Corrosion ofSilicon-Based Ceramicsin Combustion Environments NathanS.J acobson* 44 135National Aeronautics and Space Administration, Lewis Research Center, Cleveland, Ohio Silicon-based ceramicsandcomposites are primecandi- dates for heat engine and heat exchanger structural compo- nents.In such applications these materials are exposed to combustion gases and deposit-forming corrodents.In this papercombustionenvironmentsare definedforvarious applications. These environments lead to five main types of corrosive degradation:passive oxidation,deposit-induced corrosion,activeoxidation,scale/substrateinteractions, and scale volatility. Each of these is discussed in detail. The keyissues in oxidation mechanisms ofhigh-puritysilicon carbide (Sic) and silicon nitride (Si,N4) in pure oxygen are discussed. The complicating factors due to the actual com- bustionenvironmentandcommercialmaterialsaredis- cussed. These discussions include secondary elements in the ceramics; additionaloxidants,suchas waterandcarbon dioxide (COJ ;combustion environmentimpurities;long- term oxidation effects; and thermal cycling. Active oxida- tion is expected in a limited number ofcombustion situa- tions, and the active-to-passive transition is discussed. At hightemperaturesthe limiting factorsare scale melting, scale volatility, andscale/substrateinteractions.Deposit- induced corrosion is discussed, primarily for sodium sulfate (Na,SO,),but also for vanadate and oxide-slag deposits as well. In applying ceramics in combustion environments it is essential to be aware of these corrosion routes and how they affect the performance of a component. I.Introduction ANYpotentialusesofsilicon-based ceramicsand com- Mpositesinvolveexposuretocombustiongases.These applications rangefrom hot-section structural componentsof gas turbinesandpistonenginestoheatexchangertubesfor industrial furnaces. Gas turbines are used in aircraft and electric power generation and have been tested for automobiles.Poten- tialceramic componentsofthese engines include combustor liners and, perhaps some day, turbineThe location of these components in a gas turbine engine is shown in Fig.I(a). Inpistonenginesthepotentialceramiccomponentsinclude S.M. Wi edcrhorn~ontri huti ngedi tor ManuscriptNo. 1955915.ReceivedJ une15.1992; approvedSeptembcr 28, 19?2. Member, American Ccramic Society. valvesand pistonheads,4 as showninFig.I(b). Theuseof ceramic tubes as heat exchangers has been proposed for indus- trial furnaces, such as glass remelt furnaces, steel soaking pits, and aluminum reclamation furnaces.' An example is illustrated in Fig.I (c). In addition, ceramics are under considerationfor heatexchangers in coal-firedcombustors.'Amore developed applicationistheuseofceramictubesforindirectheating, where hot combustion gases pass through the tube,heatinga process on the out~i de. ~The structural components of the hot sections are subject to arange ofchemicalattackprocesses, dependingonthetemperature,pressure,andchemical environment. The focus ofthis paper is on the silicon-based ceramics sili- con carbide (Sic) and silicon nitride(SilN,,).These materials are inherently unstable inair andforma thinlayerofsilicon dioxide (SO,) in an oxidizing environment. SiO?has the lowest permeability to oxygen of any of the common oxides and forms an effective reaction barrier.'Therefore, silicon-based ceramics have the potential of substantially better high-temperature oxi- dation behavior than metals. The protective oxide scales on sili- con, Si c, and Si,N,are shownschematically in Fig. 2.Note thatSi,N,formsa silicon oxynitride (Si2N20)layer belowthe SiOz layer.'SiCmay also form an oxycarbide layer, but there is only limited evidence for this."The analogous protective layers onsuperalloys for high-temperature applications are alumina (A120Jand chromia (Cr203)." Extensive work has been done on the performance ofAI,O,and Cr,O,scales incombustion environments.Ingeneral,our knowledge ofthebehaviorof SiO, scales lags behind that of A120, and Cr20,. The purpose of this paper is to reviewour current state of knowledge of the interaction of SiO, with combustion environ- ments. Because the focus is on the interaction of SiO? with the environment, manyofthe conclusions applytoallSi0,-pro- tectedmaterials.Thisincludes