Concept of Fgm

8/3/2019 Concept of Fgm

1/11

Concept of Functionally Graded Materials for

Advanced Thermal Barrier Coating Applications

Woo Y Lee* and David P. Stinton'OakRidge National Laboratory, OakRidge, Tennessee 37831

-

6063

Christopher C. Berndt'

Department of Materials Science and Engineering, State University of New York at Stony Brook,

Stony Brook, New York 11794-2275

Fazil Erdogan and Yi-Der LeeDepartment of Mechanical Engineering and Mechanics, Lehigh University, Bethlehem, Pennsylvania 1801 5

Zaher MutasimSolar Turbines, Inc., San Diego, California 921865376

is feature article explores the concept of creating func-nally graded metal-ceramic composite microstructures

r thermal barrier coatings used in gas-turbine applica-ns. From a thermomechanical perspective, this concepters the possibility of significantly improving the life andiability of thermal barrier coatings. However, prior researchveals that progress has been somewhat limited because ofe oxidative instability exhibited by some metal-ceramicmposite microstructures. The present study addressesme of the materials criteria and research issues associatedth preparing chemically stable, yet mechanically durable,aded metal-ceramic microstructures for realistic appli-tion environments.

.A. Munir-contributing editor

Manuscript No. 191958. Received March 11,1996; approved September 16, 1996.upported by the Advanced Gas Turbine Systems Program, Office of Industrialhnologies, and in part by the High-Temperature Materials Laboratory Fellowshipgram, Office of Transportation Technologies, U.S. Department of Energy, underntract No. DE-AC05-960R22464 with Lockheed Martin Energy Researchporation.

Member, American Ceramic Society.

I. Introduction

DURING the past several decades, the gas-turbine communityhas made considerable progress in using ceramic coatings

to protect metals from high-temperature turbine environments.These ceramic coatings, commonly referred to as thermal bar-rier coatings (TBCs), are currently used, in conjunction with aircooling, to prolong the life of metallic "hot-section" turbinecomponents in revenue-generating aircraft engine services.'The use of TBCs also offers the possibility of increasing turbine

inlet temperatures and, consequently, of improving the thermo-dynamic efficiency of gas turbines used in aircraft and land-based applications.*However, in practice, such benefits havenot been realized yet, largely because the performance of cur-rently available TBCs is considered as not sufficiently reliable

and predictable for use as a "prime reliant" material. In otherwords, current TBC systems are conservatively designed to

avoid situations where the ceramic layer (or a portion of thelayer) may prematurely spa11 off, possibly causing the cata-strophic failure of the underlying metallic components.

Nevertheless, from a materials perspective, TBC technologyis currently considered as a viable near-term solution for thedevelopment of more-efficient aircraft engines and land-basedgas turbine^.^ Some of the major technical issues associatedwith the development of reliable TBCs are (i) processing


2/11

3004 Journal of the American Ceramic Society--

Lee et al. Vol. 79, No. 12

improvements from both performance and economic stand-points, (ii) understanding the failure mechanisms of currentTBCs in simulated and actual turbine environments, and(iii) developing characterization and test methods to measureand understand the time-, temperature-, and environment-dependent characteristics of TBCs for reliable life prediction.However, with the anticipation of the ultimate limitations asso-ciated with conventional TBCs, it also seems necessary toexplore the possibility of incorporating novel materials con-cepts in future coating design. In particular, this study addresses

the concept of functionally graded materials (FGMs) byreviewing prior work and reporting some confirmatory experi-mental results.

11. Conventional Thermal Barrier Coatings( I ) Design Philosophy

Most TBCs are structurally as well as functionally complexand, therefore, should be considered and studied as a materialssystem. For example, a state-of-the-art TBC system used inaircraft engines typically consists of a 125-250 pm thermallyinsulating ceramic layer and a 50-125 pm metallic bond coatlayer between the ceramic layer and the metal component sur-face. Y203-stabilizedZrO, (YSZ) is most commonly used asthe ceramic layer because of its low thermal conductivity andrelatively high coefficient of thermal expansion (CTE) for aceramic (Table I). The role of the metallic bond coat is toprotect the substrate from high-temperature oxidation becausethe transport of oxygen through the YSZ layer, via ionic diffu-sion and/or through microcracks or connected porosity, is rela-tively fast at typical turbine temperatures.

In the early stage of TBC development, ZrO, coatings par-tially stabilized with 6%-8% Y203prepared by air plasmaspray (APS) were found to be suitable for TBC application^.'^It also was observed that the incorporation of a controlledamount of porosity resulted in improved spallation behavior.''This behavior was attributed to the fact that the incorporation ofmicrostructural defects, such as porosity and microcracks, intothe APS coating effectively increased the apparent complianceof the coatings. This approach of using a compliant YSZ layerhas been practiced as an effective means of accommodating thestrains developed between the ceramic coating and metallicsubstrate due to their mismatch in material properties.

