Surface & Coatings Technology 272 (2015) 350–356

Contents lists available at ScienceDirect

Surface & Coatings Technology

journal homepage: www.elsevier.com/locate/surfcoat

The thermal behavior of CMAS-infiltrated thermal barrier coatings

Tyler R. Kakuda a, Carlos G. Levi a,b, Ted D. Bennett a,⁎
a
Department of Mechanical Engineering, University of California, Santa Barbara, 93106-5070, United States
b
Materials Department, University of California, Santa Barbara, 93106-5070, United States

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

Article history: Understanding the mechanisms by which the durability and functionality of thermal barrier coatings (TBCs) are
Received 16 January 2015 compromised by the infiltration of molten calcium–magnesium alumino-silicates (CMAS) requires an assess-
Accepted in revised form 28 March 2015 ment of the effects on the thermal and mechanical properties of the coating. This study focuses on quantifying
Available online 3 April 2015
the effect of CMAS on the thermal properties and heat transport in TBCs. The thermal properties of a 7 wt.%
yttria-stabilized zirconia (7YSZ) TBC deposited on a superalloy substrate by air plasma spray (APS) were mea-
Keywords:
Calcium–magnesium alumino-silicates (CMAS)
sured before and after CMAS infiltration. A rise in both volumetric heat capacity and thermal conductivity of
Thermal barrier coatings (TBCs) the coating was observed upon infiltration. Calculations to explain these trends were performed for a model
Thermal properties TBC system and found to be in good agreement with the measured results. The evolution of the phase constitu-
tion of the coating was analyzed by Raman spectroscopy and the integrity of the interface was characterized by
optical examination of cross sections. These tests determined that the coating remained in good contact with the
substrate and experienced no phase change after infiltration.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction promising approach is the use of Gd zirconate (Gd2Zr2O7, GZO) in ther-

Among the mechanisms limiting the durability of thermal barrier
coatings (TBCs) identified to date [1], one that has generated consider-
able concern is the infiltration of molten calcium–magnesium
alumino-silicate (CMAS) deposits [2]. The net effect is the crystallization
of the molten material within the pore structure of the TBC, degrading
its in-plane compliance and increasing the potential for delamination
upon thermal cycling [3]. Because silicate deposits can melt at temper-
atures of ~1200 °C or even lower [4,5], which are on the same order as
the surface temperature of TBCs in current technology, CMAS infiltra-
tion becomes a fundamental barrier to further increases in turbine
inlet gas temperature [2], and hence in engine efficiency. In addition
to degrading the in-plane compliance of the coating the penetration
of CMAS can attack chemically the insulating oxide, inducing de-
stabilization of the t′ phase [4,6,7] desirable for toughness [8]. Moreover,
the thermal insulation ability of the coating is reduced as the coating
porosity is filled with solid phases [9], resulting in higher temperatures
at the interface with the metallic substrate, and accelerated oxidation
rates that can activate other modes of failure such as rumpling [10,11].
Changes in the thermal properties of the coating can also reveal
critical information about the coating condition and the extent of CMAS
infiltration. The latter is especially relevant since the accumulation of
elastic strain that can drive delaminations is related to the depth of infil-
tration [3]. Hence, limiting the extent of CMAS infiltration has been a key
driver in the search for advanced TBC designs [2,12–15]. A particularly
⁎ Corresponding author. Tel.: +1 805 893 8115; fax: +1 805 893 8651.
E-mail address: bennett@engr.ucsb.edu (T.D. Bennett).
mal barrier coatings [16], wherein a chemical reaction between molten
CMAS and GZO induces rapid crystallization of the melt and inhibits
infiltration [9]. If crystallization occurs rapidly within the outer pore
structure of the coating and subsequent reaction proceeds slowly, as in
GZO TBCs, the remaining issue is whether the thickness of the stiffened
layer is sufficiently small to avoid delamination [3].
Proper characterization of the thermal behavior of an infiltrated TBC
is essential in developing and validating models to predict the onset of
delamination and the associated fail-safe condition [2,3,17,18]. A corol-
lary benefit is a better understanding of the effect of CMAS on the overall
heat transport through the coating system and temperatures at critical
interfaces, notably those influencing the growth kinetics of the TGO.
Photon transport through infiltrated coatings has been investigated pre-
viously [19]. That study demonstrated that CMAS infiltration caused an
increase in radiation transport through the coating in the spectral range
of 0.5–2.5 μm. The large increase in transmission, coupled with the
higher gas temperatures in advanced engines where CMAS may be a
problem, suggests that radiative heat transport can be significant for in-
filtrated coatings in an operating engine. The present paper addresses
the complementary mechanism of phonon transport through a CMAS
impregnated system. The results from both studies can then provide a
better overall description of coating thermal performance after CMAS
penetration.
Phase of photothermal emission analysis (PopTea) [20], differential
scanning calorimetry (DSC) and density measurements are used in
this study to determine the thermal properties of a typical thermal bar-
rier coating deposited by air-plasma spray (APS) before and after CMAS
infiltration. The experimental procedure outlined in this report yields an

