WO2007130862A2 - Thermoelectric properties of ceramic thin film thermocouples - Google Patents

Abstract

Thin film ceramic thermocouples are used to assess temperatures beyond 14000C in the hot sections of gas turbine engines. Several ceramic materials were systematically examined as thermoelements including indium tin oxide (ITO), zinc oxide (ZnO) and a NiCrCoAlY/alumina nanocomposite. These ceramic thermoelements were initially tested relative to a platinum reference electrode and the relationship between process parameters and thermoelectric properties, including Seebeck coefficient, sensitivity and voltage/temperature behavior was established. Bi-ceramic junctions comprised of ZnO and ITO generated a very large electromotive force at low temperatures but lacked high temperature stability. When ITO was combined with a NiCoCrAlY/alumina nanocomposite, a very large Seebeck coefficient with excellent emf/temperature behavior was realized. A ceramic thermocouple based on this combination was demonstrated at temperatures up to 1200°C.

Description

THERMOELECTRIC PROPERTIES OF CERAMIC THIN FILM THERMOCOUPLES

PRIORITY INFORMATION

This application claims priority to U.S. Provisional Patent Application Serial No. 60/797,302 filed on May 3, 2006, which is incorporated herein in its entirety.

BACKGROUND OF THE INVENTION

The development of advanced propulsion systems employing new materials and designs requires the continuous monitoring of engine components under typical operating conditions. Thus, structural models can be validated and newly developed materials can be evaluated in-situ with ceramic sensor elements that can withstand the harsh environments associated with gas turbine engines. Therefore, engine health monitoring is becoming essential since these temperatures can severely influence reliability, lifetime and performance issues of the respective components. For example, the assessment of the temperature distribution or pattern factor in a gas turbine engine combustion chamber is critical since the lack of proper fuel burning can severely damage the components comprising the chamber. In order to meet the long-term instrumentation needs for engine health monitoring in these harsh environments, thin film thermocouples were developed for in-situ temperature measurement.

Thin film sensors are well suited for these kinds of measurements since their negligible inertial mass has minimal impact on vibration patterns and will not be affected by the high g-loading associated with rotating components. Since these sensors are directly deposited onto a component, the sensors are in intimate contact with the component's surface and no adhesives are required. In addition, their thickness will not adversely affect gas flow through the engine.

SUMMARY OF THE INVENTION Ceramic thermocouples have certain advantages over precious metal thermocouples typically used in the gas turbine engine environment, when deposited in thin film form, these include little or no electromigration, a high melting point and chemical stability at elevated temperatures. The aforementioned ceramic thermocouples will not undergo phase changes and have a large and stable Seebeck coefficient when thermally cycled between room temperature and 1500⁰C. Furthermore, ceramics are more oxidation resistant than metals and ceramics are not as costly as platinum and rhodium based thermocouples. In addition to these properties, their superior chemical stability at elevated temperature makes ceramic sensors promising candidates for other electrical devices at temperatures up to-1500°C. Some of the potentially most capable ceramics in terms of expected thermoelectric properties, are the families of composites of indium tin oxide (ITO), tin oxide, aluminum oxided doped with zinc oxide (AZO) and a cermet (nanocomposite) based on NiCoCrAlY and aluminum oxide. Since ITO-based sensors have been developed for strain measurements of engine components at temperatures up to 1500⁰C in air ambients, there is a need to produce temperature sensors and/or remote power supplies at the same location on a turbine blade with minimal processing.

