Degradation of thermal barrier coatings on an Integrated Gasification Combined Cycle (IGCC) simulated film-cooled turbine vane pressure surface due to particulate fly ash deposition

Degradation of Thermal Barrier Coatings on an Integrated Gasification Combined Cycle (IGCC) Simulated Film-Cooled Turbine Vane Pressure Surface due to Particulate Fly Ash Deposition


CHAPTER 1: INTRODUCTION
Today, there are continuing efforts to allow for operation of gas turbines at higher temperatures since higher gas inlet temperatures are directly proportional to higher efficiencies. Since 1950, the firing temperatures for industrial gas turbines have increased from more than 700°C to over 1400°C [4] with the aero-derivative gas turbines operating at even higher temperatures. The history of the turbine operating temperatures is shown in a later section in Figure 4. A schematic of a gas turbine turbofan engine is given in Figure 1 [5], and the highest temperatures are located after the combustion chamber and prior to the turbine inlet. To handle the high temperatures, the use of cooling schemes as well as high strength, directly solidified and single crystalline nickel based superalloys are required. Additionally, ceramic thermal barrier coatings (TBCs) have also been incorporated to allow operation at higher temperatures than the melting point of the superalloys. Thermal barrier coatings are designed to withstand high temperatures and protect the component substrate below the layers of coating by lowering the surface temperature of the metallic substrate. A typical TBC system can be seen in Figure 2  Due to the low thermal conductivity (k) of the TBC system relative to the substrate, thermal insulation is provided to the metallic/superalloy substrate. This allows the turbine system to run at much higher gas temperatures than the substrates' melting point. Although an increase in operating temperatures is beneficial to the efficiency of the gas turbine, elevated temperatures have brought about several durability issues for the TBCs. One particular issue is the degradation of the TBC system from particulates and contaminants that become molten deposits. In recent years, a growing interest has been shown to molten contaminant deposition that can degrade the components in the turbines, or inhibit cooling designs. Molten deposits in the gas turbine can form from impurities that enter through the inlet air (fine sand, volcanic ash, etc.) or the upstream combustion of particulate laden alternative fuels such as coal-derived synthesis gas (syngas). Even with modern filtration systems, contaminant particles smaller than 10 μm may pass through into the engine [7]. Larger particles can also pass through as the filters wear out. These particles become molten as they pass through the combustor, and as they impact the components in the first stage of the turbine, the molten particles will deposit onto the components. Figure 3 shows an example of a turbine vane with molten deposits from volcanic ash [8]. Figure 4 contains the temperatures capabilities of nickel-based superalloys and TBCs over the past 50 years. Molten deposits become a factor as temperatures exceed roughly 1200°C [9]. Most of the common contaminants become molten around that temperature.
Understanding the degradation of the TBC from syngas particulate matter molten deposits will help extend the life of engine components and reduce down time for component repair and replacement.
The following section documents research objectives and analysis procedures performed on the thermal barrier coated articles. The test articles are post-fly ash deposited TBC coated test articles from a previous study by Murphy et al. [10] to examine the interaction between fly ash and film cooling.

1.1: Research Objectives
The purpose of the research was to characterize the effects of bituminous coal fly ash deposition on the directed vapor deposition (DVD) TBCs of non-dimensional representations first stage turbine vanes. This was accomplished through the following methods: 1) Use of a micro-indentation technique as a NDE evaluation to find the surface stiffness responses (elastic modulus) of exposed and unexposed areas and quantitatively characterize the strain tolerance differences.
2) Analysis of TBC systems using scanning electron microscopy (SEM) to provide microstructural differences between areas exposed to deposition and unexposed areas.
3) Energy dispersive X-ray spectroscopy (EDS) examinations to provide elemental mapping of cross sections that show the depth of the deposition penetration and elemental distribution of various areas within the TBC/deposition interaction zone.

CHAPTER 2: THERMAL BARRIER COATINGS (TBCS)
A thermal barrier coating consists of several layers, each with different physical, thermal, and mechanical properties, as described in Chapter 1. The system has metallic and ceramic components that form the complex insulation system. The typical TBC layers are the ceramic top coat, bond coat, thermally grown oxide (TGO), and the substrate. The descriptions of the layers are expanded upon below.

2.1: Ceramic Top Coating
The top layer of a TBC system typically contains the ceramic top coat comprised of yttriastabilized zirconia (YSZ). The top coating is typically the thickest layer and is the layer most responsible for insulating the substrate. The reduction in heat transfer is accomplished by the layer's low thermal conductivity (k). In addition to the low thermal-k, the layer should also consist of a low coefficient of thermal expansion (CTE) that closely matches that of the substrate, high resistance to oxidation, strain tolerance, microstructural/phase stability, chemical compatibility, and resistance to mechanical erosion [11].
The two most common TBC application methods are air plasma sprayed (APS) and electronbeam physical-vapor deposition (EB-PVD). Air plasma spray (APS) deposition is a low cost method that incorporates an intersplat process (successive impact of semi-molten powder particles on a substrate [12]) that lead to pores that lie parallel to the substrate surface and perpendicular to the temperature gradient.
APS coatings have a very low thermal conductivity (0.9-1 • ⁄ ) [13] and are used in stationary and rotating components of hot stage parts in industrial turbines. EB-PVD coats have a higher thermal-k (1.8- ) but are usually preferred due to the strain tolerance of the coat. The strain tolerance of the EB-PVD is provided by the resultant columnar grain microstructure. The EB-PVD method is commonly used on rotating and non-rotating components in the hot stage such as blades and vanes in jet engines.
Schematics of the two coating types are shown in Figure 4 [14] and as-deposited coatings are given in (a) (b)

2.2: Bond Coat
The bond coat (BC) plays a crucial role in the TBC system mechanically and chemically.
Mechanically, the bond coat is responsible for the adhesion between the superalloy substrate and the ceramic top coat. The BC serves as a bridge to connect the creep and yield characteristics of the two separated layers [11]. Bond coats also contain the aluminum reservoir such that the alumina, α-Al 2

2.3: Thermally Grown Oxide (TGO)
During operating conditions, a thermally grown oxide (TGO) layer develops between the ceramic top coat and the bond coat. The TGO layer is an oxidation reaction product with the bond coat and is designed to be α-alumina (α-Al 2 O 3 ) [11]. The thermally grown oxide layer has a major influence on the durability of the TBC system. It acts as a diffusion barrier and reduces further oxidation of the bond coat, and becomes thicker as thermal exposure time increases. While TGO growth is inevitable, the growth rate can be controlled by the choice of bond coat.
The primary TGO morphology is rapid inward oxygen diffusion and slow outward diffusion of the aluminum [15]. This causes the TGO to grown towards the bond coat instead of the ceramic top coat.
The three layers within the TGO microstructure consist of an Al 2 O 3 -ZrO 2 interaction zone, a developed and crystalline Al 2 O 3 , and a fined grained Al 2 O 3 layer mixed with spinel oxides formed from the bond coat such as Ni(Cr,Al) 2 O 4 [16] and can be seen in Figure 7 [17]. This growth of the TGO is referred to as growth strains and is responsible for a majority of the spallation failures in the TBC system.

