<<

of -Based in Combustion Environments

Nathan S. Jacobson* National Aeronautics and Space Administration, Lewis Research Center, Cleveland, Ohio 44 135

Silicon-based ceramics and composites are prime candi- valves and piston heads,4 as shown in Fig. I(b). The use of dates for and structural compo- tubes as heat exchangers has been proposed for indus- nents. In such applications these are exposed to trial , such as remelt furnaces, soaking pits, combustion gases and deposit-forming corrodents. In this and aluminum reclamation furnaces.' An example is illustrated paper combustion environments are defined for various in Fig. I(c). In addition, ceramics are under consideration for applications. These environments to five main types of heat exchangers in coal-fired combustors.' A more developed corrosive degradation: passive oxidation, deposit-induced application is the use of ceramic tubes for indirect heating, corrosion, active oxidation, scale/substrate interactions, where hot combustion gases pass through the tube, heating a and scale volatility. Each of these is discussed in detail. The process on the out~ide.~The structural components of the hot key issues in oxidation mechanisms of high-purity silicon sections are subject to a range of chemical attack processes, (Sic) and silicon (Si,N4) in pure are depending on the temperature, , and chemical discussed. The complicating factors due to the actual com- environment. bustion environment and commercial materials are dis- The focus of this paper is on the silicon-based ceramics sili- cussed. These discussions include secondary elements in the con carbide (Sic) and (SilN,,). These materials ceramics; additional oxidants, such as and are inherently unstable in air and form a thin layer of (COJ; combustion environment impurities; long- dioxide (SO,) in an oxidizing environment. SiO?has the lowest term oxidation effects; and thermal cycling. Active oxida- permeability to oxygen of any of the common and forms tion is expected in a limited number of combustion situa- an effective reaction barrier.' Therefore, silicon-based ceramics tions, and the active-to-passive transition is discussed. At have the potential of substantially better high-temperature oxi- high temperatures the limiting factors are scale melting, dation behavior than . The protective scales on sili- scale volatility, and scale/substrate interactions. Deposit- con, Sic, and Si,N, are shown schematically in Fig. 2. Note induced corrosion is discussed, primarily for sulfate that Si,N, forms a (Si2N20)layer below the (Na,SO,), but also for vanadate and oxide-slag deposits as SiOz layer.'SiC may also form an oxycarbide layer, but there is well. In applying ceramics in combustion environments it is only limited evidence for this." The analogous protective layers essential to be aware of these corrosion routes and how they on superalloys for high-temperature applications are alumina affect the of a component. (A120J and chromia (Cr203)." Extensive work has been done on the performance of AI,O, and Cr,O, scales in combustion I. Introduction environments. In general, our knowledge of the behavior of SiO, scales lags behind that of A120, and Cr20,. ANY potential uses of silicon-based ceramics and com- The purpose of this paper is to review our current state of Mposites involve exposure to combustion gases. These knowledge of the interaction of SiO, with combustion environ- applications range from hot-section structural components of ments. Because the focus is on the interaction of SiO? with the gas turbines and piston engines to heat exchanger tubes for environment, many of the conclusions apply to all Si0,-pro- industrial furnaces. Gas turbines are used in and electric tected materials. This includes composites of Sic and Si,N,, power generation and have been tested for automobiles. Poten- such as Sic--reinforced Sic matrices and SiC-fiber-rein- tial ceramic components of these engines include combustor forced Si3N, matrices, and SO,-forming alloys, such as molyb- liners and, perhaps some day, turbine The location of denum disilicide (MoSi,). In this paper the major chemical these components in a engine is shown in Fig. I(a). degradation routes are discussed. Particular emphasis is on the In piston engines the potential ceramic components include mechanisms of corrosion and the key questions involving them.

11. Environments S. M. Wiedcrhorn~ontrihutingeditor Combustion environments vary widely, depending on , temperature, pressure, and oxidizer. are complex mix- tures of , and they are classified by boiling points. Manuscript No. 1955915. Received June 15. 1992; approved Septembcr 28, 19?2. Lower-boiling-point fuels include automotive gasoline, and Member, American Ccramic Society. higher-boiling-point fuels include aviation fuel and fuel oils.12 4 Journul ofthe American Ceramic Society-Jacobson Vol. 76, No. 1 ,- Fan r Turbine I blades I Stator vanes \ I

Sample holder --,Flue gas

(b) (4

Fig. 1. Schematics of proposed applications lor ceramics where ceramics will be subjected to combustion environments: (a) gas turbine engine, (b) piston engine, and (c) aluininum reclamation (see Ref. 5).

Table I lists some common fuels and their approximate impurity equivalence ratio of 1 is stoichiometric, an equivalence ratio of contents. "-lh Note that all fuels have some amount of , less than 1 denotes a fuel-lean region, and an equivalence ratio which may lead to corrosion. The last category of fuel oils con- greater than I denotes a fuel-rich region. Most gas turbines tains more sodium, , and than the other operate in the fuel-lean or stoichiometric regions. These fuels. These lower-purity fuels are likely to be utilized more in regions contain large amounts of oxygen with the CO, and H,O. the future as petroleum resources decrease. Impurities have a Most corrosion studies have been performed in the fuel-lean major influence on corrosion. Table I does not include coal- region. However, some novel combustor designs may involve derived fuels, which generally have even higher impurity the fuel-rich region, which produces larger amounts of carbon levels. monoxide (CO) and gas (HJ. However, this region Combustion is the oxidation of these fuels to the stable prod- also contains the combustion products C02and H,O. Note that ucts (C02)and water (H20). Equilibrium com- these are equilibrium calculations; if equilibrium is not bustion products can be calculated by mixing the fuel and the attained, other species, such as elemental carbon, may form. oxidant together in a free--minimization computer The goal of a corrosion study is to understand the chemical code. Figure 3(a) shows the equilibrium combustion products reactions that occur between these combustion products and the for a standard aviation fuel (Jet A-CH, 91xs) as a function of proposed hot-gas-path structural materials. equivalence ratio. Equivalence ratio is defined as the fuel-to-air Figure 3(b) shows the adiabatic flame temperature as a func- ratio at a particular point divided by the stoichiometric fuel-to- tion of equivalence ratio. This temperature is calculated by air ratio for complete combustion to CO, and H20. Thus, an using a free-energy-minimization computer code and setting January 1993 Corrosion of Silicon-Based Ceramics in Combustion Environments 5

LSi02

0 .5 1.o 1.5 2.0 Equivalence ratio

2500 r SiOp L 2000 1500 1000 500 (b) I I I ~~~ I 0 .5 1 .o 1.5 2.0 Equivalence ratio

Fig. 3. Calculated gas composition and Rarne temperature 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- salts that can deposit on the heat exchanger.' In a coal- tired furnace 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. Fig. 2. Schematic of protective oxide scales: (a) on Si, (b) on Sic, Table I1 summarizes the composition of these combustion and (c) on Si,N,. environments. This list is by no means exhaustive. As noted, specific applications generate specific corrodents. For example, the net heat loss equal to zero." This is an idealized case that combustion of municipal wastes may generate hydrogen chlo- assumes no heat loss through the walls of the combustion cham- ride (HCI).'" In summary, combustion environments are com- ber. It does give the maximum temperature of the flame, how- plex, involving not only the combustion products COz and ever. In general, the wall materials and the components H,O, but a variety of other gases and possible deposits as well. downstream from the flame are at lower temperatures. Temperatures and are again quite dependent on the Corrosion occurs not only by gaseous coinbustion products, particular system. The ranges are listed in Table II. An but also by deposits. Perhaps the most common deposit is important issue to consider is thermal cycling. Some applica- sodium sulfate (Na,SO,), which forms when sodium reacts with tions, such as a utility turbine or industrial furnace, involve sulfur fuel impurities.'X The sodium may originate from a essentially isothermal exposures for long periods. Other appli- marine environment, from a salted roadway, or as a fuel impu- cations, such as an aircraft gas turbine, involve thermal rity. Corrosion by Na,SO, is termed "hot corrosion" and is dis- cycling. cussed in detail herein. Other types of deposit-induced corrosion originate from vanadium fuel impurities and from oxide slags. Ill. Qpes of Corrosive Attack and In some ways the piston chamber of an internal combustion Experimental Techniques engine is a more complex environment than the hot section of a gas turbine. Pressures and temperatures vary through the Figure 4 shows the major types of corrosive attack as a func- stroke of the piston. Adiabatic flame temperatures and combus- tion of pressure and reciprocal temperature. The major types of tion gas product compositions can be calculated by the same corrosive degradation are passive oxidation, deposit-induced free-energy-minimization program described previously with corrosion, active oxidation, oxideisubstrate interactions, and (C,H,) as a model fuel. The results of this calculation scale vaporization. The temperature boundaries between these are similar to those for Jet A aviation fuel. Current automotive types are only approximate and are dependent on the specific engines run slightly fuel rich to allow proper operation of pollu- system. Furthermore, rarely is one mechanism operative. In tion control devices. practice, several mechanisms operate simultaneously. In addition to heat engines, silicon-based ceramics are also Laboratory studies of corrosive degradation fall into two prime candidates for heat exchanger tubes in industrial fur- main categories: burner rig studies and laboratory furnace stud- naces. The environment encountered by a heat exchanger var- ies. Burners more accurately model the actual combustion situ- ies widely. Again, one would expect large amounts of COz and ation. Figure 5 shows a typical high-velocity burner rig.

Table I. Properties of Some Common Fuels Boding range H C molar S content Nd + K content V content Fuel (K)* rdtio (wt%)' (ppm)' (ppd Unleaded automotive gasoline 300-375 2.02 0.15 3.6 <0.003 Jet A (commercial aviation fuel) 450-560 1.92 0.05 -10 0.06 Fuel oils 450-6 15 1.61 0.1-1 .0 10-20 <300

~ *Reference 12. 'Refcrcnce 13. 'References 14and IS. 'Reference 15. 'References 15 and 16 6 Journal of the American Cerumic Sociep-Jucohson Vol. 76, No. 1

Table 11. Summarv of Combustion Environments Application Tcrnperaturc (K) Pressure (bar)’ Gas atmosphere Potential deposit Heat engines Gas turbine 1173-1673 1-50 Fuel lean: NZ,O?, CO,, HZO; Na2S0,, NaZV,O,. fuel rich: N2. COz,HzO, CO, H? Piston 1273-1 873’ 1-10 N2, COZ, HZO,O2 Heat exchanger 1273-1 873 -I Oxidizing Alkali halides, Na2S0,, transition-metal oxides Coal combustion 1473-1673 1-10 Reducing Acidic or basic coal slags

’ I bar = 1o‘Pa. ‘Cycling.

- NaqSOq-induced SiO, has some unique properties that influence its perfor- ‘ corrosion of SiO,- 7 mance as a protective oxide. In most oxidation experiments, SiO, forms as an amorphous film and then crystallizes to either cristobalite or tridymite. This is significant because these phases have very different physical and chemical properties.” It is generally accepted that the mobile species is oxygen, not silicon. For this reason the chemical reaction occurs at the Passive oxidation Si02/Si or SiOz/ceramic interface.”-23Transport through SiO, can occur by diffusion of molecular oxygen as interstitials or by network exchange of ionic The latter is referred to as either “network exchange diffusion” or “ionic diffusion” in 9 8 7 6 5x1 O4 this paper. Molecular oxygen diffusion coefficients are mea- Reciprocal temperature, 1/T, K sured by examining permeation through a SiO, membrane;” network exchange diffusion coefficients are measured by an isotope exchange technique."^" Table 111 lists both types of dif- 1200 1400 1600 1800 2000 fusion coefficients. The molecular oxygen diffusion coefficient -Temperature, T, K is roughly 10‘ times greater than the ionic oxygen diffusion coefficient. There have been attempts to correlate these two Fig. 4. Major types of corrosive attack and degradation as an types of diffusion coefficient^;'^ however, Cawleyj” has approximate function of reciprocal temperature (P,,,,,,= Pc,x,ddn,). reported the difficulties with such a correlation due to network/ interstitial exchange and mobile network . These issues and the disparity of the diffusion coefficient measurements (Table However, burner rig tests are expensive, and it is difficult to 111) indicate there are still some unresolved issues in these fun- precisely control all the operating parameters. Laboratory fur- damental transport quantities. nace tests offer much more accurate control of system parame- Before discussing the oxidation of SIC and Si,N,, it is appro- ters. Gas composition, pressure, and temperature can be priate to discuss the oxidation of pure silicon. Because this pro- accurately controlled. The extent of corrosion can be monitored cess is critically important to the , it has by following the weight gain with a microbalance or by measur- been extensively studied. The classic paper in this area is by ing scale thickness with various optical techniques. Figure 6 Deal and Grove.” They view oxidation as consisting of three shows a standard laboratory microbalance apparatus for iso- distinct steps, as shown in Fig. 8. The three steps are transfer of thermal oxidation studies. Standard -optical techniques the gaseous oxidant to the outer surface of the oxide film, diffu- have been useful for characterizing corroded specimens. In gen- sion through the oxide film, and reaction at the oxideisilkon eral, burner rig tests are helpful for an overall assessment of interface. From these steps they derive the following linear- behavior in the actual environment. Laboratory furnace tests parabolic equation: complement burner rig tests by permitting isolation of individ- x’ Ax = B(t ual effects so that they can be understood on a fundamental + + T) (1) basis. where x is the scale thickness, B the parabolic rate constant, t the time, and T the time shift corresponding to the presence of IV. Isothermal Oxidation an initial oxide layer. The quantity BIA is the linear rate con- stant. The parabolic rate constant is given by

(1) General Considerations and Pure Silicon Oxidation B = 2D,,,C*(02)/N,, A large amount of research has been performed on the iso- thermal, passive oxidation of silicon-based ceramics. This where D,,, is the diffusion coefficient through the film, C*(O,) research is complex because of the various secondary effects, is the equilibrium concentration of oxidant in the scale, N,, the such as other oxidants and impurities in the ceramics and the number of oxygen incorporated into the SiO, scale environment. These effects are noted in Fig. 7. Fundamental per unit volume. Note that No must be modified for Sic oxida- studies are at the center of the circle-pure materials and pure tion due to the formation of CO. For short oxidation times, oxi- oxygen environments. Eliminating the complicating factors dation follows a linear rate law. The physical interpretation of allows detailed studies of the basic oxidation mechanism. this linear region is still controversial; it has been attributed to Actual combustion environments are more complex, involving interface control” or to diffusion control which is nonparabolic the various factors on the outer circle in Fig. 7. due to strain effects in the oxide.’2 For longer times, oxidation The focus of this section is on fundamental studies of Sic and follows a parabolic law, and diffusion through the thicker oxide Si,N, oxidation. These studies provide an atomistic understand- is rate controlling. An activation energy can be determined ing of the mechanism of SiO, scale growth. These include from a plot of In B vs l/T. The magnitude of the activation determining the slow reaction step (or steps) and understanding energy can reveal useful information about the diffusion pro- the diffusion mechanism. Despite the large number of papers in cess. Deal and Grove” report that care must be taken to inter- this area, many questions remain unanswered. This paper is not pret the oxidation process at the appropriate times (i.e., intended to he an exhaustive review but rather a discussion of parabolic behavior cannot be assumed when the process is still some of the key issues involved. in the linear region). They have shown that Eq. (I) describes a January 1993 Corrosion ojSilicon-Based Ceramics in Combustion Environments 7

Ignitor Salt solution

(a)