In another significant electron-beam phys-ical vapor deposition (EBPVD) was used to produce a coatingmicrostructure that consisted of YSZ columns (several microm-eters in diameter) that were preferentially oriented perpendicu-lar to the substrate surface (Fig. 1). The spallation life of theEBPVD coating was found to be superior to that of air-plasma-sprayed coatings by some investigator^.'.'^ This result wasattributed to the ability of the YSZ columns in the EBPVDcoating to become physically separated or segmented duringthermal cycling and, therefore, to accommodate residual andthermal stresses. This type of microstructure was observed tobe unique to the EBPVD process. Dinwiddie et al.I4 showedthat the thermal conductivity of the as-deposited EBPVD coat-ing was higher than that of the as-sprayed APS coating (-1.5

vs-0.8 W/(m.K) at 1000C) but was much less susceptible todegradation by thermal aging. The EBPVD coating wasreported to be particularly useful for protecting rotating turbinecomponents in commercial aircraft applications.'

Thus, from a practical standpoint, the conventional approach

for accommodating the material mismatch between the ceramic _coating and metallic substrate is to make the ceramic layercompliant by incorporating structural defects, such as micro-cracks and porosity as well as vertically segmented columns.However, in terms of achieving long-term durability, severalmajor problems are associated with the presence of these micro-structural features. As an example, the sintering of the micro-structural defects with increased temperature or prolonged timeexposure becomes an important issue, which may result in anincrease in thermal conductivityI4and a decrease in coatingcompliance. Also, these microstructural features provide rapiddiffusion paths for oxygen and other corrosive species.

(2 ) Failure Modes

Simulated and actual aircraft engine tests have shown thatstate-of-the-art TBCs typically exhibit wide failure distribu-tions and also experience a variety of failure modes rangingfrom particulate-induced erosion, to salt-induced hot corrosion,to high-temperature oxidation, to thermal-mechanical f a t i g ~e . ~However, from a more fundamental point of view, the instabil-ity at the metal

-

ceramic interface can beconsidered as the mostcritical factor that limits the ultimate performance of currentTBCs. This bimaterial interface, inevitably required for bond-ing the ceramic coating and metallic surface, is the major sourceof time-, temperature-, and environment-dependent propertychanges as well as failures.

The YSZ coating prepared by APS tends to fail by the spall-ation of the YSZ layer near the interface between the bond coatand the ceramic layer but mostly within the ceramic layer.10.15.16The time-to-failure depends on many factors, but the two mostdominant parameters are the oxidation of the bond coat and thenumber of thermal cycles. The ultimate failure of the EBPVDcoating also is due to the spallation of the ceramic layer. How-ever, its exact failure location is clearly different from thatobserved for the APS coating. As shown in Fig. 1, the EBPVDcoating fails at the interface between the bond coat surface andits A1203~c a l e . ' ~ . ' ~As in the case of the APS coating, the life ofthe EBPVD coating mainly depends on oxidation and thermalcycle^.'^

In both APS and EBPVD systems, a thin A1203scale formson the surface of the bond coat upon exposure to oxidativeenvironments because of rapid transport of oxygen through theYSZ layer. Because Al,03 is relatively impermeable to oxygen,further oxidation of the bond coat is retarded once a continuousscale is formed. This scale formation, therefore, actually pro-vides protection to the metallic substrate from high-temperatureoxidation. Therefore, the overall performance of a TBC notonly depends on the ceramic layer itself but also is dictated byits bond coat and, more precisely, the nature of the metal

-

oxideinterface that develops upon oxidation. This is particularly truefor the EBPVD YSZ layer that appears to be sufficiently com-pliant so that the coating fails not within the ceramic layer, butat the scale-bond coat interface. Also, for the EBPVD system,

Table I. Selected Properties of Potential Coating Materials*

Material

OxygenLinear Elastic Thermal diffusivityCTE modulus conductivity at 1000C

(X lo-6 K-') (GPa) (W/(m.K)) (m2/s)YSZ (partially stabilized) 8.9-10.6 205 -2 at 1127C 10-litA1203 7.2-8.6 380-434 5.8at1127"C lo-"3A1203.2Si02(mullite) 5.7 145 3.3 at 1127C lo-''TiO, 9.4 283 3.3 at 1127C 10-l7Ca0.5Sr0.5Zr4P6024 3 70 - at 1000"~

-

-

*Measured for dense, bulk materials. Compiled from Refs. 4-9. Apparent properties of these materials, if prepared by APSand EBPVD, are expected to be widely different from the tabulated values. Also, this type of materials assessment always suffersfrom the lack of high-temperature properties and uncertainties in reported values, depending on specimen preparation, impuritylevels, measurement methods, etc. Nevertheless, this type of assessment provides a starting basis. +ForZrO,.


3/11

ecember1996 Concept of FunctionallyGraded Materials for Advanced Thermal Barrier Coating Applications 3005

g. 1. Cross-section of an EBPVD YSZ coating deposited on aperalloy substrate coated with an aluminide bond coat after exposureoxidative environments. Micrograph shows that the EBPVD coatingls by debonding at the interface between the Al,O, scale and thend coat surface. (Micrograph courtesy of Ben Nagaraj, Generalectric Aircraft Engines.)

e effects of substrate materials can be considerable becausethe diffusion of some alloying elements as well as impuritiesthe metal-scale interface. Therefore, to a large extent, thevelopment of more-reliable EBPVDTBCs necessitates a newneration of bond coat materials, which substantially improvee adhesion of the metal-scale interface.

