Thermal Barrier Coatings Design with Increased ... - Wiley Online Library

Int. J. Appl. Ceram. Technol., 3 [2] 81–93 (2006)

Ceramic Product Development and Commercialization

Thermal Barrier Coatings Design with Increased Reflectivity and Lower Thermal Conductivity for High-Temperature Turbine Applications Matthew J. Kelly,* Douglas E. Wolfe, and Jogender Singh Applied Research Laboratory, Penn State University, University Park, Pennsylvania 16801

Jeff Eldridge, Dong-Ming Zhu, and Robert Miller NASA-GRC, Cleveland, Ohio 44135

High reflectance thermal barrier coatings consisting of 7% Yittria-Stabilized Zirconia (7YSZ) and Al2O3 were deposited by co-evaporation using electron beam physical vapor deposition (EB-PVD). Multilayer 7YSZ and Al2O3 coatings with fixed layer spacing showed a 73% infrared reflectance maxima at 1.85 mm wavelength. The variable 7YSZ and Al2O3 multilayer coatings showed an increase in reflection spectrum from 1 to 2.75 mm. Preliminary results suggest that coating reflectance can be tailored to achieve increased reflectance over a desired wavelength range by controlling the thickness of the individual layers. In addition, microstructural enhancements were also used to produce low thermal conductive and high hemispherical reflective thermal barrier coatings (TBCs) in which the coating flux was periodically interrupted creating modulated strain fields within the TBC. TBC showed no macrostructural differences in the grain size or faceted surface morphology at low magnification as compared with standard TBC. The residual stress state was determined to be compressive in all of the TBC samples, and was found to decrease with increasing number of modulations. The average thermal conductivity was shown to decrease approximately 30% from 1.8 to 1.2 W/m-K for the 20-layer monolithic TBC after 2 h of testing at 13161C. Monolithic modulated TBC also resulted in a 28% increase in the hemispherical reflectance, and increased with increasing total number of modulations.

Introduction A critical goal of both the Department of Defense and Department of Energy is doubling the thrust-

This research was sponsored by the United States Navy Manufacturing Technology (ManTech) Program, Office of Naval Research, under Navy Contract N00024-02-D-6604. [email protected] r 2006 The American Ceramic Society

to-weight ratio for high-operating-temperature turbine systems with the use of lightweight turbine structures. Therefore, there is a need to develop lightweight smart structures capable of operating at 30001F. Coatings that are self-indicating of damage with tailored microstructure and chemistry may meet these challenges. The approach has to be simple, cost-effective, and compatible with current technology. Specific selection of materials will depend upon application and operating environment.

82

International Journal of Applied Ceramic Technology—Kelly, et al.

Thermal barrier coating (TBC) composed of partially stabilized zirconia ZrO2–7wt% Y2O3, (7YSZ) has been used in the turbine industry for over 50 years and has played significant role in extending the life of components. Initially, TBC was applied by plasma spray process. However, for more than a decade, the preference is in applying TBC by electron beam-physical vapor deposition (EB-PVD) as it offers superior thermo-mechanical properties and lifetime performance over plasma spray process.1 Recently, research efforts have been directed in reducing the thermal conductivity of TBC without sacrificing its high-temperature thermal stability and mechanical properties needed for turbine industry. Thermal conductivity of ceramic materials is both a phonon and photon phenomena dependent on various factors including material structure, operating environment, and temperature. Scattering of anharmonic elastic waves, either by inelastic phonon–phonon or phonon– lattice interaction, and controls phonon conductivity. Phonon–phonon scattering or ‘‘Umklapp scattering’’ occurs when two or more anharmonic elastic waves interact, which give rise to a finite thermal conductivity contribution.2 The number of phonon–phonon interactions are determined by the mean free path, and for next generation TBC operating at 30001F the mean free path is approximated to be the lattice spacing, which leads to a fixed phonon thermal conductivity contribution for a given material system.3 Lattice defects such as voids or dopants cause phonon–lattice interactions that reduce the mean free path and frequencies of oscillation permitted in the lattice thereby reducing phonon thermal transport. Photon contributions to total heat transfer are often disregarded in calculations for thermal conductivity, because of the small effect at low temperatures. However, this effect can become quite large at elevated temperatures because of the fourth order dependency on temperature. Obeying Wien’s Law, as the body increases in temperature, the emission spectrum becomes more pronounced and the peak of emission intensity shifts toward a smaller wavelength. For next generation turbines, coating material optical properties will contribute toward high temperature thermal conductivity. In particular, understanding the percentage of photons a material transmits, reflects, and absorbs for a given temperature is critical (Fig. 1). 7YSZ is 80% transparent (or transmits 80% of incoming photons) below 5 mm wavelength and nearly 100% opaque above 8 mm wavelength (or the material absorbs nearly all radiation),

Vol. 3, No. 2, 2006

Fig. 1. Seven percent Yittria-Stabilized Zirconia Electromagnetic Transmission Spectrum.

