Search

A thermal barrier coating system survived burner rig testing at 1300°C for 500 h. A 160 µm thick YSZ coating was applied to a Ti2AlC MAX phase sample by PS-PVD and tested face-on in an atmospheric Mach 0.3 jet fuel burner, using 5-h thermal cycles. No TBC spallation or recession was observed, only a 2.4 mg/sq. cm mass gain. The modest weight gain precluded severe volatility losses under high velocity burner conditions. The coating surface exhibited colonies of (111) flourite fiber-textured columns separated by craze patterns, with no visible moisture attack. The metastable tetragonal t' YSZ phase was obtained initially, transitioning to teq and cubic YSZ, but with little detrimental monoclinic. The thickness of the alumina TGO was ~21-23 m under the heated YSZ face and ~13-15 m on the uncoated, cooler backside. The backside exhibited removal of initial transient TiO2 nodules and etching of the underlying Al2O3 scale by volatile hydroxides in high temperature, high velocity water vapor. Aerodynamic forces produced some bending of the cantilevered sample due to creep. The test indicated exceptional stability of YSZ coatings on Ti2AlC under turbine conditions.

Oxidation of high-temperature aerospace materials is a universal issue for combustion-path components in turbine or rocket engines. In addition to the question of the consumption of material due to growth of protective scale at use temperatures, there is also the question of cyclic effects and spallation of scale on cooldown. The spallation results in the removal of part of the protective oxide in a discontinuous step and thereby opens the way for more rapid oxidation upon reheating. In experiments, cyclic oxidation behavior is most commonly characterized by measuring changes in weight during extended time intervals that include hundreds or thousands of heating and cooling cycles. Weight gains occurring during isothermal scale-growth processes have been well characterized as being parabolic or nearly parabolic functions of time because diffusion controls reaction rates. In contrast, the net weight change in cyclic oxidation is the sum of the effects of the growth and spallation of scale. Typically, the net weight gain in cyclic oxidation is determined only empirically (that is, by measurement), with no unique or straightforward mathematical connection to either the rate of growth or the amount of metal consumed. Thus, there is a need for mathematical modeling to infer spallation mechanisms. COSP is a computer program that models the growth and spallation processes of cyclic oxidation on the basis of a few elementary assumptions that were discussed in COSP: A Computer Model of Cyclic Oxidation, Oxidation of Metals, vol. 36, numbers 1 and 2, 1991, pages 81-112. Inputs to the model include the selection of an oxidation-growth law and a spalling geometry, plus oxide-phase, growth-rate, cycle-duration, and spall-constant parameters. (The spalling fraction is often shown to be a constant factor times the existing amount of scale.) The output of COSP includes the net change in weight, the amounts of retained and spalled oxide, the total amounts of oxygen and metal consumed, and the terminal rates of weight loss and metal consumption.

The Materials Division of the NASA Lewis Research Center has been heavily involved in the cyclic oxidation of high temperature materials for 30 years. Cyclic furnace and burner rig apparati have been developed, refined, and replicated to provide a large scale facility capable of evaluating many materials by a standard technique. Material behavior is characterized by weight change data obtained throughout the test, which has been modelled in a step-wise process of scale growth and spallation. This model and a coupled diffusion model have successfully described cyclic behavior for a number of systems and have provided insights regarding life prediction and variations in the spalling process. Performance ranking and mechanistic studies are discussed primarily for superalloys and coating alloys. Similar cyclic oxidation studies have been performed on steels, intermetallic compounds, thermal barrier coatings, ceramics, and ceramic composites. The most common oxidation test was performed in air at temperatures ranging from 800 deg. to 1600 C, for times up to 10000 h, and for cycle durations of 0.1 to 1000 h. Less controlled, but important, test parameters are the cooling temperature and humidity level. Heating and cooling rates are not likely to affect scale spallation. Broad experience has usually allowed for considerable focus and simplification of these test parameters, while still revealing the principal aspects of material behavior and performance. Extensive testing has been performed to statistically model the compositional effects of experimental alloys and to construct a comprehensive database of complex commercial alloys.

