Thick Ceramic Barrier Coatings Derived from Transition Metal Modified Polyorganosilazanes for High Temperature Applications Zur Erlangung des akademischen Grades Doktor der Naturwissenschaften (Dr. rer. nat.) Genehmigte monographische Dissertation von M. Sc. Samuel Aeneas Kredel Erstgutachter: Prof. Dr. Ralf Riedel Zweitgutachter: Prof. Dr. Robert Stark Darmstadt 2024 I Thick Ceramic Barrier Coatings Derived from Transition Metal Modified Polyorganosilazanes for High Temperature Applications Genehmigte monographische Dissertation zur Erlangung des akademischen Grades Doktor der Naturwissenschaften von M. Sc. Samuel Aeneas Kredel aus Heppenheim, Deutschland Vorgelegt beim Fachbereich Material- und Geowissenschaften, Darmstadt, Technische Universität Darmstadt Erstgutachter: Prof. Dr. Ralf Riedel Zweitgutachter: Prof. Dr. Robert Stark Tag der Einreichung: 25.10.2024 Tag der Disputation: 09.12.2024 Jahr der Veröffentlichung der Dissertation auf TUprints: 2025 URN: urn:nbn:de:tuda-tuprints-290103 Veröffentlicht unter CC BY 4.0 International https://creativecommons.org/licenses II Erklärung laut Promotionsordnung § 8 Abs. 1 lit. c PromO Ich versichere hiermit, dass die elektronische Version meiner Dissertation mit der schriftlichen Version übereinstimmt. § 8 Abs. 1 lit. d PromO Ich versichere hiermit, dass zu einem vorherigen Zeitpunkt noch keine Promotion versucht wurde. In diesem Fall sind nähere Angaben über Zeitpunkt, Hochschule, Dissertationsthema und Ergebnis dieses Versuchs mitzuteilen. § 9 Abs. 1 PromO Ich versichere hiermit, dass die vorliegende Dissertation selbstständig und nur unter Verwendung der angegebenen Quellen verfasst wurde. § 9 Abs. 2 PromO Die Arbeit hat bisher noch nicht zu Prüfungszwecken gedient. __________________________________ Samuel Aeneas Kredel Darmstadt, den 25.10.2024 III Abstract As part of the DFG-funded research training group MatCom-ComMat, this project focuses on developing novel ceramic coatings for use as environmental and thermal barriers on eutectic Mo-20Si-52.8Ti or similar substrates. This new material system aims to enable the operation of combustion engines, such as aerospace and power plant gas turbines, at temperatures exceeding 1200 °C to increase their efficiency. To achieve this, the preceramic polymer Durazane 1800 was chemically modified with varying amounts of the transition metals hafnium and tantalum as well as boron because these tailored precursors yield Six(HfaTa1-a)(Bb)CyNz ceramics with desirable properties, including low thermal conductivity, high oxidation resistance and relatively high coefficients of thermal expansion for polymer-derived ceramics. The precursors demonstrated excellent processability for coating applications, allowing for the preparation of X-ray amorphous ceramic coatings on various substrates via spin coating followed by thermal treatment. The resulting coatings exhibited good adhesion, homogeneity, evenness, and thermal shock resistance up to a critical ceramic layer thickness above 500 nm. To increase the total coating thickness, a multilayer approach was followed, which, however, has been time and energy intensive. To overcome these drawbacks, a new method for the general fabrication of wet-chemically prepared multilayer coatings was developed and registered as a patent. Using this method, filler- and crack-free polymer-derived ceramic coatings with a thickness of more than 10 μm were prepared for the first time. At 1200 °C, these ceramic multilayer coatings were able to grant excellent protection against oxidation for both silicon and Mo-20Si-52.8Ti substrates up to at least 50 h. However, for all substrate materials except silicon, severe damage to the coatings occurred due to interdiffusion, an increasing mismatch in thermal expansion coefficients or oxidation of the substrate leading to its expansion or volatilization. Thus, for application as a turbine material as described above, i) a change in the composition of the ceramic coating, ii) the implementation of an interjacent diffusion barrier, or iii) an adjustment in the composition of the substrate is considered to be required in order to transfer the good oxidation properties of the individual materials to the materials compound. Since the developed fast process is not limited to a single precursor, it is expected that a ceramic coating with a changed composition can be similarly prepared. Thereby, the described failure mechanisms may be evaded or inhibited. Furthermore, the prepared ceramic coatings are attractive candidates for environmental barrier coatings on various substrates. The implementation of fillers was explored as well, explicitly targeting the use case of a thermal barrier coating. Therein, mostly crack-free and adherent Six(Hf1)(B0.5)CyNz coatings up to thicknesses above 100 μm on a bond coat on silicon substrate could be obtained. These coatings IV showed strong oxidation when exposed to air at 1200 °C for 100 hours. Still, they remained adherent and mostly crack-free. Exploring this system with other passive and active fillers, such as Hf6Ta2O17, holds promise for advancing these findings. V Zusammenfassung Im Rahmen des DFG-geförderten Graduiertenkollegs MatCom-ComMat, sollen in diesem Projekt neuartige keramische Beschichtungen für die Nutzung als Schutzschicht gegen Hitze oder korrosive Spezies auf eutektischem Mo-20Si-52.8Ti oder ähnlichen Substraten entwickelt werden. Dieses neue Materialsystem