ars

ha02, Usity P

rbintalld) ythstabty br toyered structure. Lastly, the thermal instability of the layered system is investigated

eratingbarrierl TBC s) insula

coat which is deposited onto a bond coated

thermal expansion coefcient and high erosion resistance. The bond hindering diffusion. The second class of plausible replacement coatings

ughness which resultsrmance compared to

(ZrO2: 2 mol% Y2O3 +h pyrochlore Gd2Zr2O7

Surface & Coatings Technology xxx (2014) xxxxxx

SCT-19304; No of Pages 8

Contents lists available at ScienceDirect

Surface & Coatin

l sespinels and/or other unwanted and poorly adherent oxides, thusresulting in enhanced TBC adhesion.

The primary focus for the development of advanced TBCs is reducingthe thermal conduction through the TBC system while maintainingthermo-mechanical and thermo-chemical stability. Current 7YSZ

prevent reactions with the TGO [9] and a low toin a signicant reduction in the erosion perfostandard 7YSZ.

This study investigates rare earth doped YSZ1 mol% Yb2O3 + 1 mol% Gd2O3) and the rare eartcoat of choice is usually a Pt modied nickel aluminide, (Pt,Ni)Al,though recent interest has been focused on / type coatings [2,3].Since the top coat 7YSZ material is a very good oxygen conductor, thebond coat serves to form a stable, slow growing and mechanically ro-bust thermally grownoxide or TGO, typically-Al2O3. This TGOprotectsthe underlying substrate from severe oxidation which could produce

is the pyrochlore type rare earth zirconates, RE2Zr2O7 [4]. Thesematerials possess phase stability from 1500 C to over 2000 C whileexhibiting thermal conductivities and sintering rates signicantlylower than 7YSZ. Additionally, researchers [8] have indicated a bene-cial reduction in CMAS degradation with Gd2Zr2O7. The primary draw-back of rare earth zirconates is a requirement for a diffusion barrier tothermal barriers are inadequate due to thephase instability (N1200 C), increased sinterthermal conductivity. Thus the insulating c

Corresponding author at: The Applied Research LaboUniversity, University Park, PA 16802, USA. Tel.:+1 814 8

E-mail address: dew125@arl.psu.edu (D.E. Wolfe).

http://dx.doi.org/10.1016/j.surfcoat.2014.03.0490257-8972/ 2014 Elsevier B.V. All rights reserved.

Please cite this article as: M.P. Schmitt, et al.nickel based superalloyermal conductivity, high

than 7YSZ [7]. Zhu et al. [5] ascribed this to the formation of immobiledefect clusters which efciently scatter phonons while substantially[1]. 7YSZ is chosen due to its relatively low thRare earth

1. Introduction

As the push continues for higher opbines, research into advanced thermalcome increasingly important. A typica7 wt.% yttria stabilized zirconia (7YSZ 2014 Elsevier B.V. All rights reserved.

temperatures in gas tur-coatings (TBCs) has be-ystem is composed of ating porous ceramic top

generation TBCsmust possess lower thermal conductivity and sinteringrates than 7YSZ, while maintaining erosion resistance and phase stabil-ity at elevated (N1400 C) temperatures. Co-doped YSZ and the rareearth zirconate pyrochlores have garnered attention as potential re-placements of 7YSZ [1,47]. YSZ co-doped with rare earths haveshown substantially lower thermal conductivities and sintering ratesMultilayer and thought is given to optimization of layer thickness.

Pyrochlore 57% when utilizing a nanolaMultilayer thermal barrier coating (TBC)earth doped YSZ and rare earth pyrochlore

Michael P. Schmitt a,b, Amarendra K. Rai c, Rabi Bhattaca The Applied Research Laboratory, The Pennsylvania State University, University Park, PA 168b Department of Materials Science and Engineering, The Pennsylvania State University, Univerc UES Inc., 4401 Dayton-Xenia Road, Dayton, OH 45432, USAd NASA Glenn Research Center, Cleveland, OH 44135, USA

a b s t r a c ta r t i c l e i n f o

Article history:Received 9 December 2013Accepted in revised form 22 March 2014Available online xxxx

Keywords:Thermal barrier coatingTBC

To allow for increased gas tuto protect the underlying merare earth doped (Yb and Gwith novel nanolayered andlimit of current 7 wt.% yttriaduce the thermal conductiviGZO is shown to be an orde

j ourna l homepage: www.eir elevated temperatureing rate and inadequateeramic top coat of next

ratory, The Pennsylvania State65 0316; fax:+1 814 863 2986.