and it is translucent in the range of 5–8 mm wavelength. Photon absorption occurs at interfaces either at the material boundaries or at atomic sites in materials that permit photon transport. Absorption causes a finite increase in energy, which then dissipates by atomic oscillations (phonon transport) and photon radiation at longer wavelength. Partially stabilized zirconia has historically offered the best balance of mechanical and thermal properties needed for turbine application. Unfortunately, it is nearly transparent over a wide range of wavelengths. Because of this translucent property, in high-temperature environments significant heat can be transported through the 7YSZ to the base material in spite of its low thermal conductivity. The total amount of thermal energy transferred through TBC dictates oxidation rate and creep properties of the base material. By reducing the energy transportation rate, components may operate in higher temperature environments with the same lifetime or longer. Significant progress has been made in developing low thermal conductive TBC with the same basic composition (ZrO2–7 wt% Y2O3) having rare earth oxide dopants. It has been reported that multi-component oxide dopants enhance the thermal stability and reduces thermal conductivity by 50%.4,5 The reduction in thermal conductivity is because of presence of 5–100 nm defect clusters distributed through out the coating matrix, which limit the mean free path of both phonons and photons. New generation, low conductive thermal barrier coatings are based on: ZrO2 Y2 O3 Nd2 O3 ðGd2 O3 ; Sm2 O3 Þ Yb2 O3 ðSc2 O3 Þ Primary stabilizer Oxide dopant cluster

www.ceramics.org/ACT

TBC Design for High-Temperature Turbine Applications

Techniques for reducing the high temperature thermal conductivity focus on the rejection of incoming photons. Changing the optical properties of a material, often with a coating, will alter photon transport into the material. A common method of changing the transmission of electromagnetic waves is application of reflective coatings. Multiple interfaces throughout the thickness of the TBC cause incoming photons to be reflected. There are two approaches in creating multiple interfaces in the TBC. One approach is forming short-range alternate layers of high and low-density structures in monolithic materials (Fig. 2a), i.e., modulated microstructure with periodic density variation that will alter its refractive index properties. The second approach is creating a multilayer structure with two different materials (Fig. 2b) having different refractive indexes. Single material coatings with periodic density variations have an advantage over inhomogeneous reflective coatings such that modulated density structures are not sensitive to the incidence angle of radiation. Modulated density variation in thermal barrier coatings was demonstrated in 7YSZ and Hf-26 wt% Y2O3 by periodically interrupting the incoming vapor flux during the deposition process. Hemispherical reflectance of 7YSZ and Hf-26 wt% Y2O3 was increased from B30% to 45% (20 equal periodic vapor flux interruptions) and B55% to 65% (with 40 equal periodic vapor flux interruption), respectively.6 Significant benefit of this methodology is that there is no change in the coating chemistry, columnar morphology or crystallographic structure. The additional benefit is exhibiting lower thermal conductivity and better strain tolerance in comparison with standard 7YSZ coating. Monolithic (7YSZ) graded structures, as well as, multilayer coating materials with different refractive indexes and densities (7YSZ and Al2O3) will be discussed.

(a)

I

R

(b) I

R

High index Low index High index Low index High index T

T

Fig. 2. Two methods for reducing the photon transport through thermal Barrier Coatings. (a) single layer structure with modulated density, (b) multi-layer, multi-material structure.

83

Experimental Procedure All TBC systems were deposited in an industrial prototype Sciaky Inc. (Chicago, IL) EB-PVD unit consisting of six EB-guns (1–6) and a three-ingot continuous feeding system (A–C) as shown in Fig. 3. Before applying the TBC, 1 in. diameter platinum– nickel–aluminide plated Rene N5 discs were heat tinted for 30 min in air at 7041C and cooled to room temperature. The surface and the side of each test specimen were grit blasted in a Unihone grit blaster using high-purity 400 mm size aluminum oxide particles. The distance from the edge of the nozzle to the surface of the samples was approximately 15 cm with a pressure of 30 psi. The angle of the nozzle with respect to the sample surface was 451 in order to minimize the amount of embedded Al2O3 particles incorporated into the bond-coated surfaces. The grit blast time on each sample varied between 10–15 s and was performed until a uniform matte finish was obtained. The samples were then tack welded to strips of 304 stainless steel foil that were tack welded to a 5.08 cm diameter mandrel. The sample holder with the mounted samples was then ultrasonically cleaned for 20 min in acetone, rinsed with de-ionized water, ultrasonically cleaned for 20 min in methanol, and then dried with nitrogen gas. The sample holder was then mounted in the evaporation unit with a source to substrate distance of 30.48 cm. The vacuum unit was evacuated to a base pressure of 7.5 106 Torr with the oxygen gas lines being evacuated. The samples were positioned directly over a 4.928 cm diameter 7YSZ ingot (TransTech Inc., Adamstown, MD) rotating at 7 rpm. Using two electron beams (#2 and #5 in Fig. 3a), the samples were indirectly heated to 10001C under a graphite ‘‘A-frame’’ heater assembly. After a minimum of 20 min at temperature, the samples were allowed to soak at 10001C for an additional 20 min while injecting approximately 150 sccm of oxygen (PchamberB0.81 103 Torr) into the chamber in order to form a thermally grown oxide (TGO). Immediately following the formation of the TGO, 7YSZ was evaporated (using electron beam #3 and TBC ingot position B in Fig. 3a) at an ingot feed rate of 0.8 mm/min. Again, B150 sccm of oxygen was injected into the deposition chamber (Pchamber 5 1 103 Torr) while simultaneously evaporating the 7YSZ ingot to maintain the oxygen stoichiometry of the condensing coating.


