Date post: | 13-Apr-2017 |
Category: |
Documents |
Upload: | molly-m-gentleman |
View: | 491 times |
Download: | 8 times |
UNIVERSITY OF CALIFORNIA
Santa Barbara
High Temperature Sensing of Thermal Barrier Materials by Luminescence
A Dissertation submitted in partial satisfaction of the
Requirements for the degree Doctor of Philosophy
In Materials
By
Molly Maureen Gentleman
Committee in charge:
Professor David R. Clarke, Chair
Professor Ted Bennett
Professor Anthony K. Cheetham
Professor Anthony G. Evans
Professor Carlos G. Levi
September 2006
ii
The dissertation of Molly Maureen Gentleman is approved.
___________________________________________ Ted Bennett
___________________________________________ Anthony K. Cheetham
___________________________________________ Anthony G. Evans
___________________________________________ Carlos G. Levi
___________________________________________ David R. Clarke, committee chair
August 2006
iii
ACKNOWLEDGEMENTS
It is a privilege to acknowledge the many people who have contributed to
my graduate career.
First, I would like to thank my sister Eileen. She and I spent many years
hammering away in our Grandpa’s basement confident that we could construct
radios out of the junk that was lying around his workspace. Upon graduating from
junior-high, she embarked on a new academic journey that shaped both of our lives.
Eileen attended the Illinois Mathematics and Science Academy and soon learned
that advanced education was truly accessible to anyone regardless of their
background. Upon completion of high school, she went off to Tulane to pursue a
degree in biomedical engineering and her plan to get me to do the same. During
my senior year of high school, she demanded that I apply for college and get a
degree in engineering. It sounded like a good idea to me, and so I did.
Eileen graduated from Tulane summa cum laude as the first woman in my
family to graduate from college and went on to pursue a masters and doctorate
degree in her field. As I approached my college graduation, she again insisted that
I go to graduate school (only really good ones) and get an advanced degree. That
brings me here. Over the past five years, Eileen has been my primary source of
support for everything from the best way to approach an experiment to when and
where to meet her for vacation. I could not have made it here without her.
iv
I must also thank the many friends I have met here in Santa Barbara. The
Clarkers are a particularly wonderful group of people who I feel privileged to have
met and worked with. I would like to thank Vladimir Tolpygo specifically as he
has contributed to my development as a researcher more than any other member of
the group has. There has also been a large number of staff here at the university
who I would have been pained to do without. In particular, I would like to thank
Deryck Stave who has provided technical experience support that I have not seen
matched elsewhere.
Finally, I would like to thank my advisor David Clarke and the rest of my
committee. All of them have been a great source of technical and scientific support
over the last five years.
v
MOLLY M. GENTLEMAN
Materials Department Bldg. 503 Rm 1355 University of California, Santa Barbara Santa Barbara, CA 93106-5050 (805) 893-7509 [email protected] EDUCATION
Ph.D., Materials, University of California, Santa Barbara. Degree expected Summer 2006. Dissertation: High Temperature Sensing of Thermal Barrier Materials by Luminescence Advisor: Dr. D.R. Clarke
Bachelor of Science, Metallurgical and Materials Engineering with honors, Illinois Institute of Technology, August 1998-May 2001.
AREAS OF SPECIALIZATION
Luminescence spectroscopy, optical characterization of materials, high temperature luminescence, phase characterization, Raman spectroscopy, thermal barrier coatings.
RESEARCH EXPERIENCE
Doctoral Research, UCSB, 08/2001-present. Conceived and developed novel temperature sensing technique for
the measurement of temperature of embedded layers within a thermal barrier coating. Analysis of crystal structure/luminescence relationship as well as concentration/luminescence relationship for yttria stabilized zirconia and pyrochlore zirconate materials.
Dr. D.R. Clarke, Materials Department
Research Assistant, Argonne National Laboratory, 03/2000-08/2001. Evaluation of the relationship between acoustic emission and
mechanical measurements of elastic modulus for ceramic matrix composites with a focus on development of a non-destructive technique for measuring accumulated damage. Dr. W.A. Ellingson, Energy Technology Division.
TEACHING EXPERIENCE
Teaching Assistant, Introduction to materials, UCSB 01/2002-03/2002. Prepared questions and homework solutions. Prepared and lead
recitation lecture weekly.
vi
PUBLICATIONS Gentleman MM, Eldridge JI, Zhu DM, Murphy KS, and Clarke DR. In Press. Non-contact sensing of TBC/BC interface temperature in a thermal gradient. Surf. Coat. Technol. Gentleman MM, Lughi V, Nychka JA, and Clarke DR. 2006. Non-contact methods for measuring thermal barrier coating temperatures. Int. J. Appl. Ceram. Technol. 3 [2] 105-112. Gentleman MM and Clarke DR. 2005. Luminescence sensing of temperature in pyrochlore zirconate materials for thermal barrier coatings. Surf. Coat. Technol. 200: 1264-1269. Gentleman MM and Clarke DR. 2004. Concepts for luminescence sensing of thermal barrier coatings. Surf. Coat. Technol. 188-189: 93-100. Clarke DR, Tolpygo VK, Gentleman M. 2004. Luminescence-based characterization of protective oxides: from failure mechanisms to non-destructive evaluation. Mat. Sci. Forum 461-464: 621-629.
ABSTRACTS AND PRESENTATIONS Gentleman MM, Eldridge JI, Zhu DM, Murphy KS, and Clarke DR. 05/2006. Non-contact sensing of TBC/BC interface temperature in a thermal gradient. International conference on metallurgical coatings and thin films, San Diego, CA. Gentleman MM, Eldridge JI, Zhu DM, Clarke DR, and Murphy KS. 01/2006. Novel, direct sensing of the TBC/BC interface temperature in a thermal gradient (poster). MURI/TBC meeting, Santa Barbara, CA. Gentleman MM, Eldridge JI, and Clarke DR. 05/2005. Luminescence sensing of temperature in pyrochlore zirconate materials for thermal barrier coatings. International conference on metallurgical coatings and thin films, San Diego, CA. Gentleman MM and Clarke DR. 01/2005. Non-contact temperature measurements by Eu luminescence (poster). MURI/TBC meeting Santa Barbara, CA. Gentleman MM and Clarke DR. 12/2004. Non-contact temperature measurements by luminescence. International symposium of research students on materials science and engineering, Madras, Chennai, India.
vii
Gentleman, MM and Clarke, DR. 04/2004. Concepts for luminescence sensing of thermal barrier coatings. International conference on metallurgical coatings and thin films, San Diego, CA. Gentleman M, Sandberg D, Deemer C, Ellingson WA, and Todd J. 01/2002. Correlation of destructive and non-destructive elastic modulus measurement methods in ceramic composites. The American Ceramic Society Conference and Exposition on Advanced Ceramics and Composites, Cocoa Beach, FL.
HONORS Recipient, Madras Metallurgical Services, Best Technical Presentation Prize, International symposium of research students on materials science and engineering, Madras, Chennai, India. 12/2004. Recipient, Bunshah Award for Best Paper in Symposium A, International conference on metallurgical coatings and thin films, San Diego, CA. 04/2004. Recipient, NSF IGERT fellowship on Advanced Optical Materials. 12/2001-12/2003. Recipient, The Engineering Ceramics Division, student technical presentation First Place, The American Ceramic Society Conference and Exposition on Advanced Ceramics and Composites, Cocoa Beach, FL. 01/2002. Recipient, CAMRAS/NEXT full-tuition scholarship, Illinois Institute of Technology, Chicago, IL. 08/1998-05/2001.
viii
ABSTRACT
The basis for a number of concepts for in situ monitoring of thermal barrier
coating heath and temperature using non-contact luminescence techniques is
described. Restrictions imposed by phase compatibility and the optical properties
of the coating materials are discussed as guidelines for the selection of rare-earth
ions as chromophore dopants in these sensors. Extensive studies of type and
concentration of rare-earth doping lead to the selection of Eu-doped materials as the
preferred material for temperature sensing. Sensor layers, consisting of 1 atomic
percent europia-doped yttria stabilized zirconia were deposited and exposed to high
heat flux conditions similar to those seen today’s turbine environments and were
shown to be capable of measuring temperature at different depths within the
coating.
ix
TABLE OF CONTENTS
Foreword…………………………………………………………………………...1
1. Chapter 1: Concepts for Luminescence Sensing of…………………………..3
Thermal Barrier Materials
1.1. Introduction
1.2. Thermal barrier coating systems
1.3. Sensing concepts
1.4. Measuring temperature using luminescence
1.5. Phase compatibility of sensors
1.6. Selection of luminescent dopants
1.7. Demonstration of sensor concepts
1.8. Summary
2. Chapter 2: Experimental Investigation……………………………………38
2.1. Introduction
2.2. Materials characterization
2.2.1. Materials processing
2.2.2. Raman spectroscopy
2.2.3. X-ray diffraction
2.3. Luminescence characterization
2.3.1. Room temperature luminescence characterization
2.3.2. High temperature luminescence characterization
2.3.3. Amplification and electronics
3. Chapter 3: Luminescence sensing of temperature in………………………54
pyrochlore zirconate materials
3.1. Introduction
3.2. Materials
3.3. Room temperature luminescence
x
3.4. High temperature luminescence
3.5. Discussion
4. Chapter 4: Concentration Effect on Luminescence in……………………74
Eu-doped Yttria Stabilized Zirconia Intensity
4.1. Introduction
4.2. Luminescence spectra
4.3. Luminescence lifetime
4.4. Experimental procedure
4.5. Results and discussion
4.6. Summary
5. Chapter 5: Luminescence Thermometry of (Gd,Eu)2Zr2O7……………....95
5.1. Introduction
5.2. Experimental
5.3. Room temperature luminescence
5.4. High temperature luminescence
5.5. Discussion
5.6. Summary
6. Chapter 6: Non-contact Sensing TBC/BC Interface……………………...116
Temperature in a Thermal Gradient
6.1. Introduction
6.2.
6.3. Sensor layer deposition
6.4. Luminescence lifetime calibration
6.5. Luminescence lifetime measurement in a thermal gradient
6.6. Results
6.7. Discussion
7. Conclusions and Future Work……………………………………………134
1
Foreword
The goal of this thesis is the development of non-contact sensors capable of
measuring the temperature at the thermal barrier coating, near its interface with the
bond coat at temperatures applicable to current turbine systems. The requisite of
“prime reliance” on today’s engines is discussed as the driving force for the
development of non-contact heath monitoring tools for thermal barrier coatings.
To this end, luminescence sensing of temperature and damage is proposed as one
such non-contact technique. Chapter 1 begins with the introduction, description,
and ultimate demonstration of two basic sensor concepts, “visualization,” and
“measurement” sensors. Here the use of the optical properties and phase stability
of common thermal barrier materials (yttria stabilized zirconia and Gd2Zr2O7) as
well as the thermally grown oxide, another vital component in thermal barrier
systems, are discussed as selection criteria for the development of embedded sensor
materials. The choice of rare-earth ions as chromophore dopants is described and
the technique of luminescence lifetime decay for non-contact temperature
measurements is illustrated.
Chapter 2 departs from sensor development to discuss, in detail, the
development of the experimental apparatus used in this work and all the techniques
implemented in processing and analysis of the materials discussed. Several of the
most common problems encountered by others working in this area are reviewed
and their complications resolved. The third chapter discusses the use of Eu3+ as a
dopant ion in several thermal barrier coating hosts. Issues involving the selection
2
of excitation energy for these studies and the differences seen in the luminescence
lifetime for a series of materials with slightly different hosts, but the same doping
conditions are discussed. The work done in this chapter provides a basis for the
studies done subsequently in chapters 4 and 5.
Chapters 4 and 5 are an in-depth assessment of the effects that dopant and
defect concentrations have on the temperature sensing capabilities of the Eu-doped
yttria stabilized zirconia as well as (Eu,Gd)2Zr2O7. In these chapters, the changes
in luminescence lifetime are attributed to Eu3+ site symmetry and the extent of
disorder and chromophore ion-ion interaction in the two systems. Both of these
chapters provide a basis of composition selection for temperature sensors.
Finally, chapter 6 describes the implementation of a sensor based on the
work of the previous five chapters. In this chapter, a temperature sensor, of
optimum composition (determined in chapter four), was deposited by electron
beam physical vapor deposition. Temperature calibrations were done using the
techniques described in chapter 2. Following deposition and calibration, the sensor
was then exposed to a thermal gradient similar to that seen in current turbine
environments and, combined with pyrometry, was used to determine the thermal
conductivity and heat fluxes in the coating.
3
Chapter 1 Concepts for Luminescence Sensing of Thermal Barrier
Materials
1.1 Introduction
The use of ceramic coatings for thermal insulation of metal blades and hot
sections of aerospace and power generation turbines has become integral in modern
engine design. These coatings, commonly known as thermal barrier coatings,
enable metallic engine components to run in gases whose temperatures are often
above the melting temperature of the superalloys, resulting in greater engine
efficiency [1]. One result of increasing turbine temperatures is a risk of
catastrophic coating failure leading to substrate overheating and potentially failure
of the engine. Consequently, “prime reliance” of coatings has become an industry
standargoal for these systems. The ability to predict the remaining life of a coating
has hence become a sort of “holy grail,” allowing for component replacement to be
based on actual damage accumulation instead of scheduled replacement.
Monitoring of the metal temperature beneath the coating is a particularly promising
means to predict the life of the coating because there is an exponential relationship
between the metal temperature and the evolution of damage that leads to failure of
the coating. In this chapter, a novel technique for monitoring this temperature, as
well as damage in the coating, will be presented along with the selection criteria for
material schemes for developing such sensors.
4
1.2 Thermal barrier coating systems
Thermal barrier coatings are multilayer systems (Figure 1.2.1) that consist
of a thin overcoat of low thermal conductivity ceramic on top of a thin layer of
metallic “bond-coat” formed on a superalloy component[2-4]. During exposure to
high temperatures in air (and in typical turbine combustion environments), a thin-
layer of aluminum oxide forms at the interface between the “bond-coat” and the
thermal barrier coating. The composition of the “bond-coat” is chosen to form
aluminum oxide, commonly called the thermally grown oxide (TGO), upon
oxidation since aluminum oxide has the slowest oxygen diffusion rate of any oxide
and hence leads to the lowest oxidation rate. The most common material presently
used as a thermal barrier coating is yttria-stabilized zirconia (YSZ) but there is
some use of the rare-earth zirconates, such as gadolinium zirconate, Gd2Zr2O7,
(GZO) on account of their lower thermal conductivity at high temperature [4, 5].
Pertinent to optical sensing applications, these coatings are typically deposited by
either atmosphere plasma-spraying (APS) or electron-beam physical vapor
deposition (EB-PVD). The former results in the formation of “splat” boundaries
parallel to the coating that tend to scatter light and decrease optical transmission.
The EB-PVD coatings have a columnar microstructure, which tends to cause
preferential channeling of light into and out of the coating and scattering of light
perpendicular to the direction of coating growth. Both methods of deposition also
result in the coatings having microscopic porosity, a microstructural feature
desirable for further decreasing the low thermal conductivity of these coatings
5
below that of even the bulk material. It is important to note that YSZ coatings have
the metastable tetragonal prime crystal structure [4] and are quite distinct from
fully stabilized cubic zirconia (“synthetic diamond”) used in applications such as
optical windows and substrates for thin-film growth[6]. Their composition is also
distinct, usually 7-8 w/o Y2O3 as opposed to the 20 w/o of the cubic single crystals.
Both YSZ and GZO coatings are translucent in the visible due to extensive
entrapped porosity. The optical band-gap of YSZ is reported to be ~5 eV[7]. The
optical absorption spectrum of GZO is unknown, but is believed to be similar to
YSZ and other complex oxides, with band-gap absorption in the mid-ultraviolet
and absorption in the mid-infrared. Although the far-infrared (6 – 25 µm) optical
properties of tetragonal-prime yttria-stabilized zirconia have been reported[8] as a
function of temperature, the optical absorption properties of both materials in the
visible and near-infrared at elevated temperatures have not been reported. The
presence of porosity within the coating causes optical scattering, particularly in
plasma spray coatings. The frequency dependence of the optical absorption and
optical scattering are shown for a typical YSZ coating in Figure 1.2.2[9]. As will
be demonstrated, the difference in absorption in the visible and the UV, illustrated
in Figure 1.2.2, provides an opportunity to use lasers of different frequencies to
probe different depths into YSZ coatings for a variety of sensor applications.
Also pertinent to the issue of sensing is that the primary modes of failure in
a TBCs are spallation and delamination of the coating[10]. In both modes, the
coating separates from the component along a failure path within the ceramic
6
coating, typically close to the interface with the thermally grown oxide. (In large
thermal gradients, the failure path can shift to different planes in the coating.)
Failure thus causes a local reduction in the thickness of the TBC with the
consequence that the temperature of the metal under the spalled region increases
because of the lower thermal resistance afforded by the thinner coating.
1.3 Sensing Concepts
Several concepts for monitoring the “health” of thermal barrier coatings can
be envisaged. These range from the possibility of visually assessing damage and
wear to in situ measurement of coating temperatures. Broadly speaking the
concepts can be classified as being either “visualization” sensors or “measurement”
sensors.