composites ofSi c and Si,N,, such as Sic-fiber-reinforcedSi c matrices and SiC-fiber-rein- forced Si3N, matrices, and SO,-forming alloys, such as molyb- denumdisilicide(MoSi,).Inthispaperthemajorchemical degradation routes are discussed. Particular emphasis is on the mechanisms of corrosion and the key questions involving them. 11.Environments Combustion environments varywidely,depending onfuel, temperature,pressure,and oxidizer.Fuels are complex mix- tures of hydrocarbons, and they are classified by boiling points. Lower-boiling-pointfuelsincludeautomotivegasoline,and higher-boiling-point fuels include aviation fuel and fuel oils.12 4Journul oftheAmerican Ceramic Society-JacobsonVol. 76, No. 1 ,- Fan blades I I r Turbine blades \I Stator vanes Sample holder --,Flue gas (b)(4 Fig.1.piston engine, and (c) aluininum reclamation furnace (see Ref. 5) .Schematics of proposed applications lor ceramics where ceramics will besubjected to combustion environments: (a) gas turbine engine, (b) Table I lists some common fuels and their approximate impurity contents. " - l h Notethatallfuels havesome amountofsulfur, which may lead to corrosion. The last category of fuel oils con- tainsmoresodium, potassium,andvanadium thantheother fuels. These lower-purity fuels are likely to be utilized more in the future aspetroleumresources decrease.Impurities have a majorinfluenceon corrosion.TableIdoesnotinclude coal- derivedfuels,whichgenerallyhaveevenhigherimpurity levels. Combustion is the oxidation of these fuels to the stable prod- ucts carbon dioxide (C02)and water (H20). Equilibriumcom- bustionproducts can be calculated bymixingthe fuel and the oxidanttogetherinafree-energy-minimizationcomputer code.Figure 3(a) shows the equilibrium combustion products for a standard aviationfuel (J et A-CH,91xs)asa function of equivalence ratio. Equivalence ratio is defined as the fuel-to-air ratio at a particular point divided by the stoichiometric fuel-to- air ratiofor complete combustion to CO,and H20. Thus, an equivalence ratio of1 is stoichiometric, an equivalence ratio of less than 1 denotes a fuel-lean region, and an equivalence ratio greater thanIdenotesafuel-richregion.Mostgas turbines operateinthefuel-leanorstoichiometricregions.These regions contain large amounts of oxygen with the CO, and H,O. Mostcorrosionstudies havebeenperformedinthefuel-lean region.However,some novelcombustor designs may involve the fuel-rich region, which produces larger amounts of carbon monoxide (CO) and hydrogen gas (HJ .However, this region also contains the combustion products C02and H,O.Note that theseareequilibriumcalculations;ifequilibriumisnot attained,other species, such as elemental carbon,mayform. Thegoalofacorrosionstudyistounderstandthechemical reactions that occur between these combustion products and the proposed hot-gas-path structural materials. Figure 3(b) shows the adiabatic flame temperature as a func- tionofequivalenceratio.Thistemperatureiscalculatedby usingafree-energy-minimization computercodeandsetting J anuary 1993Corrosion of Silicon-Based Ceramics i n Combustion Environments5 LSi 02 LSiOp Fig.2. and (c) on Si,N,. Schematic ofprotective oxide scales: (a) on Si, (b) on Sic, the net heatloss equal to zero."This is an idealizedcase that assumes no heat loss through the walls of the combustion cham- ber. It does give the maximum temperature of the flame, how- ever.Ingeneral,thewallmaterialsandthecomponents downstream from the flame are at lower temperatures. Corrosion occurs not only by gaseous coinbustion products, butalsobydeposits.Perhapsthemostcommondepositis sodium sulfate (Na,SO,),which forms when sodium reacts with sulfurfuelimpurities.'X Thesodiummayoriginatefroma marine environment, from a salted roadway, or as a fuel impu- rity. Corrosion by Na,SO,is termed "hotcorrosion" and is dis- cussedindetailherein.Othertypesofdeposit-induced corrosionoriginate fromvanadiumfuelimpurities andfrom oxide slags. Insome ways the pistonchamber of an internal combustion engine is a more complex environment than the hot section of a gasturbine.Pressuresandtemperaturesvarythroughthe stroke of the piston. Adiabatic flame temperatures and combus- tiongas productcompositions can be calculated bythesame free-energy-minimization