Fig. 5. Mach 0.3 burner rig. (Courtesy of M. Cuy, NASA Lewis.) wide range of oxidation data. Calculations of the parabolic rate there is a countercurrent of gas. The following reactions are constant that are based on the molecular oxygen diffusion coef- generally accepted: ficient of Norton2'show good agreement with the measured val- SiC(s) 1.502(g)= SiO,(s) CO(g) ues, and the activation energy for parabolic oxidation is close to + + (3) that for molecular oxygen diffusion through SiO?. Thus, it is Si,N,(s) + 0.7502(g) = I.SSi,N,O(s) + 0.5N2(g) (4a) generally accepted that molecular oxygen diffusion through the SiO, layer is rate ~ontrolling.~~.~~ SizN20(s)+ 1.50z(g) = 2Si02(s) + N,(g) (46) The oxidation of Sic and Si3N4is more complex. Figure 2 However, the actual processes may involve other reactions. The illustrates some of the processes involved. In both ceramics production of CO in the oxidation of Sic has been observed 8 Journal of the American Ceramic Society-Jacobson Vol. 76, No. 1

diffusion of oxygen through the oxide film, (3) reaction at the oxideiceramic interface, (4) transport of product gases (e.g., CO and ) through the oxide film, and (5) transport of product gases away from the surface. This process is even more complex in the Si,N, case because of the duplex film formation. Many questions remain about these steps. The key question concerns the rate-controlling, or slowest, step. This has been an area of controversy for both Sic and Si,N, and is discussed in further detail herein. Another question deals with the transport Purified gas in mechanism through the SiOz scale. Is it permeation through the network by oxygen molecules or a network exchange mecha- nism of 02-?Even less is known about the transport mecha- nism of product gases (CO and nitrogen) outward through the r- SiOz scale. As noted, SiOz undergoes a number of phase changes. This section deals primarily with amorphous scales. Crystallization, which complicates the oxidation process, is discussed separately. (2) First consider the oxidation of Sic. The key observations have been summarized by several in~estigators."~"Generally, Thermocouple only one scale forms, although there is limited evidence that an L-Gas out oxycarbide may form between the oxide and the Sic."' Figure 9 shows a representative kinetic curve. Many investigators have Fig. 6. Schematic of oxidation balance apparatus observed a brief (<<1 h) linear region followed by a parabolic region."+" The focus of most investigations has been on the parabolic region, as shown in Fig. 9. Figure I0 shows the para- Other bolic rate constants for a number of Sic materials. Only the oxidants higher-purity materials (i.e., chemically-vapor-deposited (CVD) Sic and single- Sic) are discussed in this section. The key question concerns the rate-controlling step. There are three possibilities: (I) oxygen diffusion inward, (2) CO dif- Impurities fusion outward, and/or (3) an interfacial reaction. Recently, Thermal (SO,, HCI, Na, K), Luthra"' systematically examined these possibilities based on transition metals available experimental data. His general approach is taken / \ isothermal here. Table IV summarizes some of the experimental observa- oxidation, tions and their implications. pure materials, (A) Oxygen Diflusion Inward: In general, most of the pure oxygen data imply oxygen diffusion inward as rate limiting. Rates are Ceramici / parabolic and dependent on the partial pressure of oxygen additive (Po,). Motzfeld4' has reviewed the literature to about 1963 and effects crystallization carefully compared the rate constants for silicon and Sic. After correcting for the stoichiometry difference (i.e., the additional oxygen necessary to oxidize carbon), he finds that silicon and Long-term Sic have essentially the same rates and the same activation isothermal . Therefore, he concludes that the same process that effects controls silicon oxidation also controls Sic oxidation (i.e., oxygen diffusion inward). Much of the data he uses are for Sic Fig. 7. Complications to isothermal oxidation due to additive-con- powder. The difficulty in a precise determination of the surface taining ceramics and combustion environments. area and possible impurities suggests that high-purity coupons may give more precise data for Sic. Recent measurements on single-crystal give rates somewhat slower than those for with the gas ~hromatograph,.'~although the possibility of fur- silicon, as shown in Fig. 10. ther oxidation to CO? must be considered. There has been no Table V summarizes the activation energies for oxidation of clear observation of the production of elemental carbon. The high-purity Sic. Note that there is generally a low-temperature oxidation of Si,N, may lead to the production of (T < 1623 K) and a high-temperature (T > 1623 K) regime. (NO(g)) in addition to molecular nitrogen gas, as observed with Consider the low-temperature regime. Here the activation a mass ~pectrometer.~",~~Note that, even though the reactions energy is low, about 120 to 140 kJ/mol, and is similar to that for lead to a net weight gain, there is also some weight loss due to oxidation of pure silicon and molecular oxygen diffusion gas evolution. These reactions may be monitored by following through amorphous SiO,. This supports molecular oxygen dif- net weight gain, scale thickness as a function of time, or even fusion inward as a rate-controlling step for Sic oxidation below gas evolution. The following factors allow conversion of para- I623 K. bolic rate constants from scale thickness to net weight gain:" Consider further the type of transport through the growing 1 (m'is) = I .67 x lo-" (kg'/(m4.s)) for Sic (5) SiO, layer. As mentioned, the activation energy for oxidation becomes much greater after 1623 to 1673 K for controlled- I (m'is) = 3.77 X lo-" (kgLi(m4.s)) for Si,N, (6) nucleation thermally deposited (CNTD) Sic and the carbon Note that Eq. (5)is based on Eq. (3) and that Eq. (6) is based on face of single-crystal Sic. Zheng et ~1.~~.~~have explained this a combination of Eqs. (4a) and (4b), i.e., the oxidation of Si,N, phenomenon as a change in the oxygen transport mechanism. to SO,. This is appropriate since SiO, is formed in much larger Below about 1623 K, the activation energy for the parabolic amounts than Si2N,0.' rate constant is close to that for silicon oxidation, and diffusion The oxidation process for Sic and Si,N, involves five steps: of molecular oxygen through the scale is rate controlling. They (I) transport of molecular oxygen gas to the oxide surface, (2) show this by forming a scale in '"0and then reoxidizing in ''0. January 1993 Corrosion of Silicon -Rri.srd Ccrntnics in Cornbus tion Eli virotinwnrs 9

Table 111. Diffusion Coefficients for Oxygen in Amorphous SiO, Typc 01 diftuwn coctlicicnt. Ditfu\ cocflicicnt lnvestigator Twhniquc Icmpcralurc rlrnpc (K) valuc (Ill'/\)

Norton" Permeation MolecularO,, 973-1 373 2.9 x 10 'exp( - 112900iRTj Isotope exchange Network exchange. I 123- IS23 2.0 X 10 '.'exp(- 121336/RT) Sucov2' isotope exchange Network exchange, 1198-1498 1.5 X 10 "exp(-29790l/RT) Muehlenbachs and Schaeffer'" Isotope exchange Network exchange, 1423-1 703 4.4 X 10- "exp( - 82425iRT) 'E* in Jimol.

Gas C' = 3K"D::P,;:,'" where the subscripts i and R refer to the SiC/SiOl interface and SiOJgas interface, respectively; D,, is the diffusivity of oxygen; and the K' and K" are constants. Equation (9) simplifies since P,,ol >> P$,(),and there is a negative vacancy gradient as described by Eq. (76). Thus, k, shows very little dependence on external Po,. This is in accordance with the oxidation literature for scale growth on materials with this type of oxide defect structure.'" The change in Po, dependence of k,, with tempera- Gas-to-oxide Diffusion Interface ture4' is consistent with a transition from to transfer, in oxide, reaction, a larger contribution of ionic diffusion. Note also that, with a Jg = h (C' - Co) J3=kCj c0-ci defect structure different from that in Eq. (7a),the Po, depen- J2 = Deff xg dence of k, changes. (5) CO Diflusion Ourwurd: Some investigators have Fig. 8. Processes involved in silicon oxidation. (After Deal and attributed the higher activation energies at higher temperatures Grove." j to a transition to CO-diffusion-outward rate control. K'' Two factors tend to discount this. Zheng rt a/.'' have oxidized Sil'C and examined the resultant scale by using SIMS. They found no An accumulation of "0 at the SiO,/SiC and SiOjgas interface, carbon gradient, as one would expect if CO diffuses outward as determined by secondary ion mass spectrometry (SJMS), slowly. Thermodynamic arguments also support the rapid trans- supports this. Above 1673 K the activation energy becomes port of CO outward. Suppose the reverse is true and CO dif- higher, and this is attributed to a mixture of network exchange fuses outward slowly. Then Po, at the SiC/SiOz interface must diffusion and molecular oxygen diffusion. Zheng et ~1.~'~'~have be close to I, and, as required by diffusion control. Eq. (3) observed a constant distribution of "0 through the scale in their equilibrium is maintained; the Pco has a value of about loz7bar reoxidation experiment. They attribute this to the exchange (I bar = 10' Pa) at 1600 K. Extremely high pressure would be process. However, Cawley and Boyce4' have reported that, expected to blow the scale off, but this is not observed. Some without some type of concentration gradient, it is difficult to type of "CO pressure gauge" is needed at the SiCiSiO-, inter-' draw conclusions about the diffusion mechanism. face. The formation of bubbles in the scale may provide some A transition from molecular oxygen to network oxygen dif- indication of pressure buildup." However, there are two difti- fusion should be reflected in the dependence of the parabolic culties with this. First, the precise pressure to form a bubble is rate constant k, on Po,. For molecular oxygen diffusion, the rate not clear-it depends on local surface tension and a variety of constant should be pioportional to Po-, assuming that the solu- other factors. Second, the source of bubbles at high tempera- bility of oxygen in SiO, obeys Henry's law. For network tures may also be the Sic + SiOLreaction,'' as discussed later exchange diffusion, one expects k, to show a weak dependence in this paper. on P0,.4x.s0 Zheng et have found k, to be dependent on (C) Interfircia1 Rraction: The other possibility for rate (P,2y, where n varies from 0.6 at 1473 K to 0.3 at 1773 K. This control is interfacial reaction. Some investigators have reported linear reaction rates for the silicon face of single-crystal Sic, variation is generally consistent with the proposed transition which is attributed to interfacial reaction However, between the two types of diffusion. At lower temperatures one most of the data for pure Sic indicate parabolic kinetics for the might expect a value of n closer to I for molecular oxygen dif- majority of the reaction period. Table IV summarizes the exper- fusion. Zheng et al. attribute the difference to deviations from imental observations and the rate-controlling step supported. Henry's law for the of oxygen in SO2. Narushima Most of the evidence implies oxygen diffusion inward as being et al." have found k, to be proportional to (Po2)''2at 1823 to rate controlling. 1948 K. From the high activation energy, Narushima et al. con- However, the situation is not so clear. The most recent data clude that network exchange diffusion of oxygen inward is rate indicate that single-crystal Sic has rates slower than that of controlling. pure silicon, even with the stoichiometry correction. The In the case of network exchange diffusion, the expected Po, (0001) silicon face of single-crystal Sic shows rates slower dependence is important to explore further. Using standard than that of the (000T) carbon face." There is no explanation Kroger-Vink notation and assuming that oxygen vacancies are for this. Furthermore, the assumption of oxygen diffusion the primary defect, inward and CO diffusion outward assumes that one is fast and the other is slow. Although CO is polar and molecular oxygen is 0, = V;; + 2e' + 0.502 (70) not, it is difficult to imagine that they would have dramatically different permeation properties. These three facts make it dif- [V;] = K' Pi,'" ficult to accept oxygen diffusion inward as completely rate con- Do = DZPG2"" trolling. Physical phenomena do not always fall into distinct categories. As Luthra'X reports, it may be that a mixed control According to the Wagner oxidation theory'" for ionic diffusion, mechanism is operative and both diffusion and the interface JournuI of the American Ceramic Society-Jucobson Vol. 76,No. I

0 20 40 60 80 100 120 Time, hr

.I6 -

Kp = 1.3x104 mg2/cm4h

0 20 40 60 80 100 120 Time, h

Fig. 9. Sample oxidation curves for CVD Sic in dry oxygen (7 = I673 K)?' (a) weight gain versus time (I mgicm' = 10' kgim') and (b) (weight gain)2versus time (1 mg'/(cm'.h) = 3.6 X lo7kg'/(m''.s)). reaction are rate controlling. Because of the different rates of observed parabolic kinetics, suggesting that either oxygen dif- single-crystal Sic and silicon, Luthra advocates mixed inter- fusion inward is rate limiting or nitrogen diffusion outward is face and CO diffusion outward control. However, the limited rate Iimlting,9.4!.5J-55 The parabolic rate constant varies linearly evidence for high CO pressures at the Si0,iSiC interface and with oxygen pressure' at 1523 to 1673 K, and this variation lack of a carbon gradient tend to oppose CO diffusion outward. suggests a molecular oxygen diffusion mechanism. Du et al." More work is necessary to determine the rate-controlling step have performed double oxidation experiments with "OZ and (or steps). Ih O2 and suggest that permeation by molecular oxygen is the primary diffusion mechanism and that considerable exchange (3) Silicon Nitride with the network occurs at higher temperatures. The higher Next consider Si,N, oxidation. This is more complex activation energies for Si,N, as compared with silicon (Table because of the duplex layer formation. Figure 1 1 shows the par- V1) have been attributed to oxygen transport through Si2N20." abolic rate constants for a range of Si,N, materials. As in the As in the Sic case, rate control by diffusion of a gaseous prod- Sic case, the focus in this section is on the high-purity materials uct outward is unlikely because this would create a very large (i.e., CVD Si,N,). Perhaps the most obvious thing about the nitrogen pressure at the oxidehitride interface. Oxidation stud- CVD Si,N, is the steeper slope-the activation energy ies at very high temperatures show only limited bubbling." An is substantially more than that for pure silicon. Table V1 lists interfacial reaction-rate-limiting step is not likely because lin- some activation energies for CVD Si,N,. In addition, the rates ear rates have not been clearly observed. Table VI1 summarizes are slower than those for pure silicon and pure Sic for short oxi- the experimental observations and the rate-controlling steps. dation times (-5 h) below about 1700 K; above 1700 K the Again, most of the data support oxygen diffusion inward as a rates are similar. These facts suggest that the oxidation mecha- rate-controlling step, but there are some inconsistencies, as are nism for Si,N, is different from that for SIC. shown below. Luthra3' has also systematically investigated the possible The difficulty with oxygen diffusion inward as a rate-control- rate-controlling steps for Si,N,. Most investigators have ling step arises from the substantially slower rates of Si,N, as January 1993 Corrosion ofSilicon-Bu,sed Ceramics in Cornbustion Environinents

A Single-crystal silicon, Deal and Grove3' B Single-crystal Sic, Zheng et al.47 C Chemically vapor-deposited (CVD) Sic, Schiroky et aL41 D CVD Sic, Narushima et al.5i E Controlled-nucleation, thermally deposited Sic, Costello and Tressler et al.42 F Sintered a-Sic, Costello and Tressler et al.42 G Hot-pressed SiC(AI2O3,0.20-atm 02),Hinze et aLm H Hot-pressed SIC(AI2O3, Singhal el aL61

-1 5 \L \

v) -16 2 E a" P ' -17

-1 8

I I -19 I I 5 6 7 8x1 O4 Reciprocal temperature, 1/T, K Fig. 10. Parabolic rate constants for a variety of Sic materials

Table IV. Rate-Controlling Steps for Sic Oxidation Observations suDDort Oxygfn diffusion lntcrfilcial CO diffusion Observation inward reaction outward RcTerence(5) Parabolic kinetics J J 4 1 ,42,46,47,51 k,~& (0.3 k,(C face-Sic) J J 47 k,(C face) > k,(Si face) J J J 42,44,47 'Permcation of O1.'Network exchange of o? ".