Possibly the simplest “visualization” sensor is as an inspection method at
room temperature, for instance standard boroscope inspection during periodic
maintenance, to assesses whether the coating has locally spalled away. In cases
where the spalled regions are very large, they can be seen visually. This is the
basis for boroscope inspection today. However, when the spalled regions are
smaller, the concept of a “red-line” sensor facilitates greater contrast. The concept
is shown in Figure 1.3.1 and consists of an inner layer of the coating doped with
chromophore adjacent to the bond-coat. Where the coating is intact, the doped
region is buried and so can only be probed with a laser wavelength to which the
coating is transparent. Where the coating has spalled away or has been eroded, the
7
doped layer is exposed and can be illuminated with a laser in the UV as well as in
the visible.
An elaboration of the “red-line” sensor is what might be termed a
“rainbow” sensor, shown schematically in Figure 1.3.2. It consists of a series of
layers, each with a different dopant that luminescence at a different frequency.
Under optical excitation in the visible, characteristic luminescence from each of the
layers can be collected. As the thermal barrier coating erodes away, successive
doped layers will be removed and the visible luminescence will characterize the
remaining layers, enabling the remaining thickness of the coating to be assessed.
Alternatively, using UV excitation above the optical band gap of the coating, it
should be possible to probe the outermost layer in the “rainbow” sensor exclusively
and thereby directly monitor the progress of erosion.
Both sensor configurations also form the basis of temperature measurement
using dopants whose luminescence characteristics are temperature-dependent. The
“red-line” sensor configuration then provides a means of monitoring the
temperature of the coating in contact with the bond-coat. More ambitious, and also
highly desirable, is the measurement of the temperature gradient through the
thickness of the coating by monitoring the luminescence simultaneously from
layers at both the inner and outer portions of the coating.
1.4 Measuring Temperature Using Luminescence
Traditionally, three properties of the luminescence emission of a material
have been used to measure temperature: observed shifts in peak position, relative
8
peak intensities, and luminescence lifetime decay[11]. The first is the measurement
of observed line shifts of a single luminescent transition with temperature. It has
been noted that luminescence lines from chromophore ions show an observed blue
shift with increasing temperature [12] and that these shifts are attributed to
interactions between the electronic states and thermally activated lattice phonons.
At first glance, measuring temperature by peak shifts appears to be an appealing
method due to the small spectral range required (peak shifts are on the order of ~10
cm-1 for 0 – 300 K) but the technique becomes increasingly unreliable due to
extensive spectral broadening that occurs at temperatures much above room
temperature. Therefore, peak shift measurements are inappropriate for thermal
barrier coating applications where the temperature range of interest is well above
room temperature (in excess of 1000 °C) and spectral broadening and decreased
intensity would make any such measurement very difficult.
The second technique is the measurement of peak intensity ratios. This
technique monitors changes in the relative intensities of two distinct luminescence
peaks resulting from different excited states of the chromophore ion for a specific
concentration. Because the relative populations of the excited states are strongly
dependent on temperature, the relative intensities are also temperature dependent,
providing a basis for the use of relative intensities to measure temperature. By
measuring the relative intensities of two different luminescence lines over the
temperature range of interest and developing an intensity-ratio temperature
relationship, it is possible to derive the temperature of a material from observed
9
luminescence. This technique has been applied to various materials systems at
temperatures approaching those applicable to TBC systems [13-16], but caution
must be used when applying the intensity ratio technique. The technique is very
sensitive to small changes in luminescence intensity due to compositional changes
and the reabsorption of emitted light. While the application of this technique is
currently being explored for use on thermal barrier systems, it remains to be seen if
small deposits of calcium-magnesium-aluminosilicate (CMAS) or other debris
commonly found on engine hardware may affect the observed spectral intensities
and invalidate the temperature measurement.
The third scheme, luminescence-lifetime decay, is the technique chosen for
temperature measurements in this work. In this scheme, the luminescence from the
dopant ions is excited with a very short pulse of light and the intensity of the
luminescence from a single excited state is recorded as a function of time after the
end of the pulse. From this measurement, the decay time can be extracted to yield
a lifetime–temperature relationship, which can subsequently be used to calibrate the
temperature based on the material’s luminescence lifetime. In the simplest cases,
the logarithm of intensity decreases linearly with time and the decay can be
identified as the slope of the decay curve (Figure 1.4.1) fit using a simple
exponential decay
( )τtII −= exp0 Equation 1.4.1
where I is the intensity of light, I0 is the intensity of light at time t=0, and τ is the
luminescence lifetime of the material. In some of the materials studied here, the
10
luminescence decays with a more complex function, characterized by the sum of
exponential decays (Figure 1.4.1) and are fit using equation 1.4.2.
( ) ( )2211 expexp ττ tItII −+−= Equation 1.4.2
Materials used to measure temperature using the luminescence lifetime decay
technique can often be chosen to yield long lifetimes (> 1 µs) avoiding the use of
complicated high-speed electronics. The use of simple electronics, coupled with
the small spectral range required for the measurement, relative insensitivity to
thermally induced line broadening and background variations, and the techniques
limited susceptibility to uncertainties introduced by small changes in concentration
makes the luminescence lifetime technique particularly appealing for very high
temperature measurements.
The temperature sensitivity of luminescence lifetime can be described as
follows. As the temperature of a luminescing body increases, the equilibrium
populations of the excited and ground states follow a Boltzmann distribution,
( )[ ]kTEEgg
NN
abb
a
b
a −−= exp Equation 1.4.3
where N is the equilibrium population of the level, g is the dengereracy of the level,
E is the energy of the level, k is the Boltzmann constant, and a and b denote the
ground and excited states respectively. The rate of decay from the excited state, w,
is given as a sum of the rate of radiative decay for the isolated ion (wr), the rate of
non-radiative decay (wnr), and the rate of energy transfer between like ions (wET).
This can be described simply as
11
ETnrrtotal wwww ++= . Equation 1.4.4
Where τ1=w . For dilute solutions at low temperatures, the radiative rate
dominates. As the temperature of the system is increased, phonon modes within
the host material are activated and non-radiative decay plays an increasingly
important role in the total decay rate of the material. Moos and Weber [17, 18]
describe this process in detail through the mean of a multi-phonon relaxation model
where the multiphonon rate ( )Twmp varies with temperature according to
( ) ( ) pmpmp nwTw )1(0 += Equation 1.4.5
where ( )0mow is the rate of multiphonon emission at 0 K, the phonon occupancy
factor ( )[ ] 11exp −−= kThn ν , hν is the energy of the accepting phonon mode, and p
is the number of phonons required to bridge the gap between the excited and
ground states. Following activation of the phonon modes in the host and thus
multiphonon de-excitation of the excited state, the lifetime decreases rapidly with
increasing thermal energy. An example of typical luminescence lifetime
dependence on temperature is seen in Figure 1.4.2. As the thermal energy of the
system approaches the energy gap between the ground and excited state, complete
quenching of the luminescence is observed. This effect will be explored in the
following chapters.
1.5 Phase Compatibility
As a thermal barrier coating is in direct physical contact with the thermally
grown aluminum oxide, phase compatibility of the coating material with aluminum
12
oxide places very severe restrictions on the selection of possible luminescent
materials and their microstructural location within a coating system. Luminescent
materials that are not thermodynamically compatible with the coating will react and
decompose with inevitable degradation in luminescence and, possibly coating life.
For the two most common thermal barrier coatings materials, YSZ and GZO, the
pertinent phase compatible diagrams with aluminum oxide are shown in Figure
1.5.1. drawn from Leckie et al. [19] and Leckie, Kramer, Ruhle, and Levi [20].
The figure consists of two pseudo-ternary phase diagrams joined along their
common Al2O3-ZrO2 pseudo-binary at 1200 °C.
Of the compounds in the compatibility diagrams of Figure 1.5.1, the three
best known to be hosts for luminescence ions [21-23], yttria (Y2O3), yttrium
aluminate (YAlO3), and yttrium aluminum garnet (YAG), are seen to be
incompatible with the tetragonal-zirconia (YSZ) of thermal barrier coating and in
the case of yttria and yttrium aluminate, the thermally grown oxide, aluminum
oxide (Al2O3), as well. This indicates that none of these materials can be used as
stable luminescence hosts in direct contact with the thermally grown oxide formed
by oxidation of the bond-coat and the YSZ. If the thermal barrier coating were to
be made from the pyrochlore phase, such as Gd2Zr2O7, then it would be possible to
use the GdAlO3 phase as a host for the luminescence sensor, but the pyrochlore
itself is not compatible with alumina. The lack of stability in oxidizing
atmospheres at high temperatures obviously precludes the use of other well-known
phosphors, such as the oxy-sulphides.
13
These phase compatibility restrictions indicate that the coating materials
themselves must be used as the hosts for the luminescing chromophore.
Fortunately, both YSZ and GZO can accommodate trivalent luminescent ions such
as the rare-earth ions (by partial substitution for the Y3+ and Gd3+ ions respectively)
because of their similar ionic size and valence [24] as will be discussed in the next
section. Indeed, there are multiple reports of the photoluminesence of a number of
rare-earth ions in YSZ [16, 25-27]. It should be noted that the phase compatibility
restrictions are less severe for sensing the outer portions of the coating since the
luminescence sensing material does not have to be compatible with alumina as well
as the coating material. With that in mind, it would be possible for the perovskite
phases, such as YAG, to be deposited on the top of the coating for temperature
sensing.
1.6 Selection of luminescent dopants
The rare-earth ions ranging from cerium to ytterbium display a partial
filling of the 4f shell. In their elemental state, the outer 5d and 5s electrons act to
shield the inner 4f electrons. This is important because the 4f electrons are those
responsible for the optical transitions in these ions. The observed energy levels for
several of the trivalent rare-earth ions in LaCl3 are seen in the Dieke diagram in
Figure 1.6.1. Shielding by the outer shells results in sharp absorption and emission
peaks as well as splitting of the free ion energy levels by much less than the
spacing between levels (denoted by the thicknesses of the energy levels in the
Dieke diagram Figure 1.6.1).
14
As mentioned previously, the trivalent rare-earth ions have some limited
solubility in both the YSZ and GZO phases. As they are also well known to be
luminescent ions [22, 28] and are widely used in phosphors and displays, the
trivalent rare-earth ions are the natural choice for the class of dopants used in
luminescence sensing of thermal barrier coatings. For room temperature sensor
applications, such as the “red-line” sensor, there are several choices of rare-earth
dopant. For the rainbow sensors, the dopants in the different layers must be chosen
so that there is little overlap between the luminescence lines of the different
dopants. Guidance in selecting potential rare-earth ion dopants for YSZ and GZO
coatings is provided by the Dieke diagram (Figure 1.6.1) which tabulates the
luminescence from different energy levels of the indicated rare-earth ions in
lanthanum chloride [28]. This diagram indicates that for room temperature sensor
applications almost all the rare-earth ions can serve because potential luminescence
dopants as they luminesce over the entire visible spectrum and into the infrared.
For sensing applications at temperature, the selection of appropriate dopants
is more restrictive since the luminescence signal must be detectable against the
blackbody radiation from the surroundings and hot gas as well as the coating itself.
Therefore, to be detectable, the luminescence must be at wavelengths below the
high-energy edge of the blackbody curve. This is shown in Figure 1.2.2 for a
number of representative blackbody temperatures and indicates that that ions that
luminesce at wavelengths below ~650 nm are required in order to avoid being
swamped by blackbody radiation. In addition, to allow the luminescence from a
15
buried layer to reach the surface for collection, the ion must luminesce at
wavelengths longer than the band gap of the material (~ 300 nm)., thus enabling
propagation through the coating
To measure temperature, the choice of rare-earth ions is still more
restrictive, particularly if the luminescence lifetime method is used. Throughout
the course of this work, europium has been shown to be a particularly good dopant
for temperature sensing in YSZ and GZO. The selection of europium for this study
is based on it having electron energy levels with a large energy gap between the
states that can be excited in the visible (5D) and the lower energy levels (7F), and
because the resulting luminescence emission is not only in the visible, but also at a
wavelength shorter than the edge in the black-body emission curve for temperatures
up to ~1500 °C (Figure 1.6.1). The large energy gap found in Eu3+ is desirable
because the rate of non-radiative decay by multiphonon relaxation, which competes
with radiative decay limiting the maximum temperature where luminescence is
observed, depends on the size of this gap [29] .
1.7 Concentration Effects
Once an appropriate chromophore ion has been identified, it is essential to
establish its useful range of concentration for the particular sensor application. For
visualization sensing, the most desirable doping concentration is the one that
maximized the luminescence signal intensity (subject to phase stability
considerations). On the other hand, for in situ temperature measurements, although
maximizing the luminescence intensity is clearly desireable, it is important that the
16
selected concentration is one for which the lifetime decay varies over the
temperature range of interest.
The concentration at which the luminescence intensity is at its maximum is
usually set by the phenomenon of concentration quenching. This is illustrated by
the intensity of different luminescence lines from Sm-doped YSZ excited using an
argon-ion laser at 514 nm (Figure 1.7.1). The intensity data as a function of
concentration, normalized to the maximum intensity in each spectrum, is plotted in
Figure 1.7.2. The curves through the data correspond to a fit using the Johnson-
Williams equation [30].
( )( )cAccc z
−+−
=1
1η Equation 1.7.1
Where η is the quantum efficiency of the chromophore and c is the mole fraction of
luminescent ions. Notationally, the exponent z represents an effective coordination
number of isolated, like ions around the excited ion and the parameter A represents
a ratio of optical cross sections but, in practice, they are usually used as fitting
parameters since usually neither is known a priori. The values of z and A vary for
the different spectral transitions in Figure 1.7.2 and are listed in Table 1.7.1 and are
typically on the order of 38 and 0.008.
Also, as indicated by the data in Figure 1.7.2, the Sm-dopant concentration
at which the luminescence intensity peaks depends on the particular spectral
transition. Interestingly, there are dopants and combinations of excitation and
luminescence wavelengths over which no apparent concentration quenching occurs.
17
This unexpected behavior occurs, for example, for the 606 nm line of Eu-doped
YSZ when excited with a 514 nm laser (Figure 1.7.3), as well as for the 654 and
678 nm lines of Er-doped YSZ, whereas the lines at 543 and 562 nm or Er-doped
YSZ do exhibit concentration quenching (Figure 1.7.4.) In the particular case of
Er-doped YSZ, the continuing increase in intensity as a function of concentration is
believed to be due to cross-relaxation because the second set of peaks (654 and 678
nm) are significantly lower in energy and intensity than the primary transitions.
This is a process where the increase in the concentration dopant ions leads to
increased energy transfer between ions and the energy transfer between like ions
results in the luminescence lines observed at 654 and 678 nm. Consequentally, the
increase in concentration causes an increase in intensity. The reason that
concentration quenching is not seen in Eu-doped YSZ has not yet been explained.
1.8 Demonstration of Sensor Concepts
In this section, the basic concepts described in the preceding sections are
illustrated. Two types of samples were used. One was a series of 7YSZ thermal
barrier coatings deposited by electron beam deposition at UCSB. These coatings
consisted of a thin (~ 10 µm) layer of Er-doped YSZ deposited on an oxidized
metal alloy and then over-coated with ~120 µm of undoped YSZ to make a
standard thickness YSZ coating. On one of these coatings a second layer of Eu-
doped YSZ was deposited on top of the 120 µm of undoped YSZ. For this
demonstration, the composition of the Er-doped sensor layer was
Zr0.965Y0.07Er0.005O1.9625 while the Eu-doped sensor layer had a composition of
18
Zr0.96Y0.07Eu0.01O1.96. The first of the EB-PVD samples was used to demonstrate a
“red-line” sensor (Figure 1.3.1) and the second a simple “rainbow” sensor (Figure
1.3.2) consisting of only two sensor layers: one at the top of the coating an the
other at the bottom (layers 1 and 5 in Figure 1.3.2) with undoped YSZ between the
two layers. This second configuration will be referred to as a “heat-flux” sensor
since temperature measurements from the two layers taken simultaneously can be
used to measure the heat flux in the coating. The other types of samples were a
series of ceramic multilayer materials of differently doped layers created by
sequential deposition of rare-earth-doped powders and sintering at 1500 °C. Both
doped YSZ and doped GZO multilayer materials were produced this way to
simulate the “rainbow” coatings based on both material systems.
First the “red-line” sensor was examined. When the buried Er-doped sensor
was illuminated using the 514 nm excitation, the characteristic luminescence from
the Er3+ was detected through the full thickness of the coating (Figure 1.8.1). In
contrast, using the illumination at 248 nm, no luminescence from the Er-doped
layer was detectable since the absorption length at 248 nm is substantially smaller
than the coating thickness. This straightforward demonstration illustrates the
combined use of a buried luminescence layer and selection of excitation
wavelengths to differentiate between the surface and inner-layers of a thermal
barrier coating. Incidentally, the presence in the spectrum of the R-line
luminescence from the thermally grown oxide confirms that the 514 nm excitation
penetrates through the TBC, substantiating that the Er luminescence is from the
19
TBC next to the TGO. Figure 1.8.2 shows the spectra from the “heat flux” sensor
when excited in the visible. The spectra demonstrates that both the Er3+
luminescence from the buried layer as well as the Eu-doped top-layer in the “heat-
flux” sample can be excited simultaneously and that the lines do not overlap.