2.4: Superalloy Substrate
The bond coat and ceramic top coat are deposited on the metal substrate, which is usually a single or polycrystalline Cobalt or Nickel based alloy. The substrate of the turbine components must handle the stresses from the operational forces, but must also have mechanical properties necessary for high temperature endurance. Cobalt or Nickel based alloys provide excellent high temperature creep and corrosion resistance. The material properties are also enhanced by element additions and precipitation hardening.

CHAPTER 3: THERMAL BARRIER COATING FAILURE MECHANISMS
There are several challenges that arise when a thermal barrier coating system is placed into turbine operating conditions. To study the failure mechanisms of a TBC system, furnace cycle and burner rig tests are widely used. Evans et al. [18] broke down the various mechanisms affecting EB-PVD TBCs into two basic categories, intrinsic and extrinsic, which can be seen in Figure 8. Studies of the failure mechanisms of EB-PVD TBCs are presented since the type of top coat is most similar to the direct-vapordeposition (DVD) coatings analyzed in the current work. Each failure category will be discussed in the following subsection.

3.1: Intrinsic Mechanisms
The intrinsic failure mechanisms are associated with the strain differences and misfits within the constituent materials. The mechanisms are studied with the help of furnace cycle and burner rig tests.
Final TBC failure is generally characterized by buckling and/or spalling of the top coat. One mechanism involves the delaminations from the TGO formation between the BC and ceramic top coat and the grain boundary ridges on the BC surface. The bond coat exhibits imprints of the grains in the TGO, which suggests brittle failure from loss of adhesion at the metal/oxide interface [18]. Once the TGO reaches a critical thickness, delaminations around the TGO interfaces are common.
Another intrinsic mechanism is referred to as roughening, rumpling, or ratcheting. In rumpling, the BC surface along with the TGO becomes wavy upon thermal cycling [19]. Small cracks develop above the TGO and can lead to the buckling of the upper parts of the ceramic coat.
The last intrinsic failure mechanism comes from damages resulting from high temperature low cycle fatigue tests. Cracks initiate at the bond coat and TGO interface and propagate perpendicular to the load direction into the substrate [20]. Thus, integrity of the turbine components is compromised.

3.2: Extrinsic Mechanisms
Extrinsic mechanisms cannot be reproduced in furnace cycling or conventional burner rig tests.
One mechanism is foreign object damage (FOD). Large particle damage can cause spallation at the leading edge of airfoils [21]. Less severe particles may just thin the TBC via erosion. Erosion and spallation can have disastrous effects on the performance of critically loaded components when TBC is relied upon to keep the component substrate below the melting point.
Molten deposits that adhere to the surface of the TBC are also listed as an extrinsic mechanism.
The effect of the deposits will be discussed in the literature review in the next chapter and are the focus of this work.

CHAPTER 4: REVIEW OF RELEVENT LITERATURE
Understanding the degradation effects of molten particulate deposition on TBC can help increase the life of the hot stage components in an integrated gasification combined cycle (IGCC) gas turbine and aid in the design of advanced TBC systems. Molten deposition can negatively impact the thermal and mechanical properties of the protective coatings applied to the components. In addition, the film-cooling scheme can be affected from the alteration of cooling patterns both upstream and downstream of the cooling holes. If the cooling holes are blocked from the deposition, the effectiveness of the TBC system becomes even more important. Deposits in the turbine section from fly ash become molten in the combustor and adhere to the to the first stage turbine vanes [22]. Figure 9 shows the path of the deposition created by Logan et al. [22].

4.1: Degradation from Molten Deposits
There are several sources of particulate matter that may melt and adhere to the turbine components. Among the sources are sand ingestion, volcanic ash ingestion, and particulates from the combustion of alternative fuels such as coal synthesis gas (syngas).
Primarily in aircraft engines, molten deposits from calcium-magnesium-alumino-silicate (CMAS) attacks develop from siliceous debris introduced through the intake air. Once the temperatures exceed 1150°C, the ingested particles and debris become molten and adhere onto the top surface of ceramic top coat. Wellman et al. [23] found several degradation mechanisms from CMAS on the EB-PVD TBC.
Mechanisms listed in the study are severe damage to the YSZ column morphology from dissolution which reduces the TBCs insulating properties and strain tolerance, change in the YSZ crystal structure from tetragonal to monoclinic from depletion in yttria, and an increase in the erosion rate of the top coat.
Levi et al. [24] also performed studies on the thermomechanical and thermochemical aspects of the CMAS interactions with an EB-PVD YSZ coating. The study found the molten deposits de-stabilized the non-transformable, metastable tetragonal t´-YSZ top coat as it infiltrate the ceramic top coat towards the TGO. The infiltration can lower the strain tolerance after the deposit cools within the TBC. In addition, the stress mismatch from the difference in the coefficient of thermal expansion (CTE) between the YSZ and the molten deposits during the CMAS solidification can cause spallation within the top ceramic coatings. This occurs when the turbine cools and heats up during shutdown and startup (low cycle fatigue). An illustration of the delamination mechanism of the YSZ coating from Whitman and Wellman is given in Figure 10   Deposits can also form due to the intake of volcanic ash. Drexler et al. [26] found that volcanic  Figure 11a contains the general mechanism that molten volcanic ash can attack a conventional 7YSZ APS TBC [26]. Figure 11b shows that the newer TBCs prevent significant penetration of the molten ash. Padture [27] also experimented with traditional APS YSZ coatings and a newer TBC (Gd 2 Zr 2 O 7 ) and their mitigation of molten deposits from syngas simulated lignite fly ash. The lignite fly ash was found to penetrate 100% of the APS 7YSZ TBC and the nature of the damage is similar to that of CMAS and Eyjafjallajökull fly ash. The APS Gd 2 Zr 2 O 7 TBC resisted significantly more of the molten lignite fly ash deposition attack, which penetrated roughly 25% of the TBC thickness. The lack of infiltration depth appeared to be due to a formation of an impervious, stable crystalline layer at the fly ash/ Gd 2 Zr 2 O 7 TBC interface.
The molten deposits on TBC from foreign objects all change the morphology of the YSZ coatings, destabilize the crystal structure of the ceramic coating, and cause a degradation in the thermal and mechanical properties of the YSZ. The purpose of the current work was to investigate the type of degradation of molten deposits from syngas on the DVD TBC system. Many of the degradation mechanism discussed in the chapter should apply for the current study, and with the application of a micro-indentation procedure (discussed in a later chapter), the mechanical properties of the TBC system can be evaluated.