programdescribedpreviouslywith propane (C,H,)as a model fuel. The results of this calculation are similar to those for J et Aaviation fuel. Current automotive engines run slightly fuel rich to allow proper operation of pollu- tion control devices. In addition to heat engines, silicon-based ceramics are also primecandidatesforheatexchanger tubesinindustrialfur- naces. The environment encountered by a heat exchanger var- ies widely. Again, one would expect large amounts of COz and 0.51.o1.52.0 Equivalence ratio 2500r 2000 1500 1000 500 (b)III I ~~~0.51 .o1.52.0 Equivalence ratio Fig.3.Calculated gas composition and Rarnetemperature as a func- tion of equivalence ratio: (a) equilibrium gas composition and (b) adia- batic flame temperature. H20,but now the impurities play a key role. An aluminum rec- lamation furnace and a glass remelt furnace may involve alkali- metalsalts thatcan deposit on theheat exchanger.'Ina coal- tiredfurnace the atmosphere may be more reducing." In addi- tion, a mixture of oxides may form a slag deposit, which can be quite corrosive, on the tubes. TableI1summarizesthecomposition ofthesecombustion environments. This listis byno meansexhaustive. As noted, specific applications generate specific corrodents. For example, combustion of municipalwastesmay generate hydrogen chlo- ride(HCI).'"Insummary, combustion environments are com- plex,involvingnotonlythecombustionproductsCOz and H,O, but a variety of other gases and possible deposits as well. Temperatures and pressures are again quite dependent on the particularsystem.TherangesarelistedinTableI I . An important issueto consider is thermalcycling.Some applica- tions,suchas autilityturbine or industrialfurnace,involve essentiallyisothermal exposures for long periods. Other appli- cations,suchasanaircraftgasturbine,involvethermal cycling. I l l . Qpes ofCorrosive Attack and Experimental Techniques Figure 4 shows the major types of corrosive attack as a func- tion of pressure and reciprocal temperature. The major types of corrosive degradationare passiveoxidation,deposit-induced corrosion,activeoxidation,oxideisubstrate interactions,and scale vaporization. The temperature boundaries betweenthese types are only approximate and are dependent on the specific system.Furthermore,rarelyisone mechanismoperative.I n practice, several mechanisms operate simultaneously. Laboratorystudiesofcorrosivedegradationfallintotwo main categories: burner rig studies and laboratory furnace stud- ies. Burners more accurately model the actual combustion situ- ation.Figure5showsatypicalhigh-velocityburnerrig. Table I. Properties of Some Common Hydrocarbon Fuels Fuel(K)*rdtio(wt %)' (ppm)'( ppd Boding rangeH C molarS contentNd+KcontentV content Unleaded automotive gasoline300-3752.020.153.6, SiO forms according to Eq.(26). Toformastable SiOL scale, enough SiO(g) must be formed to maintainthe equilib- rium in Eq. (29). The transition PO, is the Po, necessary to form that amount of SiO(g). This is a transport problem and depends on the diffusion of oxygen and SiO through the boundary layer. Solving this problem leads to the following relationship: P;;"1"' =0.5(D(SiO)/D(02))"2Py;o (32) The diffusion coefficients can be estimated, and the expression reduces to P l y=0.64f&(31) This approachis shown schematically in Fig. 20(b). Note that this resultis close to the resultsbased on thermodynamic con- siderations alone, due to the fact that gas-phase diffusion coef- ficients are often close to unity. Theactive-to-passivetransitionfor SIC and Si,N,ismore complex because additional product gases form. A purely ther- modynamic interpretationallows fr to bedetermined from volatility diagrams as described by Heuer and Lou. I"'Singhal'IX has extended Wagner's theory and used the following analogues for Eq. (29) to derive a transition Po,: (34) ( 35)SiC(s) +2SiO,(s)=3SiO(g)+CO(g) Si,N,(s)+3Si02(s) =6SiO(g) +2N,(g) However, Hinze and Graham"'have suggested that other equi- libria may control this transition for Si c in place of Eq. (34): ( 36)(37) Their experimental data tend to support Eq. (37). This is simi- lar to the result of Gulbransen and J ansson""for the transition SiC(s) +SiO,(s)=2SiO(g)+C(g) 2SiC(s)+SiO,(s) =3Si(l,s) +2CO(g) / / / LSiO(g) + 02(g)=SO2 (smoke) (4 