The preparation parameters used to fabricate the ceramic films were optimized with respect to thermoelectric response (Seebeck coefficient). Based on the resulting thermoelectric properties of the ceramic films relative to a platinum reference element, bi- ceramic junctions were prepared and tested. The composition was optimized in an effort to maximize the thermoelectric response which is directly proportional to the applied temperature gradient according to the following equation:

V = kΔT (1)

where ΔT = applied temperature gradient and k = Seebeck coefficient. As shown in Equation 1, the output is influenced by the material constant k the Seebeck coefficient or thermoelectromotive force. Since the thermoelectric power among others things is dependent on the composition of the thermoelements, the thermoelectric properties were systematically investigated as a function of process parameters. A ceramic thermocouple for high temperature measurements described within can be integrated with a ceramic strain sensor. A thin film thermocouple with one thermoelement based on a nanocomposite (NiCoCrAlY and aluminum oxide) and the other based on an ITO alloy have been demonstrated. An integrated temperature and strain measurement can be made if one leg of the thermocouple is made from an ITO that is the exact same composition that is used for the strain gauge, thus one leg has an ITO deposition and the other leg is from the deposition of the nanocomposite. In this way simultaneous strain/temperature measurements could be made at elevated temperatures with minimal thin film processing. The thermocouples have been thermally cycled up to ten times from room temperature up to 1200°C. However they have also been brought up to higher temperatures such as 1500⁰C and higher.

These and other features and objectives of the present invention will now be described in greater detail with reference to the accompanying drawings, wherein:

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a graph of thermoelectric response of an ITO thermoelement relative to platinum;

Figure 2 is a graph of Voltage/temperature behavior of an ITO element relative to platinum;

Figure 3 a is a graph of the thermoelectric response of an ITO thermoelement relative to platinum; Figure 3b is a graph of Voltage/temperature behavior of an ITO thermoelement relative to platinum;

Figure 4 is a graph of thermoelectric response of an ITO thermoelement relative to an ITO element;

Figure 5 is a graph of thermoelectric response of a ZnO thermoelement relative to platinum with ZnO element sputtered in pure argon;

Figure 6 is a TEM micrograph of an as-deposited NiCoCrAlY/alumina nanocomposite;

Figure 7 is a graph of thermoelectric response of a nanocomposite thermoelement relative to platinum with a nanocomposite element sputtered in pure argon; Figure 8a is a graph of Voltage/temperature behavior of a nanocomposite/ITO thermocouple with the ITO element sputtered in pure nitrogen; and

Figure 8b is a graph of thermoelectric response of a nanocomposite/ITO thermocouple with the ITO element sputtered in pure nitrogen.

DETAILED DESCRIPTION OF THE INVENTION

To examine the use of ceramic thermoelements in thermocouples and thermoelectric generators, thin films were deposited on rectangular shaped high purity (98.6%) alumina substrates. Beams measuring 6 inches by 1 inch respectively were laser cut by International Ceramic Engineering. All reactively sputtered films were deposited in an MRC 822 machine (available from Materials Research Corporation in Orangeburg, New York) whereas all rf sputtered films were deposited in an MRC 8667 sputtering system (available from Materials Research Corporation). Prior to deposition of the ceramic films, the substrates were rinsed in methanol, ethanol and deionized water, dried in dry nitrogen and coated with a layer of high purity alumina (AI₂O₃). This prevents diffusion of ionic impurities from the substrate and promotes better adhesion. The electrical properties of the thermoelements were more stable as a result of the adhesion layer. The alumina was sputtered in an atmosphere of pure argon (9mtorr) using a high purity alumina target (99.99%) and had a nominal thickness of 0.2μm. After alumina deposition, the substrates were heated to 800⁰C to densify the coating and to further enhance the bonding between the coating and the substrate as well as eliminate residual stress and point defects in the films. Adherent platinum films were subsequently sputtered using a high purity 6-inch platinum target (99.99% pure) and were used as the reference electrode in all thermocouples as well as ohmic contacts to the ceramic sensors. They were prepared by placing an aluminum shadow mask over the alumina substrates to create the desired thin film patterns. The platinum elements were sputtered at an rf power of 350Watts and 2000 Volts in 9mtorr argon. All ITO films were prepared by rf sputtering at 350 watts (power density of 1.38

W/cm^) and 1800 volts. A background pressure of 10^~6 torr was maintained in the vacuum chamber prior to sputtering and a high-density ceramic target (6 inch diameter) with a nominal composition of 90 wt% In₂O₃ and 10 wt% SnO₂ was used for all ITO depositions.