Figure 1.8.3 is an example of the luminescence spectra from a prototypical
“rainbow” sensors based on rare-earth doped gadolinium zirconate powders excited
at 514 nm, consisting of ErO1.5 and EuO1.5 doped layers. As indicated by the
labeling, the spectral lines from the individual layers are distinct and discernable.
1.9 Summary
The restrictions of sensor layer phase compatibility with the thermally
grown aluminum oxide and the TBC coating itself place severe constrains on the
possible luminescence sensor materials and suggest that the best chromophores are
likely to be those that are capable of being incorporated into the crystal structure of
the thermal barrier coating. For both YSZ and the rare-earth pyrochlore coatings,
rare-earth dopants can serve as chromophores. A number of concepts for
“visualization” and “quantitative” luminescence sensors are demonstrated based on
the optical properties of these coating materials and the luminescence properties of
the rare-earth dopants. For temperature sensing, the choice of rare-earth dopants
and concentrations is much more restrictive, as will be described in the following
chapters.
20
1. Miller, R.A., NASA Report CP 3312 17. 1995. 2. Stern, K.H., Metallurgical and Ceramic Protective Coatings. 1996, New
York: Chapman and Hall. 3. Evans, A.G., et al., Mechanisms controlling the durability of thermal
barrier coatings. Prog. Mater. Sci., 2001. 46: p. 505-553. 4. Clarke, D.R. and C.G. Levi, Materials design for the next generation of
thermal barrier coatings. Annu. Rev. Mater. Res., 2003. 33: p. 383. 5. Maloney, M., Thermal barrier coating systems and materials, U.T.
Corporation, Editor. 2000: USA. 6. Heuer, A.H. and L.W. Hobbs, Science and Technology of Zirconia. 1981,
Columbus, OH: Am. Ceram. Soc. 7. Harris, D.C., Materials for Infrared Windows and Domes. 1999: SPIE
Optical Engineering Press. 8. Nychka, J.A., et al., Temperature-Dependent Optical Reflectivity of
Tetragonal-Prime Yttria-Stabilized Zirconia. J. Am.Ceram. Soc, 2006. 89(3): p. 908.
9. Wahiduzzaman, S. and T. Morel, Final Report to Oak Ridge National Laboratory, #ORNL/sub/88-22042/2 available as DE92-041384 from NTIS.
10. Hillary, R.E., NRC Report. 1996, National Academy Press. 11. Grattan, K.T.V. and Z.Y. Zhang, Fiber Optic Fluorescence Thermometry.
1995, London: Chapman & Hall. 12. Sovers, O.J. and T. Yoshioka, Fluorescence of Trivalent-Europium-Doped
Yttrium Oxysulfide. J. Chem. Phys., 1968. 49(11): p. 4945. 13. Wickersheim, K.A. and R.V. Alves, Fluoroptic Thermometry: a New RF-
Immune Technology, in Biomedical Thermology. 1982, Alan Liss: New York. p. 547.
14. Allison, S.W., L.A. Boatner, and G.T. Gillies, Characterization of high-temperature themographic phosphors: spectral properties of LuPO4:Dy(1%), Eu(2%). Appl. Optics, 1995. 34(25): p. 5624.
15. Feist, J.P. and A.L. Heyes, The characterization of Y2O2S:Sm powder as a thermographic phosphor for high temperature applications. Meas. Sci. Tech., 2000. 11: p. 942.
16. Feist, J.P., A.L. Heyes, and J.R. Nicholls, Phosphor thermometry in electron beam physical vapor deposition produced thermal barrier coating doped with dysprosium. Proc. Instn. Mech. Engrs., 2001. 215(G): p. 333.
17. Moos, H.W., Spectroscopic relaxation processes of rare earth ions in crystals. J. Lum., 1970. 1(2): p. 106.
18. Weber, M.J., Multiphonon relaxation of rare-earth ions in yttrium orthoaluminate. Phys. Rev. B, 1973. 8(1): p. 54.
21
19. Leckie, R.M.R., et al. Microstructure stabliity issues in emerging TBC materials. in International Symposium on Thermal Barrier Coatings and Titanium Aluminides. 2002. Bonn, Germany.
20. Leckie, R.M.R., et al., Thermochemical compatibility between alumina and ZrO2-GdO3/2 thermal barrier coatings. Acta Mat., 2005. 53: p. 3281.
21. Blasse, G., Luminescent Materials. 1994, New York: Springer-Verlag. 22. Henderson, B. and G.F. Imbusch, Optical Spectroscopy of Inorganic Solids.
1989, New York: Oxford University Press. 23. Kaminskii, A., Laser Crystals: Thier Physics and Properties. 1981, New
York: Springer Verlag. 24. Shannon, R.D., Revised effective ionic radaii and systematic studies of
interatomic distances in hallides and chalcogenides. Acta Crystallogr., 1976. A(32): p. 751.
25. Feist, J.P. and A.L. Heyes, Europium-doped yttria-stabilized zirconia for high-temperature phosphor thermometry. Proc. Instn. Mech. Engrs., 2000. 214(L): p. 7.
26. Sergo, V. and D.R. Clarke, unpublished research. 1994. 27. Eldridge, J., In Press. Surface & Coatings Technology, 2003. 28. Dieke, G.H., Spectra and Energy Levels of Rare Earth Ions in Crystals, ed.
H.M. Crosswhite and H. Crosswhite. 1968, New York: Interscience Publishers.
29. Riseberg, L.A. and M.J. Weber, Relaxation phenomena in rare-earth luminescence, in Progress in Optics, E. Wolf, Editor. 1976, North-Holland. p. 89.
30. Johnson, P. and F. Williams, Interpretation of the dependence of luminescence efficiency on activator concentration. J. Chem. Phys., 1950. 18(11): p. 1477-1483.
22
Figure 1.2.1 Schematic diagram of thermal barrier system. The system consists of a thin overcoat of low thermal conductivity ceramic on top of a thin metallic bond-coat on a superalloy substrate. The thermally grown oxide (TGO) results from oxidation of the bond-coat on exposure to high temperatures in air. The microstructure represented in this schematic is typical of an electron-beam physical vapor deposited (EB-PVD) coating.
23
Figure 1.2.2 The optical absorption and scattering cross-sections for YSZ
material, redrawn following reference [9]. Superimposed are the wavelengths of two of the lasers used to excite luminescence from the sensors constructed. Superimposed are the blackbody radiation curves for the indicated temperatures. The shaded band indicates the wavelength window over which any temperature sensor must operate in the presence of thermal radiation. At shorter wavelengths, absorption and the optical band-gap (Eg) of YSZ limit both luminescence excitation and emission. At longer wavelengths, sensing will be limited by the ability to distinguish the luminescence signal above the thermal radiation.
24
Figure 1.3.1 Schematic diagram of the use of a buried luminescence layer as a
“red-line” sensor. The same configuration, with a temperature sensitive chromophore, can be used to monitor the temperature of the coating in contact with the underlying bond-coat metal.
25
Figure 1.3.2 Schematic diagram of the “rainbow” sensor, consisting of a
multilayer of luminescent dopants. As individual doped layers are eroded, worn away or spalled, the spectrum changes with the characteristic luminescence form those layers disappearing as they are eroded away as shown schematically.
26
Figure 1.4.1 Typical luminescence decay curves illustrating (a) a characteristic
single exponential decay and (b) a double exponential decay fit with Equation 1.4.1 and Equation 1.4.2 respectively.
27
Figure 1.4.2 Schematic diagram of the temperature dependence of luminescence
lifetimes.
28
Figure 1.5.1 Diagram representing the thermodynamic phase stability of alumina with zirconia containing yttria or gadolinia. The symbols refer to the crystal structures of the compounds: P – perovskite, C – cubic, G – garnet, F – fluorite, Py – pyrochlore. (Redrawn after Leckie et al. [19] and Leckie, Kramer, Ruhle, and Levi [20]).
29
Figure 1.6.1 Dieke diagram for the energy levels of four of the candidate rare-
earth ions that are soluble within the crystal structure of both YSZ and GZO coatings. For reference, the energies of candidate excitation lasers and the peak in blackbody radiation at 1500 °C are also indicated.
30
Figure 1.7.1 Luminescence spectrum in the visible for YSZ doped with 0.5 – 5
m/o SmO1.5 and excited at 514 nm.
31
Figure 1.7.2 Luminescence intensity of Sm3+ ions in YSZ as a function of the
concentration illustrating the phenomenon of concentration quenching. The different curves correspond to the different luminescence lines at the wavelengths labeled. The curves through the data correspond to Equation 1.7.1. Excitation was done at room temperature with and argon-ion laser operating at 514 nm.
32
Table 1.7.1 Johnson and Williams fitting parameters for the data in Figure 1.7.2
33
Figure 1.7.3 Luminescence intensity verses concentration of Eu3+ in YSZ
illustrating the absence of concentration quenching for all lines with 514 nm excitation.
34
Figure 1.7.4 Luminescence intensity versus concentration of Er3+ in YSZ
illustrating the absence of concentration for the emission lines at 645 and 678 nm but quenching for the lines at 543 and 562 nm. The lack of concentration quenching of the 645 and 678 nm lines is attributed to cross-relaxation.
35
Figure 1.8.1 Luminescence from a buried ~10 µm layer of Er-doped YSZ in an
EB-PVD YSZ coating and excited with a 514 nm laser. The presence of the R-line luminescence from the TGO in the same spectra confirms that the excitation laser penetrated to the buried sensor layer. No luminescence is excited with the 248 nm laser as it cannot penetrate into the TBC.
36
Figure 1.8.2 Luminescence from a buried ~10 µm layer of Er-doped YSZ at the
base of an EB-PVD YSZ coating with a ~10 µm layer of Eu-doped YSZ at the top of the coating and excited with a 532 nm laser. The luminescence from the two layers illustrates how a “heat-flux” sensor could be constructed by choosing luminescence ions whose luminescence spectra do not overlap.
37
Figure 1.8.3 Demonstration of two prototype “rainbow” sensors containing
ErO1.5 and EuO1.5 on fabricated out of YSZ and the other out of GZO. Both were excited at 514 nm.
38
Chapter 2 Experimental Investigation 2.1 Introduction
As described in the previous chapter, there are many materials, materials
processing techniques, and luminescence techniques that are available for use in
development of temperature sensors for thermal barrier materials. The
experimental investigation of these materials presented in this chapter will be
broken into two categories: the characterization of the materials and the calibration
and use of luminescence to measure temperature. Characterization studies
consisted of crystal structure analyses as the materials’ dopant concentrations and
aging conditions were varied. These studies established the validity of the use of
bulk materials in such tests as they apply to thermal barrier coatings. Calibrations
of luminescence lifetime as a function of temperature provide a basis for the
sensors described in subsequent chapters. This chapter outlines techniques for
processing both bulk and coated materials as well as describes methods for
examining crystal structure and luminescence at both room and elevated
temperatures.
2.2 Materials Characterization
7-8 w/o Y2O3 stabilized zirconia is the preferred material for thermal barrier
coating, although there is some use of rare earth zirconates (M2Zr2O7 where M is
the rare-earth ion). In the YSZ system, under equilibrium conditions, the yttria
stabilizes the tetragonal phase at temperatures above around 1050 °C. After
extended exposure to high temperatures, the material partitions into the equilibrium
39
tetragonal and cubic phases, which can then transform to monoclinic upon cooling.
In TBCs, the deposition process is not an equilibrium process and results in a
metastable tetragonal (t’) crystal structure, which does not transform upon
cooling[1]. In order to develop luminescence sensors applicable to thermal barrier
coatings, the bulk materials used in these studies must have the same crystal
structure (t’) as that of the coating. The constraint is not so restrictive in the case of
the zirconates, but the extent of ordering in the materials still needs to be taken into
consideration. It is also vital that there be a random distribution of luminescence
ions in both bulk and coating TBC materials to prevent changes in the
luminescence behavior due to clustering and highly doped materials effects[2].
2.2.1 Materials Processing
The bulk rare-earth doped yttria stabilized zirconias as well as rare-earth
doped zirconates used in these studies were prepared from powders synthesized by
reverse co-precipitation of aqueous rare-earth nitrate solutions (RE(NO3)3) and
zirconium acetate (ZrO(C2H3O2)2) in dilute acetic acid solutions into ammonium
hydroxide (NH4OH) following the work of Ciftcioglu and Mayo[3]. Reverse co-
precipitation (the addition of acidic solution to ammonium hydroxide solution) was
selected for the preparation of these materials to avoid segregation of the cations
and ensure molecular mixing of the dopant into the host material. The random
dispersion of the dopant material is particularly important in the case of processing
luminescent materials because clustering of luminescent ions can change the
characteristic luminescence of the material. An ammonium hydroxide solution
40
with constant PH of nine was used to guarantee instantaneous precipitation of the
mixed cation solution. After precipitation and drying under heat lamps, the
powders were calcined at 950 °C for 2 hours to obtain the oxide, pressed into
pellets, and sintered at 1200 °C in air for 2 hours. After initial heat treatments,
some of these materials underwent secondary exposures at varying temperatures to
examine the microstructural evolution and are noted where appropriate.
Several EB-PVD coatings were made of the rare-earth doped YSZ
materials. The ingots for EB-PVD of the sensor layers were prepared by
infiltration of a standard commercially supplied 7YSZ ingot with aqueous rare-
earth nitrate (RE(NO3)3). The nitrate solution was gelled into the pores of the ingot
with NH4OH, dried under heating lamps, and calcined at 950 °C for 2 hours. This
process was repeated until the desired weight of rare-earth oxide was achieved in
the pores of the ingot. This technique for the doping of the ingot was chosen in
order to achieve a coating of homogenous composition often a problem for dual
source evaporation [4] or bored ingot deposition [5]. The layered structure of the
sensor layers was achieved in a single run of EB-PVD by layering the doped ingot
with a standard undoped 7YSZ ingot. The final coatings consisted of a series of
different layered structures with two different rare earth dopants. An example of
the final structure as illustrated by cathodoluminescence is seen in Figure 2.2.1.1.
2.2.2 Raman Spectroscopy
Zirconia exhibits one of the strongest Raman signals reported in the
literature. As a result, it is a powerful tool in the analysis of zirconia based
41
systems[6]. Room temperature Raman spectroscopy was used to confirm that the
yttria stabilized zirconia had the tetragonal crystal structures for all dopant
concentrations and heat treatments. The Raman spectra for the ordered and
disordered pyrochlore structures are also well known[7] and so were used in these
studies to determine the structure and ordering of the pyrochlore zirconates.
Raman spectra were collected using an Instruments SA spectrometer fitted with an
Argon-ion laser emitting at either 514 or 488 nm. An example of the Raman
spectra for each of these crystal structures is seen schematically in Figure 2.2.2.1
with arrows denoting the peaks used to identify the respective crystal structures.
Raman spectra were also taken for some of the materials at temperatures up to 1200
°C using a Linkam 1500 heating stage to examine the change in phonon energies as
a function of temperature (work of Vanni Lughi). These values were then used for
fitting in the multiphonon model discussed previously (Equation 1.4.5).
2.2.3 X-ray Diffraction
Powder x-ray diffraction was carried out on the YSZ materials to confirm
the tetragonal-prime crystal structure as well as to determine the c/a ratios for the
different dopant concentrations. X-ray analysis was performed over the range of 72
≤ °2θ ≤ 76, which isolates the [400] reflections and very clearly indicates the extent
of tetragonality of the material. Initial x-ray scans were done on a Panalytical
MRD PRO thin film x-ray diffractometer with a monochromic source to provide a
baseline for kα1 stripping of the subsequently taken peaks. All other scans were
completed using a Phillips XPERT powder x-ray diffractometer (these data are
42
examined in chapter 4). X-ray analysis was also attempted on the pyrochlore
zirconates, but in all cases, the peaks characteristic of pyrochlore ordering were
absent.
2.3 Luminescence Characterization
As described in the previous chapter, the luminescence lifetime of a rare-
earth doped material is sensitive to temperature, according to the multiphonon
model described in chapter 1. In order to use the temperature sensitivity of
luminescence to measure temperature of a thermal barrier material, we carried out
the following set of optimization experiments. The first, summarized below, is the
use of excitation spectroscopy to select the best possible excitation wavelength for
the illumination of the sensor material. The next step is to examine the effects of
concentration on intensity and then compare that with the effects that concentration
has on the luminescence lifetime as a function of temperature. By taking all of
these aspects into consideration, the best sensor materials were identified. In this
chapter, in addition to materials issues, several issues regarding the development of
the experimental methodology are discussed. The development of these techniques
has been vital in the development of sensors capable of temperature measurements
at temperatures applicable to thermal barrier systems.
2.3.1 Room temperature luminescence characterization
Excitation spectra of each of the materials were recorded at room
temperature using a PerkinElmer LS 55 spectrometer in order to identify the
excitation wavelengths that would maximize the intensity of the spectral emission
43
at different wavelengths. The PerkinElemer LS 55 spectrometer was capable of
excitation from 200 – 800 nm and collecting luminescence emission from 200 –
900 nm. In all the measurements, both the excitation and emission slit widths were
set to 5 nm. Cutoff filters were placed after the sample in the light path to avoid
scattering of the excitation. All major luminescence lines were examined for each
of the materials but only those for the strongest transitions are reported. These
excitation spectra were then compared with the laser wavelengths at our disposal
(248 nm KrF Excimer, 355 nm tripled YAG:Nd, 532 nm doubled YAG:Nd) to
determine the most efficient laser frequency for excitation used in both spectral and
luminescence lifetime measurements.