84

Vol. 3, No. 2, 2006

Fig. 3. (a) Sciaky Inc. electron beam physical vapor deposition (EB-PVD) unit with 6 EB-guns (1–6) and a 3-ingot continuous feeding system (A–C), (b) EB-PVD unit with multi-layer tooling configuration utilizing offset ingot (I) and vapor barrier (II).

oxygen loss and allowed to cool for 10 min before venting to atmosphere. In order to determine whether additional strain fields were incorporated into the TBC by the ‘‘shutter’’ method, the amount of deflection was measured for the various coatings deposited on PtAl-coated Haynes 188 alloys strips using a computer controlled machine (CMM). A special tooling holder was fabricated to ensure precise placement of the sample in the precoated and postcoated condition to ensure that the amount of deflection was measured accurately. Once the sample was positioned correctly, a probe traversed the surface of the strip using a six axes position scheme. The change in the vertical distance (z-position) was plotted as a function of the x–y coordinates and the difference from the original z-position was used for the deflection value.

Monolithic 7YSZ Coatings with Modulated Density

In order to introduce stable, microstructural features that would change the density, the condensing vapor flux was periodically interrupted using a graphite shutter. This periodic ‘‘shuttering’’ prevents the flux from depositing on the sample surface with little to no reduction (20–301C) in substrate temperature. This method resulted in smaller strain fields consisting of one or more of the following: more diffuse interfaces, microporosity, vacancies, and smaller lattice strains, i.e., density/refractive index variation. In order to achieve the various number of diffuse layers within the TBC, the shutter was closed for 30 s for each layer (while continuously evaporating the 7YSZ ingot) to prevent the coating from condensing on the surface of the substrates which were maintained near 10001C. Table I lists the primary deposition processing parameters for creating the microstructural changes for the 7YSZ layered coatings produced by the ‘‘shutter’’ methods. For both methods, at the end of the desired deposition time, the samples were retracted into the load lock chamber which was injected with oxygen gas to prevent Table I.

Sample # A B C D

7YSZ-A2O3 Highly Thermal Reflective TBC

Similarly, high reflective thermal barrier coatings were deposited in the industrial prototype EB-PVD unit with a modified experimental setup design as shown in

Deposition Parameters for Layered TBC Produced by ‘‘Shutter’’ Method

Total # of layers

Evaporation time (min)

Additional evaporation time for layered structure (min)

1 5 10 20

73 75 77.5 82.5

0 2.0 4.5 9.5

TBC, thermal barrier coatings.

Amount of time shutter remained open per layer (s)

Average coating thickness (lm)

4380 876 438 219

122.3 119.7 127.9 132.8



Fig. 3b. Two sets of coatings were produced with (i) constant individual layer thickness of the 7YSZ (400 nm) and Al2O3 (100 nm) throughout the coating and (ii) with alternating individual layer thickness of 7YSZ (740 nm) and Al2O3 (765 nm) near the substrate coating interface, and decreasing in 5 nm increments to 7YSZ (90 nm) and Al2O3 (115 nm) near the surface of the coating. In order to achieve the desired individual layer thickness within the multilayer coating, modifications were made to the experimental setup as shown in Fig. 3b. A fourth crucible was added to the chamber containing aluminum oxide ingot and was offset by approximately 110 mm from the center of ingot B (7YSZ). By changing the relative source to substrate distances between the aluminum oxide and 7YSZ evaporant materials and material evaporating rates, the coatings were deposited to yield a 4:1 ratio of 7YSZ and Al2O3, respectively. In order to prevent intermixing of the vapor cloud, a vapor barrier was positioned between the evaporant materials (Fig. 3b). Similarly, the graded multilayer coatings of 7YSZ and Al2O3 with different individual layer thickness throughout the total coating thickness was achieved in this manner by varying the evaporation rate and sample rotational speed in order to get the desired individual layer thickness. Fracture surfaces, cross-sections, and surface morphologies of the coated samples were examined by a Philips model PW6848/00 scanning electron microscope (SEM) to determine microstructural and morphology differences within the various coatings. Phase analysis, texture coefficients, and residual stresses of the coated samples were determined using Philips X’Pert model MPD and MRD X-ray diffractometers. Thermal conductivity of the TBC was measured by the steadystate heat flux (CO2 laser) technique at 13161C,7 and the IR hemispherical reflectance was measured in an FTIR spectrometer with an integrating sphere accessory at room temperature described elsewhere.8 Results and Discussion Monolithic Modulated 7YSZ Microstructure

The typical microstructure of a TBC produced by EB-PVD can be divided into two zones. The inner zone (zone I) is the early part of multiple nucleation and subsequent growth of the columnar microstructure having large number of interfaces, grain boundaries, microporosity, and randomly oriented grains. The inner zone