To investigate the electrical properties of the ITO thermoelements prepared under different sputtering conditions, the oxygen and nitrogen partial pressures were systematically varied from zero to 3mtorr while the argon partial pressure was maintained at 9mtorr. The effect of film thickness on thermoelectric response was investigated by varying deposition time.

The aluminum doped ZnO thermoelement was deposited from a 2-inch diameter target and the cermet consisting of 20 wt% Al₂O₃ and 80 wt.% NiCoCrAlY nanocomposite was deposited from a 4-inch diameter thermally sprayed composite target. Sputtering powers of

150 and 300 W were used for ZnO depositions to achieve power densities comparable to those used for ITO deposition.

The thickness of the deposited thin film was measured using a DEKTAK II profϊlometer, available from DEKTAK. After thin film deposition, the thermoelements were heat treated in a nitrogen-rich environment to remove residual argon trapped inside the deposited film. The ceramic samples along with a process monitor were heated to a temperature of 850⁰C at a rate of 3°C/hr. To evaluate the electrical properties, the furnace was maintained at 850⁰C at which point the temperature was held for 300 minutes and cooled to room temperature at the same rate. The resistivity was measured as deposited and heat treated condition. After heat treatment the thermocouples were then placed into the 7-inch hot zone of a Deltech tube furnace where a temperature gradient was applied along the sample. In order to increase the temperature difference between the hot and cold junctions, a heat shield was placed in the middle of the sample. The furnace was heated to temperatures on the order of 1350⁰C at a heating rate of 3°C/min and held for 60 minutes to establish thermal equilibrium and then cooled to room temperature at the same rate. All electrical measurements were done in air. The hot and cold junction temperatures as well as the generated emf were monitored by a USB data acquisition system (Personal DAQ 54 by I/O Tech) and associated software (Personal DaqView). Type S thermocouples were used to record the temperatures at the respective locations of the sample, as well as the thermocouple output (generated emf). For the testing of the ITO-Pt and ITO-nanocomposite thermocouples, the cold junction temperature was maintained at room temperature using a specially designed water-cooled aluminum block. The alumina substrate was attached to the end of the water-cooled block to insure that the cold junction temperature remained relatively constant throughout the testing protocol.

Table 1. Preparation conditions used to form ITO thermoelements in ITO/Pt thermocouples

Table 1 summarizes the preparation conditions used for the deposition of the ITO thermoelements in terms of O₂ partial pressure [mtorr], N₂ partial pressure [mtorr], and Ar partial pressure [mtorr]. An ITO/Pt thermocouple with the ITO element being prepared in an atmosphere containing 1.9mtorr O₂, 2.2mtorr N₂, and 9mtorr Argon is shown in Figure 1. The hot and cold (dark and light grey curve) junction temperatures along with the output voltage of the thermocouple (black curve) were recorded at a sample rate of 1/10 sec. The thermoelectric response shows a linear increase in voltage with respect to temperatures up to 1200⁰C, where heating rate starts to decrease, eventually reaching a peak output voltage of 62mV at a temperature of 1261°C. Even though similar Seebeck coefficients were observed during heating and cooling (i.e. delta temperature vs. voltage graph shown in Figure 2) it was apparent, that there was some hysteresis. This phenomenon was due to the interaction of electrons and phonons (quanta of lattice vibrational energy) in the ceramic elements. Phonon scattering from other phonons and from other impurities influence the flow of electrons along a thermal gradient and therefore, influence the emf produced by the device. The degree to which these phonons drag electrons is highly dependent on temperature. At low temperatures only a small voltage is generated thus, the interaction between electrons and phonons is very small and thus is represented by the negative deviation from the linear temperature profile. As temperature increases, more phonons become available and drag the electrons along the gradient. This results in an increase of the temperature/emf slope and temporarily a linear behavior. At high temperatures, phonon-phonon interactions become dominant and electrons are no longer dragged along. This inhibits the flow of electrons causing the thermoelectric power to level out. Due to this deviation from linearity, the Seebeck coefficient was approximated using a 3^rcWder polynomial for the heating and the cooling cycles of the experiment. This polynomial gives the relationship between the imposed temperature gradient (T) and the generated voltage (V) over the entire linear range during heating.