) Scale Formation and Adhesion

Currently, nickel and platinum aluminide coatings preparedgas-phase pack or chemical vapor deposition (CVD)18and

cuum-plasma-sprayed (VPS) MCrAlY (where M = Ni, Co,or NiCo) coatings1 are used as bond coats for most

BC applications. The protection mechanism provided by these

pes of A1203-formingcoatings recently has been describedBirks eta1.19Upon oxidation, a continuous a-A1203scale,nsisting of equiaxed grains, grows on the surface of thend coat after a brief transient growth period. Eventually,columnar grain microstructure develops. At this stage, theansport of oxygen mainly occurs along grain boundariestween the oxide columns. Also, the formation of small voidss large as several micrometers) has been observed underneathe scale at the metal surface as a result of interactive morpho-gical and diffusional mechanisms. The ability of the bondat to continue to support the growth of the A1203 scalepends on the activity and the total amount of aluminumailable in the coating. However, during use, the aluminumservoir of the bond coat is depleted by consumption by the

scale growth and by diffusion into the alloy substrate.hen the aluminum level of the bond coat falls below at whichcannot be formed preferentially, faster growing oxides of

e other constituents of the bond coat will form. Then, thehesion of the TBC can be significantly degraded. Anotherctor is that the continuing growth of the otherwise protective

scale at the bond coat-YSZ interface is a source ofcreasing strain.When a bond coat is used without a ceramic top coat (as ine case of many aircraft and land-based gas turbines operatingday), the A1203scale tends to spa11off during thermal cycling,d, subsequently, a new oxide scale forms on the coatingrface. This cycle of scale formation and spallation continuesthe expense of the sacrificial metallic coating layer, which

n be refurbished or repaired at regular maintenance intervals.owever, when the aluminide or MCrAlY coating is used as and coat for a TBC layer, even local scale spallation cantiate coating failure. Therefore, scale formation, which issential for the oxidation protection of the base metal, is an

mportant issue to consider in TBC design.

From a mechanical point of view, the driving force for scalespallation is the development of residual and thermal stresses atthe interface between the oxide scale and the metal surface. Thescale becomes severely strained with temperature cycling andincreased scale thickness because of the large differences inmaterial properties, such as CTE and Young's modulus. In aqualitative sense, when the level of these stresses exceeds thebond strength of the bimaterial interface, the scale and metalsurfaces are debonded. From the standpoint of fracture mechan-ics, it is important to consider factors that affect crack initiationand propagation along this metal-scale interface. For example,

the presence of small voids at the metal-

scale interface cancontribute to spalling by reducing the resistance of crackpr~pagat ion.~~

Scale adherence can be improved by several methods. First,the addition of small amounts of so-called reactive ele-ments-such as yttrium, hafnium, and zirconium-to super-alloys and coatings can significantly improve scale adhe-

Second, the removal of sulfur impurities to below sub-ppm levels also increases scale adhesion because sulfur tendsto segregate and weaken the bonding between the metal and~ c a l e . ~ ' . ~ ~A low level of sulfur also is observed to reduce thedegree of void formation at the metal-scale interface.19In aconceptual sense, these approaches can be viewed as an avenueto increase the work of adhesion at the metal-scale interface.

The concepts of adding reactive elements or removing sulfurimpurities have been used or are being pursued by modifyingcurrent casting and coating procedures.

111. Functionally GradedThermal Barrier Coatings

The basic thesis behind the FGM concept is that, by appro-priately combining two or more materials, the functionality of aparticular material system can be tailored and extended beyondwhat is possible with nongraded material systems.24For variousapplications, the FGM approach has been explored to mitigatesome of the major problems associated with the developmentof a sharp interface at the join of two dissimilar materials. Forstructural applications, compositionally and/or microstructur-ally graded interfaces, if properly designed and prepared, canbe more resistant to crack initiation and pr~pagat ion.~~Also, inprinciple, the graded interfaces can be useful for lowering peakstresses and eliminating stress singularities at certain crack-sensitive locations because of the gradual changes in materialproperties, such as CTE and Young's modulus, through thegraded interface region.25Although the FGM concept itselfis rather intuitive and simple to comprehend, there are manytechnical challenges associated with measuring and estimatingphysical and mechanical properties of graded materials and,therefore, validating their potentialbenefits.2G2s( I ) Examples of Functionally Graded Thermal BarrierCoatings

In the past, not surprisingly, the FGM concept has beenpursued as a means for managing residual stresses and toreplace the sharp YSZ-MCrAlY interface encountered in theAPS TBC system. However, demonstrating this FGM conceptgenerally has been difficult because the graded regions con-taining YSZ and MCrAlY as ceramic and metallic constituentstend to become unstable upon exposure to oxidative environ-ments. As an example, Fig. 2 shows a typical graded micro-structure containing YSZ and NiCoCrAlY, which was createdusing APS equipment configured with a single plasma torchwith separate YSZ and NiCoCrAlY powder feeders. Asrecently reviewed by Sampath et similar graded micro-

structures have been produced by various plasma spray tech-niques. Some selected properties of the graded materials, suchas hardness and elastic modulus, have been measured.29Also,several studie~~O,~'have been conducted to address the measure-ment and estimation of the physical and mechanical propertiesof this type of graded materials. The graded microstructure