85

ranges from 1 to 10 mm in thickness and exhibits lower thermal conductivity (B1.0 K/m-K).9,10 With increasing thickness, the TBC microstructure is characterized by a high-aspect ratio columnar grain with dominant crystallographic texture. The thermal conductivity increases as the outer part of the coating becomes more crystallograhically perfect (zone II) with fewer grain boundaries (grain size increases with increasing coating thickness, i.e., larger grains). In this outer zone (II), the thermal conductivity approaches that of bulk zirconia (2.2 W/m-K). Thus, modifying TBC microstructures should offer the best properties available for commercial EB-PVD coatings: namely, low thermal conductivity, high hemispherical reflectance, high strain tolerance, and good erosion resistance. By altering the macrostructure on the micrometer and submicrometer levels through periodically introducing strain fields (i.e., density/refractive index changes by the incorporation of microporosity and surface restructuring), the thermal conductivity of TBC materials can be significantly reduced and hemispherical reflectivity increased. As previously discussed, layered periodicity in the coating will significantly reduce both the phonon scattering and photon transport resulting in lower thermal conductivity and higher hemispherical reflectance. Through periodically interrupting the continuous flux of the vapor cloud by using a ‘‘shutter’’ mechanism, the temperature of the substrate remained almost constant (B301C drop in substrate temperature) during the deposition process with no deposition occurring when the shutter is closed. During this interruption, it is believed that the surface mobility of the condensed species contributes to the surface relaxation of the deposited coating through restructuring. As the vapor flux is prevented from depositing on the surface, the surface atoms have enough time, energy and surface mobility to diffuse to regions of lower energy. As a result, the surface strains change resulting in more phonon scattering because of different strain energy fields (and potentially sub-micron grains, interfaces, and microporosity, thus resulting in lower thermal conductivity and higher reflectivity. When the shutter is opened, the new flux deposits on a slightly different strained surface, but which is not so strained as to result in significant lattice mismatch and act as nucleation sites for the new grains. This newly strained surface results in a very diffuse interface which may contain microporosity and intracolumnar morphology differences, as well as different strain fields. Throughout this atom coalescence, stable

86


microporosity/strain fields and intracolumnar morphology differences develop with continued columnar growth without changing the surface morphology and columnar grain macrostructure as shown in Fig. 4. During the initial deposition, atoms form islands (Volmer– Weber growth mode) and grow until the flux rapidly increases resulting in grain coalescence and microporosity (resulting from grain coalescence) and tensile lattice strains. However, when viewed from the top surface, the grain size does not change, but the intracolumnar microstructure (i.e., microporosity, morphology, and strain field) is believed to be altered as shown in Fig. 4 for the 1-, 5-, 10-, and 20-layer 7YSZ coatings

Vol. 3, No. 2, 2006

deposited by the ‘‘shutter’’ method. From the SEM micrographs (Fig. 4), it is difficult to determine whether microporosity is present. However, microporosity was confirmed in the coating through the thermal conductivity measurements as the reduction in thermal conductivity is directly related to the amount of microporosity. As only a minimal temperature change occurs in addition to the disruption of the vapor flux, the long high-aspect ratio columnar grains continue to grow to the total length of the coating thickness similar to standard single layer 7YSZ. Comparison with a standard 7YSZ, the modified coating shows no distinct

Fig. 4. Scanning electron microscope micrographs showing the surface morphology (a–d) and fracture surface (e–l) of EB-PVD 7YSZ deposited by the ‘‘shutter’’ method having 1-layer, 5-layer, 10-layer, and 20-layers, respectively, on CoNiCrAlY-coated MARM247 alloy.



differences in macrostructure. As the growth orientation of the new flux remains the same as the underlying grain, similar crystallographic texture also occurs, but varies slightly resulting from increased strain fields. The interface between the condensed flux and the newly arrived flux is diffuse (i.e., no sharp distinct interface) which results in different strain fields near the interface and within each columnar grain. Such microstructural modifications will have an impact on the thermal conductivity, as well as residual stress (i.e., strain tolerance). Similar concept/phenomenon employed in the ‘‘shutter’’ approach can be correlated to the ‘‘featherlike’’ morphology often observed within each columnar grain of a standard TBC deposited by EB-PVD. The ‘‘feather-like’’ size and morphology are representative of the sample rotation speed as during each revolution, coating condenses and surface mobility changes. When the sample surface is opposite to the melt pool, little to no coating deposition occurs as well as a slight decrease in surface temperature as no radiant heat from the melt surface or heat of condensation occurs. As a result, micro strain fields develop on a much finer scale as compared with the ‘‘shuttering’’ method (30 s), because of restructuring or surface diffusion until the next high rate of condensation occurs when the sample surface is again facing the melt pool. In the present investigation, the sample rotation speed was 7 rpm, resulting in the surface of the samples being parallel to the melt pool approximately every 8.5 s. In comparison, using the 30 s delay with the ‘‘shuttering’’ method allows the samples B3.5 times more time for micro ‘‘restructuring’’ to occur which results in increased tensile strains, and thus, lower overall compressive stresses as discussed later. The periodicity of the ‘‘shutter’’ results in regions of the TBC with periodic differences in the lattice size resulting in modulated strain fields. As previously discussed, the degree of surface restructuring was much greater for the ‘‘in and out’’ approach as compared with the ‘‘shuttering’’ method. However, it should be noted that there are many factors to consider as large changes in the substrate rotation speed can also affect the grain size and intercolumnar porosity. In addition, the size of the component, condensation rate, and other deposition parameters (i.e., substrate temperature, pressure, etc.) can greatly affect the coating microstructure. Crystallographic Phase Analysis: X-ray diffraction (XRD) patterns were performed on the surface of