, Surf. Coat. Technol. (2014), hchitectures utilizing rare

rya c, Dongming Zhu d, Douglas E. Wolfe a,b,SAark, PA 16802, USA

e efciencies, new insulating thermal barrier coatings (TBCs)must be developedic components from higher operating temperatures. This work focused on usingttria stabilized zirconia (t low-k) and Gd2Zr2O7 pyrochlores (GZO) combinedick layered microstructures to enable operation beyond the 1200 C stabilityilized zirconia (7YSZ) coatings. It was observed that the layered system can re-y ~45% with respect to YSZ after 20 h of testing at 1316 C. The erosion rate ofmagnitude higher than YSZ and t low-k, but this can be reduced by almost

gs Technology

v ie r .com/ locate /sur fcoatas possible replacements for 7YSZ. Herein referred to as t low-k andGZO respectively, these materials possess both lower thermal conductiv-ity and sintering rates than 7YSZ. As the relatively high erosion rate ofmonolithic GZO could prove too large to adapt it into a reliable coating,unique multilayer architectures have been deposited in order to reducethe erosion rate relative to GZOwhile maintaining superior thermal con-ductivity over 7YSZ.

ttp://dx.doi.org/10.1016/j.surfcoat.2014.03.049

reection loss at the surface. A laser delivered heat ux was calibratedusing a sample of known thermal conductivity and by subtracting thelaser reection loss at the surface from the known heat ux, the passthrough heat ux can be determined. The pass-through heat ux wasfurther veried experimentally. The TBC/metal interface temperaturewas obtained using the known thermal conductivity of the Rene N5metal substrate. The overall thermal conductivity as functions of coatingthickness, temperature and testing time was determined from the heatuxes and the corresponding temperature gradients across the ceramiccoating. Further details of the thermal conductivity measurement pro-cedure are given elsewhere [11,12].

Particle erosion was performed using an in-house erosion rig. TBCspecimens were weighed and mounted into the sample holder with a0.5 mask. The sample holder was then loaded into the erosion rig di-

t1:1 Table 1t1:2 Depositionmatrix detailing the coating, composition and structure for the single layer andt1:3 multilayer coatings used in this study.

Matrix # Coating Composition Designarchitecturet1:4

M1 7YSZ ZrO2: 4Y2O3 (mol%) Monolayert1:5M2 t low-k ZrO2: 2Y2O3 + 1Yb2O3 + 1Gd2O3

(mol%)Monolayert1:6

M3 GZO Gd2Zr2O7 Monolayert1:7M4 t low-k/GZO ZrO2: 2Y2O3 + 1Yb2O3 + 1Gd2O3

(mol%)/Gd2Zr2O7Nanolayert1:8

M5 t low-k/GZO ZrO2: 2Y2O3 + 1Yb2O3 + 1Gd2O3(mol%)/Gd2Zr2O7

Thick Layert1:9

2 M.P. Schmitt et al. / Surface & Coatings Technology xxx (2014) xxxxxx2. Experimental procedure

One inch diameter (Pt,Ni)Al bond coated (Alcoa Howmet,WhitehallMI) ReneN5 (Sophisticated Alloys, Butler PA) buttonswere used as sub-strate materials. For high temperature isothermal heat treatment work,1 in 1 in polycrystalline Al2O3 substrates (CoorsTek, 99.5% pure) werealso utilized. The various coating architectures and compositions can befound in Table 1. Monolayer TBC structures of 7YSZ, t low-k and GZO(M1, M2 and M3 respectively) were deposited onto substrates viaelectron beam-physical vapor deposition (EB-PVD). A source-substratedistance of 12 was maintained while substrates rotated at 7 RPM on a2diametermandrel directly above the ingotmaterial. ThemultilayeredTBC structures fromM4 andM5were fabricated by depositing alternat-ing layers of t low k and GZO. For M4, the necessary ingots were co-evaporated and a vapor shield was used to prevent intermixing of thevapor clouds. Rotation through each vapor cloud provided a coatingwith alternating layers of t low-k and GZO at an average individuallayer thickness of ~200 nm. In the case of M5, the electron beamwas al-ternated between two ingots; thus, only a single ingot was evaporatedat a time. This technique yielded layered coatings with individuallayer thicknesses on the order of ~30 m. The overall thickness of M1M3 was ~200 m, while M4 and M5 were slightly thicker at around300 m. Prior to the deposition of M3M5, a ~25 m diffusion barrierlayer of YSZ (M3) or t-low k (M4M5) was deposited on top of thebond coat to maintain thermochemical stability between GZO and theTGO. A detailed description of the coating deposition procedure isdescribed in [10].

Thermal conductivity of the as-fabricated TBCs was measured at theNASAGlenn Research Center via a steady-state heatux technique. Passthrough heat ux and themeasured temperature gradients through theceramic coating system were used in conjunction with a one dimen-sional heat transfer model to calculate the thermal conductivity [11].Briey, the TBC surface temperature was maintained at 1316 C for 20h through a constant heat ux from a 3.0 kW CO2 laser while backsideair coolingwas used tomaintain the desired thermal gradient and spec-imen temperature. Surface and backside temperatures were measuredvia optical pyrometers while a laser reectometer measured the100 m

(a)

100 m

(b)

Fig. 1. SEM images of the polished cross sections of as deposited (a) M1YSZ, (b) M2t l

Please cite this article as: M.P. Schmitt, et al., Surf. Coat. Technol. (2014), hrectly beneath the particle acceleration tube which ensured consistencythrough all the erosion trials and also allowed for an adjustable incidentangle of erodent. After mounting, the rig was calibrated and ameasuredamount of media was placed into the feeder trough. In this study, thedosage increment was 5 g of 240 grit (50 m) Al2O3 and it was fed ata rate of ~2 g/min for a total of ~2.5 min. During testing, a vibratoryfeed system feeds the powder from the trough into the accelerationtube, where compressed air accelerates the erosion particles. Air pres-sures of 30 PSI were used to accelerate particles to speeds of 100 m/sat incident angles of 90. After one dose the samples were weighedagain to measure weight loss and the process was repeated. Weightloss was then plotted as a function of erodent fed and the coatingswere eroded until sufcient data points were present to determine anerosion rate.