4/11

3006 Journalofthe American Ceramic Society

-


shown in Fig. 2 was prepared on a Mar-M-247 substrate(-2 cm X -2 cm) as an abradable seal. Note that, in theturbine section, abradable seals are used to help minimize thegas leakage through the gap between the tips of the rotatingblades and the turbine Typically, abradable seals aremuch thicker than the TBCs used for vanes or airfoils. Althoughthe temperature requirements for abradable seals are not gener-ally as demanding as those for the coatings used for airfoils,there are other structural criteria, such as abradability, to con-sider. To some extent, abradable seal applications have beentargeted as a test vehicle to validate the usefulness of the FGM

coating concept because of the relatively lower temperaturerequirements as well as the need for a greater thickness (i.e.,-4 mm) that allows more gradual compositional grading pro-files. Figure 3 shows the compositional change of the NiCo-CrAlY and YSZ phases measured as a function of coatingthickness, using a digital image analyzer interfaced with anoptical microscope. The seal consists of a 0.13 mm NiCoCrAlYlayer, a 2.54 rnrn region graded with NiCoCrAlY and YSZ, anda 1.27 mm pure-YSZ layer.

(2) Failure MechanismofGradedMCrAN + YSZ TBCsThe stability of the graded seal shown in Fig. 2 was tested at

1000C in air in a resistance-heated tube furnace. The tempera-ture of the seal was measured using an optical pyrometer. Theseal spalled within the graded region after an isothermal treat-ment of 100 h. The lateral dimension of the spalled piece wasincreased relative to that of the remaining substrate, which wasattributed to the volume expansion caused by the oxidation ofthe metallic phase. A similar spallation behavior was observedafter 242 h of oxidation at 1000C. Figure 4 shows that spall-ation occurred within the graded region -0.5 mm, measuredfrom the initial substrate surface. Metallographic and X-raydiffraction (XRD) analyses indicated that most of the metallicphase in the spalled region was oxidized. However, the seal

structure remaining on the substrate was not extensively oxi-dized although some localized penetrations were evident. Also,preferential oxidation of the metallic phase occurred near thesubstrate interface, where it intersected with the free edges ofthe seal specimen. Because of the crude nature of this oxidationexperiment, it was not possible to resolve the relative contribu-tions to the observed spallation behavior from the oxidation of-the graded region versus that of the free edges. However, in thecontext of the present study, the critical issue to be emphasizedwas the oxidative instability of the graded microstructure of this

\particular coating design, which contained YSZ and NiCo-CrAlY as composite constituents.

Similar observations were reported previously by otherinvestigators. The work of Eaton and Novak4 showed thatgraded YSZ + CoCrAlY coating microstructures were not sta-ble in high-temperature oxidizing environments. In that study,a set of free-standing metal-ceramic composite samples wasprepared by APS for oxidation testing. CoCrAlY was used asthe metallic constituent (about 12 ~01%) .Upon oxidation at1093"C, it was observed that the size and weight of the YSZ +CoCrAlY composite samples significantly increased as a resultof the oxidation of the metallic phase in the YSZ matrix. Tiwariet ~ 1 . ~ ~also observed similar behavior with free-standing APScomposites containing NiCrAl and YSZ. In comparison to pureNiCrAl free-standing APS samples, the weight gain of theNiCrAl + YSZ composite samples after oxidation at 600" and800C was-3 times higher. The increased oxidation observedfor the metal + ceramic composite samples was attributed toseveral reasons by Tiwari et al. The presence of numerousmetal-ceramic interfaces in the composite samples providedshort-circuit paths for oxygen diffusion. Also, the rapid diffu-sivity of oxygen through YSZ, with an increased metallic sur-face area available through the dispersion in the YSZ matrix,increased the oxidation of the metallic phase. For similar rea-sons, MilleP4 recommended that plasma spray coatings with an

Graded Region

Fig. 2. Typical APS-FGM microstructure containing NiCoCrAlY and YSZ as metallic and ceramic constituents, respectively. This FGMmicrostructure was prepared on a Mar-M-247 superalloy substrate for use as an abradable seal. Lighter area is the metallic phase, and the darker areais the ceramic phase. Graded microstructure was created using an APS apparatus configured with a single plasma torch with separate YSZ andNiCoCrAlY powder feeders.


5/11

December1996 Concept of Functionally Graded Materials for Advanced Thermal Barrier Coating Applications

Substrate

Seal Thickness (mm)

Sea lSurface

Fig. 3. Compositional change measured as a function of seal thickness using a digital image analyzer interfaced with an optical microscope. Forhe analysis, 52 digital scans were made at 75 pm thickness intervals along the polished cross section of the APS-FGM seal structure. Lamellar

microstructural features developed during spraying (see Fig. 2) were responsible for relatively large fluctuations in the composition profiles observedn the graded region. Level of splat boundaries, which were created between molten particles during spraying, was estimated to be -8.42% + 1.5%n the outer YSZ layer without considering any particle pullouts during polishing. Compositional contribution from the splat boundaries was addedarbitrarily to the YSZ phase. In the graded region, the presence of the splat boundaries was not quantified because the splat boundaries and ceramic

phase could not be distinguished clearly during the image analysis.