For Eu3+ doped materials used most extensively in this work, the
luminescence studies presented were all performed using a frequency-doubled,
Continuum Model NY81C-1, Q-switched YAG:Nd laser emitting at 532 nm and
with a pulse length of ~10 ns for a Q-switch time of 250 µs, based on the excitation
spectra and the fact that the excitation yields the strongest intensity at 530 nm. The
luminescence emission from the rare-earth doped materials was collected using a
single crystal sapphire light pipe with a polished end and fitted at the opposite end
with a standard SMA fiber-optic coupler. A silica fiber optic cable with a 500 µm
core was connected to the SMA end of the light pipe and the luminescence was
directed to a series of collimating lenses with a 532 nm Semrock Raman filter in-
line to remove the scattered laser light from the signal. A second fiber-optic cable
was used to transmit the signal to an Ocean Optics USB 2000 grating spectrometer
44
fitted with a 300 nm cutoff filter with a photodiode array for spectrum collection
(Figure 2.3.1.1). The same configuration was used for luminescence collection
with other excitation frequencies, with the exception that the 532 nm Raman filter
was replaced with a filter of appropriate wavelength to block the operating laser
frequency.
2.3.2 High temperature luminescence characterization
For measurements of luminescence at elevated temperatures, a similar setup
to that for low temperatures was used. The samples were placed inside a box-
furnace with two access ports, one for the free-space propagation of the laser beam
excitation into the furnace and the other for collecting the luminescence signal
(Figure 2.3.2.1). To enable the luminescence to be collected at all temperatures up
to 1300 °C, a single crystal sapphire rod was used to gather the emitted light. The
sapphire light pipe was 3 mm in diameter and 10 cm long, polished at both ends,
and fitted with an SMA fiber-optic coupler at one end. The light pipe was
connected to a 500 µm core fiber-optic cable and passed through the same series of
collimating lenses and filters as described for the room temperature measurements.
This was coupled into a standard silica fiber-optic cable that was bifurcated along
its length so that a portion of the collected light could be spectral analyzed and a
portion used for luminescence decay measurements. An Ocean Optics USB 2000
grating spectrometer with a photodiode array was again used for spectral collection
and analysis. The luminescence lifetime-decay measurements were made using an
Acton SpectraPro 2150i monochromator fitted with a Hammamatsu R928
45
photomultiplier tube (PMT) or using a 610 nm bandpass filter and a Hammamatsu
H6780-20 photosensor module. In both cases the PMT bias was kept between 800
– 900.
The lifetime measurements were made under computer control with the
luminescence intensity collected after the laser pulse was triggered. The laser
trigger was connected directly to the oscilloscope. To ensure good signal to noise,
an average of the luminescence signal from 1000 laser pulses was recorded with the
signals accumulated in a LeCroy 9310AM digital oscilloscope. In some cases, a
trans-impedance amplifier was used to amplify the luminescence signal before the
oscilloscope recorded it.
2.3.4 Amplification and Electronics
One problem that has been experienced by other groups has been the
mastering the electronics for collection. Figures 2.3.4.1 and 2.3.4.2 illustrate
several common artifacts associated with inappropriate impedance matching and
bandwidths. Most common is the mismatching of impedance between the PMT
(typically 50 Ω) and that of either the amplifier (ranging from 500 Ω to 1 MΩ) or
the oscilloscope (50Ω or 1 MΩ). Mismatches of several orders of magnitude
between the impedance of these two components result in a significant
amplification of the signal intensity but have a detrimental effect on the speed of
the measurement. As is illustrated in Figure 2.3.4.1, the signal can be significantly
distorted when the PMT is set to 50 Ω and the oscilloscope to 1 MΩ. Although the
signal is greatly amplified when compared to that of a 50 Ω -50 Ω connection, the
46
resolution provided by the higher speed measurement, including the faster initial
decay, is lost and a time constant of ~ 300 µs is introduced. Similar results are seen
for the oscilloscope set at 1 MΩ, 100 kΩ, and 1 kΩ in Figure 2.3.4.2. This
phenomenon has been reported incorrectly in the literature as actual luminescence
lifetime behavior [8]. To reach the fastest speed capable by the PMT it is
imperative to match the impedance of the oscilloscope exactly to the PMT (50 Ω).
This was the condition used for the very fastest lifetime measurements reported in
these studies.
47
1. Clarke, D.R. and C.G. Levi, Materials design for the next generation of thermal barrier coatings. Annu. Rev. Mater. Res., 2003. 33: p. 383.
2. Eldridge, J.I., Unpublished work. 3. Ciftcioglu, M. and M.J. Mayo, Processing of Nanocrystalline Ceramics.
Mat. Res. Soc. Symp. Proc., 1990. 196: p. 77. 4. Allison, S.W., et al. Use of phosphor coatings for high temperature
aerospace applications. in 29th AIAA/ASME/SAE/ASEE Joint Propulsion Conference. 2003. Huntsville, AL: AIAA.
5. Feist, J.P., A.L. Heyes, and J.R. Nicholls, Phosphor thermometry in electron beam physical vapor deposition produced thermal barrier coating doped with dysprosium. Proc. Instn. Mech. Engrs., 2001. 215(G): p. 333.
6. Clarke, D.R. and F. Adar, Measurement of the Crystolographically Transformed Zone Produced by Fracture in Tetragonal Zirconia Containing Ceramics. J. Am.Ceram. Soc, 1982. 65: p. 284.
7. Michel, D., M. Perez Y Jorba, and R. Collongues, Study by raman spectroscopy of order-disorder phenomena occurring in some binary oxides with fluorite-related structures. J. Raman Spec., 1976. 5(3): p. 163.
8. Feist, J.P. and A.L. Heyes, Europium-doped yttria-stabilized zirconia for high-temperature phosphor thermometry. Proc. Instn. Mech. Engrs., 2000. 214(L): p. 7.
48
Figure 2.2.1.1 Cathodluminescence image of an EB-PVD coating with a ~10 µm
Eu-doped layer at the interface next to the bond-coat. The brightest contrast is due to the Cr3+ impurities in the alumina thermally grown oxide. Immediately above the TGO the luminescence from the Eu3+ ions in the sensor layer are visible as a light gray band.
49
Figure 2.2.2.1 Representative Raman spectra for (a) tetragonal-prime 7 w/o Y2O3
stabilized zirconia and (b) ordered and (c) disordered pyrochlore crystal structures. The * mark the peaks used to identify the tetragonal crystal structure and arrows denote the ordering peaks in the pyrochlore.
50
Figure 2.3.1.1 Schematic drawing of the optical arrangement used for measuring
the luminescence lifetime and spectrum of sensor materials at room temperature.
51
Figure 2.3.2.1 Schematic drawing of the optical arrangement used for measuring
the luminescence lifetime and spectrum of sensor materials at elevated temperatures.
52
Figure 2.3.4.1 The luminescence lifetime decay of 1 a/o Eu:YSZ at 600 °C using
two different impedances at the oscilloscope. Although the 1 MΩ termination increases the signal intensity, it slows the time resolution of the measurement and the fine features seen in the 50 Ω signal are not apparent.
53
Figure 2.3.4.2 The luminescence lifetime as a function of temperature for 1 a/o
Eu:YSZ using several different input impedances at the oscilloscope. Larger impedances result in time constants that limit the maximum temperature that can be measured using the luminescence technique.
54
Chapter 3 Luminescence sensing of temperature in pyrochlore zirconate
materials
3.1 Introduction
One method of determining the temperature of a component without
making physical contacts is by measurement of the luminescence lifetime decay of
a phosphor either embedded in the component or painted on its surface[1, 2]. This
method, which dates back to the work of Bradley in 1953 [3], is particularly
attractive in measuring temperature in aggressive environments but has generally
been limited to materials that are both stable and exhibit temperature-dependent
luminescence decay. Although there have been recent reports of temperature
dependent lifetimes up to 1500 °C in Tm and Eu-doped YAG (yttrium aluminum
garnet)[4], these compounds are neither phase compatible with the oxide formed on
the bond-coat in air at high temperatures, nor do they exhibit low thermal
conductivity required for thermal barrier coatings[5].
To ensure thermodynamic compatibility, it is desirable that the luminescent
species be individual ions in solid solution in the crystal structure of the thermal
barrier coating material itself so that the TBC also acts as the sensor. In this
concept, various practical issues with previous luminescent sensors, such as
attachment, adhesion, and high-temperature compatibility, are circumvented. The
first task is to identify suitable dopants that can be taken into solid solution within
the TBC material without destabilizing it and also have the luminescence
55
characteristics necessary for sensing and temperature measurement in the presence
of thermal radiation from other hot surfaces. These and related selection criteria
have been presented previously, together with results of luminescence doping of
yttria-stabilized zirconia[6, 7].
Observations of the principal luminescence characteristics of four different
europium-containing materials are presented in this chapter. Three are zirconates
with the pyrochlore structure and the fourth, for comparison purposes, is YSZ with
1 a/o of the Y3+ ions replaced with Eu3+, and having the formula
Y0.06Eu0.01Zr0.93O1.965. Two of the zirconates were doped with a small
concentration of europium, (Sm0.99Eu0.01)2Zr2O7 and (Gd0.975Eu0.025)2Zr2O7. The
third was europium zirconate, Eu2Zr2O7.
3.2 Materials
All materials used in this study were prepared as bulk ceramics by reverse
co-precipitation outlined in section 2.1. The Raman spectra recorded following
sintering at 1200 °C for 2 hours for all four materials examined in this study are
seen in Figure 3.2.1. The spectra from the zirconates not only confirm that they
were single-phase materials but also that they exhibit different degrees of ordering.
By comparison of the peaks with those reported in the literature [11], the
gadolinium zirconate appears disordered following initial heat treatments whereas
the samarium and europium zirconates both exhibit some degree of ordering as
evidenced by the presence of distinct Raman lines at approximately 310, 397, 530,
and 594 cm-1. The Raman spectrum of the Eu-doped YSZ confirms that the
56
material was also single-phase with no indication that the Eu-doping had caused
any destabilization of the tetragonal phase to monoclinic zirconia.
Excitation spectra for each of the materials were recorded at room
temperature to identify excitation wavelengths that would maximize the intensity of
spectral emission at different wavelengths. The excitation spectra for emission at
610 nm are reported in Figure 3.2.2 and indicate that several wavelengths,
including 390, 470, and 530 nm, could be used to excite luminescence from all of
these materials. Superimposed on these spectra are the reported[12] observed
energy levels for Eu3+ as well as the wavelengths of lasers available for excitation:
248 nm KrF excimer, 355 nm frequency tripled YAG: Nd, and 532 nm frequency
doubled YAG: Nd lasers.
On the basis of the excitation spectra and the fact that the excitation yields
the strongest intensity at 530 nm, the luminescence studies reported in this work
were all performed using a frequency-doubled YAG: Nd laser (section 2.1). Room
temperature as well as high temperature luminescence spectra measurements were
recorded according to the specifications outlined in chapter 2 (Figure 3.2.3). The
luminescence lifetime measurements were recorded from room temperature, until
either the decay time of the luminescence reached the detectability limit of out
current electronics or the intensity of the signal dropped to a level below the noise
floor of the measurements, making further measurements impossible.
57
3.3 Room temperature luminescence
The room temperature luminescence of the four materials from 550 to 680
nm (the wavelength region that encompasses the principal emission lines of Eu3+
ions in the visible) is shown in Figure 3.3.1. Both the three zirconates and the
doped YSZ exhibit similar luminescence, although that from the
(Sm0.99Eu0.01)2Zr2O7 zirconate is significantly weaker in intensity. The similarity of
major features of the luminescence spectra is in accord with expectation based on
the transitions of the 4f electrons responsible for luminescence in trivalent rare-
earth ions, and the relative insensitivity of the emission wavelengths is
characteristic of the 4f electrons being shielded by the outer shell 5s and 5p
electrons [13]. Thus, although there are small differences in the wavelengths of the
peaks from one compound to another, the principal transitions can be readily
identified as being due to the 5D0à7F0 (~582 nm), the 5D0à
7F1 (~592 nm), and the
5D0à7F2 (~610 and 635 nm) transitions. The most notable distinction between the
luminescence from the zirconates and the doped YSZ is that the splitting of the
5D0à7F2 results in a much weaker peak at 635 nm that is red-shifted by ~5 nm in
the YSZ.
The 5D0à7F2 (~610 nm) transition was chosen for lifetime measurements
because of both its high intensity and narrow full-width-half-maximum (FWHM),
as the intensity will decrease and the FWHM increase with temperature (Figure
3.2.3). This transition is also the only one of the observed transitions that is a
strong electric dipole only allowed transition [12] meaning it will have a strong
58
intensity in low symmetry crystals such as the D2d ( )m24 symmetry of the Eu3+ ion
in tetragonal zirconia. Its presence and relative strength in the pyrochlore materials
suggest that there is a significant deviation from the D3d ( )m23 symmetry of a
perfectly ordered pyrochlore as the presence of a inversion center in this symmetry
would forbid the 5D0à7F2 transition.
3.4 High temperature luminescence
Measurements were taken of luminescence lifetimes up to temperatures at
which the lifetime decay times fell below our current detectability limit (~10 ns) or
the intensity of the luminescence signal fell below the noise floor of the
measurement. The form of the luminescence intensity decay (as a function of time
after the excitation ceased) varied from one material to another. For both the
Eu2Zr2O7 and (Sm0.99Eu0.01)2Zr2O7 zirconates, the decays were single exponentials.
In contrast, the luminescence decays were more complex for the doped gadolinium
zirconate and YSZ. Both materials displayed a short exponential decay followed
by a longer exponential decay (Figure 3.4.1). In analyzing the luminescence
decays, they were fit using either a single (Equation 1.4.1) or double (Equation
1.4.2) exponential decay depending on the nature of the decay. The decay times
derived from the fitting are presented in the subsequent graphs.
The variation with temperature of the decay times of all the materials
exhibited the same characteristic features of a relatively constant, almost
temperature-independent, value, from room temperature up to an intermediate
temperature and thereafter an exponential decrease with increasing temperature.
59
As the focus of this work is primarily on the high temperature characterization of
the luminescence lifetime, the majority of the data shown is for the temperatures
where there is a temperature-dependent effect on lifetimes.
The luminescence intensity decay from the pure europium zirconate showed
the simplest behavior. It exhibited a single exponential decay over the entire
temperature range investigated with the decay time being almost temperature-
independent up to ~700 °C and then decreasing exponentially up to 1250 °C
(Figure 3.4.2). At the very highest temperature, the luminescence intensity had
reached the limit of detectability with our present measurement system, suggesting
that the exponential variation with temperature may well be continued to even
higher temperatures.
Rather similar luminescence decay characteristics were shown by the
europium-doped samarium zirconate ((Sm0.99Eu0.01)2Zr2O7), although the numerical
values for the lifetime decay were different. Interestingly, although the overall
luminescence intensity was lower at room temperature, luminescence decays could
be measured up to almost 1000 °C (Figure 3.4.3). In this figure, the data points
oscillate about a straight exponential line. This is attributed to difficulties
experienced with our furnace calibrations.
As mentioned above, the Eu-doped GZO and YSZ materials exhibited more
complex, double exponential luminescence decays. The high temperature variation
of both the fast and the slow lifetime decays recorded from the
(Gd0.975Eu0.025)2Zr2O7 material is plotted in Figure 3.4.4. Superimposed are the data
60
for the slow decay recorded from the Eu-doped YSZ. This comparison by
superimposition of data illustrates the surprising result that the luminescence
lifetime is essentially identical for these two materials even though they have
different crystal structures. This phenomenon will be addressed in the following
section.
3.5 Discussion
The principal finding of technological interest is that Eu-doping of the
pyrochlore zirconates investigated can be used as the basis for temperature
measurements in samarium zirconate up to at least 1000 °C and for both the
europium-doped gadolinium zirconates and YSZ up to at least 1150 °C.
Luminescence from Eu2Zr2O7 can be used to measure temperatures approaching
1300 °C. These temperatures are similar to those in current gas turbine systems
and thermal barrier coatings, but it is also likely that these materials can be used for
luminescence sensing of temperature at still higher temperatures with the
implementation of superior collector and detector equipment. Additionally, the
optimum doping concentrations in the pyrochlore have not yet been identified.
This is the subject of the following two chapters.
The lifetime data presented reveal some interesting correlations between the
crystal structures of the materials investigated (Figure 3.5.1). At first
consideration, one might conclude that because the Eu3+ ions occupy the same
position in all the zirconate structures, the luminescence characteristics would be
the same. The emission spectra are indeed very similar but the lifetime
61
characteristics clearly are not. Furthermore, the lifetime decays of 5D0 à 7F2
radiative transition for the Eu-doped YSZ and Eu-doped GZO are almost identical,
despite the fact that these two compounds have different, albeit related, structures.
We believe that part of the explanation lies in the fact that both these oxides exhibit
extensive site disorder as well as site defects such as oxygen vacancies, whereas the
Raman spectra indicate that both the Eu2Zr2O7 zirconate and (Sm0.99Eu0.01)2Zr2O7
zirconate are ordered to a larger extent. In contrast to the disordered GZO, the
Raman spectra of these last two zirconates exhibit well defined Raman lines at
approximately 310, 397, 530, and 594 cm-1, lines previously identified as being
indicative of ordering in pyrochlore structures [11].