* The two ITO compositions, had almost identical responses Table 2. Polynomials used to describe the emf/temperature behavior of different ITO/Pt thermocouples. H/ Denotes heating cycle, C refers to the cooling cycle.

Table 2 summarizes the terms in the polynomial (Equation T) used to describe the terms in the polynomial (Equation 2) used to describe the voltage/temperature behavior for the heating and the cooling cycles of the various thermocouples tested.

V(T)= A. ≠ + B. T² +C.T + D (2)

The electromotive force (emf) as a function of the imposed temperature gradient for an ITO element prepared in an atmosphere containing 1.9mtorr O₂, 2.2mtorr N₂, and 9mtorr Ar relative to platinum (Figure 2) shows that ITO prepared under these conditions showed little hysteresis between cooling and heating cycles indicating good electrical stability and a reproducible Seebeck coefficient.

Figure 3 shows the thermoelectric response of an ITO/Pt thermocouple where the ITO was prepared in an atmosphere of 1.2mtorr O₂, 3.1mtorr N₂, and 9mtorr Argon. The hot junction temperature (dark grey curve) increased and decreased linearly, and a linear thermal gradient was applied to the sample. The cold junction temperature was represented by the light grey curve. The voltage/temperature behavior shown in Figure 3b indicated an S shaped curve typical of a ceramic thermocouple with both negative and positive deviations from linearity during heating but a rather linear voltage response during cooling. The thermoelectric properties were evaluated in terms of the Seebeck coefficient using a 3^rc* order polynomial during the increase and decrease of the thermal gradient and the values for A, B, C, and D in the Equation 2 are shown in Table 2.

Table 3. Thermoelectric properties of the ITO thin film thermoelements prepared under different conditions

Table 3 summarizes the thermoelectric response of several ITO thermoelements prepared under different conditions relative to platinum. It not only includes Seebeck coefficients of the various ITO compositions but also indicates any deviation from the Type S behavior (+ and -). Also, the deviation of the sensor output from linearity is listed along with the deviation associated with poor sensor performance (- -) and small deviations from linear temperature profile (+). The assessment of thermoelectric properties in terms of Seebeck coefficients in general is very difficult since complex polynomial functions are usually used to describe the thermoelectric response to an applied temperature gradient. The calculated coefficients for the different ITO compositions varied greatly and in some cases even the sign of the coefficient changed depending on the temperature range. The Seebeck coefficients listed in Table 3 were based on a selected temperature range between 500⁰C and 1000⁰C. The choice for bi-ceramic junction was made on the basis of the high temperature performance and least deviation from the linear temperature profile. The ITO composition sputtered in pure argon had the best results, having little deviation from linearity at low temperatures and a voltage increase at elevated temperatures.

The thermoelectric response of a bi-ceramic junction comprised of two ITO thermoelements sputtered in atmospheres containing 2.9mtorr O₂, 2.7mtorr N₂, and 9mtorr Argon and in a pure Argon atmosphere (9mtorr), is shown in Figure 4. A maximum hot junction temperature (dark grey curve) of 1203⁰C was applied to the bi-ceramic junction with a maximum thermal gradient of 645°C. As can be seen from the figure, the voltage output (black curve) was very unstable and thus, was difficult to relate to the applied temperature gradient. As far as magnitude of the signal was concerned, the voltage peaked at 2mV and did not coincide with the peak thermal gradient. The relatively small peak voltage was due to the small difference in charge carrier concentrations in the two ITO films comprising the thermocouple. A similar thermoelectric response was recorded for another bi-ceramic junction whose ITO thermoelements were prepared under atmospheres containing 9mtorr Argon and 2.7mtorr O₂, 2.2mtorr N₂, and 9mtorr Argon, respectively.