This review is based on the phenomenon of improved oxide scale adhesion for desulfurized superalloys. The proposed adhesion mechanism involves sulfur interfacial segregation and scale-metal bond weakening. Sulfur surface segregation on superalloys is examined as a function of temperature and sulfur content, and is related to the classical behavior predicted by the McLean isotherm. Effective desulfurization to less than 1 ppmw can be accomplished by hydrogen annealing and is described by sulfur diffusion kinetics in nickel. Hydrogen annealing results in excellent cyclic oxidation resistance for a number of advanced superalloys. The concept of a critical sulfur content is discussed in terms of practical annealing conditions and section thicknesses.

The relationships between alloy microstructure and air oxidation kinetics and alloy microstructure and microhardness in the Al-Ti-Cr system for exposures at 800 C and 1000 C were investigated. The relevant phases were identified as tau (Ll2), gamma (LIO), r-Al2Ti, TiCrAl (laves), and Cr2AI. Protective alumina formation was associated with tau, Al-rich TiCrAl, and gamma/TiCrAl mixtures. Brittleness was associated with the TiCrAl phase and tau decomposition to A12Ti + Cr2AI. It was concluded that two-phase gamma + TiCrAl alloys offer the greatest potential for oxidation resistance and room temperature ductility in the Al-Ti-Cr system.

Some nickel based superalloys show reduced oxidation resistance from the lack of an adherent oxide layer during high temperature cyclic oxidation. The segregation of sulfur to the oxide-metal interface is believed to effect oxide adhesion, since low sulfur alloys exhibit enhanced adhesion. X ray Photoelectron Spectroscopy (XPS) was combined with an in situ sample heater to measure sulfur segregation in NiCrAl, PWA 1480, and NiAl alloys. The polished samples with a 1.5 to 2.5 nm (native) oxide were heated from 650 to 1100 C with hold times up to 6 hr. The sulfur concentration was plotted as a function of temperature versus time at temperature. One NiCrAl sulfur study was performed on the same casting used by Browning to establish a base line between previous Auger Electron Spectroscopy (AES) results and the XPS results of this study. Sulfur surface segregation was similar for PWA 1480 and NiCrAl and reached a maximum of 30 at% at 800 to 850 C. Above 900 C the sulfur surface concentration decreased to about 3 at% at 1100 C. These results are contrasted to the minimal segregation observed for low sulfur hydrogen annealed materials which exhibit improved scale adhesion.

The isothermal oxidation kinetics of the TiAl3 compound over a wide temperature range is documented, and these rates are related to exclusive alpha-Al2O3 scale growth. The specific weight change vs time curves are shown. Two abnormalities are immediately apparent. One is that a rapid initial uptake of oxygen occurs at times less than 5 h, followed by a lower oxidation rate at longer times, for tests at 900 C and below. The other is that the final weight changes for the 700, 800, and 900 C tests are not in the sequence expected with respect to temperature. Isothermal oxidation of drop cast TiAl above 1000 C was found to exhibit parabolic oxidation controlled by protective alpha-Al2O3 scale formation. TiAl is the only phase in the binary Ti-Al system that forms exclusive scales of alpha-Al2O3 in isothermal oxidation. High anomalous rates at short times and at temperatures below 1000 C resulted from the internal oxidation of a second phase of aluminum.

Microstructural and energy dispersive spectrometry evidence is produced, from transmission electron microscopy, to show that a silicon oxynitride inner layer is produced by the oxidation of silicon nitride in dry oxygen at 1350 C as proposed by Tressler et al. (1989). However, details of the microstructures at the oxide/nitride interface do not agree entirely with the rest of the Tressler model for the oxidation of Si3N4.

Gravimetric, SEM, and XRD data are presented which document the significant improvement obtainable in the oxidation resistance of Al3Ti-containing alloys through additions of Cr. The L1(2) Al(67)Cr(8)Ti25 alloy exhibited excellent cyclic oxidation resistance at 1473 K, with the primary oxide formed being the ideally protective alpha-Al2O3. The Al(67)Mn(8)Ti(25) alloy also tested for comparison exhibited poor cyclic oxidation resistance, with substantial occurrence of TiO2 in the protective scales. Catastrophic oxidation was also encountered in the quaternary alloy Al(67)Mn(8)Ti(22)V(3).