At dilute concentrations of a luminescent ion in a homogenous solid, the
luminescence lifetime is largely controlled by the probability of non-radiative
transitions of the excited state transitioning back to the ground state. The dominant
non-radiative transitions in defect-free materials are temperature-dependent
multiphonon relaxation processes, and therefore the observed radiative lifetime can
be expressed as:
mpiR τττ111
+= Equation 3.1
where iτ is the intrinsic lifetime, and mpτ is the lifetime of a multiphonon
relaxation of the excited state[14]. Equation 3.1 was used to fit the data in Figure
3.4.4, but the same fit was unsuccessful for the Eu2Zr2O7 and Eu-doped Sm2Zr2O7
materials. At higher concentrations, such as the conditions in the Eu2Zr2O7 and Eu-
62
doped Sm2Zr2O7 materials, the excited state can diffuse from one luminescent ion
to another through the material until transitioning, non-radiatively at some defect
site, to the ground state. This energy diffusion process has the effect of increasing
the observed lifetime since it is a competitive process to the multiphonon relaxation
process. In effect, it can be considered as introducing a concentration-dependent
intrinsic lifetime and effectively adding a third term ⎟⎟⎠
⎞⎜⎜⎝
⎛
ETτ1 to Equation 3.1 to
indicate the rate of energy transfer. Unfortunately, without more information about
the nature of this additional term, there are too many parameters to meaningfully fit
the data using this form. The effect of disorder is more complex. One of the major
effects at low concentrations is to create two principal populations of luminescent
ions: isolated ions near defects and clusters of ions close to one another.
Luminescence from the former population is believed to be responsible for the
initial rapid luminescence lifetime decay since defects effectively decrease the size
of the energy gap between excited and ground states, resulting in faster decay times
and increasing the probability of non-radiative transitions. In contrast, energy
transfer can occur amongst ions in the latter populations, because they are close
together and consequently result in longer luminescence lifetimes.
Many of these effects can be seen in the data presented here. The lifetime
of the ordered Eu2Zr2O7 zirconate is longer at high temperature than that exhibited
by either the YSZ or GZO materials even though the effective activation energy is
similar. Both the Eu-doped GZO and Eu-doped YSZ exhibit characteristic double
63
exponential decays and, furthermore, the lifetime versus temperature curves almost
overlap. The luminescence from the Eu-doped samarium zirconate,
(Sm0.99Eu0.01)2Zr2O7, is somewhat unusual due to the similarity in the excited state
energies of the Sm3+ and Eu3+ levels. This similarity has two consequences. One is
that the luminescence intensity from the Eu3+ ion is reduced because the excited
state energy can transfer to the non-radiative Sm3+ ions. This energy transfer path,
however, also enables the energy, with some probability, to transfer back to a Eu3+
ion and then radiate, thereby extending the observed radiative lifetime.
The results obtained so far suggest that these Eu-doped materials could be
employed in a number of different configurations for temperature sensing in
thermal barrier coating systems. For instance, a thin layer of Eu-doped YSZ could
be deposited on a component and then undoped YSZ deposited on top to form a
temperature sensor at the bond-coat interface. A second configuration, for
measuring the temperature at the surface of the TBC, could be made by depositing
a thin layer of either Eu-doped YSZ or Eu2Zr2O7 on an existing TBC. A third
configuration, for measuring the temperature difference through a coating, would
consist of two differently doped layers, one on the inside and the other on the
outside of the coating. During the course of this study, such sensors have been
deposited by electron beam evaporation and their analysis will be reviewed in
chapter 6.
64
1. Allison, S. and G. Gillies, Remote thermometry with thermographic phosphors: Instrumentation and applications. Rev. Sci. Instrum., 1997. 68(7): p. 2615-2650.
2. James, K.A., W.H. Quick, and V.H. Strahan, Control Eng., 1979. 26: p. 30. 3. Bradley, L.C., Rev. Sci. Instrum., 1953. 24: p. 219. 4. Allison, S.W., et al. Use of phosphor coatings for high temperature
aerospace applications. in 29th AIAA/ASME/SAE/ASEE Joint Propulsion Conferrence. 2003. Huntsville, AL: AIAA.
5. Clarke, D.R. and C.G. Levi, Materials design for the next generation of thermal barrier coatings. Annu. Rev. Mater. Res., 2003. 33: p. 383.
6. Gentleman, M.M. and D.R. Clarke, Concepts for luminescence sensing of thermal barrier coatings. Surface & Coatings Technology, 2004. 188-189: p. 93.
7. Eldridge, J.I., et al., J. Thermal Spray Tech., 2004. 13: p. 44. 8. Gentleman, M.M. and D.R. Clarke, Luminescence sensing of temperature in
pyrochlore zirconate materials for thermal barrier coatings. Surface & Coatings Technology, 2005. 200: p. 1264.
9. Weber, M.J., Multiphonon relaxation of rare-earth ions in yttrium orthoaluminate. Phys. Rev. B, 1973. 8(1): p. 54.
10. Riseberg, L.A. and M.J. Weber, Relaxation phenomena in rare-earth luminescence, in Progress in Optics, E. Wolf, Editor. 1976, North-Holland. p. 89.
11. Michel, D., M. Perez Y Jorba, and R. Collongues, Study by raman spectroscopy of order-disorder phenomena occurring in some binary oxides with fluorite-related structures. J. Raman Spec., 1976. 5(3): p. 163.
12. Dieke, G.H., Spectra and Energy Levels of Rare Earth Ions in Crystals, ed. H.M. Crosswhite and H. Crosswhite. 1968, New York: Interscience Publishers.
13. Henderson, B. and G.F. Imbusch, Optical Spectroscopy of Inorganic Solids. 1989, New York: Oxford University Press.
14. Moos, H.W., Spectroscopic relaxation processes of rare earth ions in crystals. J. Lum., 1970. 1(2): p. 106.
65
Figure 3.2.1 Raman spectra of the three zirconate materials and of the Eu-doped
YSZ material. Ordering in the zirconates is indicated by the presence of additional Raman peaks indicated.
66
Figure 3.2.2 Excitation spectra for the emission at 610 nm. All the spectral
features can be attributed to either electronic transitions of the Gd3+ or Eu3+ ions. Uniformly, the strongest excitation for all four samples occurs at a wavelength corresponding to the 5D0 and 5D1 transitions of the 4f electrons of the Eu3+ ions.
67
Figure 3.2.3 Spectra of Eu-doped YSZ taken at several temperatures with 532 nm
excitation.
68
Figure 3.3.1 Emission spectra for the three Eu-containing zirconates and Eu-
doped YSZ materials studied. The spectra were recorded at room temperature using excitation at 532 nm. The arrow denotes the emission line (5D0à
7F2) used for luminescence lifetime measurements.
69
Figure 3.4.1 Luminescence decay curves illustrating the double exponential for
the (Gd0.975Eu0.025)2Zr2O7 at several temperatures.
70
Figure 3.4.2 Luminescence lifetime decay time as a function of temperature for
Eu2Zr2O7. A line has been drawn though the data to aid the eye.
71
Figure 3.4.3 The variation of the luminescence decay with temperature of
(Sm0.99Eu0.01)2Zr2O7. A line has been drawn though the data to aid the eye.
72
Figure 3.4.4 Temperature dependence of the luminescence decay times for both
the fast and slow decays of Eu in (Gd0.98Eu0.02)2Zr2O7. Superimposed are the lifetime decays data for the Eu-doped YSZ material. The line through the data is a fit of the data using the multiphonon model. The scatter in the data at ~400 °C is due to difficulties in the fitting of the lifetime decays due to the emergence of the faster exponential decay.
73
Figure 3.5.1 Schematic drawing of the rare-earth zirconate and zirconia crystal
structures. The introduction of trivalent ions on the zirconia structure results in oxygen vacancies on the anion sub-lattice.
74
Chapter 4: Concentration Effect on Luminescence in Eu-doped Yttria
Stabilized Zirconia
4.1 Introduction
In the previous chapters, europia-doped YSZ was shown to be a promising
material for temperature sensing. In chapter three a composition of 1 a/o Eu:YSZ
was investigated and shown to exhibit temperature sensitivity ranging from 500 to
1150 °C[1]. This measurement was limited by both the speed of the decay time
(~10ns) and the decrease in intensity caused by thermal quenching of the
luminescence. Although this composition was not chosen arbitrarily, very little
was known about the choice of optimum dopant compositions for europium in YSZ
a priori. In this chapter, we will explore the effects of Eu3+ and Y3+ composition
on the luminescence and temperature-sensing capabilities of europia-doped yttria
stabilized zirconias.
The interest in Eu3+ as a phosphor in thermal barrier materials has grown
considerably due to its promise as efficient non-contact temperature sensor.
Traditionally, Eu3+ has been used to probe the site symmetry and chemical bond
differences of the surrounding host materials in the neighborhood of the Eu3+ ion[2-
7]. In the work of Dexpert-Ghys, site selective spectroscopy of the Eu3+ ion was
used to explore changes in symmetry induced by the partial stabilization of zirconia
by yttria[8]. In those studies it was determined not only that the luminescence from
the europium ions was characteristic of lower symmetry than expected for
tetragonal zirconias, but also that there was no significant change in the
75
luminescence spectra as the crystal structure evolved from tetragonal to cubic. In
these studies we go a step further and look at the luminescence from Eu3+ in
partially stabilized zirconias as it pertains to changes in luminescence intensity and
luminescence lifetime. The development of a luminescence temperature sensor for
use in YSZ thermal barrier coatings requires the understanding of both the
luminescence intensity as a function of temperature as well as concentration and
simultaneous calibrations of the luminescence lifetime as a function of temperature.
Both of these behaviors are intimately linked to the crystal structure of the material
as well as the intrinsic rates of radiative and non-radiative luminescence decay for
the ion in YSZ. To determine if it is possible to increase the sensor sensitivity, we
carried out a series of experiments looking at the effects of Eu3+ doping in
conjunction with the effects of stabilizer content to examine the role ion-ion
interaction and crystal structure on the luminescence behavior of the material.
4.2 Luminescence Spectra
The luminescence from the trivalent Eu ions consists primarily of
transitions from the upper 5DJ=0-3 multiplet to the lower 7FJ=0-4 levels. Of particular
interest, are the 5D0à7F1 and 5D0à
7F2 transitions. The former being allowed as a
magnetic dipole transition and forbidden as an electronic dipole transition and the
latter allowed only as a forced electronic dipole transition induced when the Eu3+
site does not contain a center of symmetry[7]. As the deviation from inversion
symmetry increases, the intensity of the electric dipole transition increases with
respect to the intensity of the magnetic dipole transition. 7YSZ used in thermal
76
barrier coatings, consists of a metastable tetragonal crystal structure, (P42/nmc).
This yields a D2d ( )m24 symmetry at the site of the Eu3+ site, which lacks a center
of symmetry. Based on selection rules on the total quantum number J [4, 7]both
the 5D0à7F1 and 7F2 transitions would be expected. Luminescence from a structure
containing D2d ( )m24 symmetry at the Eu3+ site is expected to also result in a
splitting of the 5D0à7F2 transition into two lines at ~610 and 630 nm along with a
concomitant splitting of the 5D0à7F1 level into two peaks[2]. If the symmetry is
further distorted to symmetry less than D2d ( )m24 due to the distortions caused by
oxygen vacancies in the crystal, both the 5D0à7F1 and 5D0à
7F2 transitions may
exhibit many more lines. For example, lowering to a Cs (m) symmetry for instance,
would result in one line observed for the 5D0à7F0 line, three for the 5D0à
7F1 line,
and five for the 5D0à7F2 line similar to what is seen in the spectra recorded for the
zirconia materials studied here (Figure 4.2.1).
Concentration quenching of the luminescence intensity, as discussed in
chapter 1, is a phenomenon often seen in luminescent materials where the
luminescence intensity increases linearly with increasing dopant concentration and
then begins to drop in intensity as ion-ion energy transfer results in energy transfer
to luminescence quenching sites. The decrease in intensity is the result of the
nearness of one luminescent ion to another significantly reducing the thermal
activation energy for non-radiative transitions from those ions resulting in
extinction of the luminescence. This behavior can be understood through use of
77
Equation 1.7.1 described by Johnson and Williams [9]. This equation has been
used to successfully fit the data in Figure 1.7.2 as well as the data in Figure 4.2.2
but begins to break down at high dopant concentrations when the distortion caused
by large volume fraction of dopant ions distorts the crystal symmetry of the host
material. This is important in these studies because the intensity of the signal can
significantly hinder or increase the ability to measure the luminescence lifetime and
hence the temperature of a material.
4.3 Luminescence Lifetime
Although previous work has shown that there is no significant change in the
luminescence spectra as the dopant levels and crystal structures of partially
stabilized zirconias are varied [8], this is not expected to be the case for
luminescence lifetime behavior under the same conditions. When a photon excites
an electron in a luminescent ion into an excited state, multiple competing processes
dictate the rate of de-excitation of the electron. These processes include radiative
emission, energy transfer to other ions, energy dissipation to phonons, and energy
dissipation to defects and traps and are dependent on concentration of the dopant
ion, host crystal structure, and temperature of the system[10].
The rate of radiative emission of the isolated Eu3+ ion should remain
constant for a given host material, but all other processes are expected to vary with
both composition and temperature. The rate of energy transfer to other ions was
most appropriately described by Forster [11]. The dissipation of energy to phonons
has two important components. The first is the low temperature rate of
78
multiphonon transitions. This value can be determined experimentally for an
isolated ion in a host and is exponentially dependent on the distance between the
excited state and next lower available electron level [12, 13]. The second
component depends on the number and energy of available phonon modes for the
host material and the strength of ion-phonon coupling and can be determined
experimentally from the high temperature luminescence lifetime. The processes
involved in ion-ion interaction also play a part in the luminescence lifetime of a
material. This is manifested in a change in luminescence lifetime with
concentration where 1/τ=1/τ0+1/τD where τ0 is the lifetime in an infinitely dilute
composition and τD is the lifetime of ion-ion energy transfer and dependent on the
average ion-ion distance. As the ion-ion energy transfer rate increases, there is a
corresponding decrease in the luminescence lifetime. This process may be
exaggerated with increasing temperature leading to even larger effects of ion-ion
interactions for high temperature luminescence measurements.
4.4 Experimental Procedure
All materials used in this study were made from reverse-coprecipitation of
aqueous zirconium acetate, yttrium nitrate, and europium nitrate as described in
chapter two. The materials were calcined at 950°C before being pressed into
pellets and sintered at 1200°C for 2 hours. The compositions made are represented
on the ternary phase diagram in Figure 4.4.1. Raman spectroscopy was used to
confirm the presence of a tetragonal phase in all materials (Figure 4.4.2). X-ray
analysis was then used to determine the c/a ratio for the tetragonal phase or
79
determine if the cubic phase had developed if appropriate (Figure 4.4.3). The
materials consisted of two series: one where the total dopant and hence total
vacancy concentration was held at 7a/o and a second where the yttria content was
held constant and europia was added in excess of the yttria with Eu3+ ranging from
0.05 a/o to 5 a/o. This allowed two studies: the first to look at the effect of doping
concentration while keeping the total doping concentration at the level of doping
commonly used in thermal barrier materials, and the second to examine the effects
of defects introduced by oxygen vacancies.
Luminescence lifetimes and spectra in this work were all excited using a
frequency-doubled, Q-switched YAG:Nd laser emitting at 532 nm and with a pulse
length of ~10 ns. Both room and elevated temperature measurements were carried
out according to the guidelines outlined in chapter two. All Eu-doped zirconias
studied exhibited double exponential decays and were fit using Equation 1.4.2. The
long luminescence lifetime (τ2) as a function of temperature is plotted in Figures
4.4.5 and 4.4.6 for the two doping regimes.
4.5 Results and Discussion
In this work a series of parallel experiments on two types of europia-doped
yttria stabilized zirconia systems were carried out to determine the effects of dopant
concentration and oxygen vacancies on the luminescence intensity and lifetime.
The luminescence spectra reveal several interesting features about the nature of all
of the materials used in these studies. Luminescence spectra recorded for all
materials exhibit luminescence from all three transitions discussed previously
80
(5D0à7FJ=0-2) as is illustrated in Figure 4.2.1. Specifically, in the EuxY0.07-
xZr0.93O1.965 “substitutional series”, where the total dopant concentration is
maintained at 7 a/o, the symmetry has distorted from a regular tetragonal material
resulting in as many as five peaks in the 5D0à7F2 transition in addition to the
presence of the 5D0à7F0 transition and the splitting of the 5D0à
7F1 level into three
peaks instead of the two expected. Correspondingly, the Raman and x-ray spectra
(Figures 4.4.2 and 4.4.3) illustrate that these materials are single-phase with
increasing tetragonality (Figure 4.4.4) as the europium concentration was increased.
The increase in concentration of europium ions (ionic size of 1.066 Å) for yttrium
(ionic size 1.019 Å [14]) causes an increase in the distortion in the crystal and
hence an increase in the distortion on the Eu3+ site. This is the source of the
additional peaks seen in the luminescence spectra of Figure 4.2.1.