http://dx.doi.org/10.1016/j.surfcoat.2015.03.043
0257-8972/© 2015 Elsevier B.V. All rights reserved.
 T.R. Kakuda et al. / Surface & Coatings Technology 272 (2015) 350–356
infiltration of amorphous CMAS, which is determined not to alter the
chemical nature of the coating or lead to coating degradations. In real
systems, a typical infiltration yields crystallized CMAS caused by high
temperature exposure during engine cycles. In these systems, chemical
attack can alter the coatings' thermal behavior and is extremely destruc-
tive to the coating. Based on bulk properties measurements of both
crystallized and amorphous CMAS, the significance of this difference
can be estimated for the infiltrated coating system.
2. Experimental details
2.1. Thermal measurements
PopTea is used as the primary technique to measure thermal proper-
ties in this paper. The principle of PopTea is to determine coating ther-
mal properties by interrogating the phase of thermal emission from
the coating established during modulated heating. The coating is heated
with a CO2 laser (10.6 μm and spot size of 6 mm) and thermal emission
is measured with an InSb detector in the mid infrared (5 μm). PopTea
measurements are made at 70 °C, controlled by a resistively heated
stage. At this temperature there is no risk of inducing delaminations
within a TBC or changing the coating thermal properties. Since temper-
ature excursions within the coating caused by modulated heating are
small, heat transfer is dominated by conduction and the measurement
is not influenced by radiation through the coating. An analytical model
has been developed to calculate the transient phase of thermal emission
from the TBC as a function of unknown coating properties and known
geometric parameters [20]. This model is used to evaluate candidate
values of the coating properties from experimental measurements. A
single measurement consists of recording the phase of thermal emission
as a function of laser frequency. The range of laser frequencies is select-
ed to achieve a wide range of thermal penetration depths, to allow the
coating properties to be uniquely determined.
There are only two thermal parameters important to fit results from
the PopTea method [21]:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
i ffi
rffiffiffiffiffiffiffiffiffiffi uh
u kρC p
α sub u full infiltration of the coating porosity without evidence of YSZ decom-
asub ¼ γ ¼ th i sub :
α coat kρC p
coat
Both parameters, asub and γ, reflect the contrast between the two
key thermal properties of the coating and substrate, i.e. the thermal
pffiffiffiffiffiffiffiffiffiffiffi
diffusivity (α) and the thermal effusivity ( kρC p ).
While the capabilities of PopTea are well demonstrated, several
complementary thermal measurements were used in this study to
demonstrate the reliability of the results. Independent measurements
of specific heat and density were made for comparison with PopTea.
Specific heat values were determined by differential scanning calorime-
try (DSC, Netzsch 404 C Pegasus). DSC measures specific heat by
comparing the amount of heat necessary to raise the temperature of a
sample and a known reference. The samples and reference used for
these measurements were cylindrically polished pellets 3 mm in diam-
eter and ~50 mg in weight. Sapphire was used for the reference and has
a well-defined specific heat over the temperature range scanned. To
make comparisons between specific heat (Cp) measured by DSC and
volumetric heat capacity (ρCp) measured by PopTea, the density (ρ) of
each sample was measured using Archimedes principle with water as
the fluid.
2.2. Specimen preparation and characterization
The sample used for CMAS infiltration was a 7 wt.% yttria-stabilized
zirconia (7YSZ) coating deposited by air plasma spray on a Rene 80
Nickel-base superalloy (provided by Siemens Power, Inc.). The coating
thickness was 300 μm, which is representative of TBCs deposited on
351
turbine airfoils used in power generation. Measurements were made
on this coating for three different coating conditions, as seen in Fig. 1.
The first measurement (1) was made before infiltration to acquire base-
line thermal properties needed for a pre/post infiltration comparison,
and to determine the pristine coating density (ρcoat) and corresponding
pore volume fraction (υcoat = 0.27). Part of the coating surface was then
infiltrated locally with CMAS in a high temperature furnace. A second
measurement (2) was made on an un-infiltrated area of the coating to
assess possible changes in thermal properties from the as-deposited
condition due to temperature effects alone. The final measurement
(3) was made on an area of the coating experiencing the full CMAS
infiltration.
In addition to the coating measurements, dense amorphous and
crystallized CMAS pellets were thermally characterized to help under-
stand the changes caused by coating infiltration. Both CMAS pellets
used for characterization were made from CMAS powder with the com-
position shown in Table 1. The pellets were formed by pressing the
powder into flat discs and melting them on platinum foil in a high
temperature controlled furnace. Amorphous CMAS was generated by
rapidly ramping up and down (~ 10 °C/min) from the melting point
(~ 1250 °C). To create crystallized CMAS from amorphous stock, the
pellet was held within the crystallization range previously determined
by DSC (~1080 °C) for 6 h and slowly cooled at 2 °C/min.
The amorphous pellet was used for infiltration of the TBC and
polished into a 1 cm diameter disc, 80 μm thick. The pellet thickness
was dictated by requirements for full infiltration of the coating (h =
L × υcoat) where L is the coating thickness and the void fraction υcoat.
The coating thickness was measured from micrographs of the TBC
cross-section and the void fraction was determined from measurements
on the as-deposited, un-infiltrated coating. To infiltrate the coating, the
amorphous CMAS pellet was placed at the center of the coupon and
heated uniformly in a temperature controlled furnace to 1250 °C at
10 °C/min followed by immediate cooling at 10 °C/min to 1000 °C. To
prevent thermal shock or “desktop spallation” [22] of the infiltrated
coating, the sample was cooled slowly (2 °C/min) for temperatures
below 1000 °C in order to allow the glass stresses, induced by the
thermal expansion mismatch, to relax [23]. This treatment permitted
ð1Þ
position or coating delamination, as shown below.
The integrity of the coating microstructure and bond coat is revealed
in the optical micrographs of the coating sectioned after the measure-
ments, shown in Fig. 2. The infiltrated region appears to be in perfect
contact with the substrate (i.e. no spallation, or coating separation). Fur-
thermore, the condition of the infiltrated coating is uniform and without
cracks. The extent of the infiltrated CMAS can be readily identified by a
contrast in the image between the infiltrated and uninfiltrated regions,
created by differences in polishing properties of the two regions.
Fig. 1. Top view of TBC before and after CMAS infiltration. Locations of PopTea measure-
ments are highlighted. (1) Baseline, before infiltration; (2) uninfiltrated region after
high temperature exposure; (3) infiltrated region of coating.
352 T.R. Kakuda et al. / Surface & Coatings Technology 272 (2015) 350–356