nterface region graded with MCrAlY and YSZ are inappropri-ate because of oxidation of the graded region at temperatures

Despite the oxidation problems associated with using YSZas a ceramic constituent, most FGM TBC work reported in theiterature has been performed on the APS YSZ + MCrAlY

material system.354'Successful oxidation test results generallyhave not been reported for this material system except for thework of Mendelson etaL41They have reported that the cyclicife of a dual-graded TBC containing a graded YSZ-MCrAlYnterface and a laser-glazed top surface was longer than that ofonventional two-layer TBCs. Also, for the lower-temperatureituations (


6/11

3008 Journal of the American Ceramic Society

-

Lee et al. Vol.79,No. 12

same time, the ceramic coating must be chemically andmechanically compatible with its bond coat and substrate aswell as with turbine atmospheres. Meanwhile, the concept ofgrading can be used to manage stresses and to increase thetoughness of its metal-ceramic interfacial region.

In a practical sense, it appears that no single ceramic materialis likely to meet this complex set of criteria. Even if such amaterial exists, another important issue is whether strain-tolerant ceramic microstructures will continue to be needed atthe expense of oxidation performance. To illustrate this point, anew class of ceramic materials, commonly referred to as NZPs,

can be considered as a possible alternative materialto YSZ. NZPs are named after their parent composition(NaZr2P3012).43.44AS compared in Table I, the thermal conduc-tivity of these materials (-1 W/(m.K) for slip-cast Ca,,Sr,.,-Zr4P6O2,5)is essentially lower than that reported for dense, fullyor partially stabilized YSZ ceramics (- 2 W/(m-K)6).Young'smodulus of the NZP materials also is rather low for a ceramic(-70 GPa),44which is another desired characteristic relating tocoating compliance. The melting point of NZPs is typically>1800C. Because of their crystal structure, oxygen diffusivityin NZPs (both ionic and electronic) is expected to be relativelylow, although it has not been measured a ~ c u r a t e l y . ~ ~Further-more, one of the NZP compositions, Cao.,Sro,Zr4P6024,hasbeen found to be resistant to corrosion induced by Na2S04.46However, the CTE of NZPs is very low (see Table I), particu-larly in comparison to nickel-based superalloys. Also, theirstability in turbine environments (e.g., possible dissociationinto gaseous species containing phosphorus) and compatibilitywith respect to the nickel-based alloys are highly questionable.Recently, Cao,Sro,Zr4P,024 was air plasma sprayed onto anickel-based superalloy with a MCrAlY bond coat.47Prelimi-nary isothermal oxidation test results suggest that this particularNZP material reacts substantially with the bond coat at 1000Cin air after 100 h. This observation suggests that, although theCao.5Sro,,Zr4P6024material has some desirable characteristics,its use as a direct substitute for YSZ is unlikely.

(4) Rationale for Multilayered Functionally GradedThermal Barrier Coatings

Exploring alternative thermal barrier materials is expected tobe a difficult, time-consuming, and uncertain task because ofthe complex materials criteria imposed on candidate materials.Probably a more pragmatic approach to this challenge is toincorporate additional layers in the coating structure to satisfythe multiple functionality required for more oxidation resistant,yet mechanically durable TBCs, as schematically illustrated in

Fig. 5. One plausible example of this multilayer approach is toincorporate an oxygen barrier layer that is sandwiched betweenthe thermal barrier layer and metallic surface. The role of thisinterlayer, which can be functionally graded, is to retard oxygentransport to the metal surface as well as to avoid the formationof a sharp metal-ceramic interface.

(5) Oxygen Barrier Materials and Required Morphology

Table I lists several ceramic materials for consideration as anoxygen barrier material. Zr02-based materials (i.e., stabilizedwith CeO,, MgO, SrO, etc.) are not included in Table I becausemost of them are good electrolytes and, thus, oxygen conduc-tors. Table I shows that Al,O, and mullite exhibit low oxygendiffusivity at 1000C, but their thermal conductivity is higherthan that of YSZ. Other properties, such as CTE and elasticmodulus, also should be considered because these propertiesdetermine the level of residual and thermal stresses in the coat-ing and substrate. As a qualitative approximation, the in-planeresidual stress of the coating (a)due to temperature changes orgradients is expected to be proportional to the magnitude of theCTE mismatch between the substrate and coating materials( A a ) and inversely proportional to the compliance of the coat-ing (C); i.e., a A a/ C . The compliance of a dense coatingdepends on its intrinsic elastic modulus as well as microstruc-ture. From this simple approximation, it can be argued that

a dense mullite coating deposited on a superalloy substrategenerates less residual stress than a dense A1203 coatingbecause the low modulus of mullite compensates for its lowCTE value. However, mullite's oxygen diffusivity is expectedto be higher than that ofA1203.Eaton and Novak observed that A1203,mullite, and MgA1204composites mixed with a metallic phase (CoCrAlY) exhibitedsuperior oxidative stability to YSZ + CoCrAlY compositespecimens. The observed oxidation behavior was explainedmainly by the relatively low rate of oxygen diffusion throughthese oxides. The use of A1203as a ceramic constituent forpreparing a graded abradable seal structure recently wasreported by DeMasi-Marcin and Gupta.' In this application, anabradable seal was graded in multiple steps to provide smooth

transitions going from a metallic bond coat, to A1203,to fullystabilized YSZ, to partially stabilized YSZ. Recently, Rumaneret aL4' reported that an APS coating graded with MCrAlY andmullite (or A1203)was much more crack and oxidation resistantthan conventional dual-layered MCrAlY/YSZ TBCs, but at theexpense of increased thermal conductivity. Therefore, from astandpoint of matching CTE, providing thermal resistance, andobtaining strain-tolerant microstructures, A1203or mullite is