hat das Ziel, den Betrieb von Verbrennungsmaschinen, wie Gasturbinen für die Luftfahrt und Kraftwerke, bei Temperaturen über 1200 °C zu ermöglichen, und somit deren Effizienz zu steigern. Hierfür wurde das präkeramische Polymer Durazan 1800 chemisch mit unterschiedlichen Mengen der Übergangsmetalle Hafnium und Tantal sowie Bor modifiziert, da die aus diesen maßgeschneiderten Präkursoren abgeleiteten Six(HfaTa1-a)(Bb)CyNz Keramiken geeignete Materialeigenschaften, wie geringe thermische Leitfähigkeit, hohe Oxidationsbeständigkeit und, für polymerabgeleitete Keramiken, hohe thermische Ausdehnungskoeffizienten besitzen. Die Präkursoren zeigten ausgezeichnete Verarbeitbarkeit für Beschichtungsanwendungen und ermöglichten die Herstellung röntgenamorpher keramischer Beschichtungen auf verschiedenen Substraten mittels Schleuderbeschichten, gefolgt von thermischer Behandlung. Die resultierenden Beschichtungen wiesen gute Adhäsion, Homogenität, Ebenheit und Thermoschockresistenz bis hin zu einer kritischen Schichtdicke von über 500 nm auf. Um die Gesamtdicke der Beschichtung zu erhöhen, wurde ein mehrschichtiger Ansatz verfolgt, welcher bislang jedoch zeit- und energieintensiv war. Um diese Nachteile zu überwinden, wurde eine neue Methode zur generellen Anfertigung nasschemisch präparierter mehrschichtiger Beschichtungen entwickelt und als Patent angemeldet. Mittels dieser Methode wurden zum ersten Mal füllerfreie und rissfreie polymerabgeleitete keramische Beschichtungen mit Dicken von mehr als 10 μm hergestellt. Bei 1200 °C konnten diese mehrlagigen keramischen Schichten sowohl Silizium als auch Mo-20Si-52.8Ti als Substrat für mindestens 50 h hervorragend gegen Oxidation schützen. Allerdings erlitten die Beschichtungen auf allen Substratmaterialien außer Silizium schwere Schäden aufgrund von Interdiffusion, zunehmendem Unterschied zwischen den thermischen Ausdehnungskoeffizienten oder einer mit einer Oxidation des Substrats einhergehenden Volumenzunahme oder Verflüchtigung. Für eine Anwendung als Turbinenmaterial, wie oben beschrieben, wird es daher als erforderlich angesehen, i) die Zusammensetzung der keramischen Schicht zu verändern, ii) eine Diffusionsbarriere zwischen intermetallischem Substrat und keramischer Schicht aufzubauen, oder iii) die Zusammensetzung des Substrats zu ändern, um das gute Oxidationsverhalten der einzelnen Materialien auf den Materialverbund zu übertragen. Da die entwickelte schnelle Prozessführung nicht an einen Präkursor gebunden ist, wird erwartet, dass eine keramische Beschichtung mit veränderter Zusammensetzung auf VI gleiche Weise präpariert werden kann. Somit könnten die beschriebenen Versagens- mechanismen vermieden oder gehemmt werden. Weiterhin stellen die hergestellten keramischen Schichten attraktive Kandidaten für die Anwendung als Schutzschicht gegen korrosive Spezies auf unterschiedlichen Substraten dar. Der Implementierung von Füllern wurde ebenfalls nachgegangen, wobei dieser Ansatz explizit auf die Anwendung als thermische Schutzschicht abzielt. Darin konnten größtenteils rissfreie und adhärente Six(Hf1)(B0.5)CyNz Beschichtungen mit Dicken von bis zu über 100 μm auf einer Verbindungsschicht auf einem Siliziumsubstrat hergestellt werden. Nachdem sie für 100 h Luft bei 1200 °C ausgesetzt wurden, zeigten diese Beschichtungen starke Oxidation, verblieben jedoch haftend und größtenteils rissfrei. Daher sind weitere Untersuchungen dieses Systems unter Verwendung weiterer passiver und aktiver Füller, bspw. Hf6Ta2O17, vielversprechend für die Weiterentwicklung der in dieser Arbeit vorgestellten Ergebnisse. VII Table of Contents Erklärung laut Promotionsordnung .............................................................................. II Abstract ......................................................................................................................... III Zusammenfassung ......................................................................................................... V Table of Contents ......................................................................................................... VII List of Abbreviations ..................................................................................................... IX List of Symbols ............................................................................................................... X 1 Introduction .......................................................................................................... 1 2 State of The Art ..................................................................................................... 4 2.1 Thermal and Environmental Barrier Coatings ............................................................ 4 2.1.1 Background .............................................................................................................. 4 2.1.2 Function and Requirements of Barrier Coatings ....................................................... 5 2.1.3 State of the Art and Current Developments ............................................................. 9 2.2 Polymer Derived Ceramics ........................................................................................ 13 2.2.1 The Polymer Derived Ceramics Route ..................................................................... 13 2.2.2 Polymer Derived Ultra-High Temperature Ceramic Nanocomposites ..................... 