87

the various layered 7YSZ coated samples. Normal Bragg-Brantano (y/2y) diffraction continuous scans were performed over the range of 2y 5 151 to 901 at 2 s per step with a step size of 0.0301. The diffraction patterns confirm that the coatings are polycrystalline and predominantly of the nontransformable tetragonal phase (t0 phase) and have a strong (200) orientation which is highly desirable for strain tolerance during thermal cycling.11,12 No evidence of the monoclinic or cubic phases was observed in any of the diffraction patterns. Residual Stress: The preliminary investigation of the amount of strain within the various layered coatings supports the x-ray diffraction results (peak shifting) as the degree of deflection for the 7YSZ ‘‘shuttered’’ coatings decreased with increasing number of layers up to the 20-layer coating as shown in Fig. 5a. As the strip curvature was convex, the TBC is under a state of compression. From Fig. 5a, the amount of deflection decreased with increasing number of ‘‘shuttered’’ layers suggesting that the amount of tensile stress (strain fields) incorporated into the coating from restructuring phenomenon increased, resulting in a lower total compressive residual stress. To confirm the residual stress state of the deposited coatings, a four circle Philips X’Pert diffractometer (Almelo, The Netherlands) was used to determine the ‘‘as deposited’’ residual stresses using the sin2 technique discussed elsewhere.13 Similar to the deflection measurement trend, the preliminary stress analysis (Fig. 5b) by X-ray diffraction shows the same trend of decreasing compressive stress with increasing total number of layers. This same phenomena has been observed with other multilayer materials deposited by EBPVD with increasing number of layers.14,15 The absolute values of the residual stress determined by the deflection of the strip are presently being correlated with the x-ray diffraction analysis of the coated buttons. Continued investigation of the amount of residual stresses within the various coatings is presently being performed to determine whether the strain fields/microporosity are stable as a function of time at elevated temperatures and will be presented later. Thermal Conductivity: Several researchers have discussed the theory of thermal conductivity in ceramic materials and will only be discussed briefly here.16,17 Electrons, phonons (lattice vibrations), and photons (radiation) are the three mechanisms in which heat is transported in crystalline solids. However, as 7YSZ is a


88

(a)

conductivity can be written as:

0.010

kr ¼

Deflection (inches)

0.008

0.004

0.000 0

–50

Stress (MPa)

16 2 3 sn T lr 3

where s is the Stephen–Boltzmann’s constant, n is the refractive index, T the temperature, and lr is the photon mean free path. According to the above equations, in order to reduce the thermal conductivity, a reduction in the specific heat at constant volume, density, phonon velocity, refractive index, or mean free path is required. Of these, the specific heat at constant volume is constant above the Debye temperature. As a result, only changes to r, n, v, and mean free path will change the thermal conductivity. Phonon scattering generally occurs by interactions with lattice imperfections which include dislocations, grain boundaries, other phonons, vacancies, and atoms of different masses. The phonon mean free path (lp) can be defined by

0.006

0.002

(b)

Vol. 3, No. 2, 2006

–100

–150

1=lp ¼ 1=lv þ 1=li þ 1=lgb þ 1=ls –200

–250 0

5

10

15

20

25

Total number of layers (shutter method)

Fig. 5. (a) Deflection as a function of total number of layers for 7% Yittria-Stabilized Zirconia (7YSZ) (coating side) deposited by electron beam physical vapor deposition (EB-PVD) on PtAl-coated Haynes 188 alloy strips and (b) residual stress determined by x-ray diffraction of 7YSZ TBC deposited by EB-PVD as a function of total number of layers deposited by the ‘‘shutter’’ method.

ceramic insulator, electrons have little to no contribution to the thermal conductivity. Therefore, photons (kr) and phonons (kp) are the largest contributors to the thermal conductivity and the total thermal conductivity (kt) can be written as their sum: kt ¼ kp þ kr The thermal conductivity component resulting from phonons (kp) can be further expressed by: Z 1 Cv rvlp kp ¼ 3 where Cv is the specific heat at constant volume, r is the density, v is the velocity, and lp is the phonon mean free path. Whereas the radiation contribution (kr) to thermal

where lv, li, lgb, and ls represent the mean free path contributions because of vacancies, interstitials, grain boundaries, and strain, respectively. Doped TBC materials with atoms or ions of different masses result in a distortion of the bond length which creates strain fields within the lattice and thus phonon scattering.17 This is the similar concept used for the ‘‘shuttering’’ method in which period strain fields were developed through the TBC, but without changing the composition. Therefore, imperfections within the lattice can change the mean free path of phonons (scattering), and hence the thermal conductivity. Figure 6a shows comparative thermal conductivity of a standard single layered 7YSZ, layered TBC with diffuse interface produced by ‘‘shutter’’ method. The thermal conductivity of the standard 7YSZ TBC produced by EB-PVD was found to be B1.8 W/m-K. The 10layer microstructurally ‘‘modified’’ TBC deposited using the ‘‘shutter’’ layering concept reduced the thermal conductivity to B1.6 W/m-K. In order to establish a relationship between the thermal conductivity as a function of total number of TBC layers, additional TBC coated samples were produced using the ‘‘shuttering’’ concept. It was established that the initial thermal conductivity (K0) decreased near linearly as a function of increasing total number of TBC layers from 1 to 20 as shown in Fig. 6b. With the exception of the 5-layer coating, similar results were found



Thermal conductivity (W/m-K)

(a)

2.4

ZrO2-7wt.%Y2O3 (standard-single layer) ZrO2-7wt.%Y2O3 (10-layer "shutter" method)

2.2 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0

Thermal conductivity (W/m-K)

(b)

5

10

15

20 25 Time (hrs)

30

35

40

45

2.0 Ko (initial) K2 (after 2hrs) K5 (after 5 hrs)

1.8

1.6

1.4

1.2

1.0 0

5

10 15 Total number of layers

20

25

Fig. 6. Thermal conductivity of electron beam physical vapor deposition (EB-PVD) 7% Yittria Yittria-Stabilized Zirconia (7YSZ) thermal barrier coating (TBC) determined by a steady-state laser heat flux technique at 13161C showing (a) thermal conductivity of EB-PVD 8YSZ produced by standard continuous evaporation and ‘‘shutter’’ method as a function of testing time, (b) thermal conductivity as a function of total number of layers produced by the ‘‘shutter’’ method measured at k0 5 as deposited, k2 5 after 2 h, and k5 after 5 h of testing.