X-ray diffraction (PANalytical Empyrean) and Jade 10 software wereused to characterize the crystallography in all the samples. Scanningelectronmicroscopy (FEI Quanta 2000)was used to characterize themi-crostructure in the as deposited state and after erosion testingwhile boxfurnaces were used to carry out the high temperature anneals.

3. Results and discussion

3.1. Monolayer coatings

The as deposited cross sections of the single phase TBCs fromM1 toM3 are shown in Fig. 1(ac), respectively. The columnar morphology isclearly present in all three coatings. M1 and M2 appear to have slightlysmaller column diameters than M3 whose columns appear slightlymore feathered. The diffraction patterns for coatings M1M5 arepresented in Fig. 2. The tetragonal splitting of the (004)(400) peaksof M1 and M2 is visible at values of ~73.2 and ~74.2 2, respectively.M3 exhibits a single peak at 71.9 2, indicative of a cubic phase. All ofthe single layer coatings were textured with their highest intensitypeaks corresponding to the {311} planes.

The thermal conductivities of M1M5 are presented in Fig. 3. Thechart displays the initial thermal conductivity in the darker shadewith the nal thermal conductivity after 20 h of testing at 1316 C in

100 m

(c)ow-k, and (c) M3GZO. Note that M3 also includes a ~25 m t YSZ diffusion barrier.

ttp://dx.doi.org/10.1016/j.surfcoat.2014.03.049

its lifetime. From these results, it is clear that M3 (GZO) is signicantlymore stable at high temperatures (1316 C) than the M1 (7YSZ) andeven M2 (t low-k), and thus presents a very attractive replacementTBC material.

The erosion of coatings from Table 1 is presented in Fig. 4 with SEMcross sections of M1M3 shown in Fig. 5. Focusing initially on the singlelayers (circles in Fig. 4), the rst observation is that the M1 and M2perform almost identically. When comparing Fig. 5(a) and (b), we seeminimal differences in the amount of plastic deformation and crackingat the TBC surface after erosion, indicating that these materials behavevery similar mechanically. According to Wellmann and Nicholls [13],at particle velocities of 100m/s and particle size of 50 mwewould ex-pect to observe Mode I type damage. This is consistent with our obser-vations in Fig. 5(a) and (b) where the cracking is primarily limited tothe rst 510 microns with minimal plastic deformation. Since therare earth elements have a very obvious impact on thermal properties,it is interesting that erosion remains almost identical. The overallstabilizer concentration was ~4 mol% in the t low-k coatings, which isvery similar to the ~4 mol% Y2O3 in YSZ. According to the Shannontable [14], the crystal radius of an eight coordinated Gd3+ is 1.193 while that of a similarly coordinated Yb3+ is 1.125 . The average ofthese is 1.159 , which is also the crystal radius of an eight coordinatedY3+. Thus equimolar additions of Gd and Yb should have the same effecton the lattice parameter as equivalent amounts of Y. This is important asthe addition of Yb and Gd at the expense of Y will have minimal effectson the tetragonality; then the ferroelastic toughening [15] should re-main. One may surmise that additions of only rare earths with largerradii (such as Gd in our case) at the expense of Y would act to reducethe tetragonality, and diminish the ferroelastic toughening and thus

Inte

nsity

(A.U

.)M5: t'/GZO Thick LayerM4: t'/GZO NanolayerM3: GZOM2: t' Low-kM1: 7YSZ

70 71 72 73 74 75 76

t (400)t (004)GZO (400)

3M.P. Schmitt et al. / Surface & Coatings Technology xxx (2014) xxxxxxthe lighter shade. This allows the sintering rates of the coatings to be ex-aminedwhich is important as sintering not only leads to increased ther-mal conductivity but also reduced mechanical performance. The initialthermal conductivity of M1 was 1.45 W/m-K and increased by 52% to2.2 W/m-K after 20 h. The addition of rare earths in the M2 coating de-creased the initial thermal conductivity to 1.2 W/m-K with only a 37%increase to 1.64W/m-K after 20 h. The decreased sintering rate indicatesthat these coatings are signicantly more stable, both microstructurallyand chemically, than standard 7YSZ coatings in M1. This is very impor-