Thermallv insulating ceramic-Low thermal conductivity-Compliance=Compatible with oxygen barrier material-Resistant to erosion and corrosion

FGM concept08tressmanagementlInterfacial toughness

en transport barrier%ow oxygen conductivity=Compatible with bond coat and

thermal barrier materials..Dense morphologyDiffusion barrierI_Substrate Diffusion barrierlInhibit Al and subtrate elements diffusion

Fig. 5. Materials criteria for designing a functionally graded, multilayer coating system for thermal insulation as well as oxidation resistance.


7/11

3008 Journal of the American Ceramic Society-

Lee et al. Vol.79,No. 12

same time, the ceramic coating must be chemically andmechanically compatible with its bond coat and substrate aswell as with turbine atmospheres. Meanwhile, the concept ofgrading can be used to manage stresses and to increase thetoughness of its metal-ceramic interfacial region.

In a practical sense, it appears that no single ceramic materialis likely to meet this complex set of criteria. Even if such amaterial exists, another important issue is whether strain-tolerant ceramic microstructures will continue to be needed atthe expense of oxidation performance. To illustrate this point, anew class of ceramic materials, commonly referred to as NZPs,

can be considered as a possible alternative materialto YSZ. NZPs are named after their parent composition(NaZr2P30,2).43.""AS compared in Table I, the thermal conduc-tivity of these materials (-1 W/(m.K) for slip-cast Cao,Sro,-Zr4P60,;) is essentially lower than that reported for dense, fullyor partially stabilized YSZ ceramics (-2 W/(m-K)6).Young'smodulus of the NZP materials also is rather low for a ceramic(-70 GPa),"" which is another desired characteristic relating tocoating compliance. The melting point of NZPs is typically>1800C. Because of their crystal structure, oxygen diffusivityin NZPs (both ionic and electronic) is expected to be relativelylow, although it has not been measured a ~ c u r a t e l y . ~ ~Further-more, one of the NZP compositions, Ca,, SrO,,Zr4P6O2,,hasbeen found to be resistant to corrosion induced by Na2S04.46However, the CTE of NZPs is very low (see Table I), particu-larly in comparison to nickel-based superalloys. Also, theirstability in turbine environments (e.g., possible dissociationinto gaseous species containing phosphorus) and compatibilitywith respect to the nickel-based alloys are highly questionable.Recently, Cao,5Sro.5Zr,P6024was air plasma sprayed onto anickel-based superalloy with a MCrAlY bond coat.47Prelimi-nary isothermal oxidation test results suggest that this particularNZP material reacts substantially with the bond coat at 1000Cin air after 100 h. This observation suggests that, although theCao,,Sro~5Zr4P602,material has some desirable characteristics,its use as a direct substitute for YSZ is unlikely.

(4) Rationale for MultihyeredFunctionally GradedThermal Barrier Coatings

Exploring alternative thermal barrier materials is expected tobe a difficult, time-consuming, and uncertain task because ofthe complex materials criteria imposed on candidate materials.Probably a more pragmatic approach to this challenge is toincorporate additional layers in the coating structure to satisfythe multiple functionality required for more oxidation resistant,yet mechanically durable TBCs, as schematically illustrated in

Fig. 5. One plausible example of this multilayer approach is toincorporate an oxygen barrier layer that is sandwiched betweenthe thermal barrier layer and metallic surface. The role of thisinterlayer, which can be functionally graded, is to retard oxygentransport to the metal surface as well as to avoid the formationof a sharp metal-ceramic interface.

(5) Oxygen Barrier Materials and Required Morphology

Table I lists several ceramic materials for consideration as anoxygen barrier material. ZrO,-based materials (i.e., stabilizedwith CeO,, MgO, SrO, etc.) are not included in Table I becausemost of them are good electrolytes and, thus, oxygen conduc-tors. Table I shows that A1203and mullite exhibit low oxygendiffusivity at 1000"C, but their thermal conductivity is higherthan that of YSZ. Other properties, such as CTE and elasticmodulus, also should be considered because these propertiesdetermine the level of residual and thermal stresses in the coat-ing and substrate. As a qualitative approximation, the in-planeresidual stress of the coating (u) due to temperature changes orgradients is expected to be proportional to the magnitude of theCTE mismatch between the substrate and coating materials( A a ) and inversely proportional to the compliance of the coat-ing (C); i.e., a A a l C . The compliance of a dense coatingdepends on its intrinsic elastic modulus as well as microstruc-ture. From this simple approximation, it can be argued that