after 2 and 5 h of testing at 13161C (Fig. 6b). After 2 h of testing, the thermal conductivity was found to decrease from 1.8 W/m-K for the standard TBC to B1.2 W/m-K for the 20-layer ‘‘shuttered’’ TBC. This confirms that periodic interruption of the incoming flux by the ‘‘shutter’’ method results in lower thermal conductivity as compared with a standard single layer 7YSZ. The exact mechanism which contributes to the 20–30% reduction in thermal conductivity is still not fully understood, but is attributed to the localized density modulation. As previously discussed, the periodic interruption of the vapor flux results in surface relaxa-

89

tion or surface ‘‘restructuring’’ until the flux from the next layer deposits on the substrate surface. During this restructuring phase, it is believed that the surface atoms rearrange (surface mobility) their bonds and angles in order to minimize their free energy. When this occurs, the interatomic bond distance is believed to increase resulting in increased tensile strains. As the bond distances are now greater at the end of each layer, thermal transport between the material changes as there is a greater phonon distance resulting from the increase in bond length, which results in an overall lower thermal conductivity. In addition to these tensile strains, it is believed that microporosity forms at the start of the next growing layer resulting from island coalescence as previously discussed. Continued efforts are underway in order to study the heat conduction mechanisms resulting in the desired lower thermal conductivity values. Hemispherical Reflectance: In addition to decreasing the thermal conductivity values, the TBC with modified microstructures was expected to also affect the hemispherical reflectance of the coating, and therefore, reduce radiative heat transport through the TBC. The hemispherical IR reflectivity of the layered TBC deposited by the ‘‘shutter’’ method resulted in an increase in reflectivity with increasing number of total layers as shown in Fig. 7a. The hemispherical reflectance increased from approximately 35% (1 layer) to 45% (20 layer) at the 1 mm wavelength which is approximately a 33% increase in the reflectivity as compared with the standard 7YSZ. This suggests that more heat will be reflected from the coating as the number of layers increase within the TBC, thus allowing higher engine operating temperatures and better fuel efficiency. Figure 7b shows the percent improvement in hemispherical reflectivity versus the total number of layers. The trend shows that the hemispherical reflectivity increases with increasing number of layers as expected. The slightly higher reflectivity values for the layered TBC produced by the ‘‘in and out’’ method as compared with the ‘‘shuttered’’ method is attributed to the sharp interfaces and larger coating thickness. It is quite clear that the layering concept has opened an opportunity in engineering TBC with desired higher reflectance and lower thermal conductive properties through microstructural modifications. Benefit of Interrupting Incoming Flux: As the novel method (‘‘shuttering’’) of TBC deposition used in this


90

(a)

YSZ (1-layer) reference YSZ (5-layer) by "shutter" YSZ (10-layer) by "shutter" YSZ (20-layer) by "shutter"

60 20-layer

Reflectivity (%)

50

10-layer

40

30 5-layer

20 1-layer 10 0.5

(b)

1.5 2.0 Wavelength (µ µm)

2.5

additional amount of time needed to deposit the 10- or 20-layer ‘‘shutter’’ coating is relatively insignificant when compared with the reduction in thermal conductivity (15–30%), increase in hemispherical reflectivity (28–56%), increase in thermal cyclic oxidation life (50– 100%), better strain tolerance, and thus better performance of the component life with less down time of the engine. It should also be noted that deviations in the amount of time used for the ‘‘shuttering’’ concept could be further refined to yield better properties and coating performance.

3.0

7YSZ - Al2O3 Highly Thermal Reflective TBC

35 shutter method

30 Percentage of Improvement in Relfectivity (%)

1.0

Vol. 3, No. 2, 2006

25 20 15 10 5 0 –5 0

5

10 15 Number of layers

20

25

Fig. 7. (a) Hemi-spherical reflectance of layered (1, 5, 10, and 20 total layers) electron beam physical vapor deposition (EB-PVD) 7% Yittria Yittria-Stabilized Zirconia (7YSZ) thermal barrier coating (TBC) deposited by the ‘‘shutter’’ method after thermal exposure at 9501C for 20 h. (b) Percentage of improvement in hemispherical reflectivity versus the total number of layers within the EB-PVD 8YSZ TBC produced by ‘‘shuttering’’ method after thermal exposure at 9501C for 20 h.

investigation resulted in purely a microstructural effect (as there were no alloying additions, i.e., no compositional change), the concept can be applied to the next generation TBC materials including doped ZrO2-based, HfO2-based, etc. The shuttering concept was extended to HfO2-based TBC with reproducible results and are discussed elsewhere.18 In addition, depending on the total number of layers desired in the TBC, the additional time required to deposit the TBC is minimal. For example, in order to deposit a 10- or 20-layer ‘‘shutter’’ TBC, only an additional 5 or 10 min of deposition time is required, respectively. For the end user, this small