20 30 40 50 60 70 80 90Degrees 2

Fig. 2. X-ray diffraction patterns of as deposited M1M5 with the high angle (004)/(400)region inset. The mixed phase nature of the nanolayer M4 is clearly evident.tant, since sintering of the columns leads to increased erosion whilealso reducing strain tolerance, and thus thermal cyclic life. The GZO coat-ing from M3 has the lowest thermal conductivity of the three composi-tions, with an initial conductivity of 1.13 W/m-k and a small increaseof 26% to 1.42 W/m-k after 20 h. This is a substantial decrease fromM1, whose initial thermal conductivity is even greater than the sinteredvalue of GZO. This value could be further reduced by increasing coatingporosity, but at the expense of erosion performance. The sintering rateof M3 is also about half that of M1, and about three fourths that of M2.This suggests that the GZO coating would have less column sinteringthrough its lifetime, and thus minimal increases in erosion rate through

0 0.5 1 1.5 2 2.5Thermal Conductivity (W/m*K)

As Deposited

Sintered

Matrix 1

Matrix 2

Matrix 3

Matrix 4

Matrix 5

1.45

1.20

1.13

1.14

1.10

2.20

1.64

1.42

1.22

1.18

Fig. 3. Thermal conductivity of M1M5with initial values plotted in the darker shade andvalues after 20 h of testing at 1316 C surface temperatures plotted in the lighter shade.

Please cite this article as: M.P. Schmitt, et al., Surf. Coat. Technol. (2014), herosion performance, though to the authors' knowledge there is nostudy relating the ionic radii, lattice parameter and subsequent tough-ness at a given concentration. The second important observation isthat M3 has an erosion rate that is ~10 higher than that of M1 andM2. The relatively low toughness of GZO results in a brittle materialwhich readily fractures when impacted by erodent particles. Fig. 5(c)shows a signicant fracture in the rst 1015 microns of the coating

y = 0.118x

y = 0.528x

y = 1.244x

y = 0.286x

y = 1.435x

Matrix 1: 7YSZ

Matrix 2: t' Low-k

Matrix 3: GZO

Matrix 4: t' Low-k/GZO NanolayerMatrix 5: t' Low-k/GZO Thick Layer

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0 0.02 0.04 0.06

Coa

ting

Mas

s Los

s (g)

Mass Erodent Exposure (Kg)

Fig. 4. Erosion data for M1M5with coatingmass lass plotted as a function of erodent ex-posure. Erosion rates correspond to slopes of the lines and indicate that the nanolayer M4

is able to reduce the erosion rate with respect to M3 by ~57%.

ttp://dx.doi.org/10.1016/j.surfcoat.2014.03.049

and with some even further down the columns (highlighted with ar-rows) leading to increased material loss. The morphology also plays arole, as larger column diameters tend to have high erosion rates [16].In our case, the wider columns of M3 likely increase the erosion ratesince a given spallation event of a fractured column results inmore ma-terial loss. Also, there isminimal plastic deformationwhichmeansmax-

3.2. Multilayer coatings

Multilayer coatings composed of t low-k and GZO were depositedaccording to M4 (nanolayer) andM5 (thick layer) from Table 1 to com-bine the benecial attributes of both classes ofmaterials.M4was depos-ited in an attempt to reduce the erosion rate with respect to GZO while

20 m 20 m 20 m

(a) (b) (c)

Fig. 5. SEM images of the polished cross sections after erosion testing for (a) M1YSZ, (b) M2t low-k, and (c) M3GZO. Black arrows in (c) showwhere cracking has occurred at sig-nicantly lower depths than in (a) or (b).

4 M.P. Schmitt et al. / Surface & Coatings Technology xxx (2014) xxxxxximum energy is available for fracture. When combined with inherentlylow toughness, this results in a coating which exhibits a drastic reduc-tion in erosion performance from the current state of the art 7YSZ coat-ings such as M1. This is a major drawback of the GZO system andrequires attention when considering the overall lifetime of the TBC sys-tem. One important consideration is that though the initial erosion rateof M3may be relatively high compared toM1, in the aged condition theperformance gap may be signicantly smaller. This is due to thesintering rate of GZO being much lower than YSZ which yields minimalincreases in the erosion rate over time, whereas YSZ has signicantsintering and signicant increases in the erosion rate. M2 exhibits anerosion rate almost identical to that of YSZ and much lower than GZOin the as deposited condition. Therefore, in terms of coating life time,the t low-k system is favorable over GZO as it improves the thermalproperties over YSZwithout sacricing erosion performance. The draw-backs to choosing t low-k over GZO are that GZO offers a higher maxi-mum temperature capability, better thermal insulation and possibleenhanced resistance to CMAS over t low-k. Thus, an ideal coatingwould combine these factors.

(a)100 m

100 m

(c)

Fig. 6. SEM images of a polished cross section from the as depositedmultilayered t low-k/GZO cview detailing the layered morphology of M4, (c) showing the coating microstructure of M5, a