Fig.20.Schematic of two proposed descriptions of active-to-passive transitionforsilicon: (a) thermodynamic considerations onty, SO,"smoke" formation, and (b) thermodynamic and mass transport consid- erations. Wagner theory. J anuary 1993Corrosion qf'Silicoti-Based Cerumics in Combustion Environtnents1') in the molecular flow regime. Thus, the active-to-passive tran- sition for Si c is given in terms of sufficient Po, to create enough CO to establish this equilibrium.Againusing mass transport arguments simiiar to those of Wagner, they have derived the fol- lowing expression: (38) Narushima et~ 1 . " ~ haveexaminedCVD Si c andcome toa similar conclusion. Nickel12"has reported that carbon activity needs to be consid- eredinestablishing theactive-to-passive transition.Athigh carbon activities the SiC/SiO, equilibrium is critical in estab- lishing the transition. At low carbon activities Nickel proposes that the SiO(l)/SiC equilibrium is important and thus the critical equilibria becomes (39) - - (D(CO)/D(O,)) li2P2q,, SiC(s) +SiO(l)=2Si(l)+CO(g) This approach fits some recent data quite well;""however, the existence of condensed-phase SiO is quite controversial. ''I Numerousotherstudies oftheactive-to-passive transition have been performed.Antilland Warburton'"have examined the transitionin the presence of other oxidants (C02and H,O) andhavefoundbehaviorsimilartothatinoxygen.Thisis important foracombustionenvironment,wherethe primary oxidantsmaybeCO,andH20. VaughnandMaahs"3have shown that the Po2 transitionvaries with flow rate, emphasizing themasstransportissue.Theapproximateactive-to-passive oxidant transitionpressures for Si c and Si,N,are given in Fig. 4; more precise values are given in the references. l'h-120.'22.'23Inmostcombustionenvironments,activeoxidationisnot expected to be a problem. Combustors in current gas turbines bum near the stoichiometric region. The resultant environment contains oxygen, H,O,and CO,.Novel combustor designs may involve a fuel-rich region, which would contain CO, C02,H20, hydrogen, and oxygen potentials as low as about IO-'bar(Fig. 3(a)). However, the presence of H20and CO,at about 0.1 bar each should form enough SiO to satisfy the equilibrium require- ments for a stable SiO, scale (Eqs. (29) or (34) to (37) or (39), with H20and/or CO, as oxidants). The effect of water vapor i n suppressing active oxidation has been discussed byHeuer and Lou.'"*Once formed, however,the SiOz scale may be reduced in the fuel-rich environment. Thus, combustion ofhydrocarbonfuels generallyprovides an atmosphere sufficient for forming Si 02.However, coal com- bustionsituationsandcertainheat-treatingenvironments, whichare heavily reducing,can lead to active oxidation.Ina recent study of reactions of Si c in heat-treating environments, the kinetics of SiO(g) formation are found to be dependent on the gas composition and additives in the ceramic. 124.'25 Insummary, there are two causes ofvolatility.Thefirstis SiO, vaporization. Unless the net SiO, vapor pressure is greater thanabout10-'bar,thisis not a major problem.In fuel-lean and stoichiometric combustion situations, this vapor pressure is attained near the melting point of SO,, so that volatility is not a limitingfactor. In a fuel-richsituation, SiO, volatilityis likely to be an issue at higher temperatures. The second cause of vola- tility is active oxidation. The key issue is the oxidant pressure fortheactive-to-passive transition.Activeoxidationispre- dicted for only a limited number of combustion situations, and it can lead to rapid material consumption. X.Upper Temperature Limits Referringtothecorrosiondiagram(Fig.4),theabsolute upper use temperature for silicon-based ceramics is -2000K. This is the melting point of SiO,.When the protective oxide is liquid, it begins to flow, and rapid transport rates are obtained. This is a nonprotectivesituation. Figure 21shows a sample of Si c oxidized for 1h in air at 2073 K.Note the extensive corro- sion. The question then is how close to 2000 K can one safely use this material. Severalissueslimitthe useofSi c below2000 K .These include rapidoxidation rates,volatilizationoftheSiO? film, and reaction oftheSi 02film withthe Si c or Si,N4 substrate. Oxidation and volatilityhave beendiscussed previously.Note