4.2: TBC Testing: Application of Micro-Indentation
Several non-destructive evaluation (NDE) techniques have been developed to assess the mechanical qualities of the thermal barrier coatings. Zhao and Xiao [28] measured the residual stresses in TBCs using a photoluminescence piezospectroscopy and indentation technique. The techniques which include a nano-indenter with a Berkovich (three-sided pyramid) indenter tip found that the Young's modulus decreased the further away from the TGO-TBC interface.
Tannenbaum [29] studied the use of a load-based micro-indentation method at West Virginia University to predict the failure location of the TBC before its occurrence. First, the study showed that the micro-indentation system provided accurate surface stiffness responses, and through the use of scanning electron microscopy, the evaluation technique was proven not to cause micro-cracking or debonding of the TBC system [30]. Once the NDE is verified, the study found that the area that displayed a relative increase in the surface stiffness response enabled early detection of initial TBC spallation failures (verified using SEM).
Otunyo [15] performed similar experiments at WVU to evaluate the surface stiffness responses but at room and elevated temperatures. Additionally, the elevated temperature tests were also exposed to an air or carbon dioxide (CO2)/steam environmental conditions to simulate the harsh conditions the TBC components would experience. The research showed that an increase in the stiffness response meant an increase in residual stresses from high interfacial rumpling and non-uniform oxide growth, which allowed to determination of the TBC's thermal life cycle.
The two studies by Tannenbaum [29] and Otunyo [15] show that the micro-indentation systems at WVU can be used as a non-destructive test to evaluate the mechanical properties of the TBC. For the current study, the use of the micro-indentation systems allowed for evaluation of the effects of deposition on the mechanical properties of the DVD TBC system.

CHAPTER 5: TEST ARTICLE DESIGN, FLY ASH DEPOSITION PARAMETERS, AND TEST ARTICLE SECTIONING
Chapter 5 contains the experimental set-up from the previous study by Murphy [10] and the current study. The test article design section discusses the design of the articles to simulate a first stage turbine vane and a description of the TBC layers deposited on top. The fly ash parameter describes the matching of the laboratory conditions to the engine condition for the simulated particulate matter completed by Murphy. Lastly, the sectioning section describes the fabrication of the test samples from the aforementioned test articles.

5.1: Test Article Design
The test articles were designed to simulate the pressure side of a first stage turbine vane at incline angles of 10° and 20°. HAYNES 230 (HA230) alloy was used as the substrate material because of the material's excellent high-temperature strength, resistance to oxidizing environments, and lower thermal expansion characteristics than most high-temperature alloys. The chemical composition of HA230 is listed below in Table 1 [31]. Each article has four round cooling holes with a 3.9mm diameter placed on the face at an angle 30° to the surface. The cooling holes were scaled to 8X the size of an actual gas turbine cooling hole using Reynolds similarity. A hollowed out backside is connected to a high-pressure air system that allows cooling air to be delivered evenly to the face. Figure 12 displays isometric views of the test article design [1].  Thermal barrier coatings (TBC) were applied to each test article using a Directed Vapor Deposition (DVD) process developed and applied by Directed Vapor Technologies International, Inc.
(DVTI) [32]. The DVD process, shown in Figure 13, uses an inert gas jet around the vapor to direct it toward the surface, which is a more efficient process than electron beam physical vapor deposition (EB-PVD) [33]. In addition to the efficient high deposition rate, the DVD process features non line-of-sight deposition and controlled intermixing during multiple source evaporation. These two features allow for coating deposition on complex engine components and enable novel coating architectures. The bond coat of the TBC system is a ϒ-ϒ´ platinum aluminide (PtAl) developed by Brian Gleeson at the University of Pittsburgh and Iowa State University and has an approximate thickness of 15-20 μm [34]. The ϒ-ϒ´ PtAl bond coat allows for the phase properties to closely match those of the nickel-based superalloy. The ceramic top coat layer was applied as 7% Yttrium Stabilized Zirconia (7YSZ) with an approximate thickness of 400 μm. The cooling holes were masked during the coating process, which created a shallow trench. A trench is actually employed in modern day gas turbine film-cooling schemes [35]. Figure 14 displays the two angled test articles after TBC had been applied [1]. Below in Figure 15 contains the morphology of a DVD YSZ coating during the development of the deposition process by Hass et al. [36]. The DVD YSZ layer reveals a columnar structure with fine columns with 1-2 μm diameter and coarser columns with diameters of 10-20 μm. The YSZ columns exhibit parallel growth and have perpendicular intercolumnar voids. The porosity was generally aligned to the substrate. The 7YSZ coatings produced using the DVD technique have also been found to have thermal conductivities as low as 0.8 W/mK if a zig-zag pore microstructure is used [37]. The zig-zag microstructure allows for a longer thermal diffusion path. A cross-section SEM micrograph of the asdeposited TBC system in the current study can be seen in Figure 16.