A ZnO film doped with Al₂ O₃ was used to create a bi-ceramic junction with ITO. ZnO is a p-type semiconductor and when combined with n-type semiconductors such as ITO, there is considerable potential to produce a very large thermoelectric response. Figure 5 shows the thermoelectric response of a ZnO/Pt thermocouple, where the dark grey curve represents the hot junction temperature. The ZnO elements were sputtered in pure argon (9mtorr) at an rf power of 150W. A maximum thermal gradient of 516⁰C was achieved at 1325°C with a corresponding voltage (black curve) of 13 ImV being obtained for this thermocouple pair. A very similar thermoelectric response was observed for a ZnO/ITO thermocouple. Here, a linear increase in thermoelectric response was achieved in the presence of a small thermal gradient (527°C) and a maximum voltage of 14OmV was achieved.

In order to stabilize the output signal of thermocouple comprised of ZnO thermoelements at elevated temperatures, a protective alumina layer was sputtered over the elements. The thermocouples comprised of ZnO thermoelements showed unstable behavior and various inflection points. As the thermal gradient was increased, the voltage rapidly increased but with the continued heating however, the voltage decreased and finally disappeared.

A nanocomposite based on NiCoCrAlY and aluminum oxide was one material utilized for robust ceramic thermocouples. Earlier studies of this material indicated that it exhibited poor thermal conductivity due to the large number of interfaces in the direction of heat transfer and associated phonon scattering. Surprisingly, the material had reasonably good electrical conductivity due to the large metal content and thus, was an excellent candidate for thermocouples and thermoelectric devices. The TEM micrograph of the NiCoCrA lY/alumina nanocomposite (Figure 6) shows the alumina (black phase) uniformly distributed throughout the NiCoCrAlY matrix (grey phase). The phonon-electron interaction responsible for electron drag as described earlier does not occur in this predominantly metallic material and therefore no significant S-shaped thermoelectric response was observed. Initially, the nanocomposite film was tested relative to platinum and was combined with two different ITO thermoelements. These ITO elements were prepared in pure argon and pure nitrogen atmospheres, respectively and were heat treated in a nitrogen-rich atmosphere. The nano composite material itself was sputtered in 9mtorr argon and no post deposition heat treatment was necessary.

The thermoelectric response of the cermet relative to platinum is shown in Figure 7 wherein the platinum included a nanocomposite element sputtered in pure argon (9mtorr). The emf (black curve) peaked at 95.6mV coinciding with the peak temperature of 1100⁰C and a temperature gradient of 915°C. At low temperatures, little voltage was generated by the thermocouple, suggesting that a threshold temperature of 400⁰C was necessary to achieve a response. Initially, a negative voltage was produced with significant scattering of data at low temperatures. As the temperature was increased, however, the voltage increased rapidly and the scattering in the data disappeared. Apparent in this thermocouple was the linear thermoelectric response compared to the ITO thermocouples, likely due to the large metal content in the material. However, different slopes associated with the thermoelectric response were observed as the temperature was increased and decreased. The relation between applied temperature gradient (T) and generated voltage (V) according to Equation 2 for the nanocomposite is displayed in Table 4.

Table 4. Polynomials used to describe the emf/temperature behavior of different nanocomposite/Pt and nanocomposite/ITO thermocouples. H denotes heating cycle C refers to the cooling cycle.