Please cite this article as: M.P. Schmitt, et al., Surf. Coat. Technol. (2014), hmaintaining a lower thermal conductivity relative to t low-k while M5was deposited in order to study the effects ofmodifying layer thickness.SEM micrographs of M4 and M5 are presented in Fig. 6(a,b) and (c,d),respectively. At a similar distance from the substrate, the columnwidthsof M5 appear larger and denser than those of M4. The use of a single de-position ingot at a time reduces the deposition rate forM5which in turnallows for increased surface adatom mobility and diffusion and there-fore less nucleation of new columns. The higher magnication imageof M4 in Fig. 6(b) clearly illustrates the layered nature of the coatingwith individual layer thicknesses on the order of ~200 nm. This is in-creased to ~30 m for the layers in M5. The diffraction patterns of M4and M5 are presented in Fig. 2. M4 is clearly a multiphase mixture of t low-k and cubic GZO as evidenced by the tetragonal (004)(400)peaks and cubic (400) peak. The inset shows a reduction in the peak in-tensities and a 0.2 leftward shift of the tetragonal peaks to ~74 (004)and 73 (400) 2. The cubic (400) peak of GZO also shifts by ~0.2 to aposition of 71.7 2. There are severalmechanisms that could be respon-sible for this shift. The layered nature of the coatings could cause CTEmismatch stresses, interface mist stresses and intrinsic growth

(b)5 m

(d)

5 m

oatings (a) showing the coatingmicrostructure ofM4, (b) showing a highermagnicationnd (d) showing a higher magnication image of one of the layer interfaces in M5.

ttp://dx.doi.org/10.1016/j.surfcoat.2014.03.049

stresses. The CTE of YSZ is 10.7 ppm/K [17]while GZO is 8.9 ppm/K [18],the difference of which could cause stresses upon cooling. Likewisemismatch at the layer interfaces may cause mist stresses, which, inconjunction with the relatively large interfacial volume, could have asignicant contribution to the overall stress state. Crystallite size isalso quite small in these coatings, as the layer thickness (~200 nm)limits grain size. This could also contribute to the overall stress state.We see a shift in the texturing from the {311} for the single layer coat-ings in M1M3 to the {200} in M4. For M5, the diffraction pattern onlyshows evidence of the GZO phase, likely a product of its high absorptionand the relatively large layer thickness. The coating is heavily texturedin the {311} with no peaks visible in the tetragonal region (inset). Thisconrms the increased surface mobility mentioned previously.

The initial thermal conductivity of M4 is 1.14 W/m-K which in-creases by only 7% to 1.22 W/m-K after 20 h (Fig. 3). The approximatecomposition of the coating is 50:50 t low-k:GZO. Eq. (1) outlines a sim-ple linear two phase mixing rule where Px and Py are the properties ofinterest for phase x and y, fx and fy are the phase fractions of phase x

conductivity of almost half of M1's 2.2 W/m-K. This illustrates the enor-mous potential of the multilayer system.

From Fig. 4, the erosion rate of M4 lies between that of M2 and M3.Thus, the layering of the tougher t low-k phase with the weak GZOphase is able to reduce the erosion rate with respect to GZO by 57%. Ifwe apply themixture rule to our erosion rate, wewould expect a reduc-tion of ~50% for a 50:50 mixture. The difference could be caused bychanges in morphology, porosity, layer interfaces and strain state,among others. Again, this would require more analytical evaluation toextract the cause of the differences between expected and observedvalues. For M5, the initial erosion rate is 1.435 g/kg, a value similar tothat of M3-GZO. At 0.02 kg of erodent fed, we observed a change inthe slope to a value of 0.286 g/kg which corresponds to a transitionfrom the top GZO layer in the thick layer coating to the t low-k layer be-neath. In this case the erosion rate of the t low-k layer in M5 is approx-imately twice that of pure t low-k in M2. There are several scenariosthat could cause this, likely tied to the fact that though the t low-klayers in M5 are tough; the GZO layers below are quite weak. Post ero-

nd tM1 od to

5M.P. Schmitt et al. / Surface & Coatings Technology xxx (2014) xxxxxxand y and Pc is the resulting property of the composite. Applying thismixing rule to thermal conductivity for our system, we would expecta conductivity of about 1.17 W/m*K for the nanolayer coating in M4.

Pc Px f x Py f y 1

The slight reduction in the observed thermal conductivity could beattributed to an increase in porosity and/or a morphology that moreeffectively scatters phonons. Hemispherical reection has been shownto reduce phonon conduction in multilayer coatings [19], but porositymust be consistent among coatings in order to conrm this effect. Sincethermal conductivity is proportional to porosity, the 2% reduction inthe observed thermal conductivity (1.14) from the expected value(1.17) would correspond to a 2% reduction in porosity. This is outsidethe accuracy of most image analysis software and most other commonmethods of analyzing TBC porosity, and thus we cannot conrm thepresence of hemispherical reection without further testing. The M5coating yields a slight reduction in thermal conductivity to a value of1.1 W/m-K. This decrease is likely a product of a combination of a slightlyangledmorphology and a lower coating density, though this is difcult toconrm. Interestingly, both coatings show an increase in thermal con-ductivity of ~7% over 20 h. This is substantially less than the 37% and26% increases observed for the parent phases M2 (t low-k) and M3(GZO), respectively. If we again apply a rule of mixtures, we would ex-pect an increase of ~32% increase. The observed 7% increase suggeststhat there is very little sintering in the multilayer systems under theseconditions. When comparing the multilayer coatings in M4 and M5 tothe current state of the art 7YSZ in M1, we observe a nal thermal