Table V. Activation Energies for Sic Oxidation Investigators Tcrnperdture rangc (K) Activation energy, E* (klirnol) Deal and Grove" (Si) 1073-1473 119.3 Costello and Tressler4' (CNTD Sic) 1473-1673 142 Zheng el al.'" (single-crystal Sic) 1673-1773 293 C face 1473-1623 120 C face 1623-1773 260 Si face 1473-1773 223-298 Schiroky et aL4' 1640- I 820 125.5 Narushima et ul." 1823-1948 345 12 Journal ofthe Americun Cerumic Society-Jacobson Vol. 76. No. 1

Single-crystal silicon, Deal and Grove3' Chemically vapordeposited (CVD) Si3N4, Du et ak9 CVD Si3N4, Choi et aI.= CVD Si3N4, Schiroky et a14' CVD Si3N4, Hirai et aI.% Hot-pressed Si3N4 (1 wt40 MgO), Singhala Hot-pressed Si3N4 (1 wt% MgO), 0.20-atm 02,Tripp and Graham70 Hot-pressed Si3N, (5 wtY0 MgO), Schlichting and Ga~ckler~~ Hot-pressed Si3N, (3 wtY0 La&), Schlichting and Ga~ckler~~

-12 r

-14 - GLX F

3 -16 - E ," -8 -18 - \\ - -20 -

Table V1. Activation Energies for Si,N, Oxidation Investigators Temperature range (K) Activation energy. Ev (kJ/niol) Deal and Grove" (Si) 1073-1473 119.3 Du el ul." I 373- I673 464 Choi ef a/." 1273-1573 330 Hirai 1823-1 923 420 Schiroky et 1673-1 873 276

Table VII. Rate-Controlling Steps for Si,N, Oxidation Observations suppiirt Oxygm diffusion lnterlacial Nitrogcn dil luaion Observation inward reaction outward Rctcrcncc(s1 Parabolic kinetics' J J 9,54,55 Activation energy E* > E* (Si, Sic) ,? ,? 9,54 k,, cx Po, J 9 k not flPN,) J 9 do indication of large J 41 PN,at Si021Si,N, Sonic deviations froin paraholic kinetics-wmc inllucncc of intcrfacc ('!) compared with silicon. This fact to unreasonably high pressures at the SiO,/Si,N,O interface. This has been discussed by both Du et a/." and LuthraIXand the argument follows. The oxygen pressure dependence and double oxidation experiments suggest that, at low temperatures, molecular oxygen diffusion where D is the diffusivity of oxygen in SiO', K a proportional- occurs for silicon and Si,N,. Because the SiLN20layer is only ity constant relating pressure to concentration, APo, the pres- about one-tenth of the SiO, layer, consider the parabolic growth sure difference through the SiO? scale on the designated rate of the SiO, scales on Si,N, only. This focus is then on the substrate, and No the number of oxygen atoms incorporated associated equilibrium at the Si02/Si,N,0 interface. The para- into the scale. All the constants, including D, cancel, and bolic rate constant for Si,N, is about lo-' that of silicon at 1600 aps,o1 - 10'AP;\;:N4 K, and, thus, (12) January 1993 Corrosion Silicon-Busd Crrumic~sin Combustion Environrmxt.s 13

Because the oxidation of silicon fits well with known diffusion coefficients, the pressure at the oxideisilicon interface is set by the SiO,/Si equilibrium and can be taken as zero. Thus, us, - o2 - P&O, = 1O2(P,.O2- p2y (13) If Px,o,is taken as I, then P:;;: is about 0.99 bar. A condition for diffusion control is thermodynamic equilibrium at the SO,/ Si,N,O interface: Si,N,O + 1.50, = 2Si0, + N, (14) Putting a value of 0.99 bar for Po, into the equilibrium constant for this equation gives a nitrogen pressure of 1 Oz4 bar at 1600 K, an impossibly high value. There are three possible explanations for this: (1) thermodynamic equilibrium is not attained at the interface, (2) SiOz scale formed on Si,N, may be different from that formed on silicon, and (3)network exchange diffusion may Fig. 12. Cross section ofSi,N, oxidized for48 h at 1673 K, ahowing occur in the Si3N, case at the same temperature at which rnolec- SilN,O layer. The SiO, layer has been thinned. (Courtcsy OIL.Ogbtiji. ular oxygen diffusion occurs in the silicon case. If the first is NASA 1.cwis.) true, diffusion cannot be truly rate controlling. This supports some type of mixed control, which has been suggested by Luthra.'8.58There is no clear evidence of a different type of SiO, may often crystallize because of impurities in the substrate or scale being formed on Si,N, than on silicon. Some investigators water in the environment. Figure 13 shows a micrograph of have suggested that nitrogen or nitrogen compounds diffusing spherulitic cristobalite forming on Sic. In general, transport in outward may create some sort of blocking effect to oxygen dif- cristobalite is slower than transport in amorphous Si02. Cos- fusion inward,', but this is difficult to prove experimentally. If tello and Tre~sler~~have correlated a slowing of reaction rates nitrogen has such an effect, one would expect a dependence of with crystallization of the oxide scale. The exact nature ofcrys- the rate on PN2,which has not been observed.'Double oxidation tallization is highly dependent on impurities in the specimen experiments support molecular oxygen diffusion for both sili- and/or the atmosphere. However, for sintered a-Sic with con and Si,N,, discounting the third explanation." and carbon additives, crystallization begins at the Si0,iSiC LuthraSXhas reconciled these differences by suggesting rate interface and spreads to the surface .23.h" control by mixed interfacial reaction and nitrogen diffusion out- Densification aids in Sic and Si,N, have an even more com- ward. He replots the data of Du et UI.~on a plot of logarithm of plex effect. The oxide additives, such as thickness versus logarithm of time. Assuming that the growth oxide (MgO), alumina (A120,), and yttria (YzO,), tend to rate obeys the parabolic portion of Eq. (l), the slope of such a migrate out to the SiO? scale and form the corresponding sili- plot should be 0.5. However, the slope is 0.96 at 1373 K and cate.hi74 Figure 14 is a sample of Y,O,-containing Si,N, oxi- 0.70 at 1673 K.5xThis suggests a transition from interfacial dized for 100 h. The bright, acicular at the surface reaction control through mixed interfacial reaction and diffu- are an yttria-. Clarke7' has reported that two forces drive sion control. The inclusion of interfacial reaction control may these transition metals to diffuse to the surface. The additives also explain the differences between silicon, Sic, and Si,N,, are initially present as a grain-boundary silicate glass. The ini- since the interfacial reaction is different for these systems. tially pure SiO, layer creates a gradient and a driving force for Du et ~1.'~''have explained the differences between Sic and the cations to diffuse into this glass. The free energy of silicate Si,N, on the basis of duplex layer formation on Si,N,. Figure 12 formation in the surface SiOz is a second driving force. is a cross section of oxidized Si,N, which shows the duplex In the case of Sic hot-pressed with A1,O1, an observed pres- scale of Si,N,O and SiO,. The stoichiometry of the oxynitride is sure dependence on oxygen'" suggests that oxygen diffusion most likely a solution, but it is generally identified as inward is still the rate-controlling step. However, enough alu- Si,N,O. In general, the thickness ratio of oxide to oxynitride is minum diffuses into the scale to form an aluminosili- on average about 9 to 1 .9 Du et UI.~."have proposed that oxygen cate ,42.hi.6?.hh Generally, this phase appears to accelerate diffusion inward is rate controlling and that this Si,N,O layer In the case of Sic sintered with boron and carbon sin- has a lower permeability to oxygen than SiO, and thus acts as a tering aids, boron diffuses into the It creates a lower diffusion barrier. OgbujiS9has recently performed experiments scale, which leads to somewhat higher oxidation rates oxidizing Si,N, and Sic at 1573 K and then in at 1773 K and reoxidizing at 1573 K. In the first oxidation the rates of Si,N, are much slower than those of Sic. However, the annealing removes the Si,N,O layer and the two rates become comparable. This supports the concept of Si,NzO as a barrier to oxygen diffusion inward. But the inconsistencies for rate con- trol by oxygen diffusion inward still remain. In summary, a fundamental understanding of Sic and Si,N, oxidation is emerging, but many questions are unanswered. These deal with the rate-controlling step and the transport mechanism through SiO, scales. The differences between Sic and Si'N,, which are apparent for short oxidation times and below I700 K, continue to be an intriguing problem.

V. Isothermal Oxidation-Additive-Containing Materials The discussion thus far has centered on highly pure materials oxidizing in pure oxygen for short times and forming amor- phous SO2.The situation in actual combustion environments is Fig. 13. Micrograph of Sic oxidized for 24 h at 1673 K, showing more complex. First, consider isothermal oxidation on less- formation of spherulitic cristobalite. (Courtesy of E. Opila, NASA pure materials. In a combustion situation the amorphous scale Lewis. ) 14 Journal of the American Ceramic Society-Jacobson Vol. 76, No. 1

that certain compositions of a Y,O,-containing Si,N, have been found to catastrophically oxidize at 1273 K. This has been found to be due to the oxidation of a grain-boundary phase and the associated volume e~pansion.~~The surface oxide does not form in a sufficient quantity to protect the grain boundaries at 1273 K. Related to this is the oxidation of porous Sic and Si,N, mate- rials. There are several studies on reaction-sintered Si,N,, which is about 15% poro~s.'~.'' In general, these studies show that internal oxidation of the pores is critical. Porz and Thummler7' have shown that the oxidation of Si,N, at the pore mouth tends to limit the diffusive flow of oxygen into the inte- rior of the Si,N,. Thus, material with small pores can show good oxidation behavior. Also, a preoxidation at high tempera- tures may limit oxygen penetration.

VI. Long-Term Oxidation and Thermal-Cycling Effects The studies discussed to this point deal with short-term, lab- oratory-scale studies, typically on the order of several hours. In Fig. 14. Crystalline yttria silicate formed by migration of yttria a heat exchanger or utility turbine a component may be sub- impurities into scale. jected to isothermal conditions for hundreds or thousands of hours. Information on the effects of these long-term exposures over pure Sic. It has also been suggested that the formation of is limited. Events such as crystallization may tend to slow reac- boron hydroxide is responsible for the bubbling from oxidation tion rates over initial rates obtained from amorphous scales. observed in this material.4" The high thermodynamic svdbility Some 500- to 1000-h tests in complex CO, and H,O environ- of boron hydroxide indicates it can form at substantial pressures ments indicate that parabolic rates hold for these long time^.^^.^" with a small pressure of water vapor and small amount of FoxXnhas shown that, after 100-h isothermal oxidation of Sic in boron. oxygen, the calculated parabolic rate constants are in reason- In the case of hot-pressed Si,N, with MgO and Y,O, addi- able agreement with those extrapolated from several-hour tives, Singhalh4and Cubicciotti and Lau") have seen no pressure exposures. His studies on CVD Si,N, also indicate that the dependence on oxygen. This differs from the observations for short-term rates may not extrapolate and that the long-term CVD Si,N, (Ref. 9) and indicates another oxidation mecha- rates are close to those for CVD Sic, even below 1700 K.XOThe nism. Cubicciotti and LaubX." have performed reoxidation reasons for this are not clear. experiments on MgO- and Y,O,-containing Si,N,. Here they Many long-term applications, such as an aircraft turbine or oxidized for a time interval, polished the scale off, and reoxi- an internal combustion engine, involve thermal cycling. For dized. They found that the process simply continues, as shown this reason extensive studies have been done on cyclic oxidation in Fig. 15. From this and microprobe data they concluded that of metal alloys." However, studies on the cyclic oxidation of the oxidation is controlled by migration of the magnesium or silicon-based ceramics are limited. SiOzundergoes a number of cations outward through the grain-boundary phase. phase transformations on heating and cooling. Figure 16 shows More recently, Andrews and Riley7' have shown that both cat- the thermal expansivity for polycrystalline Sic, amorphous ion outward diffusion and oxygen inward diffusion are SiO,, and crystalline SiO,.xLThis mismatch between the coef- important for some types of Si,N,. ficients of of the substrate and the scale may These additives can have other effects on oxidation. In some lead to tensile stresses and cracking. Figure 14 illustrates scale cases liquid may form." It has been shown that oxida- cracking. The question is to assess the effect of this cracking on tion rates increase with the amount of additiveP and that addi- oxidation rates. tives lead to deviations from the parabolic rate law.7" Also note The limited studies on cyclic oxidation indicate that Sic and Si,N, perform relatively well under thermal cycling. Figure 17 shows a kinetic curve for sintered a-Sic containing carbon and 2.4~1O4 boron additives tested at cycles of 5 h at 1773 K in air and then ...' cooling to room temperature.xi Also shown is a curve for iso- *..- .**- thermal oxidation. The two tests show very similar behavior. .*"' Similar results have been obtained at 1573 K. It appears that 2.0 ,..* ,.*** any cracks induced by the cycling heal immediately at these f temperatures and scale thicknesses. Further studies are under-