Table 1
Molar composition of CMAS and calculated specific heat at 70 °C.

Materials CaO MgO Al2O3 SiO2

mol% 35 10 7 48
C p@70  C (J/kg/K) [21] 806.9 1000.9 864.1 815.4
C pCMAS (J/kg/K) (est.) 835

Although there appears to be a small amount of CMAS left on the

surface, the height is less than the surface roughness of the APS. Visual
inspection of the micrograph in the un-infiltrated region reveals that
the coating has detached from the substrate on the right-hand side of
the coating. However, this occurred during sectioning and was not the
result of infiltration.
Raman spectroscopy of the coating cross-section, shown in Fig. 3,
reveals that the TBC retains the tetragonal-prime structure in the infil-
trated region. The relative width and height of the six Raman bands in
the spectrum remain unchanged, and there is no evidence of the charac-
teristic monoclinic peaks at ~200 cm−1 after infiltration. These observa-
tions suggest that any de-stabilization of YSZ by reaction with CMAS [7]
is sufficiently limited to obviate deleterious stresses ensuing from the
tetragonal-monoclinic transformation. The combination of optical
micrograph inspection and Raman spectroscopy results shows that the
full CMAS infiltration was successful and could be accomplished in a
time sufficiently short to avoid causing a detectable phase change or
delamination of the TBC.
3. Thermal property results
3.1. CMAS pellets
Thermal properties of CMAS with compositions representative of
those seen in the engine environment are measured here. Since PopTea
requires a contrast in thermal properties between a coating and sub-
strate materials, the CMAS pellets were bonded to a Ni-base superalloy
substrate (IN738). A colloidal silver suspension was used as the bonding
agent due to its high thermal conductivity, and care was taken to avoid
developing voids at the interface to minimize thermal contact resis-
tance. In this manner, the thermal conductivity and volumetric heat ca-
pacity of the amorphous and crystalline CMAS pellets were determined
Fig. 3. Raman spectroscopy curves taken on the infiltrated and uninfiltrated regions of the
coating. The similarity between the peak locations and peak widths shows that the condi-
tion and composition of the YSZ are essentially the same between the two regions.
from PopTea measurements. There is a notable shift in data between the
raw phase curves of the crystallized and amorphous CMAS measure-
ments shown in Fig. 4. Most of this shift is caused by thickness differ-
ences between the samples, but the observed change in slope is
related to the change in thermal properties. Table 2 reveals that the
crystalline CMAS has a higher thermal conductivity that is ~ 2.1 times
higher than its amorphous counterpart. However, there is less than a
10% rise in the density and specific heat resulting from crystallization
of the CMAS.
To validate the values of volumetric heat capacity made by PopTea,
independent measurements of the specific heat were made by DSC
and densities were determined by the Archimedes method. Fig. 5
shows the results of DSC on the amorphous and crystalline CMAS
pellets. The similar volumetric heat capacity of crystalline CMAS
(875 J/kg/m3) and amorphous CMAS (900 J/kg/m3) suggests that
important vibrational modes are insensitive to the structure of the
solid. Therefore, it is reasonable to assume that C pCMAS is independent
of the CMAS phase at temperatures concerned with the measurement.