a dense mullite coating deposited on a superalloy substrategenerates less residual stress than a dense A1203 coatingbecause the low modulus of mullite compensates for its lowCTE value. However, mullite's oxygen diffusivity is expectedto be higher than that ofA1203.Eaton and Novak observed that A1203,mullite, and MgA1204composites mixed with a metallic phase (CoCrAlY) exhibitedsuperior oxidative stability to YSZ + CoCrAlY compositespecimens. The observed oxidation behavior was explainedmainly by the relatively low rate of oxygen diffusion throughthese oxides. The use of A120, as a ceramic constituent forpreparing a graded abradable seal structure recently wasreported by DeMasi-Marcin and Gupta.' In this application, anabradable seal was graded in multiple steps to provide smooth

transitions going from a metallic bond coat, to A1203,to fullystabilized YSZ, to partially stabilized YSZ. Recently, Rumaneret a/.,*reported that an APS coating graded with MCrAlY andmullite (or Al,O,) was much more crack and oxidation resistantthan conventional dual-layered MCrAlYIYSZ TBCs, but at theexpense of increased thermal conductivity. Therefore, from astandpoint of matching CTE, providing thermal resistance, andobtaining strain-tolerant microstructures, A1203or mullite is

Thermallv insulatina ceramic-Low thermal conductivity-Compliance-Compatible with oxygen barrier material

mResistant to erosion and corrosion

i s r g l l l ~ I FGM conce~t- -Stress mananement-Interfacial toughness

Metallic bond coat IDiiYusionbarrier\

kOxvgen transport barrier-Low oxygen conductivity-Compatible with bond coat and

thermal barrier materials-Dense morphology

Substrate Diffusion barrierInhibit Aland subtrate elements diffusion

Fig. 5. Materials criteria for designing a functionally graded, multilayer coating system for thermal insulation as well as oxidation resistance.


8/11


9/11



3'T. Ishizuka, S. Okada, and K. Wakashima, ''In Situ Measurement ofThrough-the-Thickness Thermal Conductivities in Ceramic/Metal Composition-Graded Multilayers by Means of Thermography"; see Ref. 27, p. 453.32A.P. Matarese and G. S. Bosshart, "Method of Applying ContinuouslyGraded Metallic-Ceramic Layer on Metallic Substrates," U.S. Pat. No.4,588,607, 1986.33R.Tiwari,C. Perdikaris, C. C. Berndt, and H. Herman,"Oxidation ofNickel-Based Bond Coat Materials"; p. 11-25 in Proceedings of the 1992 Coatings for

Advanced Heat Engines Workshop (Monterey, CA, Aug. 1992). U.S. Departmentof Energy, Washington, DC.34R.A. Miller,"Assessment of Fundamental Materials Needs for Thick Ther-mal Barrier Coatings (TTBCs) for Truck Diesel Engines"; p. 11-7 inProceedingsofthe 1990Coatings for Advanced Heat Engines Workshop (Castine, ME, Aug.

1990). U.S. Department of Energy, Washington, DC.35Y.Shinohara, Y. Imai, and S. Ikeno, "Thermal Stability of Plasma-SprayedNi-Cr-Al-Y/PSZ FGM in Uniform and Gradient Temperature Fields"; seeRef. 27, p. 255.36A.Kumakawa, M. Niino, S. Kiyoto, and S. Nagata, "Thermal Fatigue ofFunctionally Graded Materials under High Heat Fluxes"; see Ref. 26, p. 213.37K.Kuroda, K. Kusaka, A. Moro, and M. Togawa, "Evaluation Tests forZr02-NiSystem"; see Ref. 26, p. 289."T.W. Clyne and Y. C. Tsui, "The Effect of Intermediate Layers on Residual

Stress Distributions and Debonding of Sprayed Thermal Barrier Coatings"; seeRef. 27, p. 129.39M.Alaya, G. Grathwohl, and J. Musil,"A Comparison of Thermal Cyclingand Oxidation Behavior of Graded and Duplex Zr02-ThermalBarrier Coatings";see Ref. 27, p. 405.T.Y. Jian, T. Hashida, H. Takahashi, N. Shimoda, and M. Saito, "An

Accelerated Testing Method of Zr02-Based FGM Coating for Gas TurbineBlades"; see Ref. 27, p. 419.

41M.I. Mendelson, T. N. McKechnie, and L. B. Spiegel, "Graded ThermalBarrier Coatings-Evaluation,"Ceram. Eng. Sci. Proc.,15,555 (1994).42M.B. Beardsley,"Application of Thick Thermal Barrier Coatings to DieselEngines"; see Ref. 34, p. 11-53.43C.-Y.Huang, D. K. Agrawal,H. A. McKinstry, and S. Y. Limaye,"Synthesisand Thermal Expansion Behavior of Bal+xZr4P6-,Si,024 and SrI+,Zr4-P6-,Si,02, Systems," J.Mater.Res., 9,2005 (1994)."T. B. Jackson, S. Y. Limaye, and W. D. Porter, "The Effects of Thermal +