The second, EuxY0.07Zr0.93-xO1.965-x/2 “additive series” of materials,
consisting of EuO1.5 added in addition to 7 a/o YO1.5, does not show an increase in
the number of observed peaks in the luminescence spectra with concentration. In
contrast to the substitutional-series, as the europium concentration increases, the
5D0à7F2 peak splitting becomes more distinct consisting of two primary peaks at
606 and 630 nm suggesting an increase, rather than decrease, in symmetry at the
Eu3+ site. Although the intensity continues to increase for all peaks with
concentration, the relative intensity of the 5D0à7F2 transition with respect to the
5D0à7F1 transition decreases with concentration also indicating that the symmetry
at the Eu3+ site is increasing with concentration. Both the Raman and X-ray spectra
81
for these materials support this assertion. The Raman spectra, although apparently
exhibiting tetragonal structure at first glance, reveals the presence of a cubic peak
emerging at ~600 cm-1 upon peak fitting and spectral analysis, which strengthens as
the total dopant level is increased and the composition moves towards the cubic
field on the phase diagram. X-ray analysis of these materials likewise confirms the
emergence of a cubic phase as the dopant concentration is increased.
Luminescence spectra for all materials is this study are characterized by
large line widths on the range of fifty to hundreds of wavenumbers. These values
are typical of line-widths measured for Eu3+ in glassy matrixes[8, 15]. This
suggests that although these Eu-doped zirconias exhibit diffraction patterns, the
fluctuations in the local environment from one europium ion to the next due to the
defect structure of the material yields a mean measurement of Eu3+ environment by
luminescence. This is opposed to specific information of the Eu3+ ion site that is
often gleaned from site selective spectroscopy of rare earth materials in crystalline
hosts. This must be taken into consideration when studying all luminescence data
from these materials because strict rules of symmetry do not apply because of this
averaging over all the Eu3+ sites.
The luminescence intensity as a function of concentration for these two
series of materials show that for the substitutional-series, with constant vacancy
concentration, has a maximum in luminescence intensity for the 5D0à7F2 transition
at approximately 2 a/o EuO1.5. The additive-series, doped to approach the cubic
field in the ternary phase diagram, does not reach a maximum in intensity even at 5
82
a/o EuO1.5. The intensity of the luminescence spectra must be considered in
conjunction with the luminescence lifetime as a function of temperature as it
appears that both are capable of being the limiting factor in the luminescence
lifetime measurement.
The luminescence lifetime as a function of temperature data for the two
series of materials also show dissimilar results (Figures 4.4.5 and 4.4.6). Both
series exhibited the same basic variation of the lifetime decay time with
temperature; a relatively constant, almost temperature independent, value from
room temperature up to an intermediate temperature followed by and exponential
decrease with increasing temperature. The low temperature luminescence lifetime
of the substitutional series shows a trend similar to that seen in the intensity as a
function of concentration studies with an increase in luminescence lifetime with
concentration up to 2 a/o EuO1.5 and then decreases with increasing concentration
(Figure 4.4.5). The high temperature behavior displayed uniform lifetime with
temperature for concentrations up to 1 a/o where upon the rate of change of the
luminescence lifetime with temperature began to increase with increasing
concentration and can be fit using a multiphonon model. The high-temperature
luminescence lifetime concentration-dependence is dependent on the increased rate
of ion-ion interactions at those elevated concentrations and temperature.
The behavior of the additive-series is simpler to explain (Figure 4.4.6). The
low-temperature luminescence lifetime increases monotonically with concentration.
This is in contrast to behavior seen in other concentration studies where the lifetime
83
generally decreases as the dopant concentration is increased[16], but may be
attributed to in increase in decay time as the excitation energy is able to transfer
from one ion to another effectively increasing the amount of time spent in an
excited state. The high temperature behavior of these materials shows no
sensitivity to concentration and has the same values of luminescence lifetime as a
function of temperature as for Eu0.05Y0.02Zr0.93O1.965, the highest dopant
concentration of the substitutional-series. This suggests that the limiting factor in
the luminescence lifetime at high temperature for these highly doped materials
appears to be not only the amount of Eu3+, but more importantly, the total amount
of trivalent ions and hence oxygen vacancies in the material. The presence of
additional Y3+ atoms appears to have the same effect as the additional distortion of
the larger Eu3+ atoms in the substitutional-series in terms of increasing the rate of
non-radiative decay and hence, decrease the luminescence lifetime.
4.6 Summary
There are several findings of technological interest uncovered in the course
of this chapter. The first is that Eu-doped YSZ zirconia systems can be used to
measure temperatures in thermal barrier systems ranging from ~500 to 1150 ºC.
These temperatures are similar to those seen in current gas turbine system and
thermal barrier coatings. Second, and more specific to the studies of this chapter, is
a basis for the choice of compositions for specific temperature sensing applications
is outlined.
84
The substitutional-series of doping provides several benefits over the
additional-series. The first is the higher temperature sensitivity of the lower
concentration materials. Compositions ranging from 0.1 a/o to 1 a/o EuO1.5 have
exhibited temperature sensitivity up to ~1150 ºC allowing temperature sensing in
that range assuming the signal is strong enough to be efficiently collected.
Considering the intensity data of Figure 4.2.2, the highest temperature sensing
capabilities would be for a materials composition of Eu0.01Y0.06Zr0.93O1.965 both in
terms of intensity as well as temperature sensitivity. Luminescence lifetime of this
composition is also insensitive to small changes in composition making it appealing
as small changes in composition due to deposition conditions or aging of the
material would not invalidate the temperature measurement.
The additive-series of materials poses some interesting questions in terms of
selecting materials for sensor applications. Although the intensity of the
luminescence increases continually as the dopant level increases, so does the
instability of the tetragonal phase vital to TBC life. This effect coupled with the
fact that there is no change in the luminescence lifetime at elevated temperature
with concentration and the temperature sensing capabilities are limited to ~1100 ºC
suggest that this series of materials may not be well suited for temperature sensing
in thermal barrier materials.
The results so far suggest that Eu-doped materials can be employed with the
proper doping concentrations as temperature sensing materials in thermal barrier
materials. The work in this chapter has shown that doping must be done in a way
85
to preserve the tetragonal prime crystal structure as well as in a doping regime with
high concentration and low lifetime sensitivity to concentration. This generally
consists of materials in the range of Eu0.005Y0.065Zr0.93O1.965 to
Eu0.015Y0.055Zr0.93O1.965. Although doping in addition to 7 a/o YO1.5 yields
materials with many appealing features for temperature sensors, they cannot be
applied to TBC systems for the reasons discussed above. Sensors of the
compositions discussed here have been deposited by electron beam physical vapor
deposition and used to measure temperature in thermal gradients and will be
discussed in detail in chapter 6.
86
1. Gentleman, M.M. and D.R. Clarke, Luminescence sensing of temperature in pyrochlore zirconate materials for thermal barrier coatings. Surface & Coatings Technology, 2005. 200: p. 1264.
2. Blasse, G. and A. Bril, On the Eu3+ flourescence in mixed metal oxides. V. The Eu3+ flourescence in the rocksalt lattice. J. Chem. Phys., 1966. 45(9): p. 3327.
3. Blasse, G., A. Bril, and W.C. Nieuwpoort, On the Eu3+ fluorescence in mixed metal oxides Part I. J. Phys. Chem. Solids, 1966. 27: p. 1587-1592.
4. Judd, B., Optical absorption intensities of rare-earth ions. Physical Review, 1962. 127(3): p. 750-761.
5. McCauley, R.A. and F.A. Hummel, Luminescence as an indication of distortion in A2 3+ B2 4+ O7 type pyrocholores. J. Lum., 1973. 6: p. 105.
6. Nieuwpoort, W.C. and G. Blasse, Linear crystal-field terms and the 5D0 - 7F0 transition of the Eu3+ ion. Sol. Stat. Commun., 1966. 4: p. 227.
7. Ofelt, G., Intensities of crystal spectra of rare-earth ions. J. Chem. Phys., 1962. 37(3): p. 511-519.
8. Dexpert-Ghys, J., M. Faucher, and P. Caro, Site selective spectroscopy and structural analysis of ytrria-doped zirconias. J. Solid State Chem., 1984. 54: p. 179-192.
9. Johnson, P. and F. Williams, Interpretation of the dependence of luminescence efficiency on activator concetration. J. Chem. Phys., 1950. 18(11): p. 1477-1483.
10. Riseberg, L.A. and M.J. Weber, Relaxation phenomena in rare-earth luminescence, in Progress in Optics, E. Wolf, Editor. 1976, North-Holland. p. 89.
11. Forster, R., Ann. Physik, 1948. 2(55). 12. Weber, M.J., Multiphonon relaxation of rare-earth ions in yttrium
orthoaluminate. Phys. Rev. B, 1973. 8(1): p. 54. 13. Yamada, Shionoya, and Kushida, J. Phys. Soc. Jap., 1972. 32: p. 1577. 14. Shannon, R.D., Revised effective ionic radaii and systematic studies of
interatomic distances in hallides and Chalcogenides. Acta Crystallogr., 1976. A(32): p. 751.
15. Dieke, G.H., Spectra and Energy Levels of Rare Earth Ions in Crystals, ed. H.M. Crosswhite and H. Crosswhite. 1968, New York: Interscience Publishers.
16. Ryba-Romanowski, W., Effect of temperature and activator concentration on luminescence decay in erbium-doped tellurite glass. J. Lum., 1990. 46: p. 163.
87
Figure 4.2.1 Emission spectra for Eu-containing YSZ materials for several
different compositions over two doping regimes (a) substitutional EuxY0.07-
xZr0.93O1.965 and (b) additional-series EuxY0.07Zr0.93-xO1.965-x/2. All materials were excited with a 532 nm frequency doubled YAG:Nd laser.
88
Figure 4.2.2 Luminescence intensity verses concentration of Eu3+ in YSZ with
different amounts of Y3+ illustrating a peak in intensity around 2 a/o for Eu-doped materials with a constant amount of 7 a/o total stabilizer while the Eu-doped zirconias with 7 a/o YO1.5 in addition to europia doping have not reached a peak in intensity by 5 a/o EuO1.5. The data are fits according to Equation 1.7.1 with fitting parameters of z=0.74 and A=38 for the substitutional series and z=0.032 and A=1.5 for the additive series.
89
Figure 4.4.1 Diagram represented the thermodynamic phase stability of the
zirconia, yttria, europia system illustrating the compositions synthesized for this work. Where t and c denote the tetragonal and cubic phases respectively.
90
Figure 4.4.2 Raman spectra of representative materials from the (a)
substitutional- and (b) additional-series showing the characteristic peaks for tetragonal zirconia. The data have been offset for comparison. There is a shoulder in the additional-series spectra at ~600 cm-1 denoting the emergence of a cubic phase in the materials with concentration.
91
Figure 4.4.3 X-ray spectra of the (a) substitutional- and (b) additional-series of
yttria stabilized zirconias illustrating the tetragonality of the materials or the presence of the cubic peak in the additional-series.
92
Figure 4.4.4 Calculated c/a ratio for the x-ray data seen in Figure 4.4.3.
93
Figure 4.4.5 Luminescence lifetime decay as a function of temperature for the
substitutional-series. The data are fit using a multiphonon model.
94
Figure 4.4.6 Luminescence lifetime decay as a function of temperature for the
additional-series. The lines through the data have been drawn to aid the eye and are not an actual fit of the data.
95
Chapter 5: Luminescence Thermometry of (Gd,Eu)2Zr2O7
5.1 Introduction
Rare earth zirconates are an alternative to yttria stabilized zirconia as
thermal barrier materials and therefore must be studied in the context luminescence
temperature sensors. Initial studies discussed in chapter 3 have revealed a
considerable divergence of the luminescence lifetime behavior between Eu-doped
gadolinium zirconate and fully concentrated europium zirconate[1]. Gadolinium
zirconate doped with 1 a/o EuO1.5 has been shown to have a luminescence lifetime
with temperature sensitivity from 400-1150°C. Europium zirconate has been
demonstrated to show temperature sensitivity of the luminescence in excess of
1280°C. The large disparity between the temperature sensitivity of these two
compounds and more generally the concentration dependence of luminescence
lifetime of the 5D0 state on temperature in (EuxGd1-x)2Zr2O7 the solid solution series
are the subject of this chapter. Spectral and Raman analysis are utilized to
examine the relationship between crystal structure, ion-ion interaction,
luminescence intensity, and the luminescence decay behavior as the composition is
systematically varied across the phase diagram from (Eu0.01,Gd0.99)2Zr2O7 to
Eu2Zr2O7.
Luminescence lifetime measurements of temperature traditionally rely on
luminescence from dilute solutions of chromophore-doped materials where
luminescent ions can be assumed to be isolated. In dilute materials such as those
96
discussed in chapter four, the temperature dependence of luminescence lifetime is
dependent on the radiative rate of the ion as well as the rate of multiphonon de-
excitation of the luminescence[2, 3]. As the concentration of luminescent ion
increases, such as in the case of moving from (Eu0.01Gd0.99)2Zr2O7 to Eu2Zr2O7,
competing processes such as ion-ion interactions and excitation energy migration
begin to play in increasing role in the temperature dependence of the luminescence
lifetime[4]. Ion-ion interactions dominated by excitation energy migration, a
resonance energy process where an excited rare earth ion shifts from the excited
state to the ground state while an unexcited coupled ion makes the opposite
transition. Because there is no removal of excitation energy between the new
excited-state and the former excited-state, no distinction can be made between the
two states. This process, similar to concentration quenching, can accelerate the
decay rate of the excitation by energy transfer to ions acting as energy sinks. The
rate of the quenching depends on the separation between interacting ions and
therefore the doping level of the material and the likelihood of finding the excited
energy landing on a defect. When the transition between ions is resonant, such as
direct energy transfer between ions with the same excited states, the rate at which
the luminescence is quenched is very high because energy transfer can occur
without the assistance of phonons. For energy transfer between states that are of
slightly different energies, such as in the case or luminescent ions in a distorted
crystal structure similar to those that will be discussed here, the process is slow at
low temperatures and increases as temperature increases and the population of
97
phonons required to bridge the energy mismatch is increased. When Gd3+ ions are
introduced into a matrix of Eu3+ ions, they can serve to distort the energies of the
Eu3+ ions slightly and hence increase the non-radiative decay rate of the material.
One would expect that this would result in a decrease in the luminescence lifetime
in these materials.
5.2 Experimental
The europium-doped zirconates used in this study were prepared from
powders synthesized by reverse co-precipitation of aqueous rare earth nitrate and
zirconium acetate solutions. Compositions were made at 10 a/o intervals along the
solid solution line between Gd2Zr2O7 and Eu2Zr2O7 (Figure 5.2.1). Following
precipitation, the powders were calcined at 950 °C for 2 hours, pressed into pellets,
and sintered at 1200 °C for two hours. Raman spectroscopy was used to determine
the extent of ordering of the pyrochlore crystal structure in the materials. Raman
spectra for compositions from (Eu0.01Gd0.99)2Zr2O7 to Eu2Zr2O7 after sintering at
1200 °C are seen in Figure 5.2.2. The arrows in Figure 5.2.2 indicate the peak
positions for an ordered europium zirconate pyrochlore[5]. The Raman spectrum
for an ordered gadolinium zirconate pyrochlore would show a similar peak
structure with the peak energies shifted to larger Raman shifts by approximately 2-
10 cm-1. The spectra from the zirconates confirms not only that all the materials
examined are single phase, but also that there are different degrees of ordering for
different compositions. All compositions appear to be disordered after initial heat
treatments, but the shift of peaks to smaller Raman shifts, as the europium
98
concentration is increased illustrate the trend towards increased ordering with the
increase in europium concentration. This is expected since the order-disorder
transition temperature is higher in Eu2Zr2O7 than in Gd2Zr2O7.
The luminescence studies reported in this work were all performed using a
frequency double, Q-switched YAG:Nd laser emitting at 532 nm and with a pulse
length of ~10 ns. Room temperature spectra were collected using a standard silica
fiber-optic cable fitted with an in-line 532 nm laser filter to block scattered laser
light and recorded using an Ocean Optics USB 2000 (Figure 2.3.2.1). High
temperature luminescence lifetime measurements were taken according to the
details outlined in chapter two.
The spectra were fit and the normalized peak intensities were recorded and are
plotted as a function of Eu composition (Figures 5.2.3 and 5.2.4).
5.3 Room temperature luminescence
Luminescence spectra of six representative compositions in Figure 5.3.1
show the three principal luminescence lines from Eu3+ illustrating the 5D0 à 7F0,
7F1, and 7F2 transitions as labeled. All compositions show similar spectra with
variation in the relative peak intensities and splittings as a function of composition.
The similarity in luminescence spectra is in accordance with expectation that
luminescence from 4f electrons is relatively insensitive to small changes in the host
as the 4f electrons are shielded from the host by the outer 5s and 5p electrons[6].
As the composition of the europium is increased, the luminescence spectra show a
decrease in the relative intensity of the 5D0à7F0 transition with respect to the much
99
stronger 5D0à7F1 and 5D0à
7F2 transitions. This is not a surprise as this transition
is highly forbidden as it is a J=0 à J=0 transition and its presence in the
pyrochlore crystal structure suggests that the symmetry around the Eu3+ site has
been broken allowing a linear term in the crystal-field expansion [7-9]. The
breaking of D3d symmetry at the europium atom site allowing the above transition
is likely associated with the disorder seen in the Raman spectra of these materials.