The transient oxidation of beta-NiAl in air at 800 C and 1100 C has been studied using electron microscopy. The oxide scale consists predominantly of metastable Al2O3 phases. Theta-Al2O3 is the major oxide phase within 10.0 hr of oxidation at 800 C and 0.1 hr at 1100 C. The scales form epitaxially on (001) sub beta and (012) sub beta specimens throughout the transient stage, whereas the degree of preferred oxide orientation decreases with oxidation time on (011) sub beta and (111) sub beta specimens. The orientation relationships reflect the small mismatch between parallel close-packed directions in the metal and in the cation sublattice of the oxides. The correlation of distinctive oxide surface morphologies with internal structural defects indicates the strong tendency of the Al2O3 scale to grow via short-circuit diffusion paths.

NbAl3 is one candidate material for advanced aeropropulsion systems because of its high melting point, low density, and good oxidation resistance. Although NbAl3 has the lowest oxidation rate among the binary Nb-Al alloys, it does not form exclusive layers of protective Al203 scales. Recently Perkin et al., have shown the feasibility of forming alumina scales on Nb-Al alloys at greatly reduced Al contents. However, the objective was to maintain the high Al content, and hence low density, while achieving the capability of growing protective alumina scales. Alloy development followed approaches similar to those used successfully for superalloys and oxidation resistant MCrAly coatings. Among the three elements examined (Ti, Si, and Cr) as ternary additions to Nb-Al3, Cr was the most effective in favoring the selective oxidation of Al. Nb-41Al-8Cr formed exclusive layers of alumina and had a k sub p value of 0.22 mg squared/cm (sup 4)/hr at 1200 C. The addition of 1 wt percent Y to this alloy was also beneficial, resulting in nearly an order of magnitude decrease in K sub p at 1200 C. Further improvements were achieved by adding about 1 wt percent Si to the quaternary alloy. The k sub p value of 0.012 mg squared/cm (sup 4)/hr for Nb-40Al-8Cr-1Y-1Si at 1200 C was identical to the best NiAl + Zr alloys. These NbAl3 alloys also exhibited excellent cyclic oxidation resistance for 100 hr at 1200 C, being nearly equivalent to NiAl + Zr.

The oxidation of Ni-Al intermetallic alloys in the beta-NiAl phase field and in the two phase beta-NiAl/gamma'-Ni3Al phase field has been studied between 1000 and 1400 C. The stoichiometric beta-NiAl alloy doped with Zr was superior to other alloy compositions under cyclic and isothermal oxidation. The isothermal growth rates did not increase monotonically as the alloy Al content was decreased. The characteristically ridged alpha-Al203 scale morphology, consisting of cells of thin, textured oxide with thick growth ridges at cell boundaries, forms on oxidized beta-NiAl alloys. The correlation of scale features with isothermal growth rates indicates a predominant grain boundary diffusion growth mechanism. The 1200 C cyclic oxidation resistance decreases near the lower end of the beta-NiAl phase field.

NbAL3 is one candidate material for advanced aeropropulsion systems because of its high melting point, low density, and good oxidation resistance. Although NbAl3 has the lowest oxidation rate among the binary Nb-Al alloys, it does not form exclusive layers of protective Al2O3 scales. Recently Perkin et al., have shown the feasibility of forming alumina scales on Nb-Al alloys at greatly reduced Al contents. However, the objective was to maintain the high Al content, and hence low density, while achieving the capability of growing protective alumina scales. Alloy development followed approaches similar to those used successfully for superalloys and oxidation resistant MCrAlY coatings. Among the three elements examined (Ti, Si, and Cr) as ternary additions to Nb-Al3, Cr was the most effective in favoring the selective oxidation of Al. Nb-41Al-8Cr formed exclusive layers of alumina and had a k sub p value of 0.22 mg squared/cm (sup 4)/hr at 1200 C. The addition of 1 wt percent Y to this alloy was also beneficial, resulting in nearly an order of magnitude decrease in K sub p at 1200 C. Further improvements were achieved by adding about 1 wt percent Si to the quaternary alloy. The k sub p value of 0.012 mg squared/cm (sup 4)/hr for Nb-40Al-8Cr-1Y-1Si at 1200 C was identical to the best NiAl + Zr alloys. These NbAl3 alloys also exhibited excellent cyclic oxidation resistance for 100 hr at 1200 C, being nearly equivalent to NiAl + Zr.