01 ..*- way for longer times.x' E 1.6 f 9 .' 0) LindbergX4.XShas examined a variety of commercial Sic and s Si,N, materials in a burner rig with 12-min cycles at 1643 K for i a total of 3500 h. Durability was assessed from retained 'S 1.2 01 strength. In general, the dense, uniform-grain-size materials E showed negligible strength reduction. Maeda et ~1.~'have stud- .-cn ied a sintered Al,O,-containing SIC for periods up to 3000 h. 5 .8 0 First oxidation Weight gains were comparable to those obtained during isother- 0 Second oxidation mal oxidation, but the actual kinetics involved at least four dif- ferent parabolic stages. They attribute these to the various .4 microstructural changes in the scale: crystallization of amor- phous silica, transformation of those crystalline phases, and viscosity changes in the oxide scale due to migration of the I additives. In the early stages a smooth scale is formed following 0 10 20 30 40 50 Time, h parabolic kinetics; as time progresses, cracks and bubbles develop, leading to more rapid rates, and then these heal over Fig. 15. Reoxidation curve for Si,N, with MgO additives.hx leading to slower rates. Andrews and RileyX7have studied a January 1993 Corrosion qf Silicon-Based Cerumics 111 Combustion Envirorirrierits IS may not always be oxygen; it may also be H20and CO?. H,O may induce crystallization and thereby change oxidation behav- ior. It may also alter transport properties through the scale. rCrystalline Consider first the effect of H20 on isothermal oxidation of pure silicon. Deal and Grove" have shown that silicon oxidizes over an order of magnitude faster in wet oxygen than in dry oxygen. They attribute this to the higher solubility of H20 in SiO, (-3.4 x loi9molecules/cm') versus molecular oxygen in SiO, (-5.5 X 10'' molecules/cm3).31~88As Eq. (2) shows, the parabolic rate constant is directly proportional to the solubility qr \qr of the oxidizing species. Deal and Grove further show that sili- a con oxidizes at the same rate regardless of whether the mixture is H,O/Ar or H,0/02. This indicates that H,O is the primary oxidizing species. Irene and GhezXYhave further shown that H,O disrupts the SiO, network by forming nonbridging SiOH groups. This modified SiO, network permits faster H,O transport. One would expect similar behavior for Sic. There is general agreement that H20 accelerates the oxidation of SiC.4S.h'.'"0"4 The major reaction is very likely Sic + 3H,O(g) = SiO,(s) + CO(g) + 3H,(g) (15) -. 2 0 400 800 1200 1600 2000 Jorgensen et a/.'" have studied the oxidation of Sic powders Temperature, K and found the rates to be proportional to log PHZ0.Cappelen et d." have observed that the rate in H20/N, and HzO/Ozis the Fig. 16. Thernial expansion of SIC, Si,N,, amorphous SO2, and same, indicating that, as in pure silicon, the primary oxidizing crystalline SiO, as a function of temperature.X? species is H,O. They also have observed parabolic rates in pure oxygen and linear rates in H20. It appears that H,O diffusion is Si,N, containing various oxide additives. Again the strength so rapid that the rate-controlling step becomes the interfacial shows relatively little change. Weight gains are similar to iso- reaction. Narushima et ~1.'~have studied Sic in wet oxygen for thermal oxidation, although again the kinetics are somewhat longer times and observed an initial linear region followed by a different. Some of these differences may be attributed to scale parabolic region. Unlike oxidation in dry oxygen, the linear cracking on cycling. region lasts long enough (- 1 h) to measure reliable linear rate constants. Apparently the rapid permeation of H,O extends the VII. Isothermal Oxidation- period of interfacial reaction control. Tressler et al.', have Carbon Dioxide and Water as Oxidants examined the effect of water vapor on the oxidation of single- crystal Sic and sintered a-Sic.In both cases steam accelerated Actual combustion atmospheres are not simply pure oxygen. oxidation by about 10 to 20 times, which is attributed to more All combustion environments contain nitrogen, CO,, and H20. rapid permeation of the H20molecules, very likely by incorpo- Fuel-lean environments contain oxygen; fuel-rich environ- ration of (OH)- into the SiO,. Recent work in purified oxygen ments contain CO and hydrogen. Thus, the primary oxidant versus 10% water vaporipurified oxygen has shown an increase in the parabolic rate constant by a factor of 2 to 3 at 1473 K.45 Particular care was taken to avoid the effects of sodium impuri- 8x1 0-5 ties, which may often effect this type of meas~rement.~'.'~ r There are some important questions on the effects of water vapor on Sic oxidation. Choi et al.'4 have examined the oxidation of Si,N, in wet 7- oxygen. They have observed somewhat different behavior than for wet oxidation of silicon or Sic. In the case of H,O/Ar, the 6- pressure of H,O has no effect on the rates. However, in the case of H,0/02 the rates show a complex dependence on PH,O.They attribute the wet oxidation to the following reactions: N 5- El Si,N,(s) + 6H,O(g) = 3Si02(s) + 4NH3(g) (16) 'a, .-c 4NHdg) + 5Wg) = 6H,O(g) + 4NO(g) (17) g4- They suggest that the effects of H,O on Si,N, oxidation are due E .-0 to the product (NH, and NO) gases blocking the outward diffu- 3- sion paths, increased solubility of H20 versus molecular oxy- 5 gen in SiO,, or different adsorption properties of gaseous 0 Isothermal molecular oxygen versus water vapor. 2- 0 5-h cycles In addition to water vapor, CO, may act as an oxidant in a combustion environment. Antill and Warburton" have studied long-term oxidation of Sic in COz. They generally para- bolic behavior and rates about an order of magnitude slower than those in pure oxygen. Warburton et ~1.'~have also exam- ined long-term oxidation of porous Si,N, containing MgO. Par- 0 10 20 30 40 50 abolic behavior was generally observed, although the sealing off of internal in this material was a critical factor. Time, h In summary, many questions remain about the effect of water Fig. 17. Cyclic oxidation curve for sintered a-Sic with carbon and vapor and CO, on the oxidation of Sic and Si,N,. It is accepted boron additives at 1773 K, 5-h cycles. that water vapor induces devitrification in SiO, scales and 16 Journal ofthe AmcJrican Ceramic Society-Jacobson Vol. 76, No. 1 accelerates the oxidation of Sic. The differences bctween SIC transport of H20through the scale. Zheng et UI.'(~~."'~have made and Si,N, are still not understood. similar observations for Sic and Si,N, at 1373 to 1573 K. In their case, sodium is ion implanted into the Sic sample. Thus, the concentration of sodium in the growing SiOz scale remains VIII. Combustion Environment Impurity Effects constant. The same oxidation mechanism is maintained, but In addition to the various gases shown in Fig. 3(a), conibus- faster rates are observed. For Sic they attribute an increase of tion environments may also contain a number of corrosive oxidation by a factor of 2 to the formation of nonbridging oxy- impurities. These include sulfur oxides SO, and SO,, carbon- gens, which allows faster molecular diffusion through the SiO, containing gases, (HCI), sodium, vanadate scale. For Si,N, the situation is less clear, but they attribute the compounds, and transition-metal atoms. Even in small quanti- acceleration of oxidation rates by a factor of 14 to a modifica- ties these impurities may have a deleterious effect on a silicon- tion of the SizN20layer. based ceramic. In addition to sodium, transition metals, such as vanadium, In most combustion environments sulfur is present as SO, , , and , may also be present from the fuel and SO,, and, thus, the environment is oxidizing-sulfidizing. It and/or the metallic parts of the engine or furnace. These too has been shown that pure SiO, is resistant to attack by SOz and may be incorporated into the scale, leading to enhanced SO,." Limited experiments by the author have shown that oxidation. exposure to SO, has no measurable effect on SIC at 1473 K over In the case of metals, small amounts of impurities can lead to several hours. In some coal combustion situations, it may be severe corrosion. In general, this information is lacking for possible to obtain more sulfidizing environments because of the ceramics. Current evidence suggests that, in oxidizing environ- presence of (H2S)and low oxygen potentials. ments, the stable SiOz scale prevents extensive attack from gas- Under these conditions a stable SiOL scale may not form, eous impurities. However, metal-atom impurities may be and forniation of SiS(g) and SiO(g) may lead to rapid incorporated into the scale and lead to more rapid oxidation. degradation.')'',y7 An actual burner rig test encompasses oxidation in the pres- The major carbon-containing species are CO and CO1, so ence of impurities and thermal cycling. These burner rigs thus that the combustion environment is also oxidizing<-arburizing. provide the most realistic test. In general, Sic and Si,N, have CO, is an oxidant, and its behavior has been discussed. SiO(g) shown good durability in fuel-lean environments to about forms in predominantly CO environments, as discussed in more I673 K.'''.l')s Significant effects noted in these studies include detail in the next section. Actual combustion environments the incorporation of metal impurities in the SiOz scale from the contain a mixture of these impurities, generally with sufficient metallic parts of the burner rig and the of the SiOz C02to be oxidizing. In some coal combustion situations there scale due to the high velocity of the burner rig. Scales of lower may be a high carbon activity and a low oxygen activity. Where hydrocarbons such as are present, a net transport of viscosity, such as those grown on additive-containing Sic and Sic has been observed."' The low oxygen potential allows Si,N,, tend to deform more than a pure Si02scale. SiO(g) to be formed, and it subsequently react IX. Volatility Issues and Active Oxidation SiO(g) + CH,(g) = SiC(s) + H20(g) + Hl(g) (18) Such a route leads to rapid degradation. (1) Silicon Dioxide Scale Volatility corrosion can be an issue due to the formation of the As Fig. 4 shows, two major degradation routes are scale vol- stable SiCI, species.'x."" In general, chlorine can be quite reac- atility and active oxidation to SiO(g). Consider first the volatil- tive with Sic in a IOW-P,,~environment. In a heavily oxidizing ity of the SiO, scales. The vapor species above SiO, are most environment, it has been- found that 5% Cl,/Oz environments conveniently represented by a volatility diagram, as shown in show enhanced oxidation by a factor of 10 ovcr pure oxygen. Fig. IS. These diagrams have a variety of applications and have This has been attributed to incorporation of chlorine in the SiOL been discussed by a number of inve~tigator~.'"~'""'O~Note that, network and the subsequent formation of nonbridging bonds at I-bar oxygen pressure, the major vapor species above SiOz is leading to a more open structure.""' It is more likely that HCI Si02(g)and the minor species is SiO(g). will form in combustion environments. Short-term exposures In a combustion situation the question to address is how of Sic to HCl/Oz mixtures show no major effects.'jx much material will be lost by vaporization into the flowing gas Alkali metals (sodium and potassium) are common impuri- stream. It can be shown that the flux of gas vaporizing through ties in heat engines and industrial furnaces. These impurities a boundary layer into a flowing gas stream is given by")'."" may be present in the fuel, in the intake air in a marine environ- ment, or as a byproduct of an industrial furnace, such as an alu- minum remelt furnace. Most alkali-induced corrosion is due to J= large deposits of Na2S0, forming on the material, as discussed in a subsequent section. However, there are some recent studies where cy is the sticking coefficient, P '& the equilibrium vapor of very-low-levcl sodium and potassium incorporation into a pressure of SiO,(g)in the atmosphere of interest, M the molecu- growing SiOl scale on SIC. Such a situation is possible in lar weight of SiO,, R the gas constant, T the temperature in combustion environments and provides some insights into degrees Kelvin, PI the total pressure, and h,, the mass transport the mechanism of modified SiO, growth in the presence of coefficient. Note that Eq. (20) has the correct limits. For low impurities. pressures or fast velocities (large h,,), c~P,/[(2rrMRT)"~h,,,]<< Pareek and Shores"" have studied the oxidation of Sic with 1. and the expression reduces to thc Langmuir equation for carefully controlled amounts of potassium. Basically, they Vaporization into a vacuun~.For high pressures or slower veloc- have used potassium (K,CO,) as their source of ities, aPI/[(2~MRT)"%,,,]>> I, and the expression is domi- potassium and have controlled thc activity by varying the nated by the mass transport coefficient. Under these conditions, to P,,,? ratio: a becomes unimportant, which is consistent since cy has mean- K,CO, + H,O = 2KOH + CO? ing only in a low-pressure situation. The mass transport coefficient can be derived from various For dry CO,/Oz mixtures, which lead to a very small amount of semiempirical correlations. Consider, for example, a combus- potassium incorporated into the scale, they observe enhanced tion chamber modeled as a hollow cylinder, with vaporizing parabolic oxidation. They attribute this to potassium modifica- walls:"' tion of the SiOz network. For C02-02-H,0 mixtures they observe linear rate laws that they suggest are due to the rapid January 1993 Corrosion of'Silic~~n-Rasc~rlCc~rornic.~ in Comhusiion Eni~ironini,nis 17

Si (I) Si02 (s) 0- A

-24 -20 -16 -12 -8 4 0 log Pop,bar

Fig. 18. Volatility diagram for the Si-0 system at 1800 K. (From Kohl pi 11I."") where D,,,,, is the diffusion coefficient of SiO, in the combus- Figure I9(a) shows the vapor pressure above SiO, in a fuel- tion atmosphere, L a characteristic dimension taken to be the lean environment. These results are basically the same as the diameter of the cylinder, v the viscosity of the combustion gas, pressures taken from the volatility diagram (Fig. 18) at high p the of the combustion gas, and v the linear gas veloc- oxygen potentials. Figure 19(a) shows that the pressure ofSi0, ity. From Eqs. (20) and (21) one can extract a relationship approaches 10 ' bar only at temperatures near the melting between flux and vapor pressure. Note that this is only an point. Thus, vaporization in a fuel-lean cnvironment is unlikely approximation because many of the terms in Eq. (20) are tem- to be a major issue. perature dependent. However, because vapor pressure shows an Figure 19(b) shows the vapor pressure above Si exponential temperature dependence and the other terms show librium fuel-rich environment. Note that now ther fractional power temperature dependence, this is a reasonable tial pressure of SiO(g). A stable SiO, film is expected to form approximation. because of the oxidants COLand H,O. However, the source of The issue then is to determine an acceptable Hux and thereby SiO(g) is reduction of SiO, in thi environment. The higher extract an acceptable vapor pressure limit. An acceptable flux total vapor pressures indicate that vaporization may be an issue depends on the application. For clarity, flux can be converted to at temperatures above 1770 K. It may be that, in a fuel-rich situ- recession rate as follows: ation, coupled oxidation and vaporization lead to paralinear kinetics, I1O.I 14 R( mis) = 1 0 './( mol/(m2.s))MWip' (22) (2) Active Oxidation where MW is the molecular weight of SiO, and p' the density of The most common route of SiO(g) generation is active oxi- SOz. There are no guidelines for acceptable recession rates; dation. Although this can lead to rapid degradation, it only however, lifetimes of 10000 h or longer are desired for future occurs at low oxygen potentials. Most of the combustion envi- combustor liners. This suggests a criterion of 10 x 10 in./ -' ronments are oxidizing with large amounts of the oxidants, 10000 h, or 7.0 x IO-"m/s. This in turn leads to a maximum molecular oxygen, CO,, and H20. However, in some special acceptable vapor pressure of 10-' bar as determined from Eqs. (21) and (22) and some approximate combustion chamber cases, such as coal combustion, the oxidant pressure may fall parameters. Other investigators have taken similar criterion."' below that needed for a stable SiO' film to form. In this case To determine when SiOz vaporization is a problem, one active oxidation would be favored: needs to know at what temperature the total pressure of the Si + 0.S02 = SiO(g) (26) vapor species above SiO, is IO-'bar. There are several equiva- lent ways to do this. First, for a simple overpressure of oxygen, Sic + 0: = SiO(g) + CO(g) (27) one can read the total pressure from the volatility diagram for a Si,N, + 1.502 = 3SiO(g) + 2N2(g) (28) range of temperatures. Second, one can solve the vaporization equations: In these cases a protective oxide film is not formed and the vola- products may lead to rapid consumption of the ceramic. The major issue with active oxidation is the Po, transition point. Again, it is appropriate to begin with pure silicon. Heuer and Lou"'x show how volatility diagrams can be used to predict the active-to-passive transition. The for a stable SiOl film can be determined from the necessary equilibrium An oxygen pressure can be put into the equilibrium constants SiOz(s) + Si(s) = 2SiO(g) (29) of Eqs. (24) and (2S), and then the three equations can be solved. Third, which is perhaps the most versatile, is to use a This can be read from point A on the volatility diagram (Fig. free-energy-minimization computer code" 'Ii for determining 18) to be about lo-' bar at 1800 K. Basically, enough oxygen to the equilibrium vapor pressures. Basically, such a code takes form 10 '-bar SiO(g) from Eq. (26) is required, which is 0.5 X solid SiOz and a particular combustion atmosphere and mini- 10 '-bar oxygen. The factor of 0.5 is from the stoichiometry of mizes the total free energy of the system subject to certain con- Eq. (26). Therefore, the condition for the active-to-passive straints. The results are the equilibrium vapor pressures in that transition is combustion atmosphere. The versatility of this approach is that volatility can be assessed in any combustion environment. Pol = 0.5P~~,:, (30) 18 Journal ofthe American Ceramic Society-Jacobson Vol. 76.No. I

3 4 5 6 7x1 O4 Reciprocal temperature, 1 IT, K

10-2

2 10-4 ci g 10-6 u) ii 10-8 B B 10-10

10-12 3 4 5 6 7x1 O4 Reciprocal temperature, 1/T, K

Fig. 19. Partial pressures of major vapor species in combustion atmospheres with Jet A fuel: (a) fuel-lean atmosphere, equivalence ratio of 0.5, P,,, = 0.10 bar and (b) fuel-rich atinosphcre, equivalence ratio of 1.75, P,. = 5 X 10 bar.