that there are relatively few measurements at these high temper- atures because i t is difficult to attainsuch temperatures in oxi - dizing environments. An important difference between silicon-based ceramics and other oxide-protected metalalloys i sthat, at sufficientlyhigh temperatures,theoxidescaleandtheSi c orSi3N4ceramic react.Such reactions can generate gas pressures at the oxide/ ceramic interface: (40)SiC(s)+2SiO,(s)=3SiO(g)+CO(g) Si,N?O(s)+SiO,(s) =3SiO(g)+N&)(41) The best way to examine these interfacialreactions is with the Si-C-0andSi-N-0predominancediagrams,whichare shown in Fig. 22. The intersectionof the SiO and CO isobars with the SiC/SiO, coexistence line (line AB) gives the gas pres- sures in the carbide system, as shown in Fig. 22(a). The inter- sectionofthe SiO and nitrogenisobarswiththe SizN,0/Si02 coexistence line (line AB) gives the gas pressuresin the nitride system, as shown in Fig. 22(b). Consider first the Si-C-0system. Figure 23(a) shows these total pressures as a function of temperature for the carbon-satu- rated,stoichiometric,andsilicon-saturatedsystems.Taking 1 atrn as the criteria for instability, carbon-saturated Si c in con- tact with SiOz becomes unstable at1800 K. Pressures approach 1atmfortheothersystemsnearthemeltingpointofSiO,. Thesetypesofreactionshavebeensuggestedasthesource ofbubblingatveryhightemperaturesinanoxidizedSi c sample . 41. 17b Recently,J acobson ef a1.I"have examined the Sic +SiO, reaction in detail. They have found that a carbon-saturated sys- temtendstoadjust toastoichiometric composition.Thisis done by releasing a large amount of CO and forming a second- ary Si c. Such a process at the carbon-saturated SiCiSiO, inter- face may lead to extensive bubbling. The Si-N-0system is more complex because two interfaces are involved. However,Eq. (41) is predicted to be themajor source ofSiO(g) and N,(g).The totalpressures are shownin Fig.23(b).The pressures predictedare less thanthose in the Si-C-0system. This may be the reason that only limited bub- bling is observed when Si,N,is oxidized at high temperatures."' Fig.21. in air. (Courtesy of D. Fox, NASA Lewis.) Chemically-vapor-depositedSic oxidizcd for Ih at2073 K Journal qfihe American Cerumic Society-JacobsonVol. 76, No. 1 20 0 -1 0-2 - 44 P 0 r -3 -25-24-23-22-21-20 log Po2, bar Si2N20(s) 0 N h 8 - -10 -ic II -19-18 Si 02 (s) - 1 "4 0 -35-30-25-20-15-10 log Pop, bar Fig.22. (b) SI-N-0system. Predominance diagrams atI500 K: (a) Si-C-0system and Insummary,theupperusetemperaturesofsilicon-based ceramics arelimitedbyscalemelting,oxidation rates,vola- tility,andscaleisubstratereaction.Scalemeltingoccursat 1996 K.Oxidationandscale volatilityare highlydependent onceramic composition and atmosphere.Thescaleisubstrate reactionis more of an issue in Si c than in Si,N,. XI.Deposit-Induced Corrosion ( I ) Formation of Deposits Asdiscussedpreviously,dopantlevelsofsodiuminSiO? scalesappeartoaccelerateoxidation.Largeramountsof sodium, either from a marine environment, a salted roadway, or fuelimpurities,canreactwithsulfurfuel impurities toform Na,SO,,which is a highly stable molecule:" 2NaCl(g)+SO&)+0.5O,(g)+HzO(,g) =Na2S0,(/)+2HCI(g)(42) TheNa,SO,forms as a deposit on engine parts.This type of deposit-inducedcorrosion is termed "hotcorrosion ," and there is a large amount of literatureon hot corrosion of metals."'~'?' Thestudiesonsilicon-basedceramicshaverecentlybeen summarized. '.'('The primaryemphasis inthissectionis on Na,SO,-induced attack. However, there are other types of deposit-induced corro- sion(e.g., vanadate-induced attackfrom lowerpurityfuels"' and oxide-slag-induced attack from coal combustion"). Similar- itiesbetweenthevarious types ofdeposit-induced attack are discussed below. In general, hot corrosionoccurs in two steps: ( 1 ) deposition and (2) corrosive attack. It is first necessary to determine when lo*r 120014001600180020002200 Temperature, T, K (b) la001;oo l b02doo2; ooTemperature, T, K Fig.23. ib) Si,N,O/SiOlinterface. Total vapor pressure at interfaces: (a) SiCiSiO, interface and a corrosive Na,SO,deposit forms. Generally,Na,SO,is corro- sive betweenits meltingpoint( 1157 K) and the dewpoint for Na,SO,deposition.Thelattercan becalculatedfrom afree- energy-minimizationcomputercode. '' Thedewpointsare