Giese Jr."The Effect of F/OH Substitution on Some Laver-Silicate and a fluoromica interphase up to very high temperatures that are TaG. H. Beall, k. Chyung, S B. Dawes, K P. Giadkare, and S N. Hoda, be established between a Mgo-buffered anorthite-matrix phase limited only by the fluoromica melting temperature, which, in the 4935387, June 19, 1990.(b)G. H Beall, K Chyung, S. B Dawes, K. P. Gadkaree
need to form more product phases is minimized, effectively protecting the fluoromica interphase from a catastrophic loss of MgO. Following this methodology, thermodynamic equilibrium can be established between a MgO-buffered anorthite-matrix phase and a fluoromica interphase up to very high temperatures that are limited only by the fluoromica melting temperature, which, in the case of fluorokinoshitalite or fluorophlogopite, is 1460° or 1388°C, respectively. These same chemical-buffering techniques should be equally applicable to alumina-rich stabilized-celsian matrices. Preliminary studies also indicate that MgO additions to “barium-stuffed” cordierite-matrix compositions can sufficiently reduce the alumina or silica activities to provide thermochemical stability against fluoromica interphases. Although not truly a feldspar mineral, the interlocking double-ring structure of cordierite is quite similar to that of framework silicates, which suggests an ideal matrix-phase mechanical response. This aspect is one focus of our continuing research. The second approach toward stable interfaces uses the chemical flexibility offered between phases exhibiting a solid-solution equilibrium. The interdiffusional attributes of K-Ba solid-solution celsian against both end-members of the fluorokinoshitalite– fluorophlogopite K-Ba solid-solution series, involving the cationic exchange of K1 and Si41 for Ba21 and Al31, provides a wide compositional and thermal stability range. The compositions characterizing the interfacial response between stabilized celsian and fluoromica establish the exact thermochemical conditions necessary for maintaining global equilibrium between the interphase and matrix, up to temperatures that are limited only by the fluoromica melting temperature. Both of these approaches offer substantial latitude in designing functional interfaces between fluoromica interphases and aluminosilicate feldspar matrices. This design flexibility can be directly attributed to the chemical and structural complexity of the fluoromicas; the manipulation of multivariable, polyphase systems, although tedious to characterize, may offer the best approach toward propertytailorable interfaces in engineering ceramic composites. Acknowledgments All fluoromica and feldspar powder precursors were prepared and provided by Drs. Kenneth Chyung and Steven Dawes (Corning, Inc., Corning, NY). The authors are truly grateful for the help of Doug Waelchli (University of Wisconsin–Madison) in preparing the reaction-couple specimens. The authors also thank Dr. Y. Austin Chang (University of Wisconsin–Madison) and Dr. Doug Swenson (Michigan Technology University, Houghton, MI) for many fruitful discussions concerning the modeling of multicomponent interdiffusion systems. Lastly, the authors also are appreciative of Mark Paquette for his patience and perseverance in the frustrating task of preparing laminate specimens for optical and electron microscopy. Namaste. References 1 M. Y. He and J. W. Hutchinson, “Crack Deflection at an Interface Between Dissimilar Elastic Materials,” Int. J. Solids Struct., 25 [9] 1053–67 (1989). 2 H. R. Shell and K. H. Ivey, “Fluorine Micas,” Bull. U.S. Bur. Mines, 647 (1969). 3 R. F. Giese Jr., “The Effect of F/OH Substitution on Some Layer-Silicate Minerals,” Z. Kristallogr., 141, 138–44 (1975). 