Multimaterial coatings with different refractive indices cause photon scattering obeying Snell’s law. Each layer in the structure causes destructive interference to wavelengths that are odd integer multiples of the halflayer thickness and constructive interference to wavelengths that are even integer multiples of the half-layer thickness. Under this scenario, selection of materials is very important and it should be thermally stable at high temperature, minimum interdiffusion, and larger difference in refractive index. Typically, the refractive index (n) of ceramic TBC materials is: ZrO2 (2.10), CeO2 (2.35), HfO2 (1.98), Al2O3 (1.60), SiO2 (1.95), Y2O3 (1.82). Similarly, by controlling the thickness of each layer, reflectance of the coating can be controlled over a wide wavelength range. It has been theorized that the thermal radiative properties of the ceramic coatings can be increased from B35% (single layer ZrO2 -7 wt% Y2O3) to nearly 100% by tailoring each coating layer thickness from 700 to 90 nm with alternating refractive index materials (ZrO2 -7 wt% Y2O3 and Al2O3), Fig. 8. In addition, if the nanolayered structure and design is selected correctly, they will offer superior creep properties and impact resistance as compared with conventional single layered ceramic coatings. It is well documented that every 251F (131C) reduction in airfoil temperature will result in 2X component life improvement.19 By increasing radiative reflectance using the nanolayered coatings, turbine blade surface temperature will be decreased up to BD1801C (Fig. 9), resulting in longer component life by 410X. Combining highly thermal radiative barrier coating concept (i.e., nanolayered) with low thermal conductive material will offer a new class of high temperature coatings. The benefit of the tailored nanolayered coatings is enormous including no design constrain, low cost



Stack Sequence Bond coat

H

91

Coating Design 111 µm · Total thickness Total # layers 262 layers · · · First layer next to substrate n= 2.1 · Second layer from substrate n= 1.62

substrate

L

87.86 nm thick 115 nm thick

H= High index of refraction (2.1 YSZ) L= High index of refraction (1.62 Al2O3)

Predicted Reflectance

Fig. 8.

Predicted optical reflectance of multi-layer TBC with individual layers varying from 87.86 nm to 7.2 mm.

component use, reduction in engine weight and size, increasing power density (by 15%) or increasing Mach number (by 50%) and significant fuel savings. In addition, application of tailored nanolayered coatings will change the design concept for all propulsion engines including small single engines, aircraft, ship and landbased vehicle power plant.

1500 TBC

Temperature (°C)

1400

Metal

1300 1200 Reduced T 1100 standard TBC standard TBC with surface reflective layer

1000 900 0

0.25

0.5

0.75

1

1.25

Distance Below TBC Surface (mm)

Fig. 9. Measured reduction in metal temperature because of modulated density (red line) and alternating material coating (green line).

1.5

To validate the nano-layered TBC concept and produce coatings composed of ZrO2 -7 wt%Y2O3 and Al2O3 by EB-PVD, multilayer coatings of 7YSZ and Al2O3 were deposited. As shown in Fig. 8, the individual layer coating thickness near the coating surface was approximately 88 nm (Al2O3) and 115 nm (ZrO27wt% Y2O3) and increased towards the substrate/coating interface. Results and Discussion: For alternating layer coatings, the thickness of each Al2O3 layer was found to vary from B75 to 100 nm (black color) while the thickness of 7YSZ remained constant at 400 nm (white color) when substrates were rotated at a constant rate shown in Fig. 10. Changing the rotation rate during deposition produced variable layer thickness coatings with a constant ratio of each material layer thickness. The corresponding hemispherical reflectance of each coating is displayed in Fig. 11. Thermal conductivity of the TBC was measured by the steady-state heat flux (CO2 laser) technique at 13161C8 and the IR hemispherical reflectance was measured in an FTIR spectrometer with an integrating sphere accessory at room temperature described elsewhere.9 The reflectance of the fixed layer had a maxima of 73% reflection in the wavelength range of 1.85 mm , while variable layer coat-


92

Fig. 10. light.

Scanning electron microscope micrographs of variable layer thickness coatings. Al2O3 layers dark, Yittria-Stabilized Zirconia layers

ings had an increased reflection spectrum over a 1 to 2.75 mm range. This preliminary experiment clearly showed that the reflectance of the coatings could be tailored to achieve increased reflectance over a desired wavelength range by controlling the thickness of individual Al2O3 and 7YSZ layers.

Multilayered TBC Increases IR Reflectance fixed vs. variable spacing multi layers 3 hr @ 1000C

80 Fixed spacing gives higher reflectance over narrow wavelength range

Reflectance (%)

70 60

Variable spacing gives higher reflectance over wider wavelength range

50 5 mill 8YSZ single layer

40 30

5 mil 8YSZ 5 mil fixed 8YSZ/Al O multilayers 4 mil variable 8YSZ/Al O multilayers

20 10 0 0.5

Vol. 3, No. 2, 2006

1

1.5

2

2.5

3

3.5

Wavelength (µm)

Fig. 11. Hemispherical reflectance of modulated 7% YittriaStabilized Zirconia (7YSZ)/Al2O3 multilayer structures containing alternating 7YSZ/Al2O3 layer with uniform individual layer thickness and alternating layered structures through coating thickness.