5.2: Particulate Composition, Scaling, and Seeding
Unlike previous molten deposit and TBC studies where the molten deposit was formed by applying a paste onto the TBC and heat treated, in the current study, the ash particulate was injected into a high pressure, high temperature facility and impacted onto the TBC coated articles. The aerothermal facility operates at temperatures ranging from 1350-1560 K, a pressure of 45 psig, and has a mass flow rate of 12 kg/s [10]. It converts the circular flow from the combustor to a rectangular channel at the test section and allows backside cooling with front side optical access to the test articles. An image of the aerothermal facility and a cross-sectional schematic of the facility can be seen in Figure 17  Fly ash was processed by Murphy et al. [1] prior to injection into the high-pressure combustion facility. Preparation steps included baking for moisture removal and grinding and sieving to achieve the particle size distribution desired for the scaling of particle inertial characteristics from engine to laboratory conditions. After filtration, an LS Particle Size Analyzer measured the mean particle size of the processed fly ash to be approximately 13 μm. Table 2 below displays the composition of the bituminous fly ash in the current study [1]. The size distributions of the processed fly ash can be seen in Figure 18 [38] and was compared to the results from a previous study completed by Smith et al. [39].

Figure 18: Comparison of size distributions between the study by Murphy and previous literature [38]
To scale the particle inertial characteristics, the Stokes number was matched between engine conditions and laboratory, which set the desired mean particle diameter [40]. The Stokes number determines the behavior of the particles in a fluid flow, whether it follows the fluid streamlines or their own inertial path. For the study by Murphy et al. [1], the Stokes numbers were calculated for fly ash particles that range from 0.5 μm to 47 μm for the engine and laboratory conditions and shown in Table 3 [1]. For the calculations, the following assumptions were made: 1) the fly ash particles have zero slip velocity meaning that they traveled with the same velocity as the mainstream hot gas path. The engine mainstream velocity, 2) U ∞ = 150 m/s, which was calculated assuming a turbine inlet Mach number of 0.2, and 3) a turbine inlet temperature of 1509 K [41]. Because the study by Murphy et al. [1] was designed to determine the effects of particle interaction with and deposition around film cooling holes, the film-cooling hole diameter was chosen as the obstacle length scale for calculating Stokes number. The Stokes number was plotted relative to the fly ash particle diameter in laboratory and engine conditions and is shown in Figure 19 [1]. The results showed that, to match the inertial characteristics of an approximately 3 μm particle in the engine, the laboratory fly ash particle was required to be 13μm.
This meant that particle sizes greater than 13 μm generated for the experiments would interact with the scaled-up film cooling jets in the same way that smaller particles would interact with true scale film cooling jets in an actual engine. Figure 19: Stokes numbers plotted with respect to particle diameter for fly ash particles in engine and laboratory conditions [1] Lastly, to inject the fly ash particulate into the high pressure combustion rig, a PS-20 Scitek pressurized particle seeder was used. The laboratory study was desired to match the particulate loading in parts-per-million by weight hour (ppmw-hr) that exist in actual gas turbine engines that operate over a period of 10,000 hours. To achieve this, the particulate concentration (ppmw) was increased which allowed the required hours to be decreased by the same scale. The particulate loading comparison is shown in Table 4 [1]. Bons et al. found that bituminous fly ash that is injected in a similar fashion experiences a volumetric decrease when it becomes molten and from sintering [42]. The fly ash had a similar chemical composition as the one used in this study. This meant that the fly ash impacting the test articles would be slightly smaller than when it was injected using the particle seeder.

5.3: Deposition Results
In the study by Murphy et al. [1], deposition formed on only one side of the test articles. Figure   20 contains the side view of the test articles from the study performed by Murphy et al. [10]. The images are from a comparison between the face angles of the test articles at an operating temperature of 2315°F and a blowing ratio (ratio of the coolant mass flux over the freestream mass flux) of 0.25. Murphy et al. [10] found that the impaction angle led to an increase in deposition from this particular trial, which is consistent with the results from the study by Jensen et al. [43]. Additionally, Murphy found that the amount of deposition increased as the freestream temperature and particulate loading increases and when the blowing ratio decreases [10]. Deposit formation on only one side of the test articles was due to a large deposit structure forming on the transition piece of the test section [38]. This transition piece is located upstream of the test articles. Figure   Additional deposit testing was performed after those seen in Figure 12. The test articles that are used for material analysis in this study are displayed in Figure 22 in the next chapter. The deposition stratification allows for the study to test and compare different sections within the same test article.

5.4: Test Article Sectioning
To determine the effect of the fly ash deposition on the mechanical properties and microstructure of the TBC, the test articles had to be sectioned and trimmed into materials testing samples. A number of defects may be generated during the cutting process such as the formation of cracks. In order to section larger elements with ceramic coverings, devices such as abrasive saws and diamond precise saws are used [44]. The samples used in this study were sent to Westmoreland Mechanical Testing & Research, Inc. to be cut using an abrasive saw method. Three of the test articles that were labeled and sent to be cut are shown in Figure 22. Test article 1 is a 20-degree angled article with the heaviest amount of deposition adherence. Test article 2 is also a 20-degree angled article, but has been unexposed to testing. The last test article, test article 3, has an angle of 10-degree and experienced significantly less deposition than test article 1.  Removing the angle of all the test articles was the first step in creating test samples, Figure 23.
After the angle cut, a substrate thickness of approximately ¼ inches remains. The purpose of the cut was to produce flat samples for system compliance during the micro-indentation test, discussed in the next chapter. Once the angle was removed, a cut grid was created for the flat test articles to produce the test samples. Figure 21 details the cut list for the articles and labels for each test sample. The heavy deposition on test article 1 is outlined in Figure 21a.  After completion of the sectioning, samples devoid of defects (delamination, cracking, etc.) were used for micro-indentation test and were then mounted and polished for SEM/EDS analysis.

5.5: Elemental Composition of the Molten Deposits
Prior to any micro-indentation, EDS spectrum analysis was used to confirm that the molten deposit on top of the YSZ coating has a similar elemental composition as the fly ash injected into the combustion facility (see Table 2). Figure 25 shows an SEM micrograph of a cross-section of sample 3-1 near the columnar tips of the YSZ coating. A "Point&ID" X-ray spectrum was acquired at a single point placed in the deposition layer (circled in Figure 22). Table 5 has the comparison between original processed fly ash and the results obtained from the EDS spectrum analysis of Figure 22 by weight percentage. Silicon, aluminum, and oxygen are chosen elements of interest since they have the highest weight percentage in Table 2. Table 5 confirms that the adhering deposits are the molten and cooled processed fly ash based on the elemental distribution results of the molten deposits.