shownin Fig. 24 for a number of differentcombustion condi- tions. These factors tend to limit the region of hot corrosion, as shownin Fig. 4. There are two issues to note from this. First, although the region is limited, it is important to be aware of hot corrosion because it can be quite severe. Second, the dewpoint tends to increase with pressure. Future engines are expected to operate at higher temperatures and pressures. Higher tempera- turesdecreasethelikelihoodofdepositionbyexceedingthe dewpoint;however,higherpressureswillincreasethedew- point.Thus, a knowledgeofengineoperating parametersis essential for predicting the presence ofa deposit. S content,Na 1700I-percentcontent. PPm Y 0.5 2Ll ! 1500 L= .05,ywA-A .L 0)a E1300 c c C 'g1100 i! n 900.__ 1 001 0 ' 1 02 Pressure, P, bar Fig.24.Dewpoints for Na2S0, deposition J anuary 1993Corrosi on ofSilicon-Bused Cerurnics i n ComhiistioriEnvironments21 (2)Thermodynamics ofDeposit-Induced Corrosion Figure 25 shows a coupon of Si c treated in a burner rig both withandwithoutaddedsodium."'Notethe dramaticeffect. Elementalmaps ofa polished cross section are shown in Fig. 26. Note the even distribution ofsodium, oxygen, and silicon in the scale, suggesting the formation of sodium silicate glass. Sulfur was not detected in the scale, indicating that any depos- ited Na,SO,had decomposed. The likelyreactionscheme for this is SiC(s) +1.502(g) =SiO,(s)+CO(g) (43) (44) xSiO,(s)+Na,SO,(I)=Na,O.x(SiO,) ( I ) + SO&) These are the key reactions in hot corrosion. A solid, protective SiO, scale has been converted to a liquid scale. Transport rates through the liquid are rapid, leading to accelerated attack. Equation (44) can best be understood as an acid-basereac- tion.'x,'2xIt can be written more fundamentally as Na,SO,(I)=Na,O(s)+SO,(g) xSiO,(s)+Na,O(s)=Na,O.x(SiO,)( I )(45) (46) Itiswell-knownthatSiO,isanacidicoxide."'Following Lewis'acid-baseconcepts,SiO,reactswiththebasicoxide Na,Oto form a salt, sodium silicate. The thermodynamic activ- ity of Na,O(uNaZ0)can be determined from the partial pressure of SO, and the free energy change (AGO)of Eq. (45): (47) When Na,SO,has a high activity of Na,O,set by a low Pso,, it is called a basic molten salt. When Na,SO,has a low activity of Na,O,set by a high PsO,,it is called an acidic molten salt. From purelythermodynamicconsiderations one can deter- mine when Eq. (46) will occur. A threshold uNa,"for SiO, dis- solution can be calculated from Eq. (46) as AG&",l,,,"(45)=- 2.303RT log (Pso,a,,p) AC&a~~on(46)=2.303RT log uN;i 20(48) This border-line uNa,Ofor SiOz dissolution canbe relatedto a corresponding PsOlbyEq. (47). Thus, one can calculate, for a giventemperature,thethresholdPsO,for dissolution.Thisis showninFig.27.Notethetemperatureboundariesarethe meltingpointof Na2S04and the dewpoint for Na2S0, deposi- tion. The values of PsO,from combustion oftwo typicalfuels are alsoshown,as calculatedbyafree-energy-minimization program.l7.IXThe intersection of this line with the shaded region inFig.27 shows whenscale dissolution is expected. Experi- ments with quartz coupons in a burner rig verified these predic- tions,as showninFig.28. The couponstreatedinthelow- sulfurfuelshowedmorecorrosion,andchemicalanalysis revealed the presence of sodium silicate. The coupons treated in the high-sulfur fuel showed less corrosion, and chemical analy- sis revealed no sodium silicate. I'These calculations illustrate the value of the acid-baseinter- pretationand provide generalguidelines for predictingcorro- sion. However, as is often the case, the actual situation is more complex. Secondary elements can drive the moltensaltmore basic. It has been shown that SIC with excess carbon corrodes quite extensively in a situation where the PsO,is high enough to limit corrosion. The reason is that carbon tends to drive Na,S04 basic.lx An electrochemical cell can be used to measure uNalOin a melt:'x,'34 Pt,O,(r)lZrO,lNa,SO,,SO,lmullitel10-mol%-Ag2S04,Na,SO,,Ag (49) Zirconia acts as a membrane for oxide anions, and the mullite acts as a membrane for sodium cations. They combine, and the net reaction is (50) Na,O+Ag,SO,=0.50,+Na,SO,+2Ag E=1.498 +0.116 log uNa,() At1173 K, uNa,"is related to voltage ( E) by (51) Figure 29 shows the aNa,"measurements for Na,SO,/(O.OISO, +0,) and Na,SO,/(O.OISO,+0,) withadded carbon. Note that the latter swings strongly basic. The most likely interpreta- tion of this is the following sequence of reactions: ( 52)( 53)Itissignificantthat carbon tends to drive Na,SO,more basic because many types of Si c have excess carbon. These are more susceptible to hot corrosion. 