As Po, is increased above the active-to-passive transition point, volatility diagrams as described by Heuer and Lou. I"' Singhal'IX SiO, first forms as a "smoke": has extended Wagner's theory and used the following analogues for Eq. (29) to derive a transition Po,: SiO(g) + 0.50,(g) = SiO,(smoke) (31) SiC(s) + 2SiO,(s) = 3SiO(g) + CO(g) (34) This approach is similar to that of Turkdogan et ~1."~and is shown schematically in Fig. 20(a). Si,N,(s) + 3Si02(s) = 6SiO(g) + 2N,(g) (35) As Heuer and Lou"'x have reported, mass transport consider- However, Hinze and Graham"' have suggested that other equi- ations are involved in this transition. Hinze and Graham'16 have libria may control this transition for Sic in place of Eq. (34): reported that it is important to specify whether the low Po- is attained by lowering the total pressure (molecular flow regime) SiC(s) + SiO,(s) = 2SiO(g) + C(g) (36) or by lowering Po, (viscous flow regime). 2SiC(s) + SiO,(s) = 3Si(l,s) 2CO(g) In a combustion situation, oxidant pressures are part of a + (37) high total pressure and the flow is in the viscous regime. The Their experimental data tend to support Eq. (37). This is simi- classic work in this area is by Wagner."' He has considered a lar to the result of Gulbransen and Jansson"" for the transition silicon sample in contact with an OJHe mixture. At very low Po>, SiO forms according to Eq. (26). To form a stable SiOL scale, enough SiO(g) must be formed to maintain the 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 / / Ply = 0.64f& (31) / This approach is shown schematically in Fig. 20(b). Note that LSiO(g) + 02(g)= SO2 (smoke) this result is close to the results based on thermodynamic con- (4 siderations alone, due to the fact that gas-phase diffusion coef- ficients are often close to unity. Fig. 20. Schematic of two proposed descriptions of active-to-passive The active-to-passive transition for SIC and Si,N, is more transition for silicon: (a) thermodynamic considerations onty, SO, complex because additional product gases form. A purely ther- "smoke" formation, and (b) thermodynamic and mass transport consid- modynamic interpretation allows fr to be determined from erations. Wagner theory. January 1993 Corrosion qf'Silicoti-Based Cerumics in Combustion Environtnents 1') in the molecular flow regime. Thus, the active-to-passive tran- Sic oxidized for 1 h in air at 2073 K. Note the extensive corro- sition for Sic is given in terms of sufficient Po, to create enough sion. The question then is how close to 2000 K can one safely CO to establish this equilibrium. Again using mass transport use this material. arguments simiiar to those of Wagner, they have derived the fol- Several issues limit the use of Sic below 2000 K. These lowing expression: include rapid oxidation rates, volatilization of the SiO? film, and reaction of the Si02film with the Sic or Si,N4 substrate. - - (D(CO)/D(O,)) li2P2q,, (38) Oxidation and volatility have been discussed previously. Note that there are relatively few measurements at these high temper- Narushima et ~1."~have examined CVD Sic and come to a atures because it is difficult to attain such temperatures in oxi- similar conclusion. dizing environments. Nickel12"has reported that carbon activity needs to be consid- An important difference between silicon-based ceramics and ered in establishing the active-to-passive transition. At high other oxide-protected metal alloys is that, at sufficiently high carbon activities the SiC/SiO, equilibrium is critical in estab- temperatures, the oxide scale and the Sic or Si3N4ceramic lishing the transition. At low carbon activities Nickel proposes react. Such reactions can generate gas pressures at the oxide/ that the SiO(l)/SiC equilibrium is important and thus the critical ceramic interface: equilibria becomes SiC(s) + 2SiO,(s) = 3SiO(g) + CO(g) (40) SiC(s) SiO(l) = 2Si(l) CO(g) + + (39) Si,N?O(s) + SiO,(s) = 3SiO(g) + N&) (41) This approach fits some recent data quite well;"" however, the The best way to examine these interfacial reactions is with the existence of condensed-phase SiO is quite controversial. ''I Si-C-0 and Si-N-0 predominance diagrams, which are Numerous other studies of the active-to-passive transition shown in Fig. 22. The intersection of the SiO and CO isobars have been performed. Antill and Warburton'" have examined with the SiC/SiO, coexistence line (line AB) gives the gas pres- the transition in the presence of other oxidants (C02and H,O) sures in the carbide system, as shown in Fig. 22(a). The inter- and have found behavior similar to that in oxygen. This is section of the SiO and nitrogen isobars with the SizN,0/Si02 important for a combustion environment, where the primary coexistence line (line AB) gives the gas pressures in the nitride oxidants may be CO, and H20. Vaughn and Maahs"3 have system, as shown in Fig. 22(b). shown that the Po2 transition varies with flow rate, emphasizing Consider first the Si-C-0 system. Figure 23(a) shows these the mass transport issue. The approximate active-to-passive total pressures as a function of temperature for the carbon-satu- oxidant transition pressures for Sic and Si,N, are given in Fig. rated, stoichiometric, and silicon-saturated systems. Taking 4; more precise values are given in the references. l'h-120.'22.'23 1 atrn as the criteria for instability, carbon-saturated Sic in con- In most combustion environments, active oxidation is not tact with SiOz becomes unstable at 1800 K. Pressures approach expected to be a problem. Combustors in current gas turbines 1 atm for the other systems near the melting point of SiO,. These types of reactions have been suggested as the source bum near the stoichiometric region. The resultant environment of bubbling at very high temperatures in an oxidized Sic contains oxygen, H,O, and CO,. Novel combustor designs may sample .41.17b involve a fuel-rich region, which would contain CO, C02,H20, Recently, Jacobson ef a1.I" have examined the Sic + SiO, hydrogen, and oxygen potentials as low as about IO-'bar (Fig. reaction in detail. They have found that a carbon-saturated sys- 3(a)). However, the presence of H20and CO, at about 0.1 bar tem tends to adjust to a stoichiometric composition. This is each should form enough SiO to satisfy the equilibrium require- done by releasing a large amount of CO and forming a second- ments for a stable SiO, scale (Eqs. (29) or (34) to (37) or (39), ary Sic. Such a process at the carbon-saturated SiCiSiO, inter- with H20and/or CO, as oxidants). The effect of water vapor in face may lead to extensive bubbling. suppressing active oxidation has been discussed by Heuer and The Si-N-0 system is more complex because two interfaces Lou.'"*Once formed, however, the SiOz scale may be reduced are involved. However, Eq. (41) is predicted to be the major in the fuel-rich environment. source of SiO(g) and N,(g). The total pressures are shown in Thus, combustion of hydrocarbon fuels generally provides Fig. 23(b). The pressures predicted are less than those in the an atmosphere sufficient for forming Si02.However, coal com- Si-C-0 system. This may be the reason that only limited bub- bustion situations and certain heat-treating environments, bling is observed when Si,N, is oxidized at high temperatures."' which are heavily reducing, can lead to active oxidation. In a recent study of reactions of Sic 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 In summary, there are two causes of volatility. The first is SiO, vaporization. Unless the net SiO, vapor pressure is greater than about 10-' bar, this is 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 limiting factor. In a fuel-rich situation, SiO, volatility is likely to be an issue at higher temperatures. The second cause of vola- tility is active oxidation. The key issue is the oxidant pressure for the active-to-passive transition. Active oxidation is pre- dicted for only a limited number of combustion situations, and it can lead to rapid material consumption.

X. Upper Temperature Limits Referring to the corrosion diagram (Fig. 4), the absolute upper use temperature for silicon-based ceramics is -2000 K. This is the melting point of SiO,. When the protective oxide is liquid, it begins to flow, and rapid transport rates are obtained. Fig. 21. Chemically-vapor-depositedSic oxidizcd for I h at 2073 K This is a nonprotective situation. Figure 21 shows a sample of in air. (Courtesy of D. Fox, NASA Lewis.) 20 Journal qf ihe American Cerumic Society-Jacobson Vol. 76, No. 1

0 -1 lo*r

0 -2 P 0 -4

4

r -3 -25 -24 -23 -22 -21 -20 -19 -18 log Po2, bar

1200 1400 1600 1800 2000 2200 Si2N20(s) 0 Temperature, T, K

Si02 (s)

N h 8 - -10

-ic I I -1" (b) 40 -35 -30 -25 -20 -15 -10 log Pop, bar la00 1;oo lb0 2doo 2;oo Temperature, T, K Fig. 22. Predominance diagrams at I500 K: (a) Si-C-0 system and (b) SI-N-0 system. Fig. 23. Total vapor pressure at interfaces: (a) SiCiSiO, interface and ib) Si,N,O/SiOl interface.

In summary, the upper use temperatures of silicon-based a corrosive Na,SO, deposit forms. Generally, Na,SO, is corro- ceramics are limited by scale melting, oxidation rates, vola- sive between its melting point (1 157 K) and the dewpoint for tility, and scaleisubstrate reaction. Scale melting occurs at Na,SO, deposition. The latter can be calculated from a free- 1996 K. Oxidation and scale volatility are highly dependent energy-minimization computer code. '' The dewpoints are on ceramic composition and atmosphere. The scaleisubstrate shown in Fig. 24 for a number of different combustion condi- reaction is more of an issue in Sic than in Si,N,. tions. These factors tend to limit the region of hot corrosion, as shown in Fig. 4. There are two issues to note from this. First, although the region is limited, it is important to be aware of hot XI. Deposit-Induced Corrosion corrosion because it can be quite severe. Second, the dewpoint tends to increase with pressure. Future engines are expected to (I) Formation of Deposits operate at higher temperatures and pressures. Higher tempera- As discussed previously, levels of sodium in SiO? tures decrease the likelihood of deposition by exceeding the scales appear to accelerate oxidation. Larger amounts of dewpoint; however, higher pressures will increase the dew- sodium, either from a marine environment, a salted roadway, or point. Thus, a knowledge of engine operating parameters is fuel impurities, can react with sulfur fuel impurities to form essential for predicting the presence of a deposit. Na,SO,, which is a highly stable :"

2NaCl(g) + SO&) + 0.5O,(g) + HzO(,g) S content, Na 1700 I- percent content. = Na2S0,(/) + 2HCI(g) (42) Y 0.5 PPm The Na,SO, forms as a deposit on engine parts. This type of 2Ll! 1500 L= .05 ,ywA -A deposit-induced corrosion is termed "hot corrosion ," and there .L is a large amount of literature on hot corrosion of metals."'~'?' 0)a The studies on silicon-based ceramics have recently been E 1300 c summarized. '.'(' c C The primary emphasis in this section is on Na,SO,-induced 1100 attack. However, there are other types of deposit-induced corro- 'g sion (e.g., vanadate-induced attack from lower purity fuels"' ni! and oxide-slag-induced attack from coal combustion"). Similar- 900.__ ities between the various types of deposit-induced attack are 1 00 1 0' 1 02 discussed below. Pressure, P, bar In general, hot corrosion occurs in two steps: (1) deposition and (2) corrosive attack. It is first necessary to determine when Fig. 24. Dewpoints for Na2S0, deposition January 1993 Corrosion of Silicon-Bused Cerurnics in Comhiistiori Environments 21

(2) Thermodynamics of Deposit-Induced Corrosion This border-line uNa,Ofor SiOz dissolution can be related to a Figure 25 shows a coupon of Sic treated in a burner rig both corresponding PsOlby Eq. (47). Thus, one can calculate, for a with and without added sodium."' Note the dramatic effect. given temperature, the threshold PsO,for dissolution. This is Elemental maps of a polished cross section are shown in Fig. shown in Fig. 27. Note the temperature boundaries are the 26. Note the even distribution of sodium, oxygen, and silicon melting point of Na2S04and the dewpoint for Na2S0, deposi- in the scale, suggesting the formation of glass. tion. The values of PsO,from combustion of two typical fuels Sulfur was not detected in the scale, indicating that any depos- are also shown, as calculated by a free-energy-minimization ited Na,SO, had decomposed. The likely reaction scheme for program.l7.IXThe intersection of this line with the shaded region this is in Fig. 27 shows when scale dissolution is expected. Experi- ments with quartz coupons in a burner rig verified these predic- SiC(s) + 1.502(g) = SiO,(s) + CO(g) (43) tions, as shown in Fig. 28. The coupons treated in the low- sulfur fuel showed more corrosion, and chemical analysis xSiO,(s) + Na,SO,(I) = Na,O.x(SiO,) (I) + SO&) revealed the presence of sodium silicate. The coupons treated in (44) the high-sulfur fuel showed less corrosion, and chemical analy- These are the key reactions in hot corrosion. A solid, protective sis revealed no sodium silicate. I' These calculations illustrate the value of the acid-base inter- SiO, scale has been converted to a liquid scale. Transport rates pretation and provide general guidelines for predicting corro- through the liquid are rapid, leading to accelerated attack. sion. However, as is often the case, the actual situation is more Equation (44) can best be understood as an acid-base reac- complex. Secondary elements can drive the molten salt more tion.'x,'2xIt can be written more fundamentally as basic. It has been shown that SIC with excess carbon corrodes Na,SO,(I) = Na,O(s) + SO,(g) (45) quite extensively in a situation where the PsO,is high enough to limit corrosion. The reason is that carbon tends to drive Na,S04 xSiO,(s) + Na,O(s) = Na,O.x(SiO,) (I) (46) basic.lx An electrochemical cell can be used to measure uNalOin a melt:'x,'34 It is well-known that SiO, is an acidic oxide."' Following Lewis' acid-base concepts, SiO, reacts with the basic oxide Pt,O,(r)lZrO,lNa,SO,,SO,lmullitel10-mol%-Ag2S04, Na,SO,,Ag Na,O to form a salt, sodium silicate. The thermodynamic activ- (49) ity of Na,O (uNaZ0)can be determined from the partial pressure of SO, and the free energy change (AGO) of Eq. (45): Zirconia acts as a membrane for oxide anions, and the mullite acts as a membrane for sodium cations. They combine, and the AG&",l,,," (45) = - 2.303RT log (Pso,a,,p) (47) net reaction is When Na,SO, has a high activity of Na,O, set by a low Pso,, it Na,O + Ag,SO, = 0.50, + Na,SO, + 2Ag (50) is called a basic molten salt. When Na,SO, has a low activity of At 1173 K, uNa,"is related to (E)by Na,O, set by a high PsO,,it is called an acidic molten salt. From purely thermodynamic considerations one can deter- E = 1.498 + 0.116 log uNa,() (51) mine when Eq. (46) will occur. A threshold uNa,"for SiO, dis- solution can be calculated from Eq. (46) as Figure 29 shows the aNa,"measurements for Na,SO,/(O.OI SO, + 0,) and Na,SO,/(O.OISO, + 0,) with added carbon. Note AC&a~~on (46)= 2.303RT log uN;i20 (48) that the latter swings strongly basic. The most likely interpreta- tion of this is the following sequence of reactions: NaZSO,(l) + 2C = Na2S + 2C02 (52) NazS + 3Na,SO, = 4Na,0 + 4S02 (53) It is significant that carbon tends to drive Na,SO, more basic because many types of Sic 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 Na2C0, = NazO + CO&) (54) Na,CO, tends to decompose more than Na,SO,, so that it is a more basic molten salt. McNallan et ~1.'~'have shown that Na,O can also form from the reaction of (NaCI) and water vapor: 2NaCl + H,O = Na,O(s) + 2HCI (55) In both these reactions, if uNa,()is sufficiently high, SiOz is expected to dissolve. Deposit-induced corrosion can also originate from vanadium compounds, 131.136-11X Vanadium pentoxide (V20,) is an acidic oxide and therefore not expected to react with SiOz. However, the shows a limited solubility of SiO, in V,O,, which leads to accelerated rates. lih.liX Mixtures of Na,SO, and V,O, have led to severe corrosion of Si,N,."7 The concept of dissolution by basic molten salts can be extended to oxide slags as well. In the case of an oxide slag, it is more difficult to define and measure basicity. Generally, ~i~.25. Optical micrographs of sintered sic with carbon and boron basicity is approximated by the ratio of basic to acidic oxides. additives, treated in burner rig at 1273 K (leading edge is on the left): Ferber et a/." have identified three corrosion mechanisms from (a) 46 h with no sodium and (b) 13.5 h with 4-ppm sodium. oxide slags. These are by SiO,, dissolution of SiO,, 22 Journal of thc American Ceramic Sociefy-Jacobson Vol. 76, No. I

Fig. 26. Image and elcmental maps of polished cross section of glassy products formed by burner rig corrosion at 1273 K with Jet A fuel for 13.5 h with 4-ppm sodium. and formation of metal at low oxygen potentials. Pas- who saw rapid reaction until the Na,O-SiO, liquidus was sivation is analogous to the reaction in Eq. (44) not proceeding; reached, at which point a stable tridymite film formed below the dissolution is analogous to the reaction in Eq. (44) proceeding. melt. The second case is illustrated in Fig. 32. It has been shown (3) Kinetics of Deposit-Induced Corrosion that Na,SO, with an overpressure of oxygen is sufficiently The kinetics of corrosion in Na,SO, are consistent with this acidic so that SiO, is not expected to dissolve. Figure 32 shows interpretation of dissolution and enhanced oxidation.'"' Kinet- a kinetic curve for Na2S0,/02 on Si3N,. The long period of ics of hot corrosion are studied in the laboratory by coating the weight loss corresponds to vaporization of Na,SO,; the rates are specimen with Na,SO, and following the kinetics on heating. the same as those for Na,SO, vaporizing from a platinum cou- This can be accomplished with a thermogravimetric analysis pon. However, the period of weight gain corresponds to (TGA) system, such as that illustrated in Fig. 6. A one-time enhanced oxidation. Blachere and Pettit"' have shown that even deposition of salt is not as realistic as continuous deposition in an acidic deposit can enhance oxidation by inducing devitrifi- the burner; however, it does provide mechanistic information on cation and scale cracking or by the Si02scale. the first stages of reaction. The third case leads to the most extreme corrosion.'"' This Three situations can occur. The first is a basic molten salt, is illustmted by the data in Table VIII. In this case Na,SO,/ such as Na,CO, or Na,SO, in contact with carbon. This situa- (0.01S0,+02), a very acidic salt, is used. Chemical analysis tion leads to the type of kinetic curve illustrated in Fig. 30. The of the scale indicates limited attack in the presence of this salt early stages are characterized by a period of weight loss, which on the Sic and Si,N, with oxide additives. However, for the Sic corresponds to the dissolution reaction. In this case the amount with carbon additives, cxtensive attack occurs. This is of weight loss corresponds very closely to that predicted from explained by the basic effect of carbon at the bottom of the niclt the reaction"" and the acidic effect of NaZSO,/SO, at the top ofthe melt. These effects create the reaction sequence illustrated in Fig. 33. This Na,CO, + SiOz = Na2Si0, + CO, (56) self-sustaining series of reactions can lead to severe corrosion. This is followed by a period of enhanced oxidation due to liquid It is termed "fluxing" and is well-known from the hot corrosion scale. Finally, a protective SiO? layer forms below the melt, and of mctals.'?x~"" the reaction slows. This final layered is shown in In summary, hot corrosion occurs only in a limited regime. Fig. 3 1. This is consistent with the study of Mayer and Riley,I4" However, it can be quite severe, and, therefore, it is important January 1993 Corrosion of Silicon-Based Cerumics in Combustion Environments 23