CHAPTER 6: MICRO-INDENTATION
Since the 1800s, indentation has been used as a mechanical testing process to determine the mechanical properties of an unknown sample. A hard indenter tip, in which the geometry and properties are known, is submerged into an unknown sample's material with a predetermined applied load and amount of time. The sample's response to the load is dependent upon its material properties, and by measuring the depth and area of the impression with various optical technologies, the unknown mechanical properties can be derived.
Indentation testing was developed to evaluate alloys, but has been adopted up by other industrial sectors such as lumber, ceramics, and glass production. Smaller and more precise indenters have been developed to accommodate the increasing production of nanotechnology. With the ability to obtain mechanical properties from small volumes of the materials, the small, precise indentation testing can be labeled as a non-destructive evaluation (NDE). With the use high resolution displacement or optical sensors monitoring throughout the indentation procedures, the load-depth plots produced can accurately determine the material's mechanical property information.

6.1: Spherical Indentation
Heinrich Hertz isn't just known for his contributions to the electrodynamics field. Two of his articles became crucial in the field of contact mechanics. While studying the Newton's ring phenomenon, he decided to study the deflection occurring at the points of contact between the two lenses [45]. In particular, Hertz was concerned with the how to quantify the localized deformation and pressure between the two spherical surfaces [29], seen in Figure 26 [46].

Figure 26: Contact deformation between two spheres [46]
He sought to assign the contact area and material relationship with a quadratic function. The relationship he developed became the basis in which all modern day contact mechanics theories are derived. In Equations (1) and (2), P is the applied load, d i is the diameter of the spheres, a is the contact area, E R is the reduced modulus, and E and ν are the Young's modulus and Poisson's ratio respectively. (1) Hertz established that the stresses and defections that arise from contact between two elastic solids, regardless of the shape, were subjected to geometrical deformation [47]. If the diameter of the second sphere is assumed to approach infinity, a function between a rigid sphere (with radius R) and flat surface is developed, Equation (3). For spherical indentation under idealized conditions and shallow depth, h=a 2 /R, Lurie developed Equation (4) where h e represents elastic depth only [48].

6.2: Instrumented Indentation
Instrumented indentation tests can be used to obtain mechanical properties of materials such as the elastic modulus and hardness values. While the hardness test measures the indentation size directly, instrumented indentation calculates the value based on the indenter's geometry and indentation depth [29]. As load is applied to the indenter, the material of the specimen directly beneath the indenter tip deforms plastically leaving a residual impression upon removal of the load. The contact area is too small for optical techniques, so it is calculated using the known indenter geometry and depth of penetration, from which the hardness value is estimated. The elastic modulus calculation is more complicated and requires additional parameters. The elastic modulus calculation theory and details with single point and multiple partial unloading are discussed in the following sections.

6.2.1: Single Point Unloading
Single point unloading indentation tests use one unloading slope in the load-displacement curve, shown in Figure 27, to find the reduced modulus [29]. This method provides an accurate and theoretically sound measurement of the material's elastic response. By using the slope of the load-displacement curve, the reduced elastic modulus (E R ) can be calculated by Equation (5) where the contact area is calculated using a known indenter geometry and the indentation depth (A=πa 2 ). In the equation, the derivatives of the load (P) and indentation depth (h) are used.
The greatest advantage of this testing technique is the shallow penetration depths required. The testing technique can be deemed relatively non-destructive has it doesn't get interference from the substrate's material properties.

6.2.2: Multiple Partial Unloading
Developed at West Virginia University, the first generation of instrumented indentation analysis data reduction techniques required the use of a transparent sapphire spherical indenter. This transparent indenter measurement (TIM) system, along with phase shifting Twyman-Green and Moiré interferometer techniques can directly measure the out-of-plane deformation and contact radius throughout the process [29]. From the deformation and radii measurement, the elastic modulus and true stress strain relationship can be found. Since the technique requires use of sophisticated optical system, a second generation TIM data reduction technique was developed employing a partial unloading procedure shown in Figure 28 [29].

Figure 28: Load displacement curve for multiple partial unloading algorithm [29]
The basic principle of the technique operates under the assumption that the total displacement is a function of the indentation depth and the compliance of the system. Equation (6) shows that the total displacement (ℎ ) is the sum of the total indentation depth (ℎ) and the system deflection (ℎ ).
Furthermore, if the response is assumed to be solely elastic, the total indentation compliance ( ℎ⁄ ) is equivalent to the elastic compliance ( ℎ ⁄ ). Taking the derivative of Equation (6) with respect to the indentation load (P), Equation (7) is achieved. Finally, replacing the indentation compliance with the elastic compliance from Equation (5) yields Equation (8).
The reduced elastic modulus (E R ) is still to be determined. Under the assumption that the system compliance is constant throughout the indentation test ( ℎ = 0 ⁄ ), Equation (8) can be rearranged to show the elastic modulus is a function of the differences in the contact area and the slope between the partial unloading, Equation (9).
This second generation TIM method still requires the use of an optical area measurement.
Although the technique is proven to be accurate, the evolution of load based equations has helped to reduce the cost of the elastic modulus acquisition. Thus, a third generation of the TIM method is developed to employ a variation of the multiple partial unloading technique from the second generation but without any contact area measurement. If Equation (7) is rewritten with = ℎ , where =

6.3: Instrumented Indentation System
Despite the appearance of a simplistic design, the indentation process requires quite a bit of planning and experimental resources. The indentation chain components are susceptible to vibrations and drastic temperature changes. To alleviate the sensitivity of the system to vibrations, the use of the optical table as a steady test platform and a rigid loading frame are incorporated in the system. The rigidity of the loading frame is essential in attaining the boundary conditions for the ideal contact mechanics equations.
The tests performed in the experimental set-up in this study were performed at room temperature to allow the thermal expansion effects of the component materials to be neglected. The indentation system employed in this study is described in the next section.

6.3.1: System Description
A  [15]. The system is shown below in Figure   29.  Figure 30 shows the typical loading/unloading pattern of the table top indentation test. There was minimal difference in this pattern between tests performed in areas with deposition and areas without deposition. This shows that system compliance is achieved throughout the micro-indentation tests.

6.3.2: Load Cell
The load cell in the indentation system was used to convert a load response to an electrical signal.
Stress analysis is often performed by measuring strains on the surface of deformed elements with the electrical resistance strain gauges [49]. Strain gauge load cells are accurate and reliable, and when applied to a known material with a known geometry, a respective load can be measured. This is the principle on which a strain gauge load cell operates. Lastly, due to the small variations in voltage from the electrical load signals, a Wheatstone bridge configuration was used to amplify the signals acquired throughout the indentation test [50].