18 2.2.3 Preparation of Polymer Derived Ceramic Coatings ................................................. 21 2.2.4 Overcoming Critical Coating Thickness ................................................................... 25 3 Experimental ....................................................................................................... 27 3.1 Precursor Synthesis ................................................................................................... 27 3.2 Substrate Preparation ............................................................................................... 28 3.3 Thermal Treatment .................................................................................................... 29 3.4 Coating Process ......................................................................................................... 31 3.4.1 Coating of Filler-Free Coatings ................................................................................ 31 3.4.2 Coating of Filler-Based Coatings ............................................................................. 32 3.5 Sample Preparation and Analysis Methods .............................................................. 35 4 Results and Discussion ........................................................................................ 37 4.1 Towards Thick Ceramic Coatings .............................................................................. 37 4.1.1 Synthesis and Properties of the Modified Precursor ............................................... 37 4.1.2 Single Layer Filler-Free Coatings .............................................................................. 43 4.1.3 Multilayer Filler-Free Coatings (Conventional) ........................................................ 60 4.1.4 Filler-Based Polymer Derived Ceramic Coatings ...................................................... 67 4.1.5 Fast Processing of Polymer-Derived Ceramic Coatings ............................................ 71 4.1.6 Multilayer Filler-Free Coatings (Fast Processing) ..................................................... 78 4.2 Thermomechanical and Thermophysiochemical Properties of the Coatings ........... 92 4.2.1 Water Contact Angle .............................................................................................. 92 4.2.2 Thermal Shock Behavior ......................................................................................... 93 4.2.3 Physical Properties .................................................................................................. 96 4.2.4 Oxidation of Filler-Based Coatings on Silicon at 1200 °C....................................... 100 4.2.5 Thermophysiochemical Properties of Multilayer Coatings .................................... 103 VIII 5 Conclusions and Outlook .................................................................................. 129 6 References ......................................................................................................... 132 A. Appendix ........................................................................................................... 149 A.1 PDC Coated Mo-20Si-52.8Ti Exposed to 1400 °C ..................................................... 149 A.2 Annealing of PDC Coated Molybdenum ................................................................. 151 A.3 PDC Coated Mo-20Si-52.8Ti Based Substrates Oxidized at 1200 °C ....................... 152 A.4 Mo-20Si-52.8Ti based substrates Oxidized at 1200 °C ............................................ 155 List of Figures .............................................................................................................. 160 List of Tables ................................................................................................................ 170 Acknowledgments ...................................................................................................... 171 Complete List of Publications ..................................................................................... 172 Curriculum Vitae .......................................................................................................... 173 IX List of Abbreviations APS Atmospheric plasma spraying ATR Attenuated total reflectance BF Brightfield BMS Borane dimethyl sulfide complex BSE Backscattered electron CMAS Calcia-magnesia-aluminosilicate CMC Ceramic matrix composite CTE Coefficient of thermal expansion DTG Differential thermogravimetry EB-PVD Electron beam physical vapor deposition EBC Environmental barrier coating EDS Energy-dispersive X-ray spectroscopy EPMA Electron probe microanalysis FIB Focused ion beam FT-IR Fourier-transform infrared spectroscopy GI Grazing-incidence GPC Gel permeation chromatography NC Nanocomposite OLM Optical light microscopy PDC Polymer derived ceramic PDMAT Pentakis(dimethylamino)tantalum PTC Polymer-to-ceramic SAED Selected area electron diffraction SE Secondary electron SEM Scanning electron microscopy SiCf Silicon carbide fibers SSP Single source precursor TBC Thermal barrier coating TDEAH Tetrakis(diethylamino)hafnium TEM Transmission electron microscopy TGA Thermal gravimetric analysis TGO Thermally grown oxide UFF Ultrafast furnace UHTC Ultra-high temperature ceramic WCA Water contact angle XPS X-ray photoelectron spectroscopy XRD X-ray diffraction YSZ Yttria-stabilized zirconia nCP n times coated, then crosslinked, then pyrolyzed CnP coated and crosslinked n