0 Pso3 limit for corrosion 0 PSO, for 0.5 percent S fuel A PsO3for 0.05 percent S fuel

-3 r Melting point I Dewpoint

1100 1200 1300 1400 1500 Temperature, T, K

iioo 1200 1300 1400 1500 Temperature, T, K

Fig. 27. Calculated corrosion regimes for Si02:(a) 0.5% sulfur fuel, 2-ppm sodium, 0.025 fuel-to-air ratio, 1273 K, and (b) 0.05% sulfur fuel, 2-ppm sodium, 0.025 fuel-to-air ratio, 1273 K. Fig. 28. SiOz coupons treated in burner rig in 2-ppm sodium at 1273 K: (a) No. 2 diesel fuel (0.5% sulfur), 5 h (point A on Fig. to be cognizant of when it occurs. The dissolution reactions in 27(a)), and (b) Jet A fuel (0.05%sulfur), 1 h (point B on Fig. 27(b)). hot corrosion allow corrosion regimes to be predicted. These reactions are best interpreted in terms of the acid-base theory of oxide reactions. short length of the fiber. They also show that a thinner fiber coating is desirable, since it leads to healing more readily than thicker coatings. Improvement in oxidation and corrosion prop- XII. Composites erties is an important driving force in composite development. This review has concentrated on the near-surface durability of the Si0,-protected Sic and Si,N4 systems. Thus, the conclu- sions apply to monolithic Sic and Si,N, and composites of these materials. Currently there is much interest in SiC-fiber- 0 Na,SO,/O.Ol SO3 reinforced Sic and Si,N, matrices. These composites may have 0 Na2S04/0.01 SO3 + C (C at bottom of melt) some unique oxidation and corrosion properties, and it is important to mention them. At their current state of development, these materials may be porous and contain secondary elements, such as refractory oxides and/or carbon. These effects have been discussed. In addition, composites require fiber coatings to prevent bonding of the to the matrix. Loosely bonded fibers permit effec- tive crack bridging and crack deflection, which lead to a tougher ceramic. However, the current fiber coatings, carbon or (BN), are quite susceptible to oxidation if a path for oxygen is available. Thus, oxidation may lead to a deg- radation of the critical fiber interface and loss in properties. Some composites have exhibited a low-temperature oxidation problem due to inadequate coverage of the passivating Si02 scale, which then allows oxygen penetration to the fiber coat- ing~.'~*Recent work by Filipuzzi and Na~lain'~~.'~critically examines oxidation of composites from both an experimental and theoretical point of view. Their study presents important I 0-15 results on the effect of temperature and carbon coating thick- 0 t 2 3 ness on Sic fibers. Lower temperatures lead to slower oxidation Time. h rates of the carbon coating, but along a long length of the fiber; higher temperatures lead to more rapid oxidation, but along a Fig. 29. Effect of carbon on basicity of Na2S0,. 24 Journal o/ the American Ceramic Soc,iet~-Jac.ohJon Vol. 76, No. I

.1 Xl0-6 .1 xlO-6 .16xl O4 N 0 L 00 N E -.l -.l - 0 Na2S0, vaporization; 111

0 Oxidation !E -.32 0 0 - 0, p -.3 Q 0 -.48 1 -.4 .-;F 0 16 32 4% 64 80 0 QI Time, hr Q -.5 Q rn Fig. 32. Kinetics for Si,N, + Na,SO,. -.6 -.7 -.7 fl 0 .2 .4 0 16 32 4% 64 XIII. Corrosion and Degradation of Time, h Time, h Mechanical Properties The focus of this paper has bcen on chemical mechanisms of degradation. A major issue has been the assessment of this deg- Fig. 30. Kinetics for Sic + Na,CO, radation. An assessment in terms of microstructural changes

Fig. 31. Polished cross section showing final microstructure for Sic + Na2C0,alter 48 h of reaction. January 1993

Table VIII. Quantity of Corrosion Products' lo Na,O-r(S10.) NLSO, SIO. Material ('llgicln') (m$ciii') 1 (Illg/crl1 J Single-crystal Sic 1.64 t 0.86 <0.01 1.82 ? 0.96 Hot-pressed Sic with AI20, 2.01 t 0.03 <.Ol 0.39 5 0.02 Sintcred Sic with B and C 0.49 t 0.06 0.97 2 0.28 1.6 ? 0.9 10.50 2 0.66 Sintered Si,N, with YzO, 0.23 t 0. I I 1.3 * 0.2 and A120, 'Na,SO,I(O 01S0,+02):48h. 1273K.

so,/o, Na2Si03 + SO, = Si02 + Na2S0, intensity for crack growth.'" The corrosion reactions at or near Acidic the crack tip are quite critical,'"."' analogous with moisture- assisted crack growth in ceramics. Henager and JonesiiJ have shown that the presence of molten Na2S0, on Si,N, increases the velocity of slow crack growth by a factor of 2 over air at I573 K. Relatively little is known about the effects of corrosion on creep in Sic. In Si,N,, grain-boundary phase and/or conipo- sitional changes due to oxidation are expected to alter the creep rate. '" The discussion thus far has centered on monolithic ceramics. Free C -J In the case of fiber-reinforced composites, other issues, such ;IS oxidation-induced liberimatrix bonding, must be considered. '"' Fig. 33. Schematic of fluxing mechanism operative in carhon-satu- The issue here is how much fiberimatrix bonding can be toler- rated Sic + Na2S0, case. ated before an unacceptable loss in composite properties occurs. In summary, the synergistic linkage of mechanical properties may be sufficient in a fundamental study, but the designer needs and corrosion is an important area of research toward applying to know how these changes affect properties such as strength, silicon-based ceramics as structural components. Only a few , fatigue, and creep. A thorough discussion key issues have been discussed hcre; the reader is referred to of this is beyond the scope of this paper. However, a few com- other rcviews for a more thorough discussion. "'"" ments are appropriate. Many of the studies in this area deal with the change in XIV. Summary and Conclusions mechanical properties after corrosion. Is' However, applied and corrosion often act synergistically. For example, the Many proposed applications of silicon-based ceramics oxidation rate of reaction-sintered Si,N,/SiC material increases involve exposure to combustion gases. These environments with applied stress.'"' In a recent review, Tressler"' has dis- have been defined and shown to vary depending on the applica- cussed corrosion-induced effects on critical flaw generation, tion. All involve high temperatures and most are highly oxidiz- crack growth, and creep behavior. ing. Some have components that may lead to condensed-phase It has been shown that limited oxidation tends to blunt cracks deposits. and may actually increase ~trength.'".'~~.'~"However, long-term Five types of corrosive degradation have been identified: pas- oxidation'47and molten salt corrosion'"~'5'1~'s'lead to extensive sive oxidation, deposit-induced corrosion, active oxidation, pitting, as shown in Fig. 34. In a monolithic ceramic, pitting scale/substrate interactions, and scale volatilization. A funda- leads to substantial strength reductions. As more flaw-resistant mental understanding of the oxidation process is emerging from ceramics and composites are developed, the issue of surface studies on pure materials in pure oxygen. The key issues are flaws should become less important. transport through the SiO? scale. rate-controlling steps, and the Corrosion also changes mechanical properties by influencing differences between Sic and Si,N,. Commercial materials in crack growth.'52This is particularly critical in the case of mate- combustion environments add a number of complexities to the rials with a second phase at the . High-tempera- fundamental oxidation process. These include crystallization of ture oxidation may lead to crystallization and compositional the SiOz film, additive effects, H,O and CO, as oxidants, ther- changes in this phase, which may lower the threshold stress mal cycling effects, and combustion gas impiirity effects. In

Fig. 34. Sequence showing sintered Sic with boron and carbon additives beforc corrosion, after corrosion with NalSO,/(O.OISO,+ 01)at 1273 K for 48 h, showing glassy product layer, and with glassy product layer removed by HF etch, revealing extcnsive pitting of Sic. 26 Journul ofthe Americun Cerumic Societ~~-Jucobson vol. 76, No. 1

addition, some low-oxidant atmospheres can lead to active ”M. P. Murrell, C. J. Sofield, and S. Sugden, ”Silicon Transport During Oxi- oxidation. dation,” Philos. Mrcg. B, 63 [6] 1277-87 (1991) ”J A. Costello and R. E. Tressler. ”Oxidation Kinetics of Hot-Pressed and The use of silicon-based ceramics is limited at very high tem- Sintercd a-SiC,”./. Am. Ceram. Soc., 64 [6]327-31 (1981). peratures because of rapid oxidation, volatility, scale-melting, ”R. L. Meek, “Diffusion Coefficient for Oxygen in Vitreous SiO,,” J. Am. and scaleisubstrate reactions. Volatility is an issue in some Cerurn. So<.., 56 [6134142 (1973). atmospheres. Scale melting is a limitation at 1996 K. Finally, ”A. G. Kevcsz and H. A. Schaeffer, “The Mechanism of Oxygen Diffusion in Vitreous SiO?.”J. E/ec,troc.hem.So(.., 129 [2] 357-61 (1982). scale/substrate reactions may lead to extensive bubbling and are ”‘F. J. Norton. ”Gas Permeation Through the Vacuum Envelope”; pp. X-16 in an issue for carbon-rich Sic. Ti.crn.\rrction,\ offbeEighth Nufio17alVucuum S?.inposium. Edited by L. E. Preuss. Condensed-phase deposits are an issue in a narrow region at Pergainon Press, New York, 1962. lower temperatures but can be quite severe. These reactions can ”E. W. Sucov, “Diffusion of Oxygen in Vitreous Silica,” J. Am. Ceram. So<,.. 46 [I] 14-20 (1963). be understood from the acid-base theory of oxides. ”E. L. Williams, “Diffusion of Oxygen in Fused Silica,” J. Am. Cerum. In general, our understanding of the response of SiO, scales So(.. .48 [4] 190-94 ( 1965j to combustion environments lags behind that of A120, and ‘”K. Muehlenbachs and H. A. Schacffer, “Oxygen Diffusion in Vitreous Sil- Cr,O, on metallic alloys. Many important aspects remain to be ica-Utilization of Natural Isotopic Abundances.” Can. Minerd, 15, 179-84 (I977). explored both in the chemistry of these interactions and in the ”‘J.D. Cawley, “Oxygen Tracer Diffusion in Vitreous Silica”; presented at the correlation of corrosion with mechanical properties. 94th Annual Meeting of the American Ceramic Society. Minneapolis. MN, April 14. 1992 (A. R. Cooper Symposium, Paper No. 2-JXI 11-92), J. Am. Acknowledgments: It is a pleasure to acknowledge the many helpful Cenrin. Soc., in review. discussions I had in preparing this paper. In the combustion area. I would likc to ”B E. Deal and A. S. Grove, ”General Relationship for the Thermal Oxida- thank D. Blint of GM Research Laboratories and G. Seng of NASA Lewis. In tion of Silicon.” J. Appl. Phjs.. 36 1121 3770-78 (1965). the oxidation area, I would like to thank J. Smialek, L. Ogbuji, D. Fox, and “R. H. Doremus. “Oxidation of Silicon: Tests of Mechanisms”; pp. 17-23 in E. Opila of NASA Lewis K. Luthra of CR&D, Schenectady, The irnd Chernrstry qfSiO, and /he Si-SiO, Inrrrfuce. Edited by C. R NY; R. Tressler of The Pennsylvania State University, University Park, PA: and Helms and B. E. Deal. Plenum Press, New York, 1988. J. Porter II of MSNW, Inc.. San Marcos, CA. ”J. A. Costcllo and R. E. Tressler, “Isotope Labeling Studies of the Oxidation of Silicon at I000”C and 1300”C,” J. Electrochem. Soc., 131 [XI 194447 ( 1984). References ”J. D. Cawley, J. W. Halloran, and A. R. Cooper, ”Oxygen-18 Tracer Study ‘G. I.. Boyd and D. M. Kreiner, “AGT10liAdvanced Turbine of the Passive Thermal Oxidation of Silicon,” Oxid. Met., 28 [1/2] 1-16 (1987). Applications Project (ATTAP)”; pp. 275-80 in Proceeding.\ ofthe Titwry-Sirrh ’$E.Fitzer and R. Ebi, “Kinetic Studies on the Oxidation of Silicon Carbide”: Auromotive Te(,hnoli)gy Developmenr Conrr-uctors’ Coor-diriufion Mreriiig. pp. 320-28 inSilicon Carbide IY7.3. Edited by R. C. Marshall, J. W. Faust, Jr., P-2 19. Society of Automotive , Warrendale, PA, 1989. and C. E. Ryan. University of South Carolina Press, Columbia, SC, 1974. ’ti. H. Callum and T. E. Schmid, “Ceramic Composites for Aircraft Gas Tur- “‘S. Lin, “Mass Spectrometric Analysix of the Vapors in Oxidation of Si,N, bine Enginer,” AlAA Paper No. 9 I- 1892. American Institute of Aeronautics Compacts,” J. Am. Crrcim. Soc., 58 13-41 160 (1975). and Astronautics, Washington, DC, June 199 I ”M. Wakamatsu, N. Takeuchi. and T. Hattori, “Oxidation of Sintered Si,N, at ‘A. Bennett, “Requirements for Ceramics in Gas Turbine 1400OC in Oxygen Atmosphere”; pp. 1694-703 in UNITECR ‘89. Proceedings Engines,” Muter. Sc.i. Techno/., 2, 895-99 (1986). of the United International Technical Conference on . Edited by L. J. ‘A. E McLean, “Matcrials Approach to EngineiComponent Design”; pp. Trostel, Jr. American Ceramic Society, Westerville, OH. 1989. 1023-34 in Crrcrmic Mutrriuls cmd Comp(inmrafor Engines, Proceedings of the ‘XK.L. Luthra, “Some New Perspectives on Oxidation of Sillcon Carbide aiid Second International Symposium. Edited by W. Bunk and H. Hausner. Verlag Silicon Nitride,”J. Am. Cerum. Soc., 74[5] 1095-103 (1991). Deutsche Keramische Gesellschaft, Bad Honnef, Germany, 1986. “’J Schlicting. “Siliciumcarbid Als Oxidationhestandiger Hochtemperatur- ’J. I.Fcdercr, T. N. Ticgs, D. M. Kotchick, and D. Petrak. “Analysis of Can- didate Silicon Carbide Recuperalor Materials Exposed to Industrial Furnace werkstoff; Oxidations-und Heisddkorrosions-Verhalten I,” Ber. Dtsch. Keram. Environments,” Rept. No. ORNLiTM-9677, Oak Ridge National Laboratory. Ges., 56 [XI 196-200 (1979). Oak Ridge, TN, July I985 “‘J. W. Fergus and W. L. Worrell, “Oxidation of Chemically Vapor Deposited “M. K. Ferber, J. Ogle, V. J. Tennery, and T. Henson, “Characterization of Silicon Carbide”; pp. 43-52 in Ceramic Transactions, Vol. 10, Corrosion and Corrosion Mechanisms Occurring in a Sintered SIC Exposed to Baaic Coal Corrosive Uegradution ($Ceramics. Edited by R. E Tressler and M. McNallan. Slags,”J. Am. Crrtrin. So(..,68[4] 191-97 (1985). Aincrican Ceramic Society, Westerville, OH, 1990. ’D. I? Butt, R. E. Tressler, and K. E. Spear, “Durability of SIC Materiala in “G. H. Schiroky, R. J. Price, and J. E. Sheehan. “Oxidation Characteristics Gaseous N,-H,-CO Heat Treat Environments: I,” lid. Heor., 58 [9] 44-48 of CVD Silicon Carbide and Silicon Nitride,” Rcpt. No. GA-A 18696, GA Tech- (1991). nologies, Inc., San Diego, CA, 1986. ”N.J. Shaw ef a/., “Matcrials for Engine Applications Above 3000°F-An “J. A. Costello and R. E. Tressler, “Oxidation of Silicon Carbide and Overview,” NASA Rept. No. TM-100169, NASA Lcwis Research Center, Ceramics: I, In Dry Oxygen,“ J. Am. Cerum. Soc., 69 191 674-81 (1986). Cleveland, OH, 1987. “C. D. Fung and J. J. Kopanski, “Thermal Oxidation of 3C Silicon Carbide ‘H. Du, R. E. Tressler, and C. G. Pantano, “Oxidation Studies of Crystalline Single-Crystal Layers on Silicon,” Appl. Phys. Lcrr., 45 [7] 757-59 (1984). CVD Silicon Nitride,” J. Etectrochem. Soc., 136 [S] 1527-36 (1989). “L. Muehlhoff. M. J. Bozack, W. J. Choyke, and J. T. Yates, Jr., “Conipara- “‘R. Painpuch, W. Ptak, S. Jonas, and J. Stoch. “Formation of Ternary Si-0- live Oxidation Studies of SiC(O00l) and SiC(0001) Surlices,” J. Appl. Phyr.. C Phase(h) During Oxidation of Sic,” Muter. %i. Monogr., 6,43548 ( 1980). 60 171 2558-63 ( 1986). “J. L. Smialek and G. H. Meier, ”High-Temperature Oxidation”; pp. 293- “’E J. Opila, “The Oxidation Kinetics of Chemically Vapor Deposited Silicon 326 in Superallow /I. Edited by C. T. Sirns, N. S. Stoloff. and W. C. Hagel. Carbide”: presented at TMS Fall Meeting, Chicago, IL, 1992. Wiley, Ncw York, I 988. ‘“K. Motzfcld. “On the Rates of Oxidation of Silicon and Silicon Carbide in “J. H. Gary and (i. E. Hardwick, Petroleum Kefining: Tec~hirchgyand Eco- Oxygen and Correlation with Permeability of Silica Glars,”Acta Chem. Scund , nomrcs; p. 32. Marcel Dckker, New York, 1975. 18 [7] 1596-606 (1964). “North American Combustion Handbook. 2nd ed.: p. Ih. North Aincricaii “Z Zheng, R. E Tressler. and K. E. Spear, “Oxidation of Single-Crystal Sill- Co.. Cleveland, OH. 1978 con Carbide, Part I. Experimental Studies,’’J. Elecrrochem. Soc., 137 131 854- ‘““StandardSpecifications for Automotive Gasoline,” ASTM Designation D 58 (1990). 439-89. 1YYI Book ofASTM Standurds. American Society for Testing and Mate- “Y.Zheng, R. E. Tressler, and K E. Spear, “Oxidation of Single-Crystal Sili- rials, Philadelphia, PA. con Carbide, Part 11. Kinctic Model,” J. Electrochem. Soc., 137 [9] 2812-16 ”V. Valkovic, Trace Elements in Perroleurn; pp. 62-101. PPE Books. Tulaa. (1990). OK, 1978. “’J. D. Cawley and R. S. Boyce, “A Solution of the Diffusion Equation for ‘“R. H. Jungers, R. E. Lee, Jr., and D. J. von Lehmden, “The EPA National Double Oxidation in Dry Oxygen Including Lazy Exchange Between Network Fuels Surveillance Network. I. Trace Constituents in Gasoline and Commercial and Interstitial Oxygen,” Philos. Mng. A, 58 [4] 589-601 (1988). Gasoline Fuel Additives,”Environ. Health Perspect. , 10, 143-50 (1975). “‘I? Kofstad, High-Temperature Oxidorion of Metcrls; p. 125. Wiley, New ”S. Gordon and B. J. McBride, “Computer Program for the Calculation of York, 1966. Complex Chemical Equilibrium Compositions, Rocket Performance, Incident ”T Narushiina. T. Goto, and T. Hirai, “High-Temperature Passive Oxidation and Retlected Shocks. and Chapman-Jouget Detonations,” NASA Rept. No. of Chemically Vapor Deposited Silicon Carbide,” J. Am. Ceram Soc., 72 181 SP-273, NASA Lewis Research Center. Cleveland. OH, 1971. 1386-90 (1989). ‘’N. S. Jacobson, “Sodium Sulfate: Deposition and Dissolution of Silica,” ”D M. Mieskowski. T. E. Mitchell, and A. H. Heuer, “Bubble Formation in Oxid. Met.,31 [1/2]91-103(1989). OxideScaleson SiC,”J. Am. Ceram. Soc.,67[I]C-l7-C-18 (1984). ‘”J. B. Heywood, “Pollutant Fomation and Control in Spark-Ignition ”R. C. A. Harris, “Oxidation of 6-H Sic Platelets,”J. Am. Ceram. SM.,58 Engines,” frog. Energy Combust. Sci., 1, 135-64 (1975). [ 1-21 7-9 (1975). ”‘M. K. Hossain, A. C. Noke, and S. R. J. Saunders, “Oxidation of a Ni-Cr- “D. J. Choi, D. B. Fischbach, and W. D. Scott, “Oxidation of Chemically- Al at 850°C in Air Containing HCI Gas,” Oxid. Met., 12 [5] 451-71 Vapor-Deposited Silicon Nitride and Single-Crystal Silicon,” J. Am. Cerum. (I978). Soc.., 72 [7] 1 118-23 (1989). ”W. D. Kingery, H. K. Bowen, and 11. R.Uhlmann, Inrroducrion to Cerum- “T. Hirai, K. Nihara, and T. Goto, “Oxidation of CVD Si,N, at 1550°C to its, 2nd ed.; p. 595 Wiley, New York, 1976. 1650”C,”J. Am. Crrom. Soc.,63[7-X]419-24 (19x0). January 1993 Corrosion of Silicon-Based Ceramics in Combustion Environmc>nts 27