Cycling on the Physical and Mechanical Properties of [NZP] Ceramics";pp. 63-80 in Ceramics Transactions, Vol. 52,Low-Expansion Materials. Editedby D. P. Stinton and S. Y. Limaye. American Ceramic Society, Westerville,OH, 1995."W. Y. Lee, D. P. Stinton, and D. L. Joslin, "Interaction of Low-ExpansionNZP Ceramics with Na,SO, at 1000C.'' J.Am. Ceram. Soc.,79,484 (1996).'%. Limaye, LoTEC, Salt Lake City, UT, and D. Agrawal, Pennsylvania StateUniversity, University Park, PA, 1995; private communications.47W.Y. Lee, K. M. Cooley, C. C. Bemdt, D. L. Joslin, and D. P. Stinton,"High-Temeprature Chemical Stability of Plasma-Sprayed Ca,5Sro,Zr4P60,Coatings on NicalonJSiC Ceramic Matrix Composite and Ni-Based SuperalloySubstrates,"J.Am. Ceram. Soc., 79 [lo] 2759-62 (1996).48L.E. Rumaner, D. M. Gray, R. L. Mehan, and J. R. Rairden, "ThermallyProtective Composite Ceramic-Metal Coatings for High-Temperature Use,"U.S.Pat. No. 5,281,487,1994.49T.E. Strangman, "Ceramic Thermal Barrier Coating with Alumina Inter-layer,"U.S. Pat. No. 4,880,614, 1989.50J.H. Sun, E. Chang, C. H. Chao, and M. J. Cheng,"The Spalling Modes andDegradation Mechanism ofZr02-8-wt%-Y203/CVD A1203/Ni-22Cr-1OAl-1YThermal Barrier Coatings,"Oxid.Met., 40,465 (1993).5'W. Y Lee, Y. W. Bae, and K. L. More,"Synthesis of Functionally GradedMetal-Ceramic Microstructures by Chemical Vapor Deposition,"J.Mater.Res.,10,3000 (1995). 0

Woo Young Lee is a Research Staff Member in the Metals and Ceramics Division ofOak Ridge National Laboratory. He received his B.S. at the University of Maryland(1985), M.S. at Auburn University (1987), and Ph.D. at the Georgia Institute ofTechnology (1990). Prior to joining ORNL, he worked at United TechnologiesResearch Center as an Associate Research Scientist from 1990 to 1992. His researchinterests include novel materials synthesis by chemical vapor deposition, corrosion-resistant coatings for high-temperature applications, and interfacial behavior betweendissimilar materials.

David Paul Stinton is a Lead Engineer in the Metals and Ceramics Division at the OakRidge National Laboratory. He graduated from Virginia Polytechnic Institute withB.S. and M.S. degrees in ceramic engineering in 1973 and 1974. Since joining ORNLin 1974, he has authored more than 100 technical papers and received eight patentsand three R&D 100 Awards. His research interests include the utilization of chemicalvapor deposition for the fabrication of composite materials and specialized coatings.


10/11

December1996 Concept of Functionally Graded Materials for Advanced Thermal Barrier Coating Applications 3011

Christopher C. Berndt received a B.App.Sci. in secondary metallurgy from the Uni-ersity of South Australia in 1978. He received his Ph.D. in materials engineering in981 from Monash University, Melbourne, Australia. He has held fellowships at State

University of New York at Stony Brook and NASA-Lewis Research Center. Berndtas pursued a career in academic R&D in Australia and, for the past seven years, inhe United States, currently as Professor in the Materials Science and Engineering

Department of SUNY at Stony Brook. He is a Faculty Fellow of Oak Ridge NationalLaboratory and Brookhaven National Laboratory. Bemdt has numerous publicationsn the thermal spray, powder preparation, and deposit characterization fields, and has

research emphasis on the mechanical properties of coatings.

Fazil Erdogan is the G . Whitney Snyder Professor of Mechanical Engineering andMechanics at Lehigh University. He received his Ph.D. in mechanical engineeringrom Lehigh University, joined the Lehigh faculty in 1957, and has been teaching andonducting research in the general area of fracture mechanics since then.

Yi-Der Lee received his M.S. and Ph.D.of applied mechanics from Lehigh University,where he also conducted three years of post-doctoral work on the fracture behavior ofhermal barrier coatings and abradable seals with functionally gradient materials usinghe finite-element technique. Presently, he is working as research engineer in the

Materials Engineering Department of Southwest Research Institute, San Antonio,Texas.


11/11



Zaher Mutasim is a Senior Principal Research Engineer and a Group Leader for theSurface Engineering Group in the Department of Materials and Processes at SolarTurbines Incorporated. He has eight-years experience in the research, development,and application of cold and hot section protective coatings and processes to industrialgas turbines. Zaher has a B.S. in chemistry from the American University of Beirut,Lebanon, and M.S. and Ph.D. degrees in chemical engineering from the University ofWales, Swansea, U.K. Mutasim has responsibility for development of aqueouscorrosion-resistant coatings, abradable and rub-tolerant seal coatings, wear-resistantcoatings, thermal barrier coatings, and oxidation/hotcorrosion-resistant coatings. Hehas more than 25 technical publications in the area of coatings and three U.S. patents.Before joining Solar Turbines, Mutasim was a research associate in the Center for thePlasma Processing of Materials at Drexel University, from 1988 to 1991.

Concept of Fgm

Documents

Transcript of Concept of Fgm