6.3.3: Actuator
A piezoelectric actuator was used in the system for its depth sensing indentation application with a high resolution. It is comprised of multiple piezoelectric discs, and the displacement is a function of the applied electric field. As voltage is applied to the materials, a quick but sensitive mechanical strain (displacement) is produced. The active material of the positioning element is composed of piezoelectric discs. These disks are separated by thin metallic electrodes and provide stability from its stiffness at high pressures [45]. Small factors such as thermal drift and overheating must be considered in the use of a piezoelectric actuator as they may affect the precision of the results.

6.3.4: Indenter
The indenter material and geometry in the system was chosen based on the type of information desired during the test. Sharp indenters, such as Vickers, Berkovich, and Knoop, induce plasticity early in the process. These are common in determining the hardness or modulus of the top coat of thin films, independent of a response from the substrate. Spherical indenters offer a gradual transition from elastic to elastic plastic response. The gradual transition also allows for characterization of the material's strain hardening.
The material of the indenter also must be rigid and strain resistant. They must have a relatively high elastic modulus. Sapphire, tungsten carbide, and diamond are common as the material choice [50].
Sapphire indenters, although brittle, are desirable for their optical capabilities and low cost.

6.3.5: Loading Frame
The loading frame may be the most important part of the instrumented indentation system. The accuracy of the results is directly proportional to the compliance of the system. Due to this, great time and effort were required for the design and construction of the frame. In addition, the loading frame must also accommodate various sizes and geometries specified by the specimens while maintaining its rigidity. In order to minimize the compliance, a rigid stabilizing or clamping element must be utilized at each joint [51]. Issues such as weight and size must also be considered during the design.

6.4: Factors Affecting Instrumented Indentation
Experimental instrumented indentation testing has a range of errors usually arising from environmental or instrumentation malfunctions throughout the process [50]. In addition, there are errors related to the specimen material. Of the many errors, system compliance may be the most important.
Large deflections in the loading frame result in shallow indentation depths and inaccurate material property measurements. The error from the deflection significantly affects the slope of the fitted line in Equation (11). Additional parameters that affect the accuracy of the load depths sensing material characterization are discussed in the following sections.

6.4.1: Residual Stress
In the instrumented indentation analysis, the specimen was assumed to be stress free. However, residual stresses are often present from the processing of the specimen and/or the surface preparation. The level of residual stress within the material can be found by examining the shape of the pile-up at the edge of the indentation. General reports state that compressive residual stresses exist at smaller indentation depths and tensile residual stresses are present for larger indentation depths. Later reports recognized that for elastic contact, any residual stresses will alter the indentation stress and strain distributions [52].
Recently, researchers have demonstrated that residual stresses affect the indentation based mechanical property measurements as well [53]. Hardness and elastic modulus results using standard nanoindentation techniques and data analysis procedures were found to increase when the material had uniform compressive stresses and decrease under uniform tensile stresses, Figure 31 [54].

6.4.2: Surface Roughness
Since contact area was measured indirectly from the depth of penetration, the surface roughness of the material can affect the measurement of the elastic modulus measurement. Surface roughness is found to affect an instrumented indentation test if values greater than 0.05 are observed [55]. The overall effect of the surface roughness on the indentation test is an increase in the calculated radius with a reduction of the mean contact pressure. This will result in a lower measured elastic modulus.

6.4.3: Indentation Size
For isotropic homogeneous materials, the total indentation depth can cause a variation in the elastic modulus measurement. The variation in the modulus may be from the presence of thin oxide films whose material mechanical property substantially differs from the property of the substrate property the test was intended to measure [50]. Also, if the material exhibit indentation size effects, the conditions of plastic flow may not only depend on the strain but also the magnitude of strain gradients within the material. These strain gradients may occur in the vicinity of a crack or near an edge. Defects of this nature will result in the significant reduction in the surface stiffness response as indentation depth increases, particularly if shallow cracking is present [29].

6.5: Uncertainty Calculations
An uncertainty analysis was performed on the strain tolerance results in order to show the effect of measurement uncertainties on the accuracy of the elastic moduli results [56]. A root-sum-square (RSS) combination was described by Moffat as one of the basic forms used for combined uncertainty contributions in both single-sample and multiple-sample analysis [57]. The following equation is presented to describe one variable.
X i = X i (measured) ± δX i (20: 1) The equation should be interpreted to mean the following: is the best estimate of X i  ±δX i is the uncertainty in X i  (20:1) are the odds the uncertainty of X i is greater than ±δX i For the multiple-sample data in this study, X i (measured) was mean of the Young's modulus found in each sample set. The uncertainty, δX i , was represented by tS (N) /√N where S (N) is the standard deviation of the set of N observations and t is the Student's t statistic appropriate for the number of samples N and the confidence level required. A 95% confidence was used for the current study.
The second form of uncertainty came from the system/equipment measurement errors. Typically, these measuring errors come from the tolerances within the system. The current study has measurement tolerances in two variables: 1) ±0.15% in the load from the Honeywell load cell and 2) typical tolerance for the piezoelectric actuator length [58]. A medium tolerance of ±0.3 μm was chosen for an actuator travel distance of 50 μm [59]. The tolerance errors were inserted into the calculations and sequentially perturbated to calculate the uncertainty intervals for the measurement errors.
To complete the RSS analysis, the multiple-sample variable uncertainty and the measurement uncertainty were placed into Equation (14) to obtain the overall uncertainty.
The calculated uncertainties for the micro-indentation procedures can be found in the appendices.
The overall range of uncertainties is roughly between 3 and 10 GPa. The uncertainties are denoted by the error bars of the figures in Chapter 7.