The corrosion of SiC by thin films of Na2CO3 and Na2SO4 at 1000 C is characterized by a severe pitting attack of the SiC substrate. A range of different Si and SiC substrates were examined to isolate the factors critical to pitting. Two types of pitting attack are identified: attack at structural discontinuities and a crater-like attack. The crater-like pits are correlated with bubble formation during oxidation of the SiC. It appears that bubbles create unprotected regions, which are susceptible to enhanced attack and, hence, pit formation.

The most severe type of corrosion encountered in heat engines is corrosion by molten sodium sulfate, formed by the reaction of ingested sodium chloride and sulfur impurities in the fuel. This problem was studied extensively for superalloys, but only recently examined for ceramics. This problem is addressed with laboratory studies to understand the fundamental reaction mechanisms and with burner studies to provide a more realistic simulation of the conditions encountered in a heat engine. In addition the effect of corrosion on the strengths of these materials was assessed. Each of these aspects will be reviewed and some ideas toward possible solutions will be discussed.

Sintered alpha-SiC was examined in both oxidation and hot corrosion with a burner rig at 400 kPa (4 atm) and 1000 C with a flow velocity of 94 m/s. Oxidation tests for times to 46 h produced virtually no attack, whereas tests with 4 ppm Na produced extensive corrosion in 13.5 h. Thick glassy layers composed primarily of sodium silicate formed in the salt corrosion tests. This corrosion attack caused severe pitting on the silicon carbide substrate and led to a 32 percent decrease in strength, compared to the as-received material. Parallel furnace tests of Na2SO4/air-induced attack yielded basically similar results, with slight product composition differences. The differences are explained in terms of the continuous sulfate deposition which occurs in a burner rig.

Thin films of Na2SO4 and Na2CO3 at 1000 C lead to severe pitting of sintered alpha-SiC. These pits are important as they cause a strength reduction in this material. The growth of product layers is related to pit formation for the Na2CO3 case. The early reaction stages involve repeated oxidation and dissolution to form sodium silicate. This results in severe grain boundary attack. After this a porous silica layer forms between the sodium silicate melt and the SiC. The pores in this layer appear to act as paths for the melt to reach the SiC and create larger pits.

An oxidation resistant coating is produced on carbon-base materials, and the same processing step effects an infiltration of the substrate with silicon containing material. The process comprises making a slurry of nickel and silicon powders in a nitrocellulose lacquer, spraying onto the graphite or carbon-carbon substrate, and sintering in vacuum to form a fused coating that wets and covers the surface as well as penetrates into the pores of the substrate. Optimum wetting and infiltration occurs in the range of Ni-60 w/o Si to Ni-90 w/o Si with deposited thicknesses of 25-100 mg/sq. cm. Sintering temperatures of about 1200 C to about 1400 C are used, depending on the melting point of the specific coating composition. The sintered coating results in Ni-Si intermetallic phases and SiC, both of which are highly oxidation resistant.

The hot corrosion of sintered alpha-SiC by thin films of Na2SO4 and Na2CO3 was studied at 1000 C in controlled gas atmospheres. Under all conditions, corrosion led to 10 to 20 times the amount of SiO2 formed in pure oxidation after a 48-h exposure. In addition, small amounts of sodium silicate formed. Melts of Na2SO4/SO3 caused uniform pitting of the SiC substrate; Na2CO3/CO2 melts caused localized pitting and grain-boundary attack. In all cases, the protective SiO2 layer dissolved to form silicate, leading to corrosion. In the sulfate case, free carbon in the SiC promotes this process. In all cases the presence of liquid films is responsible for rapid transport rates and the subsequent rapid reaction.