Fig. 2. Optical micrographs of the cross section of the TBC coupon after infiltration. The im-
ages show two distinct regions: infiltrated region (1 cm over the coupon center) and
uninfiltrated regions (coupon sides). The difference in contrast (as seen in transition re-
gion) is the result of polished discrepancies between the two coating conditions. Images
also show that the coating and bondcoat interface are still intact under the regions of
the measurements. Fig. 4. Phase of emission for an amorphous CMAS pellet and crystallized CMAS pellet.
 T.R. Kakuda et al. / Surface & Coatings Technology 272 (2015) 350–356 353

Table 2 large change in thermal properties of the region infiltrated by CMAS,

Thermal properties of dense: amorphous CMAS, crystallized CMAS and 7YSZ. observed in measurement (3). The measured phase of emission, as a
CMAS Dense 7YSZ function of heating frequency, is plotted in Fig. 6 for the three different
Amorphous Crystallized Δ
measurements to highlight the dramatic effect of infiltration on the raw
data.
α (m2/s) 4.01 × 10−7⁎ 8.10 × 10−7⁎ 102% 8.8 × 10−7
A visual comparison of the thermal conductivity and volumetric heat
[29,30,35]
k (W/m/K) 0.83⁎ 1.78⁎ 114% 2.5 [35] capacity for the three states of the coating is provided in Fig. 7. The
ρCp (J/m3/K) 2.07 × 106⁎ 2.20 × 106⁎ 6% 2.83 [29,30] “error” bars represent the uncertainty in the fit of thermal properties,
Cp (J/kg/K) 900⁎⁎ 875⁎⁎ 3% 475 [30,29] as determined by investigating the sensitivity of the least-square error
@ 70 °C to forced departures of asub from the best-fit value. Details of the calcu-
ρ (kg/m3) 2300⁎ and 2450⁎⁎⁎ 2440⁎and 2635⁎⁎⁎ 6%–8% 5950 [29]
average = 2375 average = 2540
lated uncertainty are covered in Ref. [26]. Fig. 7 illustrates a much more
substantial rise in the thermal conductivity (~110%) than in the volu-
⁎ PopTea measurement (ρCp/Cp).
⁎⁎ DSC measurement.
metric heat capacity (~ 18%) when the coating is infiltrated with
⁎⁎⁎ Archimedes principle. CMAS. From an initial conductivity of 0.9 W/m/K, CMAS infiltration
causes the coating conductivity to increase to 2.1 W/m/K, a value
exceeding 80% of that reported for dense 7YSZ.
Additionally, measurement results are within 7% of the calculated spe- DSC and density measurements were also performed on a fragment
cific heat value that is based on the known CMAS composition and the of the infiltrated coating material taken off the substrate after PopTea
specific heats of the major components provided by Barin and Knacke measurements were performed. The nominal specific heat of the infil-
[24] at 70 °C. trated coating measured by DSC is shown in Fig. 5 (C pinfilt = 513 J/kg/K).
The densities of the CMAS pellets measured by the Archimedes From the Archimedes principle, the density of the infiltrated fragment
method (ρCMAS) were determined to be 2440 kg/m3 for amorphous was measured to be ρinfilt = 4800 kg/m3. Therefore, the volumetric heat
and 2635 kg/m3 for crystallized CMAS. Using the values of density and capacity measured by independent means is ρC pinfilt = 2.5 × 106 J/m3/K
the specific heat obtained through DSC, the volumetric heat capacity which is essentially the same as that determined from PopTea measure-
of the CMAS pellets (ρC pCMAS ) are 2.2 × 106 J/m3/K (amorphous) and ments (Table 3). Baseline measurements of the uninfiltrated coating
2.4 × 106 J/m3/K (crystalline). These values are within 10% of those de- heat capacity and density could not be made in a non-destructive fashion
termined by PopTea (Table 2), which is within the typical experimental and the coating needed to be preserved for infiltration. For this reason,
error, providing confidence in the fidelity of the PopTea derived values density was determined by dividing the diffusivity (PopTea) by specific
for the volumetric heat capacity. heat values for a 7YSZ APS coating taken from literature [27]. These values
are reported in Table 3.