The decrease in intensity of the 5D0à7F0 transition corresponds with the ordering
of the pyrochlore with composition as seen in the Raman spectra.
In the pyrochlore crystal structure a D3d symmetry at the site of the trivalent
rare-earth ion results in a splitting of the 7F1 level into two components of A2g and
Eg symmetry resulting in a singlet and doublet respectively from the 5D0 level (A1g)
as observed in Figure 5.3.1. Consequently, this splitting can then be used as a
direct indication of the distortion of the pyrochlore[10]. In Figure 5.3.1, the
splitting of this transition decreases as the total amount of europium is increased
indicating a total decrease in the distortion at the Eu3+ in the as the europium
concentration is increased. This confirms the results of the same effect observed by
the ordering in the Raman spectra.
The 5D0à7F2 transition, like the 5D0à
7F0 transition, is allowed only as a
forced electric-dipole transition in the pyrochlore crystal structure. The presence
and high intensity of this transition in all spectra recorded in this study suggest that
there is a deviation from inversion symmetry at the Eu3+ site resulting in a lowering
of symmetry to D2d. This explains not only the intensity of the peaks, but also the
100
splitting of the 7F2 level. The intensities of the 5D0à7F2 peaks as a function of
concentration are seen in Figure 5.2.3 and are fit using the Johnson and Williams
equation (Equation 1.7.1). The data shows an increase in luminescence intensity as
a function of concentration up to 30-40% europium where the intensity quickly
begins to drop with concentration. The initial rise in intensity with concentration is
due to increase in Eu3+ concentration coupled with a high level of disorder in the
material as seen by the Raman analysis. The decrease in intensity following this
maximum is not fit well using Equation 1.7.1 and is likely related to the ordering of
the pyrochlore that leads to a diminished intensity as the distortion in the
pyrochlore decreases forbidding the transition as evidenced by both the
luminescence and Raman spectra. The high concentration region of this data can
more accurately be fit using an exponential decay Figure 5.2.4.
5.4 High temperature luminescence
Measurements of luminescence lifetime of the 610 nm line as a function of
temperature were made for all compositions from room temperature until the
luminescence lifetime decay was faster than the detectability limit (~10 ns) of our
system or the intensity of the luminescence signal fell below the noise floor of our
current detection system. The 610 nm line was chosen for luminescence lifetime
measurements for its strong intensity even in the materials that exhibited ordering
in the Raman spectra. All materials, with the exception of europium zirconate,
exhibited a double exponential decay and were fit according to Equation 1.4.2.
Europium zirconate exhibited a single exponential decay over the entire
101
temperature range examined. The longer of the two exponentials (τ2), or the single
decay time in Eu2Zr2O7, are plotted as a function of temperature for all the
materials in Figure 5.4.1 a and b.
The luminescence lifetime as a function of temperature exhibited the same
characteristic features for all materials examined. From room temperature to some
intermediate temperature, the luminescence lifetime was relatively independent of
temperature. The range over which the lifetime was independent of temperature
varied with composition having the greatest extent for pure europium zirconate
ranging from room temperature to ~700 °C. At higher temperatures, the
luminescence lifetime displayed an exponential dependence of lifetime on
temperature. The rate of change of the luminescence lifetime as a function of
temperature in this region also displayed dependence on europium concentration.
The slope of the luminescence lifetime as a function of temperature varied without
any discernable trend for compositions consisting of less that 50 percent europium,
but for compositions from 60 to 100 percent the rate of change of the luminescence
lifetime as a function of temperature decreased with increasing concentration
(Figure 5.4.2). While for 60% europium the lifetime detection limit of ~10 ns was
reached by 1100 °C, pure europium zirconate retained lifetimes as long as 1 µs at
temperatures above 1200 °C.
To evaluate the effects of ordering on the luminescence lifetime and
luminescence intensity, the europium zirconate sample was examined as a function
of Raman ordering. The as-sintered pellet was given a secondary heat treatment at
102
1400 °C for 5 hours. Figure 5.4.3 clearly illustrates the increase in ordering of the
pyrochlore materials. The pyrochlore Raman peaks 311 and 397 cm-1 are seen to
have sharpened significantly upon aging and the peak at 530 cm-1 has emerged
while the shoulder at 650 cm-1, characteristic of the disordered structure, has
receded. Following the heat treatment, the luminescence lifetime as a function of
temperature was also re-examined Figure 5.4.4. The most significant change in the
data from the as-sintered state was the diminished intensity of the luminescence
signal. This is illustrated by the increase in scatter of the data at elevated
temperatures where the signal to noise ratio was decreased by ordering of the
material. Surprisingly, while the overall luminescence intensity of the signal was
decreased for intermediate temperatures, it was possible to extend the luminescence
lifetime measurement by approximately 100 °C following heat treatment allowing
temperature sensitivity up to ~1300 °C.
5.5 Discussion
The luminescence spectra reveal some interesting correlations with the
crystal structure of the materials. As the total amount of europia is increased in the
materials both the Raman spectra as well as the luminescence spectra show that
there is a continuing decrease in the distortion and disorder of the pyrochlore
crystal structure as is illustrated by the narrowing of the Raman peaks and the
decrease in the splitting of the 5D0 à 7F1 peak. This is not a surprise as the order-
disorder transition temperature is higher in Eu2Zr2O7 than in Gd2Zr2O7, but neither
luminescence intensity nor luminescence lifetime as a function of temperature
103
shows a corresponding monotonic trend. As seen in Figure 5.4.1 compositions
with between 1 and 50 a/o europia in the (EuxGd1-x)2Zr2O7 the luminescence
lifetime as function of temperature varies without a discernable trend while the
intensity increases almost linearly over that range (Figure 5.3.2). Part of the
explanation for this is likely due to the fact that these oxides are exhibiting
extensive site disorder as well as site defects whereas the more ordered materials
discussed later show more distinct trends. In this range of compositions, we
believe that there are several processes competing for control of the luminescence
lifetime as a function of temperature. These include the radiative rate of the Eu3+
ion in the host material, ion-ion interaction caused by the addition of more Eu3+
ions, and the multiphonon relaxation of luminescence and the changes in that
process as the crystal structure evolves changing phonon energies. This results in a
very complicated picture that will take further work to resolve more completely.
For compositions above ~60 a/o EuO1.5 the phenomena appears to be
simplified as the luminescence lifetime is dominated by the ion-ion interaction as a
function of temperature. This results in a linear relationship between the slopes of
the luminescence lifetime as a function of temperature over the exponential region
with composition (Figure 5.4.2). This linear relationship allows for the engineering
of compositions capable of measuring temperature over specific temperature ranges
within this range. The simplest of these oxides is europium zirconate. As
discussed in chapter two, this material, in addition to having temperature sensitivity
to the highest temperatures recorded in this study also exhibited the simplest decay
104
behavior of a single exponential decay for all temperatures observed. This suggests
that the presence of Gd3+ may be the source of the defects that cause the fast initial
decay. We propose the Eu3+ ions that are nearest to Gd3+ ions act as ions sitting
next to a defect and have a faster decay time as observed in the fast initial
luminescence decay observed in all Gd3+ containing materials. This may also
explain the decrease in luminescence lifetime as a function of temperature for these
materials as the introduction of defects would be expected to decrease the observed
lifetime by providing addition paths for non-radiative de-excitation.
These results suggest that the use of europium zirconates is necessary to
achieve the highest temperature sensitivity of from the materials. Conversely, as
show in the case of Eu2Zr2O7 this may not be the best choice in term of intensity
either because of concentration quenching as discussed in chapter one or as the
result of changes in intensity due to the evolution of the crystal’s site symmetry.
Both of these factors must be considered when choosing a material for a particular
application. For example, when choosing a sensor for the outermost surface of the
TBC, the luminescence will not need to pass through the coating and therefore
intensity will be less important than the ability to measure the high temperatures
experience by this part of the TBC therefore, the use of Eu2Zr2O7 would be the
preferred material for this sensor. Conversely, a sensor located deeper within the
coating, requiring higher intensity to pass its signal through the thickness of the
coating but with less stringent temperature sensing requirements, may be better
served with a lower Eu-containing composition.
105
5.6 Summary
The principal findings of technical interest of this work are that there is a
significant dependence of both the luminescence lifetime and intensity of these
materials on europium concentration. Gadolinium zirconate doped with 1 a/o
europia can be used to measure temperatures ranging from ~500 to 1150 °C while
pure europium zirconate has a temperature sensitivity range of ~ 700 to 1300 °C.
These temperatures are similar to those seen in current gas turbine systems and
thermal barrier systems, but as discussed in chapter three there is potential for
temperature sensing with these materials to even higher temperature through the
use of higher sensitivity detection and collection equipment. The trends governing
the temperature sensitivities of the intermediate compositions are extremely
complicated. There are two significant considerations when choosing a dopant
concentration for the temperature sensing of a (EuxGd1-x)2Zr2O7 material. The first
is the temperature range of interest and the second is the importance of the intensity
of the signal in the measurement as this study has shown that these two features of
the measurement are at odds in this system.
106
1. Gentleman, M.M. and D.R. Clarke, Luminescence sensing of temperature in pyrochlore zirconate materials for thermal barrier coatings. Surface & Coatings Technology, 2005. 200: p. 1264.
2. Moos, H., Spectroscopic relaxation processes of rare earth ions in crystals. J. Lum., 1970. 1(2): p. 106-121.
3. Weber, M.J., Multiphonon relaxation of rare-earth ions in yttrium orthoaluminate. Phys. Rev. B, 1973. 8(1): p. 54.
4. Ryba-Romanowski, W., Effect of temperature and activator concentration on luminescence decay in erbium-doped tellurite glass. J. Lum., 1990. 46: p. 163.
5. Michel, D., M. Perez Y Jorba, and R. Collongues, Study by raman spectroscopy of order-disorder phenomena occurring in some binary oxides with fluorite-related structures. J. Raman Spec., 1976. 5(3): p. 163.
6. Henderson, B. and G.F. Imbusch, Optical Spectroscopy of Inorganic Solids. 1989, New York: Oxford University Press.
7. Dieke, G.H., Spectra and Energy Levels of Rare Earth Ions in Crystals, ed. H.M. Crosswhite and H. Crosswhite. 1968, New York: Interscience Publishers.
8. Blasse, G. and A. Bril, On the Eu3+ fluorescence in mixed metal oxides II. the 5D0 - 7F0 emission. Phillips Res. Repts., 1966. 21: p. 368-378.
9. Blasse, G., A. Bril, and W.C. Nieuwpoort, On the Eu3+ fluorescence in mixed metal oxides Part I. J. Phys. Chem. Solids, 1966. 27: p. 1587-1592.
10. McCauley, R.A. and F.A. Hummel, Luminescence as an indication of distortion in A2 3+ B2 4+ O7 type pyrochlores. J. Lum., 1973. 6: p. 105.
107
Figure 5.2.1 Diagram representing the compositions synthesized for these
studies.
108
Figure 5.2.2 Raman spectra of the zirconate materials examined in this work after
heat treatment at 1200 ºC for 2 hours. Ordering in the zirconates is indicated by the presence of additional Raman peaks denoted by the arrows. The spectra illustrate an increase in ordering of the materials as the composition of EuO1.5 is increased.
109
Figure 5.2.3 Luminescence intensity of Eu3+ ions in (Eu,Gd)2Zr2O7 as a function
of concentration illustrating concentration quenching. The high concentrations at which quenching occurs coupled with the ordering data colleted in the Raman data suggests that pyrochlore ordering is primarily responsible for the quenching of luminescence at high concentrations.
110
Figure 5.2.4 Luminescence intensity of of Eu3+ ions in (Eu,Gd)2Zr2O7 as a function of concentration. The data suggests that the intensity decreases
exponentially with concentration.
111
Figure 5.3.1 Luminescence spectra in the visible for six representative
compositions of the (EuxGd1-x)2Zr2O7 system showing the three principal luminescence transitions of the trivalent europium ion.
112
Figure 5.4.1 Luminescence lifetime decay time as a function of temperature for
(EuxGd1-x)2Zr2O7 for (a) x = 0.01 to x = 0.50 and (b) x = 0.60 to x = 1.0. The lines have been drawn to aid the eye and are not a fit of the data.
113
Figure 5.4.2 Slope of the luminescence lifetime as a function of temperature
plotted as a function of concentration.
114
Figure 5.4.3 Raman spectra of Eu2Zr2O7 taken before and after heat treatment at
1400 ºC for 5 hours. Following aging the Raman spectra shows clear evidence of increased ordering of the pyrochlore structure.
115
Figure 5.4.4 Luminescence lifetime decay time as a function of temperature for
Eu2Zr2O7 before and after heat treatment at 1400 ºC for 5 hours. Aging resulted in decreased luminescence intensity resulting in the increased scatter in the data.
116
Chapter 6 Non-contact Sensing of TBC/BC Interface Temperature in a Thermal Gradient
6.1 Introduction
The use of phosphors to measure temperature dates back to the work of
Bradley [1] and has been reviewed in detail by Allison and Gillies [2]. More
recently the concept has been shown to be applicable to TBC materials and
coatings [3-6]. The previous chapters of this work explored the use of rare-earth
ions as temperature sensing dopants in thermal barrier materials. Luminescence
lifetimes of the 5D0à7F2 transition of trivalent europia ions in yttria-stabilized
zirconia exhibited the ability to measure temperature from 500-1150 °C[3, 4]. The
luminescence decay times of these materials at high temperatures have also been
shown to be sufficiently short that the measurements can be made at speeds
considerably faster than the rotational speed of gas turbines[7].
Of particular interest is the deposition of temperature-sensing layers
consisting of rare-earth doped TBC materials at the coating/bond-coat interface
used to measure the temperature at that interface. In situ temperature
measurements from such a layer can provide vital information about the remaining
life of the coating as the oxidation life of a TBC system is exponentially dependent
on the temperature. In chapter 1, it was described that thermal barrier coatings with
rare-earth doped layers were deposited at Howmet using electron-beam physical
vapor deposition (EB-PVD) and the luminescence was excited through undoped
zirconia coatings with several laser frequencies [4]. In this chapter, thin (~10 µm)
117
layers of europia-doped yttria-stabilized zirconia were deposited by EB-PVD onto
bond-coated CMSX-4 superalloy buttons to achieve sensor layers located next to
the TBC/BC interface. These coatings were then used to measure the interface
temperature while under thermal gradients similar to those in typical turbine
environments. The use of superalloys and bond coats was chosen to mimic the
conditions of a turbine environment as closely as possible. Combined with
pyrometric measurements of the coating-surface temperature and metal-surface
temperature, continuity of heat flux calculations yielded the thermal conductivity of
the coating (1.5 W/mK) and heat flux (~1 MW/m2) experienced in the tests. These
values were then used to confirm the validity of the sensor temperature
measurement. These measurements are a vital component in the integration of
sensor layers into commercial turbine coatings for health monitoring.
6.2 Sensor Layer Deposition
The coatings were deposited by electron beam evaporation at Howmet. The
ingot for EB-PVD of the sensor layers was prepared by infiltration of a standard
7YSZ ingot with aqueous Eu(NO3)3. The nitrate solution was gelled into the pores
of the ingot with NH4OH and calcined at 950°C to yield a composition of
Y0.06Eu0.01Zr0.93O1.965 as described in chapter two. The layered structure was
achieved in a single run of EB-PVD by layering the doped ingot with a standard
undoped 7YSZ ingot. The final coating consisted of two layers, an Eu-doped layer
approximately 10 µm in thickness at the base of the coating, and an undoped layer
on top, approximately 140-170 µm thick (Figure 6.1.1). The coatings were
118
deposited on CMSX-4 buttons, each 2.54 cm (1 inch) diameter, 3mm thick with
MDC-150L bond coats.
For comparison, bulk ceramic materials of the same composition were
fabricated by reverse co-precipitation with final sintering at 1200°C for 2 hours.
Raman spectroscopy was used to confirm that the bulk material was single phase
and had the tetragonal-prime crystal structure found in the coated material.
6.3 Luminescence Lifetime Calibrations
On the basis of the work in chapter three [3], the luminescence emission
spectra were excited for both the bulk and coated materials using a frequency
doubled, Q-switched, YAG:Nd laser emitting at 532 nm and having a pulse length
of ~10 ns. To develop the calibration curves relating luminescence lifetime-decay
to temperature, luminescence measurements were first recorded under isothermal
conditions inside a box furnace with a thermocouple placed next to the coatings and
bulk samples to record temperatures. The luminescence lifetimes were recorded
using the high temperature luminescence setup described in detail in chapter two
(to measure the luminescence lifetime of the 606 nm line of the Eu3+
luminescence). Isothermal measurements of luminescence intensity as a function
of time were taken for temperatures between room temperature and the
detectability limit of the collection electronics (approximately 0.01µs decay time).
These measurements were repeated for selected temperatures after both the thermal
gradient testing discussed below as well as after a series of one hour cycles at 1150
°C. These subsequent studies were done to confirm that there was no change in the
119
luminescence lifetime as a function of temperature as a result of the thermal
gradient testing or upon more extensive aging of the materials used in this study.