”H. Du. C. A. Houser. R. E. Tressler, K. E. Spear, and C. G. Pantano, “Iso- ”P. Andrews and F. Riley, “Oxidation of Silicon Nitride during High-Temper- tropic Studies ofOxidation of Si,N, and Si Using SIMS.”J. Electrochem. Sue.. aturc Thermal Cycling“: pp. 259-70 in Arltvinct’d Engitzcrrin‘y with Ccrurnic.,~. 137[2174142(1990). British Ceramic Society Proceedings. No. 46. The Institute of Ceramics, Shel- ”H. Du, R. E. Tressler, and K. E. Spear, “Thermodynamics of the Si-N-O ton. U.K.. 1990. System and Kinetic Modeling of Oxidation of Si,N,,” J. Elet.tnx.henr. Soc., 136 “A. J. Moulson and J. P. Rohcrts. “Water in Silica GI [I I] 3210-15 (1989). Soc., 57, 1208-16(1961). ’XK. L. Luthra, “A Mixed Interface ReactioniDiffusion Control Model for ”E. A. Irene and R. Ghez, “Silicon Oxidation Studies: The Rolc of H,O,” Oxidation of Si,N,>,”J. Electrochem. Soc., 138 [ 101 3001-3007 (1991). J. Elcctroc~hcw.Sue., 124[11] 1757-61 (1977). “’L.U.J.T. Ogbuji, “Role of Si,N,O in the Passive Oxidation of Chenltcally- “‘P. 1. Jorgensen. M. E. Wadsworth, and I. B. Cutler, “Effects of Water Vapor Vapor-Deposited Si,N,,” J. Am. Cerum. Soc., 75 [I11 2995-3000 (1992). onoxidationof Siliconcarbide,” J. Am. Cercrm. SOC.,44(6] 258-61 (1961). “’A. tl. Heuer. L. U Oghuji, and T. E. Mitchell, “The Microstructure of ”‘H. Cappelen, K. H. Johannen, and K. Motzfeld. “Oxidation of Silicon Car- Oxide Scales on Oxidized Si and Sic Single Crystals.” J. Am. Cerum. Soc., 63 bide in Oxygen and Watcr Vapor at 150O0C,”Acta Chem. Scund. A, 35, 247-54 15-61 354-55 (1980). (1981). ‘IS. C. Singhal, “Oxidation Kinetics of Hot-Pressed Silicon Carbide.” ‘”T Narushiina. T. Goto, Y. Iguchi, and T. Hirai, “High-Temperature Oxido- J. Muter. Sci., 11, 1246-53 (1976). tion of Chemically Vapor-Deposited Silicon Carbide in Wet Oxygen at 1823 to ‘3.C. Singhal and F. F. Lange, “Effect of Alumina Content on the Oxidation 1923 K,”J. Am. (’erum. Soc.,73[I2] 1580-84(1990). of Hot-Pressed Silicon Carbide,” J. Am. Cerum. Soc., 58 [9-101 133-35 ”M. Maeda, K.Nakamura, and T. Ohkubo, “Oxidation of Silicon Carhide in ( 1975). a Wet Atmosphere,” J. Muter. Sci.. 23, 3Y3.3-38 (1988). “’S. C. Singhal, “Oxidation and Corrosion-Erosion Behavior of Si,N, and ”-IR. E. Tressler, J. A. Costello, and Z. Zheng. “Oxidation olSilicon Carbide Sic”; pp. 53348 in Ceramics ,fbr High-Performunce Applicutions. Edited by Ceramics”; pp. 307-14 in IndustriulHeutExchun,~er.s.Edited by A. J. Haye\, et J. J. Burke. A. E. Gorum, and R. N. Katz. Metals and Cerainics Information id. American Society of Metals, Warrendale, PA, 1985. Center, Columbus, OH, 1974. ‘”J. R. Blachere and F. S. Pettit, “High-Temperature Corrosion of Ceramics,“ “S. C. Singhal, “Thermodynamics and Kinetics of Oxidation of Hot-Pressed Rept. No. DOEIER-45117.1 for the U.S. Department of Energy under Contract Silicon Nitride,” J. Murer. Sci., 11, 500 (1976). No. DE-FG02-84ER-451 17 by the Department of Metallurgical and Material\ “V. A. Lavrenko, I. G. Gogotsi, A. B. Goncharuk. A. F. Alekseev, 0. N. Gri- Engineering, University ofpittsburgh, Pittsburgh, PA, 1987. gorev, and 0. D. Shcherbina, “High-Temperature Oxidation of Reaction-Sin- “‘R. J. Fordhain, R. A. McCauley, K. M. Friederich, and U. Dworak, “Hot tered Silicon Nitride with Various Additions,” Poroshk. Metnll., 267 [3]35-39 Gaseous Corrosion of Silicon in Carburising, Sulphidising, and Oxidis- (1985). ing Environments”; pp 745-52 in Ccruniic Muteriuh und Compunent,\ jilr “J. W. Hinze, W. C. Tripp, and H. C. Graham, “The High-Temperature Oxi- En,Tines, Proceedings of the Second International Symposium. Edited by dation of Hot-Pressed Silicon Carbide”; pp. 405-19 in Mass Trunsport Phenon- W. Bunk and H. Hausner. Verlag Deutschc Kerdmische Gesellschaft. Bad Hon- enu in Ceramics. Edited by A. R. Cooper and A. H. Heucr. Ainerican Ceramic nef, Germany, 1986. Society, Columbus, OH, 1975. ”’F. C. Oliveira. R. A. H. Edwards, R. J. Fordham, and F. L. Riley, “High- ‘”5. A. Costello, R. E. Tressler, and I. S. T. Tsong, “Boron Redistribution in Teinperature Corrosion of Si,N,-Y,O,-AI,O, Ceramics in H,S/H,O/HI Reduc- Sintered a-SiC During Thermal Oxidation,” J. Am. Crwnn. Soc., 64 [6] 332-35 ing Environments”; pp. 53-68 in High-Tmprrutrrre Corrosion of Technicul ( I98 I). Crruinic,s. Edited by R. J. Fordham. Elsevier, New York, 1990. hXD.Cubicciotti and K. Lau, “Kinetics of Oxidation of Hot-Pressed Silicon ’“J. E. Marra, E. R. Kriedler, N. S. Jacobson. and D. S. Fox, “Reactions of NitrideContainingMagnesia,”J. Am. Ccrum. Soc., 61 [I 1-12] 512-17 (1978). Silicon-Bawd Ceramics in Mixed Oxidation Chlorination Environments,” ‘“D. Cubicciotti and K. H. Lau. “Kinetics of Oxidation of Yttria Hot-Pressed J. Am. Cerum. Soc., 71 [I21 1067-73 (1988). Silicon Nitride,”J. Electrochem. Soc., 126 [lo] 1723-28 (1979). ““S. Y. Ip, M. J. McNallan. and D. S. Park, “Active Oxidation of Sic-Based “‘W. C. Tripp and H. C. Graham, “Oxidation of S,N, in the Range 1300” to Cerainics in Ar-2% CI,-O, Gas Mixtures at 1000°C,” J. Am. CercIni. So(... 75 1500°C,”J. Am. Cerum. Soc., 59[9-101 399403 (1976). [7] 194248 (1992). ”P. Andrcws and F. L. Riley, “The Microstructure and Composition of Oxide ““Z. Zheng. R. E. Trcssler, and K. E. Spear, “The Effects ofCl, on the Oxida- Films Formed During High-Temperature Oxidation of a Sintered Si,N,,” J. Eur. tion of Single-Crystal Silicon Carbide,” Corros. Sci., 33 141 557-67 (1992). Cerum. Sor., 5, 245-56 (1989). ““V. Pareek and D. A. Shore\, “Oxidation of Silicon Carbide in Environments 72J,Schlichting and L. J. Gauckler, “Oxidation of Some (J-Si,N, Materials,” Containing Potassium Salt Vapor,”J. Am. Cerum. Soc.., 74 131 556-63 (199 I). PowderMetull. lnt., 9[1] 36-39 (1977). “”Z. Zheng, R. E. Tressler, and K.E. Spear, “The Effect of Sodium Contanii- ”D. R. Clarke, “Thcnnodynaniic Mechanism for Cation Diffusion Through nation on Oxidation of Single-Crystal Silicon Carbide,” Corrm. %i.. 33 [4] an lntergranular Phase: Applications to Environmental Reactions with Nitrogen 545-56 (1992). Ceramics”; pp 421-26 in Progrers in Nitrqen Ctramics, NATO AS1 Series. ““Z. Zheng, R. E. Tressler. and K. E. Spear, “A Comparison ofthe Oxidation Series E: Applied Sciences, No. 65. Edited by F. L. Riley. Martinus Nijhoff, of Sodium-Implanted CVD Si,N, with the Oxidation of Sodium-Implanted SiC- The Hague, Netherlands, 1983. Crystals,” Corro,!.Sci., 33 [4] 569-80 (I 992). “J. Chen, J. Sjoger, 0. Lindqvist. C. O‘Meara, and L. Pejryd, “The Ratc- ““W. A. Sanders and J. R. Johnston, “High-Velocity Burner Oxidation and Controlling Processes in the Oxidation of HlPed Si,N., with and without Sin- Thermal Fatigue Behavior of Si,N,- and Sic-Base Ceramics to I370”C,” NASA tering Additives.” J. Eur. Ceruni. Soc., 7, 319-27 (1991). Rept. No. TM-79040, NASA Lewis Research Center, Clevcland, OH, 1978. 75J. K. Patel and D. P Thompson, “The Low-Temperature Oxidation Problem “”J. D. Cawley and R. F Handschuh, “Phenomenological Study of the in Yttria-Densilied Silicon Nitride Ceramics,” Br. Crrum. Trans. J.. 87. 70-73 Behavior of Some Silica Formers in a High-Velocity Jet Fuel Burner,” NASA (1988). Rept. No. TM-87127. NASA Lewis Research Center, Cleveland, OH, 1985. ’“A. G. Evans and R. W. Davidge, “The Strength and Oxidation of Reaction- “‘“E. A. Gulbransen and S. A. Janswn, “The High-Temperature Oxidation, Sintered Silicon Nitride,”./. Muter. %i., 5, 314 (1970). Reduction, and Volatilization Reactwns of Silicon and Silicon Carbide,’‘ Orid. ”F. Porz and F. Thummler. “Oxidation Mechanism of Porous Silicon Nilride.” Met., 4[31 181-201 (1972). J. Mater. Sci., 19, 1283-95 (1984). “”F. J. Kohl, D. M. Leisz, G. C. Fryburg, and C. A. Steams, “Thermochein- ”J. E. Antill and J. B. Warburton, “Oxidation of Silicon and Silicon Carbide cal Analyses of the Oxidative Vaporization of Metals and Oxides by Oxygen in Gaseous Atmospheres at l000”-1 300’C’: in Reucrions Between and Molecules and Atoms,” NASA Rept. No. TM X-73682, NASA Lewis Research Gases, AGARD CP-52. Advisory Group for Aerospace Research and Dcvelop- Center, Cleveland, OH, 1977. ment, Paris, France, 1970. ““A. H Heuer and V. L. K. Lou. “Volatility Diagrams for Silica, Silicon ’J. B. Warburton, J. E. Antill, and R. W. M. Hawes, “Oxidation of Thin- Nitride. and Silicon Carbide and Their Application to High-Temperature Sheet Reaction-Sintered Silicon Nitride,” J. Am. Ceruni. Sue., 61, 67-72 Decomposition and Oxidation,”J. Am. Cerum. Soc., 73 [lo] 2789-803 (1990). (I 978). ““‘R. W. Bartlett, “Platinum Oxidation Kinetics with Convective Diffusion ‘“D. S. Fox, “Oxidation Kinetics of CVD Silicon Carbide and Silicon and Surface Reaction,” J. Elec,trochem. Suc., 114 [6] 547-50 (1967). Nitride,” Cerum. Eng. Sci. Proc., 13 [Y-101 83643 (1992). ““N. S. Jacobson, “High-Temperature Durability Considerations for the “C. E. Lowell, J. L. Smialek, and C. A. Barrett, “Cyclic Oxidation ofsuper- HSCT Combustor,” NASA Rept. No. TP-3 162, NASA Lewis Research Center, alloys”; pp. 21 9-26 in High-Tkmperuture Corrosion, Proceedings of the Inter- Cleveland, OH, 1991 national Conference. Edited by R. A. Rapp. NACE, Houston, TX, 1983. ”‘G. H. Geiger and R. H. Poirier, Trunsport Phenomrnu in Mctullur8j.v: x2Y. S. Touhkian, R. K. Kirby, R. E. Taylor, and T. Y. R. Lce, Thermul p. 244. Addison-Wesley, Reading. MA, 1974. Expunsion ofNhnineful/icSolid,s, Thennophysical Properties of Matter, Vol. 13; “’W. B. Hillig, “Prospects for Ultra-High-Temperature Ceramic Compos- pp. 350,358,873, and 1140. IFIiPlenum, New York, 1970. ites”; p. 699 in TuilorinR Multiphase and Composite Cerumics. Edited by R. E. ”E. J. Opila, D. S. Fox, and C. A. Barrett, “Cyclic Oxidation of Monolithic Tressler, G. L. Messing, C. G. Pantano, and R. E. Newnham. Plenum, New SIC and Si,N, Materials”; to be presented at the 17th Annual Conference on York, I986 Composites and Advanced Ceramics (Paper No. ECDCR 93-44), Cocoa Beach, “‘T. M. Besmann, ’~LGASMIX-PV: A Computer Program to Calculate Equi- FL, 1993. librium Relationships in Complex Chemical Systems,” Rept. No. ”L. J. Lindberg, ”Durability Testing of Ceramic Materials for Turbine Engine ORNLiTM-5775, Oak Ridge National Laboratory, Oak Ridge, TN, 1977. Applications”; pp. 149-6 I in P rocredings oj the T’venh-Fourth Automotive “T,S. Tedmon, Jr., “The Effect of Oxide Volatilization on the Oxidation Technology Development Contractors’ Coordination Meeting, P- 1 97. Society of Kinetics of Cr and Fe-Cr Alloys,” J. Electrochem. Soc., 113 [XI 76648 (1966). Automotive Engineers, Warrendale, PA, 1986. “’E. T. Turkdogan, P. Grieveson, and L. S. Darken, “Enhancement of Diffu- ”L. Lindberg, “Cyclic Oxidation of Advanced Silicon Nitrides”: pp. 79-86 in sion-Limited Rates of Vaporization of Metals,” J. Pli-yx. Chem., 67, 1647-54 Proceedings oj the Annuul Automotive Technology Development Contractors’ (1963). Meeting. Society of Automotive Engineers, Warrendale, PA, 1991. “‘J. W. Hinze and H. C. Graham, “The Active Oxidation of Si and Sic in the X“M.Maeda, K. Nakamura, T.Ohkubo, J. Ito, and E. Ishii, “Oxidation Behav- Viscous Gas-Flow Regime,”./. Electrdzem. Soc., 123 [7] 1066-73 (1976). ior of Silicon Nitride under Static and Cyclic Conditions,”Cerum. Int.. 15,247- “’C. Wegner, “Passivity during the Oxidation of Silicon at Elevated Tempera- 53 (1989). tures,” J. Appl. Phjs., 29 [Y] 1295-98 (19%). 28 Journul of’ihe American Cerumic Society-Jucobson Vol. 76, No. 1