3.2. Coating measurements 4. Discussion

Thermal property measurements of the TBC for the as-deposited and
CMAS infiltrated conditions are summarized in Table 3. Only a small
change is observed in the thermal properties of the uninfiltrated coating
between measurements made before (1) and after (2) heat treatment,
consistent with the short exposure at temperatures where sintering be-
comes active [25]. Since prior to heat treatment the thermal properties
at locations (1) and (2) were indistinguishable, it is inferred that the
high temperature treatment alone does not alter the thermal character-
istics of the coating at the center of the specimen. However, there is a
To understand the changes in thermal properties of the infiltrated
coating it is advantageous to compare the experimental results to calcu-
lations of thermal properties for a model TBC system based on effective
medium theory.
4.1. Verification and calculation of infiltrated specific heat and density
The specific heat of an infiltrated coating can be predicted from the
specific heat, volume fraction, and density of the constituent parts
[28]. For this purpose, the specific heat and density of bulk 7YSZ were
taken from the literature [29,30] and measured for the amorphous
CMAS pellet as described in Section 3.1. Properties of both are summa-
rized in Table 2. For the infiltrated coating, the specific heat can be cal-
culated from:

C pinfilt ¼ C p7YSz ω7YSZ þ C pCMAS ωCMAS ð2Þ

where, ωi is the mass fraction of the ith constituent:

X2
ωi ¼ f i ρi = f j ρi ð3Þ
j¼1

Fig. 5. Specific heat results from differential scanning calorimetry (DSC) for amorphous/
crystalline CMAS pellets and a fragment of infiltrated coating. PopTea measurements
were made at 70 °C (vertical dashed line).
fi and ρi are the volume fraction and density of the ith constituent,
respectively.
It is assumed that the void fraction of the coating before infiltration
υcoat becomes the volume fraction filled by CMAS, fCMAS = υcoat =
0.27. The volume fraction of 7YSZ in the coating is f7YSZ = 1 − υcoat =
0.73. Using the specific heat and density values for amorphous CMAS re-
ported in Table 2 (C pCMAS = 900 J/kg/K and ρCMAS = 2375 kg/m3), and
using the specific heat and density of dense-7YSZ (C p7YSZ = 475 J/kg/K
and ρ7YSZ = 5950 kg/m3, Table 2), the specific heat of the CMAS infiltrat-
ed coating is calculated from Eqs. (2) and (3) to be 530 J/kg/K. This
354 T.R. Kakuda et al. / Surface & Coatings Technology 272 (2015) 350–356

Table 3
Fitting results for 3 measurements made on a TBC coupon with PopTea. The infiltrated thermal properties are compared to theoretically calculated values based on the fitting results of
measurement.

Measurement (1) Measurement (2) Measurement (3) Infiltrated calculations Δ

Baseline Annealed 1300 °C Infiltrated

α (m2/s) 4.46 × 10−7 4.54 × 10−7 8.46 × 10−7 7.71 × 10−7 9%

8.13 × 10−7 4%
k (W/m/K) 0.91 0.88 2.09 1.94 (Maxwell) 8%
2.04 (Rayleigh) 3%
3 6 6
ρCp (J/m /K) 2.04 × 10 1.94 × 10 2.47 × 106 (530)(4985) = 2.64 × 106 6%
Cp (J/kg/K) @ 70 °C 475 [30] 475 [30] 513⁎⁎ 530 b4%
ρ (kg/m3) 4295⁎ 4084⁎ 4800⁎⁎⁎ 4985 b4%
υcoat 27% 31%
⁎ PopTea measurement (ρCp / Cp).
⁎⁎ DSC measurement (sample fragment).
⁎⁎⁎ Archimedes principle (sample fragment).