The atom superposition and electron delocalization molecular orbital theory and large cluster models have been employed to study cation vacancy diffusion in alpha-Al2O3 and the bonding of alpha-Al2O3 to nickel, aluminum, and yttrium surfaces. Al(3+) diffusion barriers in alpha-Al2O3 by the vacancy mechanism are in reasonable agreement with experiment. The barrier to Y(3+) diffusion is predicted to be much higher. Since addition of yttrium to transition metal alloys is known to reduce the growth rate and stress convolutions in protective alumina scales, this result suggests the rate-limiting step in scale growth is cation vacancy diffusion. This may partially explain the beneficial effect of yttrium dopants on scale adhesion. The theory also predicts a very strong bonding between alumina and yttrium at the surface of the alloy. This may also be important to the adhesion phenomenon. It is also found that aluminum and yttrium atoms bond very strongly to nickel because of charge transfer from their higher lying valence orbitals to the lower lying nickel s-d band.

Sintered alpha-SiC was examined in both oxidation and hot corrosion with a burner rig at 400 kPa (4 atm) and 1000 C with a flow velocity of 310 ft/sec. Oxidation tests for times to 46 hr produced virtually no attack, whereas tests with 4 ppm Na produced extensive corrosion in 13-1/2 hr. Thick glassy layers composed primarily of sodium silicate formed in the salt corrosion tests. This corrosion attack caused severe pitting of the silicon carbide substrate which led to a 32 percent strength decrease below the as-received material. Parallel furnace tests of Na2SO4/air induced attacked yielded basically similar results with some slight product composition differences. The differences are explained in terms of the continuous sulfate deposition which occurs in a burner rig.

Additional hot stage Auger experiments have provided surface segregation data for NiCrAl + or - Y or Zr alloys in agreement with other investigations. This data, combined with experimental and theoretical evidence of the Al2O3-metal bond strength, is presented in support of a chemical mechanism of Al2O3 scale adhesion. Both the detrimental effects of sulfur segregation and the beneficial effects of dopant segregation may be important. Chemical features of the dopants are compared in light of these proposed mechanisms, namely delta H sub f (sulfide), delta H sub f (oxide), electron orbital configuration, and insolubility in Ni.

The oxidation of single crystal beta-NiAl has been studied primarily using electron microscopy. Oriented metastable Al2O3 phases form during transient oxidation at 800 C. Specific orientation relationships exist on all metal orientations studied and are a result of the small mismatch along aligned close-packed directions in the cation sublattices of the metal and oxide. Transformation of the metastable Al2O3 phases at 1100 C results in an oxide morphology described as the 'lacey' structure of alpha-Al2O3 scales. This structure results from impingement of oriented patches of alpha-Al2O3 as the transformation initiates and moves radially parallel to the surface. Scale growth occurs by diffusion along high angle grain boundaries. A drastic reduction in oxidation rate accompanies the change in oxide morphology.

New process uses paint sprayer and vacuum furnace to produce silicon carbide outer layer. In cross section of silicon and silicon carbide reaction zone, top layer of silicon adheres to silicon carbide layer. Crystals prominent on melted top surface of slurry coating. Process especially useful in coating repair.

Sintered alpha SiC was corroded by thin films of Na2SO4 and Na2CO3 molten salts at 1000%. This hot corrosion attack reduced room temperature strengths by as much as 50%. Strength degradation was porportional to the degree and uniformity of corrosion pitting attack as controlled by the chemistry of the molten salt. Extensive fractography identified corrosion pits as the most prevalent source of failure. A fracture mechanics treatment of the strength/pit depth relationship produced an average K sub IC equal to 2.6 MPa sub m 1/2, which is consistent with published values.

Microstructural characteristics of Al2O3 scales grown as protective coatings on NiCrAl alloys used in jet engines are described. The alloys were pure or doped with 0.3 percent Zr or Y and oxidized in 1 atm air at 1100 C for 0.1, 1 or 20.0 hr. The scales were then examined under a microscope. Transient epitaxial scales, formed during the 0.1 hr treatment and containing Ni, Cr and Al, consisted of a mosaic of subgrains and precipitates of different phases. The Y and Zr dopants had no effect on the nucleation site locations. The appearance of intergranular porosity at 0.1 hr was exacerbated after the 1 hr treatment. A bimodal void distribution appeared after 20 hr, when no porosity was evident. The detection of local areas of preferred orientation is taken as a spur to further studies of scale growth to gain control of the grain size or even to produce single crystal scales.