'I" The reactive species NazO can form from other salts as well. Under some conditions Na,CO,can deposit and decompose to NaZSO,(l) +2C=Na2S +2C02 NazS +3Na,SO,=4Na,0+4S02 Na2C0, =NazO +CO&)(54) Na,CO,tends to decompose more thanNa,SO,,so thatit is a morebasicmoltensalt.McNallanet~ 1 . ' ~ ' haveshownthat Na,Ocanalsoformfromthereactionofsodiumchloride (NaCI) and water vapor: ( 55)2NaCl+H,O=Na,O(s)+2HCI Inboththesereactions,ifuNa,()issufficientlyhigh,SiOz is expected to dissolve. Deposit-induced corrosion can also originate from vanadium compounds, 131.136-11XVanadiumpentoxide(V20,)isanacidic oxide and thereforenot expected to react with SiOz. However, the phase diagram shows a limitedsolubility of SiO,in V,O,, whichleads to accelerated rates. lih.liXMixtures of Na,SO,and V,O,have led to severe corrosion of Si,N,."7 Theconceptofdissolutionbybasicmoltensaltscanbe extended to oxide slags as well. In the case of an oxide slag, it ismoredifficulttodefineandmeasurebasicity.Generally, basicity is approximated by the ratio of basic to acidic oxides. Ferber et a/." have identified three corrosionmechanisms from oxide slags. These are passivation by SiO,, dissolution of SiO,, ~ i ~ . 25.Optical micrographs of sintered sic with carbon and boron additives, treated in burner rig at 1273 K (leading edge is on the left): (a) 46 h with no sodium and (b) 13.5 h with 4-ppm sodium. 22Journalofthc American Ceramic Sociefy-JacobsonVol. 76, No. I Fig.26. with 4-ppm sodium. Image and elcmental maps of polished cross section of glassy products formed by burner rig corrosion at 1273 K with J et Afuel for 13.5 h and formation of metalsilicides at low oxygen potentials. Pas- sivation is analogous to the reaction in Eq. (44) not proceeding; dissolution is analogous to the reaction in Eq. (44) proceeding. (3)Kinetics of Deposit-InducedCorrosion The kinetics of corrosion i n Na,SO,are consistent with this interpretationof dissolutionand enhanced oxidation.'"'Kinet- ics of hot corrosionare studied in the laboratory by coating the specimen withNa,SO,and followingthe kinetics on heating. This canbeaccomplished withathermogravimetric analysis (TGA) system, suchasthatillustratedinFig.6. Aone-time deposition of salt is not as realistic as continuous deposition in the burner; however, it does provide mechanistic information on the first stages of reaction. Three situations canoccur.The first is a basicmoltensalt, such as Na,CO,or Na,SO,in contact with carbon. This situa- tion leads to the type of kinetic curve illustrated in Fig. 30. The early stages are characterizedby a period of weight loss, which corresponds to the dissolution reaction. In this case the amount of weightloss corresponds very closely to that predictedfrom the reaction"" ( 56)This is followed by a period of enhanced oxidation due to liquid scale. Finally, a protective SiO? layer forms below the melt, and the reaction slows. This final layered microstructure is shown in Fig. 3 1 .This is consistent with the study of Mayer and Riley,I4" Na,CO,+SiOz =Na2Si0, +CO, whosawrapidreactionuntiltheNa,O-SiO,liquiduswas reached, at which point a stable tridymite film formed below the melt. The second case is illustratedin Fig. 32. I t has beenshown thatNa,SO,withanoverpressureofoxygenissufficiently acidic so that SiO, is not expected to dissolve. Figure 32 shows akineticcurveforNa2S0,/02 onSi3N,.Thelong periodof weight loss corresponds to vaporization of Na,SO,;the rates are the same as those for Na,SO,vaporizingfrom a platinumcou- pon.However,theperiodofweightgaincorrespondsto enhanced oxidation. Blachere and Pettit"' have shown that even an acidic deposit can enhance oxidationbyinducingdevitrifi- cation and scale cracking or by doping the Si 02scale. The thirdcase leads to themostextreme corrosion.'"' This isillustmtedbythedatainTable VIII.Inthis caseNa,SO,/ (0.01S0,+02), a veryacidic salt, is used. Chemical analysis of the scale indicates limited attackin the presenceof this salt on the Si c and Si,N, with oxide additives. However, for the Si c withcarbonadditives,cxtensiveattackoccurs.Thisis explained by the basic effect of carbon at the bottom of the niclt and the acidic effect of NaZSO,/SO, at the top ofthe melt. These effects create the reactionsequence illustratedinFig. 