prediction agrees to within 4% of the DSC measured value of 513 J/kg/K
(Table 3, measurement 3).
Assuming that the volume of the coating is unchanged after infiltra-
tion (i.e. no swelling) then the density of the CMAS infiltrated coating
can be calculated by:
ρinfilt ¼ f 7YSZ ρ7YSZ þ f CMAS ρCMAS :
Using f7YSZ = 0.73, ρ7YSZ = 5950 kg/m3, fCMAS = 0.27 and ρCMAS =
2375 kg/m3 the density of the infiltrated coating (ρinfilt) is determined
to be 4985 kg/m3. For comparison, the density of the infiltrated coating
can be calculated from the PopTea measurement of volumetric heat
capacity using the DSC measured value of specific heat (513 J/kg/K).
By this method, the infiltrated density of the coating is ρinfilt =
4800 kg/m3, which is in close agreement (4%) of the value calculated
based on the initial pore volume fraction (fcoat) of the coating.
The change in density from the initial coating condition is larger than
the change in specific heat. Therefore, these calculations suggest that
density accounts for the shift in volumetric heat capacity of the fully
infiltrated coating (C pinfilt = (4985)(530) = 2.64 × 106 J/m3/K) from
the initial volumetric heat capacity value.
4.2. Infiltrated coating thermal conductivity
Due to the complexity of coating morphology, predicting the ther-
mal conductivity of a coating from first principles is difficult. Composi-
tion and morphology (a description of pore fraction, distribution and
shape) are important characteristics to an effective media model of
the coating [31]. Many studies have been performed in the past to calcu-
late the effective properties of a composite material and these vary in
degree of complexity [32]. Two classical and relatively simple ap-
proaches, developed by Maxwell [33] and Rayleigh [34] are employed
in this study to provide a rough estimate of the thermal conductivity
ð4Þ of a CMAS infiltrated coating.
In Maxwell's analysis [33], randomly placed inclusions are consid-
ered to be spherical and non-interacting (i.e. inclusions are sufficiently
separated such that heat transfer in the proximity of one inclusion
does not influence another). For this system, the effective conductivity
(kinfilt) determined by Maxwell's model is given by:
  −1 
γþ2
kinfilt ¼ k7YSZ 1 þ 3f CMAS −f CMAS ð5Þ
γ−2
where fCMAS is the fraction of CMAS as defined earlier, and γ = kCMAS/k7YSZ
is the ratio of conductivity between amorphous CMAS and bulk 7YSZ.
Using Eq. (5), with γ = 0.83/2.5 = 0.33, to calculate the conductivity
of the infiltrated coating yields, kinfilt (Maxwell) = 1.95 W/m/K. This
calculated thermal conductivity is within 10% of the measured con-
ductivity of the coating reported in Table 3 (Measurement 3), kinfilt
(PopTea) = 2.09 W/m/K.
Despite the relatively small difference between Maxwell's result and
the PopTea measurement, closer agreement can be obtained by using an
effective model developed by Rayleigh [34]. In his cylindrical model,