YS.C. Singhal, “Thermodynamic Analysis of the High-Temperature Stahil- ‘TSato. S Tcrauchi, T. Endo. and M. Shimada, “Corrosion of Si,N, and ity of Silicon Nitride and Silicon Carbide,” Cwutn. In/.,2 [3]123-30 (1976). in VIO, Melts,” J. Mutcr. Sci., 25, 1231-36 (1990) ““I. Narushima, T Goto, and ‘I. Hirai, “Active-to-Passive Transition in the ‘Yv~.K. Fcrber and V. J. Tcnnery, “Evaluation of Tubular Ceramic Heat Oxidation of CVD-Sic’; pp. 295-300 in C‘OJ~~I,V~/~~.S:C~I,I-ro.sron/C‘oariJ~,~.% of Exchanger Materials in Acidic Coal Ash for Coal-Oil-Mixture Combustion.” Adwncc,d Mcr/cwu/.v.Editcd by S. Kimur;i arid M. I)oyama Material\ Resciirch Rept. No. ORNL/TM-7958, Oak Ridge National Laboratory, Oak Ridge, TN. Society, Pittsburgh, PA, 1989. 1981. ““K. Nickel, “The Role of Condensed in thc Activc-to- ‘“‘M.I. Mayer and F. L. Riley. “Sodiurri-Assisted Oxidation of Reaction- Passive Oxidation Transition of Silicon Carbide,” J. Eitr. C‘rrunt. Soc. I 9, 3-8 Bonded Silicon Nitride,” J. Murrr. Sci., 13, 1319-28 (1978). (1992). “‘A. G. Evans, “Perspective on thc Development of High-Toughness Ccram- ‘“I? Rocabois, C. Chatillon. and C. Benard, “Vapor Pressure and Evaporation J. Am. Certinz. So(.., 73 [2] 187-206 (1990). Coefticicnt of SiO(ainorphoua) and SiO,(\) + Si(.\)Mixtures by the Multiple thod’: to appeal- in Proceedings of the 7th R. T. Bhatt, “Oxidation Elfccti on the Mechanical Properties of SiC-Fiber- erials Properties Conference (IUPAC), forced Reaction-Bonded Silicon Nitride Matrix Composites.” NASA Rept. No. TM- 102360, NASA Lewi\ Research Center. Cleveland, OH, 1989. ‘??J. E. Antill and J. 8.Warburton, “Active-to-Passive Transition in the Oxi- “’L. Filipur/.i, G. Carnus, R. Naslain, and J. Thcbault, “Oxidation Mccha- dation ofSiC,”Corro.\.Sci., 11, 33742 (1971). nisiiis and Kinetics of 1 D-SiCICISiC $: I, An Experimental “‘W. L Vaughn and H. G. Maahs, “Active-to-Passive Transition in the Oxi- Approach.” J. Ani. C‘croiri. Soc., in review dation of Silicon Carbide and Silicon Nitride in Air.” J. Ant. (’twin. So(,. , 73 [6] ‘“1.. Filipuui aitd R. Naslain, “Oxitlation Mechanisms and Kinetics of 11)- 154043 (1990). SiCICISiC Coiripo\ite Materials: It, Modcling.” J. Am. Curom. Soc.. in review,. P Butt. R. E. Trcsslcr, and K. E. Spear. Corrosion Sic Materials in “T. of “’T. E Easlcr, R. C. Bradt, and R. E. Tressler, “Strength Diytributions of Sic CO Gaseous Environments: I, Thcrinodynamics and Kinetics of Reac- ./. Am. Ctwm. Sot.., 75 [ 121 3257-07 (1992). Ceraiiiics After Oxidation and Oxidation Under Load.” J. Am. Cemm. Sac. .. 64 . l? Butt, K. E. Tressler, and K. E. Spear, “Corrosion dSiC Materials in [ 121 73 1-34 ( 1981). CO Gascou\ Environments: II, Durability iind Mechanical Properties.” ““K. Jaku\, J. E. Ritter, Jr., and W. P Kogera. “Strength of Hot-Pressed Sili- C‘mi~~i.~uc..75(1213268-77 (iW2). con Nitride After High-Temperature Exposure.” J. Aw. Cwtim. Soc.. 67 [7] ’%. H. Schiroky, “Oxidation Behavior of Chemically Vapor-Deposited Sili- 172-75 ( 1984). con Carbide,”Adv. C‘crum. Muter.. 2 121 13741 (1987): ‘“P E Becher, “Strength Retention in SIC Ceramics After Long-Term Oxida- “’N. S. Jacobson, K. N. Lee, and D. S Fox, “Reactions of SiC and SiO, at tion,“./. Am. C‘crum So(c.66 [X]C-l20-C-l21 (1983). Elcvated Temperatures.“./. Atn. C’o.rim. So(-., 75 161 1603-1 (1992). I “‘J L. Sniialck and N S. Jacobson, “Mechanijni of Strength Degradation for ”W A Rapp, “Chcnristry and Electrochemistry ofthe Hot Corrosion of Met- Hot Corrosion of a-Sic,” J. AM. C‘ertrm. Soc.. 69 [lo] 741-52 (1986). aIs.”Curros. Sci.. 42 [I01 568-77 (1986). “”F. S Pettit and C S. Giggins. “Hot Corrosion”; pp. 327-54 in Sir/xw/k)x.\ “”G. A. Gogotsi. Yu. G. Gogotri. V. P Zavada. and V. V. Tnskovsky, ”Stress I/. Edited hy C. T. Simh. N. S. Stololl, and W. C. Hagcl. Wiley. New York, Corrosion of Silicon Nitride Based Ceramics.” Ceruin. 1,171.. IS, 305-10 (19x9). 1987. ‘”‘J. J. Swab and G. L. Leatherman. ”Strength of Structural Ceramics after “”N S. Jacobson, J. L. Sniialck, and D Fox. “Molten Salt Corrosion of Exposure to Sodium Sulfate,” J. Eur. Co-an.So( .. 5, 33340 (1989) SIC and Si,N,”; pp. 99-136 in Hundliook of rmir s uiid Cotnposi/e,v.Vol. I , ‘“W. J. Tomlinson and I). M Caslin, ”Strength of Sic and Sic-7.5 vol% Synthesi\ and Properties. Edited by N. P Chcrcmisinoff. Marcel Dckker. New TiB? Composite after Corrosion with Na,SO, and NaCl Deposits.” CP~UJ~.In/., York, 1990. 17,6l-66(1991) and “‘W. C. Say. J. K Wu, W. L. Chen, “Hot Corrosion 01 a-Sit Ceramics ”?R. E Tressler, “Environmental Effect\ on Long-Tern1 Reliability of Sic and by V,O, Mclt,”J. Mutm ki., 25, 1514-17 (1990). Si,N, Ceramics”; pp. 99-123 in Ceramic Transactions, Vol. 10, Corrosion und “‘N. S. Jacobson. C. A. Stearns. and J. L. Siiiialek, “Burner Rig Corrohion of fhgrut/u/iotic!fCerumi<~\. R. SIC at 1000T.“Adv. C‘erum. Muwr.. 1121 154-61 (1986). Corrmii? Editcd by E. Tressler and M. McNallan. ”’D.-Z. Shi and R. A. Rapp, “The Solubility 01 SiO, in Fuaed NalSO, at American Ceramic Society, Westerville. OH, 1990. 900”C,” J. E/ec./rochrtn.Sor., 133 [4] 849-50 ( 1986). ”‘K. G. Nickel, R. Ilanscr, Schncider, and G. Pctzow. “Corrosion and ‘WW. Watt, R. E. Andrcsen. and R. A Rapp, “A Comparison of Rcrerence Oxidation of Advanced Ceraini Porvder Merd!. /M., 21 [3]29-34 (1989). Electroder in Molten Sodium Sulfate”; p. XI in Molrcw7 SU/~.Y.Edited by ”‘C. H. Hcnager, Jr.. iind R. H Jonea, “Molten Salt Corrosion of Hot-Pressed J. Braunstein and J. R. Selman. The Electrochemical Society, Pennington, NJ, Si ,N ,!Sic-Reinforced Composites and Effects of Molten Salt Exposure on Slow 1981. Crack Growth of Hot-Preshcd Si,N,”: pp. 197-2 10 in Ceramic Transactions. ‘I’M. McNiillan. M. van Roodc, and J. R. Price. “The Mechanism of HiFh- Temperature Corrosion 01 SIC in Flue Case\ Ironi Aluminum Remelting Fur- Vol 10. C‘orrosiori nnd Corroswr Lk~firudutiorzof Ceramic..%.Edited by R. E naces”: pp. 181-91 in Ccramic Transactions, Vol. 10, Corrotioti um! Corro\ive Trcssler and M McNallan. American Ceramic Society. Westerville, OH, 1990. D~gr~dtiti~~iof’(‘c,r(rmic..\ Edited by R. E. Trcraler and M. McNallan. American ‘”S. M. Wiedcrhorn, “Moisture-Assisted Crack Growth in Ceramics.” /n/.J. Ceramic Society, Westcrville. OH, 1990. /“rucf. Mwh., 412) 171-77 (1968). ”“M. A. Lamkin, E L. Riley. and R. J. Fordham, “Hot Corrosion of Silicon ”“L. Filipu7zi. G. Camus, J. Thehault, and R. Naalain, “Effect of High-Tem- Nitride in the Prtwnce of Sodium and Vanadium Compounds”: pp. 181-9 I in perattire Ageing Treatments on the Mechanical Behaviour of Unidirectional SIC/ Hifiii-~~rn/)errrrurr,ru/nr~,(‘orro.vinn ofrC.( hnimi Ceromicr. Edited by R. J. Fordham. Elscvier Applied Science. London, U.K.. 1990. Sic Fibrous Composites”; in Proceedings of the I I th Rino International Sympo- “’S. Brooks and D. B. Meadowcroft. “The Corrosion of Silicon-Based siiini on Metallurgy and : Structural Ccramics-Processing, Ceramic\ in a Re\idual Oil Fired Envirrinment,” Prtx.. Hr. Ccrutrt. .So(,. . 26, Microstructure, and Pmpcrties. Edited by J. J. BentLen PI u/. Riso National Lah- 237-50 (197X). oratory. Roskilde, Denmark, 1990. r8

Nathan S. Jacobson is a senior research scientist in the Materials Division at the NASA Lewis Research Center. He received his B.A. degree in chemistry from the University of California at San Diego in 1975. He received his M.S. and Ph.D. degrees in materi- als science lrom the University of California at Berkeley in 1978 and 1981, respec- tively. From 1981 to 1983, Dr. Jacobson was a post-doctoral fellow at the University of Pennsylvania. Since 1983, he has been with the Environmental Durability Group of the NASA Lewis Research Center, doing research on the oxidation and corrosion of ceram- ics and metals at high temperatures. His current research is focused on high-tempera- ture interfacial reactions and vaporization processes, as limiting factors in applications of ceramics and composites. In 1992, he received a NASA Exceptional Scientific Achievement Medal for his contributions to the corrosion of ceramics. Dr. Jacobson has authored or coauthored more than 50 technical papers.