20 m

100 m

100 m(a)Fig. 7. SEM image of a polished cross section from the multilayered coatings fromM4 (a) a(a) indicate where cracking has occurred at signicantly lower depths than observed in forin theGZO layer below. In the region of (c), the cracks are beginning to coalescewhich has le

the next GZO layer has been exposed and is severely cracked. Notably, no cracking occurs in th

Please cite this article as: M.P. Schmitt, et al., Surf. Coat. Technol. (2014), hsion SEM cross sections are presented in Fig. 7(ad). Fig. 7(a) shows thatcracking primarily occurs in the top 10 m for the coating fromM4, verysimilarly to M1 and M2. On the other hand, there is almost no plasticdeformation, similar to that observed inM3,which leads to some crack-ing at depths N10 m as shown by the black arrow. This is further evi-dence of the combination of behaviors from both classes of materials.Fig. 7(bd) shows the SEM cross section images of the eroded samplefrom M5 at various locations along the surface. In the region ofFig. 7(b) we see that there is minimal cracking in the top t low-klayer yet there is signicant cracking in the GZO layer beneath. Thiscracking can easily lead to large scale erosion of t low-k region whencracks begin to coalesce. Indeed, in a different region of the sampleshown in Fig. 7(c) we see a signicant interface cracking below the tlow-k layer to the point that in part of the image the cracks have coa-lesced resulting in partial spallation and exposure of the next layer,GZO. Fig. 7(d) shows an entire regionwherewehave eroded completelythrough the t low-k layer and into the secondGZO layer. Again, the onlycracking that is observed occurs in the GZO layer. Interestingly, we caneven observe a fracture occurring below the t low-k layer in the nextGZO layer (black arrows). Thus, it appears that crack initiation occurs al-most exclusively in the GZO layers and mostly near the layer interfacesand that it beginswell before any erosion reaches the overlaying t low-k layer. This explainswhy the second slope observedwas not as low as apure t low-k layer, since some localized spallation occurs when suf-cient cracks have been nucleated. Overall, the low erosion rate demon-strates that nanolayering GZO together with t low-k allows us tomaintain much lower thermal conductivities and sintering rates ascompared to 7YSZ and even t low-k single layers, while improvingthe erosion resistance with respect to GZO. Admittedly, a multiphase

100 m

(b)

(c) (d)hree different regions along the surface of M5 (bd) after erosion testing. Black arrows inrM2. In (b) the t low-k layer has been partially erodedwith signicant cracking occurringdelamination of the part of the t low-k layer. In (d), the t low-k layer has delaminated and

e underlying t low-k layer but in the next GZO we see cracking (arrows).

ttp://dx.doi.org/10.1016/j.surfcoat.2014.03.049

nanolayered TBC would add cost and complexity to the TBCmanufacturing process but in critical components for high performanceengines, there may be applicability. The thick layer coating exhibitedsimilarly benecial thermal performance but with a modied erosionmechanism which reduces overall performance.

3.3. Thermal stability of multilayer t low-k/GZO TBCs

The thermal stability of the multilayer systems was investigated viaa series of isothermal heat treatments at 1316 C. Fig. 8(ad) displayspost heat treatment SEM image of M4 (a) andM5 (bd) while Fig. 9 de-tails the corresponding pre and post heat treatment X-ray diffractionpatterns forM4. Themicrostructures in Fig. 8 all display the classical col-umn sintering and densication with the feathery porosity coarseninginto spherical pores. Of particular note though, is that the nanolayeringis no longer evident for M4. This suggests that there was signicant in-terdiffusion between the layers during the heat treatment. From the dif-fraction pattern in Fig. 9, it is immediately obvious that the nanolayersystem has undergone a phase change. The three peaks (two split te-tragonal peaks of the t phase and a single cubic peak of the GZOphase) havemerged into a single cubic peak at 72.7 degrees 2. Interest-ingly, there is no evidence of any remaining t or GZO peaks whichindicates that complete interdiffusion has occurred during the 20 hourheat treatment. This is ascribed to a compositional gradient that createsa thermodynamic driving force for interdiffusion which is exacerbatedby a combination of non-homogenous layer thicknesses and nondistinct layer interfaces. The nominal individual layer thickness is~200 nm in these coatings, though this varies as a function of column in-

is some degree of intermixing. As such, the diffuse interface likely cre-ates some small volume that is cubic phase (but not GZO) even in theas-deposited state. This is not apparent in the as-deposited diffractionpatterns, likely because the volume is quite small and the phase is so

10 m

10 m

(d)

t treatment at 1316 C for 20 h (a,b), 100 (c) and 500 (d) h. The porosity has spherodized for all

20 30 40 50 60 70 80 90

Inte

nsity

(A.U

.)

Degrees 2

M4: NanolayerHeat TreatM4: NanolayerAs-Dep

70 71 72 73 74 75 76

c

t

GZO

Fig. 9.X-ray diffraction patterns of as deposited (solid line) and heat treated (dashed line)M4 with the high angle (004)/(400) region inset. The mixed phase M4 transforms into asingle cubic phase after heat treatment, conrming SEM observations of instability.