33. This self-sustaining series of reactions can lead to severe corrosion. I tis termed "fluxing"and is well-known from the hot corrosion of mctals.'?x~"" Insummary, hot corrosion occurs only ina limited regime. However, it can be quite severe, and, therefore, i t is important January 1993Corrosion of Silicon-Based Cerumics in Combustion Environments23 0 Pso3limit for corrosion 0PSO, for 0.5 percent S fuel APsO3for 0.05 percent S fuel -3r Melting pointIDewpoint 11001200130014001500 Temperature, T, K i i o o 1200130014001500 Temperature, T, K Fig. 27.Calculated corrosion regimes for Si02:(a) 0.5% sulfur fuel, 2-ppmsodium, 0.025 fuel-to-air ratio,1273 K,and (b) 0.05% sulfur fuel, 2-ppmsodium, 0.025 fuel-to-air ratio, 1273 K. to be cognizant of when it occurs. The dissolution reactions in hotcorrosion allow corrosion regimes to be predicted.These reactions are best interpreted in terms of the acid-basetheory of oxide reactions. XII.Composites This review has concentrated on the near-surface durability of the Si0,-protectedSi c and Si,N4 systems. Thus, the conclu- sionsapplytomonolithicSi c andSi,N,andcompositesof these materials. Currently there is much interest in SiC-fiber- reinforced Si c and Si,N,matrices. These composites may have someuniqueoxidationandcorrosionproperties,anditi simportant to mention them. At their current state of development, these materials may be porousandcontainsecondaryelements,suchasrefractory oxides and/or carbon. These effectshavebeendiscussed.In addition, composites require fiber coatings to prevent bonding of the fibers to the matrix. Loosely bonded fibers permit effec- tivecrackbridgingandcrackdeflection,whichleadtoa tougher ceramic.However, the current fiber coatings, carbon or boron nitride (BN), are quite susceptible to oxidation ifa path for oxygen is available. Thus, oxidation may lead to a deg- radation ofthe criticalfiber interface andloss inproperties. Some composites have exhibited a low-temperature oxidation problemdue toinadequate coverage ofthepassivatingSi02 scale, which then allows oxygen penetration to the fiber coat- i ng~.'~*Recent workbyFilipuzziandNa~l ai n' ~~. ' ~critically examines oxidation ofcomposites from both an experimental and theoretical point ofview. Their study presents important results on the effect of temperature and carbon coating thick- ness on Si c fibers. Lower temperatures lead to slower oxidation rates of the carbon coating, but along a long length of the fiber; higher temperatures lead to more rapid oxidation, but along a Fig.28.SiOz couponstreatedinburnerrigin2-ppmsodiumat 1273 K:(a)No.2 diesel fuel(0.5%sulfur), 5h(point AonFig. 27(a)), and (b) J et Afuel (0.05%sulfur), 1h (point B on Fig. 27(b)). shortlength ofthefiber. Theyalso show thata thinner fiber coating is desirable, since it leads to healing more readily than thicker coatings. Improvement in oxidation and corrosion prop- erties is an important driving force in composite development. 0Na,SO,/O.OlSO3 0 Na2S04/0.01 SO3 +C (C at bottom ofmelt) I 0-15 0t23 Time. h Fig. 29.Effect of carbon on basicity of Na2S0,. 24Journal o/the American Ceramic Soc,iet~-Jac.ohJonVol. 76, No. I .1 Xl 0 - 6 .1 xlO-6 0 L 0 0N E -.l 0 p -.3 0 0 QI Q-.5Q rn ;F -.4.- -.6 -.7 111 -.l- -.7fl 0.2.4016324%64 Time, hTi me, h Fig.30.Kinetics for Si c +Na,CO, .16xl O4 N 0Na2S0, vaporization; 0Oxidation !E-.32 - 0 0, Q -.481 016324%6480 Ti me, hr Fig.32.Kinetics for Si,N,+Na,SO,. XIII.Corrosion and Degradation of Mechanical Properties The focus of this paper has bcen on chemical mechanisms of degradation. Amajor issue has been the assessment of this deg- radation.Anassessmentintermsofmicrostructuralchanges Fig.31.Polished cross section showing final microstructure for Si c +Na2C0,alter 48 h of reaction. J anuary 1993 Acidic Table VIII.Quantity of Corrosion Products' lo SIO. Na,O- r(S10.) NL SO, Material('llgicln')(m$ciii')1(Illg/crl1 JSingle-crystal Si c1.64 t0.86