Fig. 7. Comparison between the uninfiltrated and infiltrated thermal conductivity and
Fig. 6. Phase of emission data for the 3 measurement locations in Fig. 1 and corresponding volumetric heat capacity results. The sharp rise in thermal conductivity is contrasted by
PopTea fits. a small rise in the volumetric heat capacity. Measurement locations as given in Fig. 1.
 T.R. Kakuda et al. / Surface & Coatings Technology 272 (2015) 350–356
Rayleigh assumes that cylindrical shaped pores are situated on the
corners of a square lattice. For this system, the effective thermal conduc-
tivity along the axis of the pores is determined from Rayleigh's model to
be:
kinfilt ¼ k7YSZ ð1 þ f CMAS ðγ−1ÞÞ: ð6Þ
Using Eq. (6) gives an effective conductivity of kinfilt (Rayleigh) =
2.05 W/m/K, which is still slightly lower than the measured value of
2.09 W/m/K but in better agreement than Maxwell's model.
The effective thermal conductivity kinfilt is plotted in Fig. 8 as a func-
tion of initial coating porosity for the two different effective media
models of Maxwell and Rayleigh. The fact that the cylindrical model
more closely describes the measured thermal conductivity suggests
that the CMAS vertical connectivity is an important characteristic of
the infiltrated system. As CMAS seeps through the coating porosity it
is realistic to assume that vertical connectivity is established throughout
the coating thickness as opposed to the CMAS conglomerating into
isolated inclusions.
The effective media model, as applied, assumes that CMAS is fully
infiltrated into the porous structure of the TBC. On a large scale, this
assertion is supported by the optical micrographs shown in Fig. 2. On
a small scale, evidence that the porous structure is completely filled
with CMAS is obtained from the calorimetry data. The PopTea measure-
ment provides the volumetric heat capacity (ρCp) of the coating, and
through this measurement the original porosity of the TBC was deter-
mined from an independent knowledge of both the specific heat and
the fully dense specific volume of 7YSZ. With separate measurements
of the specific heat and specific volume of amorphous CMAS, the volu-
metric heat capacity of the TBC in the fully infiltrated CMAS state can
be predicted from the effective media model. Agreement between this
result and the independent PopTea measurement of volumetric heat
capacity of the infiltrated TBC confirms that the porosity of the coating
is completely filled with CMAS.
As stated in the Introduction, infiltrations of CMAS in real engine sys-
tems can appear in crystalline form [36]. Using the effective medium
theory it is possible to estimate the thermal properties of a coating infil-
trated by crystalline CMAS, to contrast with the amorphous state that
has been measured directly.
The thermal properties of bulk crystallized CMAS were measured by
PopTea and are summarized in Table 2. Using Eq. (3) with the density of
crystallized CMAS (ρCMAS = 2540 kg/m3, Table 2), it is found that ρinfilt =
5065 kg/m3. Therefore, the volumetric heat capacity of the infiltrated
Fig. 8. Plot of the effective thermal conductivity as a function of second phase volume
fraction using Maxwell and Rayleigh effective media models. All the calculations are
below the measured value of conductivity for the fully infiltrated coating.
355
coating under crystalline conditions is ρC Pinfilt = 2.67 × 106 J/m3/K, this
is only an 8% increase seen from the amorphous infiltration.
Using Eq. (6) with the conductivity of crystallized CMAS (kCMAS =
1.78 W/m/K) the conductivity of the infiltrated coating is predicted to
be kinfilt = 2.31 W/m/K. This is a 13% increase from the case of amor-
phous CMAS infiltration, and is a 114% higher thermal conductivity
than the initial coating value.
5. Conclusion
This study revealed that the phase of photothermal emission analy-
sis (PopTea) can be successfully applied to characterize the thermal
properties of APS TBCs fully infiltrated with CMAS while attached to
the superalloy substrate. The largest change caused by CMAS infiltration
was an increase of the thermal conductivity of the coating by a factor of
~2.3. Measured changes in thermal conductivity and density compared
well to model calculations that assumed that the initial pore volume
fraction of the coating was completely infiltrated with CMAS. PopTea
measurements revealed that crystallization has only a marginal effect
on the volumetric heat capacity of the CMAS constituent, but increases
its thermal conductivity by a factor of ~ 2.1. Experimental measure-
ments were shown to agree more closely with calculations based on
the Rayleigh/Maxwell model and amorphous/crystalline CMAS. In ser-
vice conditions, infiltration is accompanied by the crystallization of
CMAS, which is measured to have a 13% higher thermal conductivity
than the amorphous state. This factor is predicted to increase the rise
in thermal conductivity of the coating by 114%. The insight is particular-
ly relevant because in practice the TBCs can be found to contain both
crystalline and amorphous CMAS.
Demonstrating the ability to measure changes in thermal properties