6 M.P. Schmitt et al. / Surface & Coatings Technology xxx (2014) xxxxxx10 m

Fig. 8. SEM images of polished cross sections fromM4 (a) andM5 (bd) after isothermal heaclination. A more detailed description of the process leading to this isgiven in Fig. 10. This leads to areas where the layer thickness is approx-imately 50 nmwhich, evenwith small driving forces, may be too thin toprevent signicant interdiffusion at 1316 C. Secondly, TEM imagingand EDS mapping have indicated that the t low-k to GZO interface isrelatively diffuse (Fig. 11). This is an artifact of the deposition processas complete separation of the vapor plumes is difcult and thus there

(c)

10 m

(a)coatings and the nanolayers are no longer evident in the M4 coating. Diffusion progresses with

Please cite this article as: M.P. Schmitt, et al., Surf. Coat. Technol. (2014), hclose in lattice parameter to GZO/t (004) that it is very difcult to dis-tinguish with a lab instrument. Interestingly, this interdiffusion doesnot appear to signicantly reduce the lifetime of the coating. Initial ther-mal cycling results (55 min in, 5 min fan cooled) at ~1188 C haveshown that a standard 7YSZ and the GZO coating fromM3 have the lon-gest lifetimes of the single layer coatings. This cyclic lifewasmatched bythe nanolayer coating fromM4 in our initial study. The durability of the

(b)time in M5 and is extensive along the column boundaries.

ttp://dx.doi.org/10.1016/j.surfcoat.2014.03.049

Graphite Divider

Substrate

Underlying old TBC

Vapor Cloud #2

Vapor Cloud #1

Mandrel

Initial sample position

Underlying old TBC

New TBC material (blue)

Sample continues through vapor cloud 1 region (blue) and wayward side of column tip is shadowed leading to differences in thickness

Sample enters vapor cloud 2 region (red) and leeward side of column tip is shadowed leading to deposition

Resulting layered TBC has non-uniform layer thickness

7M.P. Schmitt et al. / Surface & Coatings Technology xxx (2014) xxxxxxUnderlying New TBC

primarily on wayward sidethick layer coating, however, is markedly decreased and is currentlybeing investigated. Since YSZ is typically used as a diffusion barrier toprevent GZO from reacting with Al2O3, the observed complete interdif-fusion is troubling as formation of a high RE content cubic phase at theTBC/TGO interface must now be considered. If this phase has a RE con-tent larger than some critical concentration X* [9], it could potentiallyreact with the TGO to form a perovskite (in the case of GZO) whichwill negatively affect the TBC adhesion.

The interaction of t low-k-GZO is more easily studied in the thicklayer system fromM5. Therefore, samples fromM5 have also undergone

old TBCmaterial (red)

Fig. 10. Schematic showing the growth of the TBCwhen utilizing the co-deposition nanolayer seand so deposition is favored on a particular side of the column leading to non uniform layer th

(a)Fig. 11. TEM image (a) and EDS linescan (b) of a FIBed section from the cross section ofM4. Theinterfaces is relatively diffuse while one is quite sharp. Coating growth direction is diagonally b

Please cite this article as: M.P. Schmitt, et al., Surf. Coat. Technol. (2014), hheat treatments at 1316 C for periods of 20, 100 and 500 hwith the cor-responding SEM images presented in Fig. 9(bd), respectively. Notably,even at the 500 hour mark complete interdiffusion has not occurred inthese samples. This conrms that the complete phase change observedin the nanolayer system was primarily due to the layer thickness. Fromthe 20 hour heat treatment image, we observe a ~0.5 m interdiffusionregion at the layer interfaces, well in excess of the ~200 nm layer thick-ness in the nanolayer system. This increases to ~1 m in the case of the100 hour treatment and N2 mafter 500 h. These values help to establishan approximate diffusion path length which could assist in optimization

2 m

tup. As the columns rotate over a given vapor cloud, part of the column surface is shadowedickness as shown in the micrograph on the bottom right.

0 100 200 300 400

Cou

nts

Distance (nm)

Oxygen ZirconiumGadolinium YtterbiumYttrium

(b)red line represents the linescan region and corresponding EDS data suggests that one of theottom right to top left.

ttp://dx.doi.org/10.1016/j.surfcoat.2014.03.049

of the layer thickness for future systems. In the ideal case, the layer thick-ness would be minimized such that the erosion either a) exhibits amechanism similar to M5 but removes much smaller amounts due tosmaller layer thicknesses or b) changes froma layer-by-layermechanismto a standard sub-surface cracking as seen inM1M4. Thisminimal layerthicknesswould be obtained by using the diffusion path lengths estimat-ed from the current study. Notably, the diffusion extends along thecolumn edges for ~5 m in the 20 hour sample, and this gradually in-creases as heat treatment continues. This again brings up concerns of re-actions with the TGO once the composition reaches X*.