4 (a)G. H. Beall, K. Chyung, S. B. Dawes, K. P. Gadkaree, and S. N. Hoda, “Fiber-Reinforced Composite Comprising Mica Matrix or Interlayer,” U.S. Pat. No. 4 935 387, June 19, 1990. (b)G. H. Beall, K. Chyung, S. B. Dawes, K. P. Gadkaree, Fig. 4. Interfacial morphology of a stabilized-celsian/fluorophlogopite reaction couple processed at 1300°C. Figure 4(a) is a BEI micrograph illustrating the interfacial morphology. Within the fluorophlogopite, one easily can see the incorporation of heavier-Z elements near the interface; spectroscopy reveals the incorporation of Ba21 into the mica (see Fig. 5). The arrow in the figure indicates the incomplete interdiffusion in one mica grain whose basal plane is approximately parallel to the reaction interface. Figure 4(b) is an SEI micrograph identifying the interfacial reaction products. As confirmed via WDS, points “B” and “C” in the figure correspond to spinel and leucite, respectively; these phases indicate the presence of excess alumina within the feldspar substrate originally. Points “A” and “E” in the figure correspond to unreacted stabilized celsian and fluorophlogopite, respectively, whereas point “D” in the figure indicates a mica grain into which ion exchange (K1 and Si41 for Ba21 and Al31) has occurred. Fig. 5. EDS linescan across the stabilized-celsian/fluorophlogopite interface reacted at 1300°C for 10 h. The annotation of the plot notes the schematic representation of the interdiffusion process between the solidsolution feldspar and the solid-solution trioctahedral mica. September 2000 Thermochemical Reactions and Equilibria between Fluoromicas and Silicate Matrices 2295
ournal of the American Ceramic Society-King et al and S. N. Hoda, "Fiber-Reinforced Composite Comprising Mica Matrix or Inter- IF. H. S. Vermaas,"A New Occurrence of Barium-Feldspar at Otjosondu ayer;"U.s.Pat.No.4948758,Aug14,1990. K. Chyung and S. B. Dawes, "Fluoromica Coated Nicalon Fiber Reinforced Hyalophane,Am. Mineral, 38, 845-57(1953). Glass ceramic c mp pt c. Halle- eactiongs between Synthetic Mica and Simple Subsolidus Relationships of Barum microcline, Hyalophane and Albite in the P Gay and N. Roy," The Mineralogy of the Potassiurm-Barium Feldspar Series 57 Com eldspar from the Noda-Tamagawa Mine, Iwate Prefec Ter Minera.J,97409-16(1979) ImM Cooper, -:, Am. Ceram. Soc, 77[7]1699-705(1994). 2M. Yoshii and K Maeda, "Relations between Barium Content and the physic T.T. King and R F nted-Temperature, Variable-Rate Mechanical ptical Properties in the Mangaoan Phlogopite-Kinoshitalite Series, Mineral. L. Brennan and K. M. prewo "Silicon carbide fibre reinforced glass-cCerami 858-65(1975) Dasgupta, S Chakraborti, P Sengupta, P K. Bhattacharya, H. Banergee, and Matrix Composites Exhibiting High Strength and Toughness,J. Mater. Sci., 17, 4. Fukuoka, "Compositional Characteristics of Kinoshitalite from the Sausar Group, 2371-83(1982) India"Am. Mineral.,74,200-202(1989) L. Munoz,“F- OH and CH Mineralogy J S. Kirkaldy and L C. Brown, "Diffusion Behavior in Ternary, Multiphase IK. Chyung, R. F. Cooper, K. P. Gadkaree, R, L. Stewart, and M. P. Taylor, Systems,Can. Met. 0, 2, 89-115(196 ment of Alkaline Earth Aluminosilicate Glass-Ceramics, U.S. Pat. No Y. A. Chang, "A Combined Thermodynamic and Kinetic Approach to the 4615987,Oct.7,1986 Metallization of GaAs, Mater. Res. Soc. Symp. Proc, 260, 43-52(1992). 2G. H. Beall, K. Chyung, and H. J. Watkins, "Mica Glass-Ceramics, "U.S. Pat. No. R. Freer, "Diffusion in Silicate Minerals and Glasses: A Data Digest and Guide the Literature, Contrib. Mineral. Petrol, 76, 440-54(19 3B. Yoshiki and K. Matsumoto. "H perature Modification of Barium H. Schmalzried, Solid State Reactions, 2nd Ed, pp. 143-47, Verlag Chemie, Feldspar, " J. Am. Ceram Soc., 34[91283-86(1951). Germany, 1981 nd D R. Uhlmann, Introduction to Ceramics Ed. pp. 381-401. Wiley, N 1976 ISN. Frety, A. Taylor, and M. H. Lewis, " Microstructure and Crystallization Solids, Liquids, Gases; pp. 68-69 Behavior of Sol-Gel Derived ySro-yBa0-Al2O3-2SiO, Glass-Ceramic,J.Non- London, U.K., 19 Cryst. Solids,195,28-37(1996) D. Swenson, C. H. Jan, and Y A Chang, "The Formation of Ohn leE. F, Osborn and A. Muan, "The System Mg0-Al2O-SiO2" Plate 3 in Phase Enhanced Contacts to Ill-V Compound Semiconductors via the Exi nae M Equlibriun Diagrams of Oxide Systems. American Ceramic Society and Edward nism: A Combined Thermodynamic and Kinetic Model, J. Appl. Plrys, 84[8] Orton Jr. Ceramic Foundation, Columbus, OH, 1960. 4332-42(1998) Segnit,"Barium-Feldspars from Broken Hill, New South Wales,Mineral. Mag,27192]166-75(1946 " F.S. Spear, Metamorphic Phase Equilibria and Pressure-Temperature a Paths, p. 612. Mineralogical Society of America, Washington, DC, 1993