6.6: Literature Review of Elastic Modulus for Typical TBC Layers
This section includes the Young's modulus and Poisson's ratio values found for the layers in this study. Since DVD is a novel deposition process and relatively recent, EB-PVD 7YSZ coat values are used in the review because of a similar columnar structure. The results at room temperature are compiled in Table 6 below. For the fly ash values, values observed for fly ash concrete are used. The modulus values for the bituminous fly ash are significantly less than Young's modulus of CMAS (E CMAS ≈ 90 GPa [24]), so an increase in the 7YSZ/fly ash zones should be less pronounced than previous CMAS studies. Since the indentation depth used in the study is only 50 μm, there should not be any influence

CHAPTER 7: MICRO-INDENTATION RESULTS
In order to perform a normalized elastic modulus analysis, the Young's modulus was determined for the unexposed Test Article 2 using sections of samples 2-3 and 2-4. Neither sample has been subjected to fly ash deposition testing. The samples were secured in a clamping holder and undergo the multiple partial unloading procedure. Once the procedure was completed, the TIM system outputs a file containing two columns: p (load) and dp/dh (change in load over change in indentation depth). The data in the columns were imported into the Excel sheet shown in Figure 32 along with the known geometry and material properties of the indenter (indenter radius, R, indenter modulus of elasticity, E 0 , and indenter Poisson's ratio, ν 0 ). The C value is the slope described in Equation (11). A Poisson's ratio value of 0.25 was chosen for the ceramic top coat since it lies in between the range in Table 6. TBC modulus of elasticity was recorded when the R 2 value in the plot is greater than 0.99. The closer the R 2 value is to 1, the closer it is to a linear relationship. The same analysis is completed for samples 1-3 and 1-4 and for samples 3-1 and 3-2. The results in for these samples are shown in Figure 35 and Figure 36 respectively. The deposition infiltrated YSZ areas within sample 1-3 and 3-1 also exhibit a higher surface stiffness values than their relatively clean counterparts. The area in sample 3-1 without heavy deposition was not evaluated due to system compliance issues. A summary of the micro-indentation testing is shown in Table 7.  In order to develop a trend for the molten fly deposits effects on the strain tolerance of the TBC, the modulus of elasticity with respect to the length across the TBC pressure surface face was plotted. In order to normalize the length scale, the width of the top face of all the TBC coated articles is measured at roughly 2 inches (see Figure 34). Uncertainty is not plotted for the individual results in an effort to prevent cluttering of the plot. Only measurement errors were applicable, and the errors were in the range of 0.5%. The same type of plots are constructed for sample 1-3 and 1-4 (see Figure 39) and 3-1 and 3-2 (see Figure 40). The individual micro-indentation elastic modulus values for YSZ coating areas with deposition (x/W ≈ 0) are higher than test areas without significant deposition (x/W ≈ 0.45-1). Areas with a low amount of deposition approach the surface stiffness value of the unexposed test article. In order to compare Figure 35, 36, 37, all the normalized individual micro-indentation indentation results were plotted in Figure 38. The trendlines in Figure 41 show that the samples produced from test article 1 (20° angled) have a higher increase in the Young's modulus from the deposition than the samples from test article 3 (10° angled). The level of modulus of elasticity increase can be attributed to a higher level of fly ash deposition adherence and infiltration.

CHAPTER 8: MICROSTRUCTURE ANALYSIS
Once the test samples have completed the micro-indentation testing, scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) examinations were performed.
Several samples were sent to Metallurgical Technologies, Inc to be placed in a conductive mounting media which aids in SEM examinations and polished to achieve a smooth reflective surface. Once the samples are returned, metallurgical testing was performed using a JEOL JSM-7600F Scanning Electron Microscope or Hitachi S-4700 Field Emission Scanning Electron Microscope. The top view of the unexposed and exposed YSZ coatings are seen in the SEM micrographs in  By having the hardened fly ash adhering onto the top surface of the ceramic top coat, the lateral strain compliance of the coating was reduced. This was the biggest contribution to the increase in surface stiffness throughout the micro-indentation tests.