4. Conclusions and future work

Thermal barriers composed of t low-k andGZOhave beendepositedand compared to standard 7YSZ in terms of thermal conductivity anderosion. While the GZO coating exhibited superb thermal conductivityand sintering resistancewith respect to 7YSZ, the erosion rate increasedby an order of magnitude. For design and lifetime considerations, thiscould preclude the use of GZO type coatings in certain areas of enginedesign. The t low-k materials exhibited an erosion rate nearly identicalto YSZ yet with signicantly better thermal conductivity. The multi-layers of these two materials have been shown to reduce thermal con-ductivity and sintering rate to values very similar to GZO with onlyhalf the erosion debit of a pure GZO coating. Altering the layer thicknesschanges the erosion behavior due to the brittle nature of GZOwhich al-lows for sub surface layer cracking yielding a higher erosion rate in

Conict of interest

There is no conict of interest in this study.

Acknowledgments

The authors would like to thank Anna Stump for help with samplepreparation and microscopy and Ethan Lucas for help with coating ero-sion testing. This work was funded in part under DOE STTR Phase IIaward No. DE-SC0005356 and the Eric Walker fellowship program ofthe Applied Research Laboratory at the Pennsylvania State University.Any opinions, ndings, conclusions, or recommendations expressed inthis material are those of the authors and do not necessarily reectthe views of the DOE.

References

[1] C.G. Levi, Curr. Opin. Solid State Mater. Sci. 8 (2004) 7791.[2] N.Mu, L. Izumi, B. Gleeson, in: R.C. Reed, et al., (Eds.), Superalloys 2008, TMS, Champion,

PA, 2008, pp. 629637.[3] T. Izumi, N. Mu, L. Zhang, B. Gleeson, Surf. Coat. Technol. 202 (2007) 628631.[4] R. Vassen, X. Cao, F. Tietz, D. Basu, D. Stover, J. Am. Ceram. Soc. 83 (2000) 20232028.[5] D. Zhu, Y.L. Chen, R.A. Miller, NASA/TM-2004-212480, NASA Glenn Research Center,

Cleveland, OH, January 2004.[6] J.Wu, X.Wei, N.P. Padture, P.G. Klemens, M. Gell, P. Miranzo, et al., J. Am. Ceram. Soc.

85 (2002) 30313035.[7] D. Zhu, R.A. Miller, NASA/TM-2002-211481, NASA Glenn Research Center,

Cleveland, OH, May 2002.[8] S. Krmer, J. Yang, C.G. Levi, J. Am. Ceram. Soc. 91 (2008) 576583.[9] R.M. Leckie, S. Krmer, M. Rhle, C.G. Levi, Acta Mater. 53 (2005) 32813292.

8 M.P. Schmitt et al. / Surface & Coatings Technology xxx (2014) xxxxxxthicker layer coatings. It has been observed that the nanolayer systemundergoes a phase change from a t/cubic GZO mixture to a singlecubic-uorite phase which is attributed to a chemical potential gradientand layer thickness issue. There was still signicant diffusion down thecolumn edges in the thick layer systemwhich indicates after long timesand high temperatures, even a relatively thick diffusion layer may notsufciently prevent GZO interaction with the TGO. This brings intoquestion the long term suitability of low-k YSZ as a diffusion barrierfor GZO and other coatings with similar composition. The stability andinteraction of GZO and YSZ/low-k type coatings will be examined ingreater detail in a forthcoming publication.Please cite this article as: M.P. Schmitt, et al., Surf. Coat. Technol. (2014), h[10] D.E. Wolfe, M.P. Schmitt, D. Zhu, A.K. Rai, R. Bhattacharya, in: D. Zhu, et al., (Eds.),Adv. Ceram. Coatings Mater. Extrem. Environ, ACERS, Daytona, FL, 2013, pp. 318.

[11] D. Zhu, R.A. Miller, B.A. Nagaraj, R.W. Bruce, Surf. Coat. Technol. 138 (2001) 18.[12] D. Zhu, R.A. Miller, J. Therm. Spray Technol. 9 (2000) 175180.[13] R.G. Wellman, J.R. Nicholls, J. Phys. D. Appl. Phys. 40 (2007) 293305.[14] R.D. Shannon, Acta Crystallogr. A32 (1976) 751767.[15] C. Mercer, J. Williams, D. Clarke, A. Evans, Proc. R. Soc. A 463 (2007) 659670.[16] R.G. Wellman, M.J. Deakin, J.R. Nicholls, Wear 258 (2005) 349356.[17] X.Q. Cao, R. Vassen, D. Stover, J. Eur. Ceram. Soc. 24 (2004) 110.[18] K.S. Lee, K.I. Jung, Y.S. Heo, T.W. Kim, Y.-G. Jung, U. Paik, J. Alloys Compd. 507 (2010)

448455.[19] D.E. Wolfe, J. Singh, R.A. Miller, J.I. Eldridge, D.-M. Zhu, Surf. Coat. Technol. 190

(2005) 132149.ttp://dx.doi.org/10.1016/j.surfcoat.2014.03.049

Multilayer thermal barrier coating (TBC) architectures utilizing rare earth doped YSZ and rare earth pyrochlores1. Introduction2. Experimental procedure3. Results and discussion3.1. Monolayer coatings3.2. Multilayer coatings3.3. Thermal stability of multilayer t low-k/GZO TBCs

4. Conclusions and future workConflict of interestAcknowledgmentsReferences