and S. N. Hoda, “Fiber-Reinforced Composite Comprising Mica Matrix or Interlayer,” U.S. Pat. No. 4 948 758, Aug. 14, 1990. 5 K. Chyung and S. B. Dawes, “Fluoromica Coated Nicalon Fiber Reinforced Glass-Ceramic Composites,” Mater. Sci. Eng. A, A162, 27–33 (1993). 6 R. F. Cooper and P. C. Hall, “Reactions between Synthetic Mica and Simple Oxide Compounds with Application to Oxidation-Resistant Ceramic Composites,” J. Am. Ceram. Soc., 76 [5] 1265–73 (1993). 7 T. T. King and R. F. Cooper, “Ambient-Temperature Mechanical Response of Alumina–Fluoromica Laminates,” J. Am. Ceram. Soc., 77 [7] 1699–705 (1994). 8 T. T. King and R. F. Cooper, “Elevated-Temperature, Variable-Rate Mechanical Response of Alumina–Fluoromica Laminates”; unpublished work. 9 J. J. Brennan and K. M. Prewo, “Silicon Carbide Fibre Reinforced Glass-Ceramic Matrix Composites Exhibiting High Strength and Toughness,” J. Mater. Sci., 17, 2371–83 (1982). 10J. L. Munoz, “FOOH and ClOOH Exchange in Micas with Applications to Hydrothermal Ore Deposits”; pp. 469–93 in Reviews in Mineralogy, Vol. 13, Micas. Edited by S. W. Bailey. Mineralogical Society of America, Blacksburg, VA, 1984. 11K. Chyung, R. F. Cooper, K. P. Gadkaree, R. L. Stewart, and M. P. Taylor, “Reinforcement of Alkaline Earth Aluminosilicate Glass-Ceramics,” U.S. Pat. No. 4 615 987, Oct. 7, 1986. 12G. H. Beall, K. Chyung, and H. J. Watkins, “Mica Glass-Ceramics,” U.S. Pat. No. 3 801 295, Apr. 2, 1974. 13B. Yoshiki and K. Matsumoto, “High-Temperature Modification of Barium Feldspar,” J. Am. Ceram. Soc., 34 [9] 283–86 (1951). 14R. E. Newnham and H. D. Meagaw, “The Crystal Structure of Celsian (Barium Feldspar),” Acta Crystallogr., 13, 303–308 (1960). 15N. Frety, A. Taylor, and M. H. Lewis, “Microstructure and Crystallization Behavior of Sol–Gel Derived 1⁄2SrO–1⁄2BaO–Al2O3–2SiO2 Glass-Ceramic,” J. NonCryst. Solids, 195, 28–37 (1996). 16E. F. Osborn and A. Muan, “The System MgO–Al2O3–SiO2”; Plate 3 in Phase Equilibrium Diagrams of Oxide Systems. American Ceramic Society and Edward Orton Jr. Ceramic Foundation, Columbus, OH, 1960. 17E. R. Segnit, “Barium-Feldspars from Broken Hill, New South Wales,” Mineral. Mag., 27 [192] 166–75 (1946). 18F. H. S. Vermaas, “A New Occurrence of Barium-Feldspar at Otjosondu, South-West Africa, and an X-ray Method for Determining the Composition of Hyalophane,” Am. Mineral., 38, 845–57 (1953). 19P. Gay and N. Roy, “The Mineralogy of the Potassium-Barium Feldspar Series III: Subsolidus Relationships,” Mineral. Mag., 36, 914–32 (1968). 20S. Nakano, “Intergrowth of Barium Microcline, Hyalophane and Albite in the Barium Containing Alkali Feldspar from the Noda-Tamagawa Mine, Iwate Prefecture, Japan,” Mineral. J., 9 [7] 409–16 (1979). 21M. Yoshii and K. Maeda, “Relations between Barium Content and the Physical and Optical Properties in the Mangaoan Phlogopite–Kinoshitalite Series,” Mineral. J., 8 [1] 58–65 (1975). 22S. Dasgupta, S. Chakraborti, P. Sengupta, P. K. Bhattacharya, H. Banergee, and M. Fukuoka, “Compositional Characteristics of Kinoshitalite from the Sausar Group, India,” Am. Mineral., 74, 200–202 (1989). 23M. F. Brigatti and L. Poppi, “Crystal Chemistry of Ba-rich Trioctahedral Micas-1M,” Eur. J. Mineral., 5, 857–71 (1993). 24J. S. Kirkaldy and L. C. Brown, “Diffusion Behavior in Ternary, Multiphase Systems,” Can. Met. Q., 2, 89–115 (1963). 25Y. A. Chang, “A Combined Thermodynamic and Kinetic Approach to the Metallization of GaAs,” Mater. Res. Soc. Symp. Proc., 260, 43–52 (1992). 26R. Freer, “Diffusion in Silicate Minerals and Glasses: A Data Digest and Guide to the Literature,” Contrib. Mineral. Petrol., 76, 440–54 (1981). 27H. Schmalzried, Solid State Reactions, 2nd Ed.; pp. 143–47. Verlag Chemie, Weinheim, Germany, 1981. 28W. D. Kingery, H. K. Bowen, and D. R. Uhlmann, Introduction to Ceramics, 2nd Ed.; pp. 381–401. Wiley, New York, 1976. 29W. Jost, Diffusion in Solids, Liquids, Gases; pp. 68–69. Academic Press, London, U.K., 1952. 30D. Swenson, C. H. Jan, and Y. A. Chang, “The Formation of Ohmic and Schottky Enhanced Contacts to III–V Compound Semiconductors via the Exchange Mechanism: A Combined Thermodynamic and Kinetic Model,” J. Appl. Phys., 84 [8] 4332–42 (1998). 31F. S. Spear, Metamorphic Phase Equilibria and Pressure-Temperature–Time Paths; p. 612. Mineralogical Society of America, Washington, DC, 1993. M 2296 Journal of the American Ceramic Society—King et al. Vol. 83, No. 9