Conclusions Microstructural enhancements were used to produce low thermal conductive and high hemispherical reflective thermal barrier coatings deposited by EB-PVD in which the coating flux was periodically interrupted creating modulated strain fields within the TBC. These TBCs showed no macrostructural differences in the grain size or faceted surface morphology at low magnification as compared with standard TBC. The residual stress state was determined to be compressive in all of the TBC samples, and the amount of compressive stress was found to decrease with increasing number of modulations. The thermal conductivity value was shown to decrease primarily resulting from increased phonon scattering from the incorporation of strain fields caused by restructuring and microporosity. The average thermal conductivity was shown to decrease approximately 30% from 1.8 to 1.2 W/m-K for the 20-layer monolithic TBC after 2 h of testing at 13161C. The monolithic layered TBC also resulted in a 28% increase in the hemispherical reflectance, and increased with increasing total number of modulations. Multilayer 7YSZ and Al2O3 coatings with fixed layer spacing showed a 73% IR reflectance maxima at 1.85 mm wavelength. The variable 7YSZ and Al2O3 multilayer coatings showed an increase in reflection spectrum from 1 to 2.75 mm.



Preliminary results suggest that coating reflectance can be tailored to achieve increased reflectance over a desired wavelength range by controlling the thickness of the individual layers.

7.

8.

9.

Acknowledgments 10.

Any opinions, findings, conclusions, or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the U.S. Navy.

11.

References

14.

1.

2. 3. 4.

5. 6.

S. M. Meier and D. K. Gupta, ‘‘The Evolution of Thermal Barrier Coatings in Gas Turbine Engine Applications,’’ Trans. ASME 116 [1] 250–256 (1994). W. D. Kingery, H. K. Bowen, and D. R. Uhlmann, (eds.), Introduction to Ceramics, 2nd Edition, Wiley, New York, NY, 1976. D. R. Clarke and S. R. Phillpot, ‘‘Thermal Barrier Coating Materials,’’ Mater. Today, 8 [6] 22–29 (2005). D. Zhu and R. A. Miller, ‘‘Development of Advanced Low Conductivity Thermal Barrier Coatings,’’ National Aeronautics and Space Administration, Cleveland, OH, E-14433, July 2004. D. Zhu and R. A. Miller, ‘‘Development of Advanced Low Conductivity Thermal Barrier Coatings,’’ Int. J. Appl. Cerm. Tech, 1 [1] 86–94 (2004). D. E. Wolfe, J. Singh, R. A. Miller, J. I. Eldridge, and D. Zhu, ‘‘Tailored Microstructure of EB-PVD 8YSZ Thermal Barrier Coatings with Low Thermal Conductivity and High Thermal Reflectivity for Turbine Applications,’’ Surf. Coat. Technol., 190 132–149 (2005).

12.

13.

15.

16.

17. 18.

19.

93

D. Zhu, R. A. Miller, B. A. Nagaraj, and R. W. Bruce, ‘‘Thermal Conductivity of EB-PVD Thermal Barrier Coatings Evaluated by a Steady-State Laser Heat Flux Technique,’’ Surf. Coat. Technol., 138 1 (2001). J. I. Eldridge, C. M. Spuckler, K. W. Street, and J. R. Markham, ‘‘Infrared Radiative Properties of Yttria-Stabilized Zirconia Thermal Barrier Coatings,’’ Ceram. Eng. Sci. Proc., 23 417 (2002). J. R. Nicholls, K. J. Lawson, D. S. Rickerby, and P. Morrell. Advisory Group for Aerospace Research and Development (AGARD)-R-823, Aalborg, Denmark, 6.1–6.9, 1997. J. F. Bisson, D. Fournier, M. Poulain, O. Lavigne, and R. Mevrel, J. Am. Ceram. Soc., 83 [8] 1993 (2000). R. A. Miller, ‘‘Current Status of Thermal Barrier Coatings—An Overview,’’ Surf. Coat. Technol., 30 1 (1987). F. C. Toriz, A. B. Thankker, and S. K. Gupta, ‘‘Flight Service Evaluation of Thermal Barrier Coatings by Physical Vapor Deposition at 5200 H,’’ Surf. Coat. Technol., 39/40 161–62 (1989). P. S. Prevey, ASM Handbook 10: Materials Characterization. American Society of Metal, Materials Park, OH, 950 pp, 1990. D. E. Wolfe, J. Singh, and K. Narasimhan, ‘‘Synthesis of Titanium Carbide/ Chromium Carbide Multilayers by the Co-Evaporation of Multiple Ingots by Electron Beam Physical Vapor Deposition,’’ Surf. Coat. Technol., 160 206 (2002). D. E. Wolfe, J. Singh, and K. Narasimhan, ‘‘Synthesis and Characterization of Multilayered TiC/TiB2 Coatings Deposited by Ion Beam Assisted, Electron Beam-Physical Vapor Deposition (EB-PVD),’’ Surf. Coat. Technol., 165 206 (2003). J. R. Nicholls, K. J. Lawson, A. Hohnstone, and D. S. Rickerby, ‘‘Methods to Reduce the Thermal Conductivity of EB-PVD TBCs,’’ Surf. Coat. Technol., 151–152 383 (2002). R. B. Peterson, ‘‘Direct Simulation of Phonon Mediated Heat Transfer in a Debye Crystal,’’ Trans. ASME J. Heat Transfer, 116 815 (1994). J. Singh, D. E. Wolfe, R. A. Miller, J. I. Eldridge, and D. Zhu, ‘‘Thermal Conductivity and Thermal Stability of Zirconia and Hafnia Based Thermal Barrier Coatings by EB-PVD for High Temperature Applications,’’ Mater Sci Forum, 455–456 579–586 (2004). O. Friedrich Soechting, NASA Conference on ‘‘Thermal Barrier Coatings Workshop’’ NASA’s Publication 3312, 1995.