8.2: Degradation of TBC from Fly Ash Deposition
Several degradation mechanisms were found with the SEM and EDS analysis. The mechanisms recorded are broken down into delamination, yttria migration from the YSZ columnar tips to the fly ash deposition, and a dissolution-reprecipitation mechanism of the columnar tips embedded in deposition.
Delamination and spallation of the YSZ coating is a very common failure mechanism in TBC systems. Deposition from foreign objects can infiltrate the coatings and make the TBC more susceptible to delamination, and the fly ash deposition in this study is no different. After the sectioning of the test articles, a large diagonal crack, seen in Figure 44, developed in sample 1-7. During sample preparation procedures, the crack caused spallation of the top coat on the sample. Figure 45 shows delamination of roughly 100 μm of the YSZ coating from the edge of sample 1-3. A high concentration of deposition was present in the area. In deposition adhered, but delamination-free YSZ coating areas, the fly ash deposition did not infiltrate deeply into the top coating. The infiltration of the deposition into the TBC of sample 1-3 can be seen in Figure 46. Figure 46a shows the cross-sectional micrograph of the YSZ coating on sample 1-3. To show yttria migration from the YSZ coating into the deposition, an EDS chemical spectrum analysis is completed for various zones within the fly ash and TBC system. Cross-sectional SEM micrographs of sample 1-3 with heavy deposition are shown in Figure 47. The SEM micrograph in Figure   47a is divided into four regions: epoxy, molten deposits, an interaction zone, and unaffected YSZ coating. Table 7 contains the distributions of the elements based on weight percentage of the latter three regions.
The numbers from the table are populated by averaging the EDS "Point&ID" spectrum results at the columnar centers and infiltration depth locations. The interaction zone region was labeled as Zone 1 and Zone 2 in Figure 47b. Zone 3 represents an YSZ area which was relatively unaffected by the fly ash infiltration depth. Zirconia and yttria are chosen to characterize the YSZ coating, and silicon, aluminum, and iron are chosen to show deposition infiltration into the top coating layer. The molten deposits that lie on top have a comparable elemental distribution as the processed fly ash. A similar change in weight percentage between the fly ash and the molten deposits can be found in a study by Bons et al. [7].
However, the detection of yttria and zirconia in the molten deposition layers indicates that some of the YSZ coating has migrated into the fly ash deposition. Zone 1 data was recorded at the tips of the YSZ coatings and shows that although zirconia was the predominant element in the zone, molten deposits presence was found. The yttria weight percentage level was roughly 5%, which is lower than the typical 7-8 % normally found in 7YSZ or 8YSZ. Zone 2 and Zone 3 data is taken at 10 μm and 20 μm from the columnar tips respectively. The level of fly ash deposition influence decreases as the depth gets deeper. Table 7 results agree with the infiltration depth from the EDS maps in Figure 46. A study by Peng et al.
completed a similar type of EDS analysis comparing CMAS infiltration depth and the elemental distribution within the EB-PVD coating [69]. Even though the CMAS penetrated roughly all of the 80-120 µm YSZ, which is as much as 12 times the infiltration depth of our top coat, both studies show yttria migration from the YSZ columnar tips into the molten deposits. This migration causes a destabilization within the ceramic top coating. By having a 5YSZ instead of a 7YSZ or 8YSZ, the top coating will have a higher tendency to change into one of the less desirable crystal phases (the monoclinic phase) during startup and shutdown.  The last degradation mechanism found in this study is the dissolution-reprecipitation mechanism of the yttria-stabilized zirconia and is associated with the yttria migration. The tips of the unexposed YSZ coating are shown in Figure 48. The tops of the 7YSZ coating have thin columns with jagged tips. Figure   49 is an illustration from a study by Levi et al. [24] that shows the dissolution-reprecipitation mechanism of YSZ in CMAS. The ceramic coating material dissolves in the molten glass and reprecipitates as one of the modified YSZ crystalline phases. The SEM micrographs in Figure 50 show the columnar tips on  [70]. Their study applied a CMAS layer with a thickness of 150 µm on an EB-PVD TBC for 4 hours at 1300 °C. Figure 51 contains a cross-sectional SEM micrograph of the exposed EB-PVD TBC [70]. Figure 51a shows the severe attacking of the CMAS, which includes the appearance of globular particles near the top surface. Figure 51b shows the relatively unaffected EB-PVD TBC microstructure near the substrate. The columnar tips of the current study, such as those in Figure 50, would be expected to change into the globular particles in 51a with further penetration and dissolution and reprecipitation. Also from the study by Krämer et al. [70], a highly magnified view of the conglomerate of globular particles embedded in CMAS can be seen in Figure 52. The purpose of the study was to analyze the effects of the syngas particulate molten deposits on the TBC coating systems on gas turbine first stage vane. The material analysis of the TBC was completed following the study by Murphy [10] on the effects of gas turbine vane film cooling parameters on the adherence of particulate deposition on first stage vane components. Molten fly ash adhered and infiltrated the YSZ coatings on simulated TBC coated film cooled test articles in Murphy's study. The approximate infiltration depth was 20 μm. Degradation of the TBC was found using a load-based multiple-partial unloading micro-indentation system to evaluate the mechanical properties of the YSZ and scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) to characterize the microstructure of the TBC system.
The results from the micro-indentation testing found an increase in the modulus of elasticity in the DVD YSZ coating due to the deposition, and subsequently a reduction in the strain tolerance. Several studies have listed a loss of strain tolerance in the YSZ top coating as a result of foreign object molten deposits, however not many studies were able to quantitatively show this. Selected areas of varying deposition quantities were tested to evaluate the surface stiffness response. These areas included areas with heavy deposition, light deposition, and unexposed. The indentation results of the areas with heavy deposition infiltration and adherence experienced higher Young's modulus values than the unexposed YSZ and YSZ on the same test samples but with a very low level of deposition. The higher modulus of elasticity values are due to the fly ash deposits cooling within the 7YSZ and causing a "stiffening" of the columnar microstructure. In terms of the loss of strain tolerance, the stiffening of the microstructure limits the lateral movement of the columns, thus reducing the strain compliance of the coating. The increase was found to be by as much as almost of 9 GPa in sample 1-3, which would be an increase of 43% in that particular case. The higher surface stiffness values make the YSZ coating more likely to spall off.
With the use of an SEM, mechanical and microstructure degradation could be seen in the YSZ top coating. Delamination in the YSZ of an area that had heavy deposition was observed on a sample produced from test article 1. In addition, a dissolution-reprecipitation mechanism could be detected in the molten deposit interaction zone. The YSZ columnar tips embedded in the fly ash deposition were reprecipitated and lost their identity. The reprecipitation of the YSZ is part of a phase change within the ceramic top coating material. This physical change can be associated with the yttria migration found by the EDS.
The EDS feature in the scanning electron microscopes allowed for chemical characterization and mapping of the TBC. The molten deposits on the test articles were found to match the processed fly ash of from Murphy's study. In addition, the mapping feature found the infiltration depth of the most heavily fly ash deposited test articles. The highest infiltration depth of the fly ash was found to be roughly 10-20 μm.
Lastly, EDS was used to find the yttria migration in the reprecipitated YSZ columnar tips. The fly ash absorption of the stabilizing yttria can cause phase change within the YSZ. This yttria migration is the leading factor to the dissolution-reprecipitation mechanism described in the SEM work. A phase transformation can change the thermochemical and thermomechanical properties of the ceramic top coating.
To summarize the work in the study, a micro-indentation procedure and microscopic examinations found several TBC degradation mechanisms associated with the adherence and infiltration of the syngas fly ash deposition. The load-based multiple-partial unloading micro-indentation system allowed the study to evaluate the "loss of strain tolerance" of the YSZ associated with molten deposits that many studies have found. The increase in the YSZ modulus of elasticity from molten fly ash deposition is due to the stiffening of the DVD 7YSZ columnar structure and can detect where delamination and spallation may occur in the future. Through microscopic SEM and EDS examinations, the work found mechanisms that have been associated with molten deposits from CMAS, volcanic ash, and lignite fly. These mechanisms include delamination, yttria migration, and the dissolutionreprecipitation that accompanies the yttria migration.

9.2: Recommendations for Future Work
Further study is needed in the fabrication of more test samples with fly ash deposition and cyclic thermal loading testing. The unique geometry of the simulated film cooled test articles made generating test samples quite difficult. Although the purpose of the original combined study was to look at the film cooling and syngas particulates, additional fly ash application to TBC coated 1" diameter samples would allow for further and alternative materials testing due to its more accepted geometry for research. Cyclic thermal loading was desired in the original scope of the study, however a lack of additional heavily fly ash infiltrated test samples prevented the thermal study. The cyclic thermal loading would simulate transient engine conditions of a gas turbine. The mechanical and microstructural properties could be evaluated versus the number of cycles the samples experienced before failure from spallation.
Future work could be completed on the same type of study with a different processed particulate injected in the high pressure high temperature facility. Performing the same tests using CMAS or volcanic ash would allow for another comprehensive study on molten deposition, film cooling, and material analyses of gas turbine components.