A Seebeck coefficient of 750μV/°C was calculated for the heating /cooling cycles of a nanocomposite/ITO thermocouple with the ITO element prepared in pure nitrogen (9mtorr)

(Figure8a). A thermocouple was produced with repeatable thermoelectric response after a large number of thermal cycles. In comparison, the Seebeck coefficients calculated for

ITO/Pt thermocouples were considerably smaller, suggesting that the thermocouples comprised of a NiCrCoA lY/alumina nanocomposite were extremely responsive. Figure 8b shows the thermoelectric response of a nanocomposite/ITO thermocouple with the ITO element prepared in pure nitrogen (9mtorr). A temperature difference between the hot and cold junctions of 1050⁰C generated an emf (black curve) of 225mV. A threshold thermal gradient of 400⁰C and a hot junction temperature of 520⁰C were necessary to obtain a signal. The output voltage tracked the Type S thermocouples and showed a linear increase and decrease corresponding to the heating and cooling cycles (Figure 8a). A Seebeck coefficient of 750μV/°C for the heating cycle indicated that the combination of the nanocomposite and ITO did not only improve the high temperature performance but also improved the sensitivity at lower temperatures. A systematic investigation of the thermoelectric properties of ITO as a function of its process parameters was made in an attempt to produce an all ceramic thermocouple. Initially, ITO films were tested relative to platinum as a reference element. After initial screening, bi- ceramic junctions based on different ITO compositions were prepared and investigated as potential thermocouples that could survive the harsh environment associated with gas turbine engines.

The testing of ITO relative to platinum indicated that there was some deviation from the linear applied thermal gradient with thermoelectric response in the range of 45mV - 63mV. Very stable results in terms of emf signal and sensitivity were observed at elevated temperatures, and several ITO compositions tracked the temperature gradient well. However, the bi-ceramic junctions comprising the two most promising ITO compositions produced relatively small signals and thus poor sensitivity.

A nanocomposite comprised of NiCoCrAlY and aluminum oxide was utilized in bi- ceramic thermocouples. The surfaces of the NiCoCrAlY exposed to air were passivated with AI₂O₃ giving the composite excellent high temperature stability in air. Unlike the ITO/ITO bi- ceramic junctions, the nanocomposite/ITO thermocouple tracked the temperature gradient at elevated temperatures (Figures 8a and 8b) and had Seebeck coefficients orders of magnitude larger than that of ITO/ITO and ZnO/ITO junctions. Seebeck coefficients on the order of 750μV/°C have been observed. In light of the foregoing, it will now be appreciated by those skilled in the art that various changes may be made to the embodiment herein chosen for purposes of disclosure without departing from the inventive concept defined by the appended claims. For example, the nanocomposite is not a semiconductor but it may be used as a thermoelectric leg. The nanocomposite has poor thermal conductive properties but is a good electrical conductor. The thermocouples may be used in other electrical applications, such as generators or in other applications where electricity is produced by having a constant temperature difference. We claim:

Claims

1. Thin film ceramic thermocouple to assess temperatures greater than 1200⁰C, said thermocouples comprises a first thermoelement, said first element comprising conductive oxides, said conductive oxides comprising indium tin oxide, zinc oxide doped with alumina and a second thermoelement based on a cermet nanocomposite of NiCoCrAlY and aluminum oxide.

2. The thin film thermocouple of claim 1, wherein bi-ceramic junctions comprise aluminum oxide doped with zinc oxide and indium tin oxide.

3. The thin film thermocouple of claim 1 wherein the thermocouple.

4. The thin film thermocouple of claim 1 wherein a Seebeck coefficient was calculated for a heating/cooling cycle of 750μV.

5. The thin film thermocouple of claim 1 wherein the thermocouple having little hysteresis after between heating and cooling curves during repeated cycling.

6. The thin film thermocouple of claim 1 which after being cycled through heating and cooling, produces repeatable results with each cycle.

7. The thin film thermocouple of claim 1, wherein after the thermocouple is cycled through heating and cooling, the thermocouple produces repeatable and reproducible results.

8. The thin film thermocouple of claim 1, wherein the thermocouple has a large Seebeck coefficient with a low thermal conductivity of the nanocomposite.

9. The thin film thermocouple of claim 1, wherein the thermocouple is sensitive.

10. The thin film thermocouple of claim 1, wherein said nanocomposite is not a semiconductor.