of an infiltrated coating is the initial step for quantitative studies on
practical TBC systems that have experienced CMAS attack. PopTea is
particularly promising in this regard because it has demonstrated the
possibility for measuring the thermal properties directly on coated air-
foils without the need to section the component. One can then envision
mapping of the conductivity as an indicator of health and remaining
functionality of the coating. Further investigation is needed to study
the change in thermal properties of partially infiltrated coatings,
which can arise when CMAS is introduced to the coating in the presence
of a thermal gradient. This situation poses new challenges for the
PopTea measurement, since the coating becomes a layered system
with an unknown extent of CMAS infiltration.
Acknowledgments
This work was supported by the University Turbine Research Pro-
gram (UTSR), SCIES Project 07-01-SR125. Support for C.G. Levi was pro-
vided by the Office of Naval Research under Grant N00014-08-1-0522,
monitored by Dr. David Shifler. The authors would like to thank Dr.
Andi Limarga for his insights and help with Raman spectroscopy, Dr.
Stephan Krämer for his advice and assistance with the microstructural
images, and Elisa Zaleski and Raymond Valdes for their assistance in
creating crystallized CMAS samples. TBC coupons used in this work
were kindly provided by Dr. Anand Kulkarni and Dr. Anirudha Vaidya
from Siemens Energy, Inc.
References
[1] A.G. Evans, D.R. Mumm, J.W. Hutchinson, G.H. Meier, F.S. Pettit, Prog. Mater. Sci. 46
(5) (2001) 505–553.
[2] C.G. Levi, J.W. Hutchinson, M.-H. Vidal-Setif, C.A. Johnson, MRS Bull. 37 (10) (2012)
932–941.
[3] A.G. Evans, J.W. Hutchinson, Surf. Coat. Technol. 201 (18) (2007) 7905–7916.
[4] M.P. Borom, C.A. Johnson, L.A. Peluso, Surf. Coat. Technol. 86 (1) (1996) 116–126.
[5] E.M. Zaleski, C. Ensslen, C.G. Levi, Melting and Crystallization of Silicate Systems Rel-
evant to Thermal Barrier Coating Damage, J. Am. Ceram. Soc (1–8) (2015).
[6] D.J. de Wet, R. Taylor, F.H. Stott, J. Phys. IV 3 (C9) (1993) 655–663.
356 T.R. Kakuda et al. / Surface & Coatings Technology 272 (2015) 350–356
[7] S. Krämer, J.Y. Yang, C.A. Johnson, C.G. Levi, J. Am. Ceram. Soc. 89 (10) (2006)
3167–3175.
[8] A.G. Evans, D.R. Clarke, C.G. Levi, J. Eur. Ceram. Soc. 28 (7) (2008) 1405–1419.
[9] S. Krämer, J. Yang, C.G. Levi, J. Am. Ceram. Soc. 91 (2) (2008) 576–583.
[10] D.R. Mumm, A.G. Evans, I.T. Spitsberg, Acta Mater. 49 (12) (2001) 2329–2340.
[11] V.K. Tolpygo, D.R. Clarke, Acta Mater. 52 (2004) 5115–5127;
Acta Mater. 52 (2004) 5129–5141.
[12] W. C. Hasz, M. P. Borom, and C. A. Johnson, “Protection of thermal barrier coating
with an impermeable barrier coating”, U.S. Patent 5,871,820, (1999).
[13] R. Darolia and B. A. Nagaraj, “Method of forming a coating resistant to deposits and
coating formed thereby”, U.S. Patent 6,720,038, (2004).
[14] R. Darolia, B.A. Nagaraj, D.G. Konitzer, M.D. Gorman, and F. Ming, “Layered thermal
barrier coatings containing lanthanide series oxides for improved resistance to
CMAS degradation”, U.S. Patent Application 200770160859-A1, (2007).
[15] A. Aygun, A.L. Vasiliev, N.P. Padture, X. Ma, Acta Mater. 55 (20) (2007) 6734–6745.
[16] M.J. Maloney, “Thermal barrier coating systems and materials”, U.S. Patent
6,177,200, (2001).
[17] A.G. Evans, J.W. Hutchinson, Surf. Coat. Technol. 201 (18) (2007) 7905–7916.
[18] S. Sundaram, D.M. Lipkin, C.A. Johnson, J.W. Hutchinson, J. Appl. Mech. 80 (2013)
011002.
[19] L. Li, D.R. Clarke, Int. J. Appl. Ceram. Technol. 5 (3) (2008) 278–288.
[20] T.D. Bennett, T.R. Kakuda, A. Kulkarni, J. Appl. Phys. 105 (8) (2009) 83501–83506.
[21] F. Yu, T.D. Bennett, J. Appl. Phys. 98 (10) (2005) 103501–103505.
[22] J.L. Smialek, D. Zhu, M.D. Cuy, Scr. Mater. 59 (1) (2008) 67–70.
[23] G.W. Scherer, Relaxation in Glass and Composites, Wiley-Interscience, New York,
1986.
[24] I. Barin, O. Knacke, Thermochemical Properties of Inorganic Substances, Springer-
Verlag, Berlin Heidelberg, New York, 1973.
[25] S.R. Choi, D. Zhu, R.A. Miller, J. Am. Ceram. Soc. 88 (10) (2005) 2859–2867.
[26] T.R. Kakuda, A.M. Limarga, D.R. Clarke, T.D. Bennett, Acta Mater. 57 (8) (2009)
2583–2591.
[27] H. Wang, R.B. Dinwiddie, Adv. Eng. Mater. 3 (7) (2001) 465–468.
[28] S.G. Zhirov, J. Eng. Phys. Thermophys. 31 (3) (1976) 1058–1060.
[29] X.G. Capdevila, J. Folch, A. Calleja, J. Llorens, M. Segarra, F. Espiell, J.R. Morante,
Ceram. Int. 35 (2009) 1219–1226.
[30] S. Raghavan, H. Wang, R.B. Dinwiddie, W.D. Porter, M.J. Mayo, Scr. Mater. 39 (8)
(1998) 1119–1125.
[31] F. Cernuschi, P. Bison, A. Moscatelli, Acta Mater. 57 (12) (2009) 3460–3471.
[32] F.C. Chen, C.L. Choy, K. Young, J. Phys. D. Appl. Phys. 10 (4) (1977) 571–586.
[33] J.C. Maxwell, A Treatise on Electricity and Magnetism, Clarendon Press, Oxford,
1892. 194–216.
[34] Lord Rayleigh, Philos. Mag. 34 (481) (1892).
[35] A. Kyongjun, S. Kakkaveri, S. Ravichandran, E. Rollie, S.L. Semiatin, J. Am. Ceram. Soc.
82 (2) (1999) 399–406.
[36] S. Krämer, S. Faulhaber, M. Chambers, D.R. Clarke, C.G. Levi, J.W. Hutchinson, A.G.
Evans, Mater. Sci. Eng. A 490 (2008) 26–35.