6.4 Luminescence Lifetime Measurement in a Thermal Gradient
Several coatings were subjected to high heat-flux exposure at NASA Glenn
Research Center and luminescence lifetime measurements were made under
different heat flux conditions. In the NASA rig, a 3.0 kW CO2 laser (10.6µm
wavelength) was used to heat the outer surface of the coating and the back of the
superalloy was cooled with a high-pressure compressed air jet (Figure 6.3.1). This
generated a heat flux through the coating and the underlying superalloy. The
instrumental details of the high heat-flux laser rig have been described elsewhere
[8]. The temperature at the top surface of the coating was monitored using an 8 µm
infrared pyrometer while a series of two-color pyrometers were used to monitor the
temperature of the metal surface. The CO2 laser delivered a constant power of
1060 W to the top surface of the coating and by adjusting the compressed air flow,
the metal temperature, the top surface temperature, and the heat flux could be
varied.
To interrogate the sensor layer while the sample was exposed to the thermal
gradient, a solid-state frequency doubled YAG:Nd laser emitting at 532 nm was
directed to illuminate the center of the coating surface while mounted in the high
heat-flux rig. As in the isothermal furnace tests, the end of a sapphire light pipe
was positioned to collect the excited luminescence signal and the luminescence
intensity was recorded and averaged in the same way as described in the previous
120
section. The one difference was that for safety reasons the measurements were
made remotely under computer control.
6.5 Results
First, we present the luminescence lifetime as a function of temperature
calibration data. As described in the previous section, the calibration curves of
luminescence lifetime as a function of temperature were determined for both bulk
ceramic samples and coatings. Both forms of the sensor material exhibited double
exponential decays over the entire temperature range and were therefore fitted
using the function:
2121
ττ tt eIeII −− += Equation 1.4.2
where I is the intensity and τ the luminescence lifetime for the individual decays
(Figure 6.4.1.). The second, longer luminescence lifetime (τ2) was used for
calibration purposes and its value is shown as a function of temperature for both the
bulk and coatings in Figure 6.4.2. The second, longer decay was chosen for its
length, which allowed measurement to higher temperatures than the shorter decay,
and because it was not dependent on the excitation intensity. The luminescence
lifetime data was consistent with a temperature dependent multi-phonon relaxation
model[9]:
( ) pmprtotal nwww )1(0 ++=
Equation 1.4.5
where wr is the rate of radiative emission, wmp(0) is the rate of multiphonon
121
emission at 0 K, ( )[ ] 11/exp −−= kThn ν is the phonon population density, v is the
frequency of the phonon, and p is the number of phonons required to bridge the
energy gap corresponding to the energy of the emitted luminescence, 16,500 cm-1.
The phonon energy was taken to be that of the highest energy Raman band for
tetragonal YSZ. Its energy and temperature dependence have been determined
recently [7] and this information was used in fitting the calibration data. The fitting
of the data to the multiphonon model was then used to represent the calibration
curve in converting lifetime measurements made under the thermal gradient tests to
the TBC/BC interface temperature. The fit of the data resulted in values of
wr=561 s-1 and wmp=0.111 s-1 for the sensor layers deposited here. Superimposed
on Figure 6.4.2 are the lifetime as a function of temperature data taken after the
heat flux measurements at NASA as well as after thermal cycling done at UCSB.
They show that there is no change in the luminescence lifetime as a result of the
thermal cycling indicating that the sensor will remain accurate under similar turbine
cycling. Additionally, the sample failed after ~550 1 hour cycles at 1150 °C which
is typical of similar materials without sensor layers under the same conditions.
Two examples of our temperature measurements are shown in figure 6.4.3
for a variety of heat flux conditions produced by changing the cooling rate on the
back of the superalloy. One plot is for a coating that is 127 µm thick and the other
for a 178 µm thick coating. In all cases, the sensor layer is 10 µm thick. In these
plots, the pyrometric measurements of the top surface of the coating and the
122
luminescence measurements are both plotted as a function of the pyrometric
measurements of the alloy surface temperature.
In addition to the measurements of the TBC/BC temperature, the steady
state, one-dimensional heat flow in the NASA rig enabled the thermal conductivity
of the coating, κTBC, to be determined. Assuming one-dimensional heat flow, the
heat flux, q, through the coating and the superalloy are the same:
alloy
alloyalloy
TBC
TBCTBC h
ThTq
Δ=
Δ= κκ .
Equation 6.3.1
ΔT is the temperature drop across either the coating or alloy, κ is the thermal
conductivity of the layer, and h is the thickness of the layer. The temperature drop
across both the coating and metal were determined using the temperature
measurement of luminescence at the interface and the pyrometry measurements of
the top and bottom surfaces. Using the reported temperature dependence of the
thermal conductivity of CMSX-4 alloy [10], the thermal conductivity of the coating
was calculated to be 1.5 W/mK over the range of 850 to 1150oC. The
corresponding heat fluxes were determined from equation 3. These ranged from
0.4 to 1 MW/m2.
6.6 Discussion
The principal finding of this work is that luminescence of a Eu-doped YSZ
sensor layer buried under a standard YSZ coating can be used to measure the
temperature at the TBC/BC interface in the presence of a thermal gradient using
123
typical thermal gradients and temperatures of coatings in use in engines today.
These results demonstrate that the luminescence lifetime method of measuring
temperature can be implemented using europium ions incorporated within the
crystal structure of current EB-PVD thermal barrier coatings and the temperature
measurement can be localized by controlling the deposition process. The fact that
the luminescence was localized to the TBC/BC interface was confirmed by
observations of the cross-section of one of the coatings after the thermal gradient
measurements were taken. This is illustrated in Figure 6.1.1, which is a
superimposition of a normal optical micrograph, to reveal the coating and the alloy,
over a luminescence image recorded using the luminescence from the Eu-ions.
There is also no apparent diffusion of the Eu3+ ions into the rest of the coating after
either the isothermal treatments or cycling in the high heat flux rig. This verifies
that the temperatures measured by the luminescence lifetime decay technique
provide us with the temperature of the TBC/BC interface and not some other depth
of the coating. This is vitally important since the technique is an average of the
luminescence from all ions excited. If the ions had been distributed at different
depths then the luminescence lifetime would have been a convolution of the
lifetimes at different depths within the thermal gradient. In principal, this can be
detected, as the luminescence decay would have been a “stretched” exponential
rather than a pure exponential. This and other higher magnification micrographs
also illustrates that there is no discontinuity in the coating caused by the switch
from the Eu-doping to the undoped ingot during deposition. The presence of
124
uninterrupted single crystal columns suggests that the addition of the sensor layer
did not disturb the integrity of the coating.
Our current detection system limits us to measurements of interface
temperatures at about 1150°C. With more efficient collection optics and detectors
as well as higher speed electronics, higher interface temperatures can be measured,
allowing the technique and the Eu-doped coatings to be used at still higher
temperatures. The fitting to the multi-phonon model indicates that increasing the
detection limit by only a factor of ten would enable us to measure temperatures as
high as 1250°C. The other attribute of the Eu-doped coatings is that the
luminescence lifetime is sufficiently short that the technique has the potential for
measuring the temperature of rotating blades and other components[3].
The lifetime data presented also shows several interesting features
apparently peculiar to the Eu3+ luminescence in yttria-stabilized zirconia. First, the
luminescence lifetime of the second, longer exponential decay, τ2, is the same for
both the bulk and coated materials over the temperature range explored up to
1150oC. Rare-earth ion luminescence is known to have very little sensitivity to the
host material because the optically active 4f electrons of the ions are shielded by the
outer 5s and 5d shells[11]. This shielding may also create an insensitivity of the
luminescence lifetime in the slightly different environments of the bulk and coated
material. By contrast, the fast decay, τ1, which is usually attributed to
luminescence quenching by energy transfer to defects, shows sensitivity to the
changes in local environment from bulk to coated material [12]. Specifically in the
125
coating, the intensity of the fast decay is higher with respect to the that seen in the
bulk and the luminescence decay time from the faster decay in the coating is
approximately 25% faster than the decay recorded from the bulk. The origin of this
difference is not known but may be related to the presence of the fine porosity
within the coatings. As mentioned earlier, the temperature dependence is attributed
to non-radiative transitions associated with the excitation of lattice phonons. In this
mechanism, the energy of the excited state, instead of being radiated as a detectable
photon, excites phonons through an ion-phonon coupling mode. The simplest
model ascribes this to phonons all of the same energy and assumes that this is the
highest energy phonon. In the case of YSZ, the B1g phonon mode, with room
temperature energy of 648 cm-1 determined from Raman spectroscopy, is likely to
be the primary phonon mode responsible for the multiphonon relaxation. Using the
measured temperature dependence of the Raman spectra, between 25 and 27
phonons are required for the non-radiative transition over the temperature range
examined. Although the energy of the B1g mode provides a good fit of the data for
the dilute concentrations of Eu3+ used in the sensors described in this work, work
done in chapter 4 illustrates that ion-ion energy transfer is likely more important
than the multiphonon relaxation rates in the case of more concentrated sensor
materials. Nevertheless, the fact that the lifetime data reported here fits the multi-
phonon relaxation model suggests that the Raman spectra provide a valuable guide
to selecting other combinations of luminescent ions and coating materials for future
high-temperature thermometry.
126
Another interesting feature of the luminescence of these sensor layers is that
there may be an effect on the emissivity of the TBC and hence the ability to
measure temperature of the TBC using infrared pyrometry when these sensor layers
are in place. Figure 6.4.4 illustrate the measurements taken by both pyrometry as
well as luminescence for a 146 µm coating deposited in the same run as other
samples examined. Superimposition on this data are the calculated values of top
surface temperature assuming the luminescence measurements are correct and the
thermal conductivity is 1.5 W/m/K as confirmed by Ted Bennett’s group by the
phase shift method[13]. This calculation illustrates that at lower temperature the
pyrometer underestimates the temperature of the surface of the TBC and
overestimates that temperature at higher temperatures. The underestimate at lower
temperatures is likely due to an increase in the optical penetration depth of the
pyrometer at those temperatures. At higher temperatures the overestimate may be
due to thermal activation of 7F levels of the europium ions in the sensor layer
leading to luminescence at wavelengths where the pyrometer is reading. This
would result in an increase in the measured temperature by pyrometery and an
increasing error as the temperature as is seen in Figure 6.4.4. This phenomenon is
currently under examination.
127
References 1. Bradley, L.C., Rev. Sci. Instrum., 1953. 24: p. 219. 2. Allison, S.W., L.A. Boatner, and G.T. Gillies, Characterization of high-
temperature themographic phosphors: spectral prperties of LuPO4:Dy(1%), Eu(2%). Appl. Optics, 1995. 34(25): p. 5624.
3. Gentleman, M.M. and D.R. Clarke, Luminescence sensing of temperature in pyrochlore zirconate materials for thermal barrier coatings. Surface & Coatings Technology, 2005. 200: p. 1264.
4. Gentleman, M.M. and D.R. Clarke, Concepts for luminescence sensing of thermal barrier coatings. Surface & Coatings Technology, 2004. 188-189: p. 93.
5. Allison, S. and G. Gillies, Remote thermometry with thermographic phosphors: Instrumentation and applications. Rev. Sci. Instrum., 1988. 68(7): p. 2615-2650.
6. Feist, J.P., A.L. Heyes, and J.R. Nicholls, Phosphor thermometry in electron beam physical vapor deposition produced thermal barrier coating doped with dysprosium. Proc. Instn. Mech. Engrs., 2001. 215(G): p. 333.
7. Gentleman, M.M., et al., Non-contact methods for measuring thermal barrier coating temperatures. Int. J. Appl. Ceram. Technol., 2006. 3(2): p. 105.
8. Zhu, D.M. and R.A. Miller, Thermal conductivity and elastic modulus evolution of thermal barrier coatings under high heat flux conditions. J. Thermal Spray Tech., 2000. 9(2): p. 175.
9. Moos, H.W., Spectorscopic relaxation processes of rare earth ions in crystals. J. Lum., 1970. 1(2): p. 106.
10. CMSX Property Data. 1994, Cannon-Muskegon Corporation: Michigan. 11. Henderson, B. and G.F. Imbusch, Optical Spectroscopy of Inorganic Solids.
1989, New York: Oxford University Press. 12. Kallendonk, F. and G. Blasse, Luminescence and energy transfer in
EuAl3B4O12. J. Chem. Phys., 1981. 75(2): p. 561. 13. Yu, F.L. and T.D. Bennett, Phase of thermal emission spectroscopy for
properties measurements of delaminating thermal barrier coatings. JOURNAL OF APPLIED PHYSICS, 2005. 98: p. 10.
128
Figure 6.1.1 Optical image of the cross-section of a thermal barrier coating
illustrating the location of YSZ:Eu sensor layer appearing red between the Superalloy and YSZ coating. The image is a superimposition of a white light image of the coating and a luminescence image from the Eu-layer obtained with UV excitation of the same region.
129
Figure 6.3.1 Experimental arrangement used to measure the temperature in contact
with the superalloy using the luminescence sensor in a temperature gradient.
130
Figure 6.4.1 Representative luminescence lifetime decays for the bulk and coated
samples at 700°C. While the long decays have the same slope, the faster decay varies in the two materials. The longer decay is used for temperature measurement.
131
Figure 6.4.2 Comparison of the luminescence lifetime decays as a function of
temperature for the bulk and coated materials. The decays are fit using a multiphonon relaxation model, equation 1.4.5 in the text. Also included are lifetime data after thermal gradient testing and thermal cycling illustrating that the luminescence lifetime as a function of temperature does not change with aging of the material.
132
Figure 6.4.3 Temperatures measured at the top of the TBC and bottom surface of
the CMSX4 alloy by pyrometry and the TBC/BC interface temperature by luminescence for three different coatings. (a) 127 µm and (b) 178 µm thick coatings.
133
Figure 6.4.4 Temperatures measured at the top of the TBC and bottom surface of
the CMSX4 alloy by pyrometry and the TBC/BC interface temperature by luminescence for a 146 µm coating. The blue line illustrates the calculated values for the 8 µm pyrometer assuming a thermal conductivity of the coating at 1.5 W/m/K.
134
Chapter 7 Conclusions and Future work
The first and most compelling conclusion of this thesis is that it is possible
to develop non-contact heath monitoring and temperature sensors for thermal
barrier coatings using luminescence techniques. Thermal barrier materials doped
with rare-earth ions have been shown to not only exhibit an ability to substitute into
thermal barrier material without negative effects on the coating life as seen in the
thermal cycling data in chapter 6, but also that they are capable of being used to
measure temperatures up to those observed in typical turbine environments. Yttria
stabilized zirconia doped with europia can be used to measure temperature up to at
least 1150 °C while Eu2Zr2O7 has temperature sensing capabilities approaching
1300 °C. This work has also shown that measurements made on bulk materials are
consistent with those made on coatings of the same composition and can therefore
be used to examine new dopant compositions easily through precursor routes
without the use of expensive deposition of all compositions of interest. Finally, the
measurement of temperature from a buried sensor layer within an EB-PVD coating
while under a thermal gradient was demonstrated resulting in not only a way to
monitor the temperature at the TBC/BC interface, a major driving force in this
work, but also a direct measurement of the thermal conductivity and heat fluxes in
the thermal barrier coating. Simple calculations have also shown that the
luminescence lifetime decays of these Eu-doped sensors are consistent with those
needed for in situ measurement on rotating engine components.
135
Future work in this area should focus on understanding the sources of
uncertainty in these experiments and on ways to diminish them. Although the
current temperature measurement techniques available for thermal barrier coatings
(infrared pyrometry) have an error similar to that of this technique, luminescence
provides us with an opportunity to decrease the error through a series of
improvements to the technique. This thus makes luminescence temperature
measurements much more valuable in terms of their potential for high sensitivity.
Possible avenues for these improvements include the exploration of more efficient
optics and electronics as well as room for the development of a more precise means
for the automated collection and analysis of the luminescence lifetime signal.
Additionally, the use of other dopant ions, including Tb3+ and Dy3+ show
promise and in each case need to be examined more thoroughly. In early studies,
terbium exhibited a tendency to substitute into the YSZ system as a tetravalent
species as opposed to the desired trivalent ion. This resulted in a weak
luminescence signal with a relatively short luminescence lifetime decays (~1 µs) at
room temperatures. It has been suggested that the use of a chloride based precursor
as opposed to the nitrates used in these studies, or co-doping with pentavalent ions
may be helpful in resolving this issue and should be explored further. In the case of
Dy-doped systems, the main concern is finding an efficient excitation source for the
luminescence. For these studies, short of purchasing new lasers of appropriate
laser frequencies (~440 nm), it may be possible to use broadband flash lamps for
excitation.
136
The effects of the sensor layer on both the oxidation characteristics of the
TGO as well as the life of the coating must be examined. Preliminary work has
suggested that the presence of the Eu-doped layer next to the TGO may have
changed the degree or rumpling observed during oxidation of the material, but
insufficient data was collected to quantify the extent of this effect if any.
Moreover, the long-term stability of the sensors in terms of phase stability and
luminescence lifetime must be established before implementation.
Finally, as Figure 7.1 illustrates, there is a significant change in the
luminescence lifetime as a function of temperature for the same materials with two
different excitation sources. This should be examined to determined if the effect
can be used to further increase the temperature sensing capabilities of rare earth
doped zirconias.
137
Figure 7.1 1 a/o Eu-doped YSZ luminescence lifetime as a function of
temperature for two different excitation energies 532 nm and 248 nm. The reason for the increase in lifetime in the case of 248 nm excitation has yet to be determined.