times, then pyrolyzed CPn coated, crosslinked, and pyrolyzed n times Dura-b Preceramic precursor to yield a Six(HfaTa1-a)(Bb)CyNz ceramic X List of Symbols A Area ψ Change of relative power over time α Coefficient of thermal expansion c Concentration const Constant LA Crystallite size LD Defect distance l Depth ∆ Difference D Diffusivity η Efficiency 𝐸 Elastic modulus f Frequency q Heat flux I Intensity ∇² Laplace operator LEQ Length including tortuosity Me Metal M Molecular mass ∇ Nabla operator F Normalization factor ∂ Partial derivative 𝜈𝑠 Poisson’s ratio of the substrate R Radius of curvature Ψ Relative Power ε Strain 𝜎 Stress T Temperature k Thermal conductivity L Thickness t Time μ Viscosity λ Wavelength v Withdrawal speed Introduction 1 1 Introduction The anticipated persisting demand for combustion engines and chemical plants, coupled with increasing energy consumption projected to continue until at least 2050, make combustion processes relevant for at least several decades [1]. However, the combustion of fossil fuels not only contributes significantly to CO2 emissions – accelerating global warming – but also causes environmental damage related to mining and the byproducts of combustion [2,3]. Furthermore, the ever-increasing amount of combusted fossil fuels will eventually deplete these deposits. Increasing the efficiency of the combustion of fossil fuels would reduce the amount of resources consumed, thereby alleviating the resulting damages. For instance, raising the operating temperature of jet turbines to 1300 °C without active cooling of the turbine blades could increase the output power by nearly 50% [4]. Beyond fossil fuels, increased efficiency translates to reduced energy consumption, lower costs, and decreased production demands for green fuels, thereby benefiting resource management and minimizing potential environmental impacts throughout their material cycles. The most straightforward way to increase efficiency in these applications is known from the second law of thermodynamics: increasing the maximum operating temperature. Currently, the most advanced high-temperature systems utilize nickel-based superalloys, prized for their exceptional toughness, creep resistance, and oxidation resistance at elevated temperatures. Their excellent mechanical properties at high temperatures arise from γʹ- precipitates within the alloy microstructure. However, the γʹ phase also imposes a significant limitation on the maximum operating temperature due to its solidus temperature of approximately 1350 °C, as without their precipitation hardening effect, toughness, and creep resistance drastically fall off [5,6]. As a result, the surface temperature of nickel-based superalloys is capped at around 1150 °C, which corresponds to roughly 80–90% of their solidus temperature [4]. Due to this restriction, these alloys are commonly combined with a thermal barrier coating (TBC) defined by a low thermal conductivity, that causes a sharp temperature decline over its thickness. The currently most widely used TBC system in this field is yttria-stabilized zirconia (YSZ). Pure zirconia undergoes a monoclinic-to-tetragonal phase transition within the application temperature range, which is accompanied by a volume change that can cause coating failure. By incorporating yttria, a metastable tetragonal t’ phase is stabilized, preventing phase transitions and ensuring coating integrity up to approximately 1200 °C [7]. However, operation Introduction 2 beyond this temperature exceeds the stabilized regime, accelerates sintering, and increases susceptibility to calcia-magnesia-aluminosilicate (CMAS) attack. Consequently, the operational temperature limit of yttria-stabilized zirconia is around 1200 °C [8–10]. This means that to enable a higher operating temperature and thereby increase efficiency, a novel material system must be developed from the ground up. This work focuses on one component of such a system: a ceramic topcoat designed for use on the intermetallic substrate Mo-20Si-52.8Ti. This substrate is provided by a partnered institute within the research training group Materials Compounds from Composite Materials. The development of the preceramic precursors for this purpose was accelerated through close cooperation with another project within the same research training group. Due to its high melting point as well as good oxidation resistance up to ca. 1200 °C, Mo-20Si-52.8Ti is a promising candidate as a substrate material and, therefore, is through to benefit both from an EBC (environmental barrier coating), a TBC, or a combination of both [11,12]. Additionally, the use of an interjacent diffusion coating may further improve adhesion and oxidation resistance. SiCN-based ultra-high-temperature polymer-derived ceramic (PDC) nanocomposites have demonstrated promising properties in previous studies, offering high-temperature stability and low thermal conductivity [13,14]. These features make them highly suitable as topcoat materials. To maximize internal homogeneity and to keep the complexity of the system and processing as low as possible within the first stage of this project, top coats derived from single- source precursors were selected as the object of this study. Beyond their excellent properties, PDCs offer unique advantages due to their processing route: The commercially available preceramic polymer precursor used in this study, Durazane 1800, can be wet-chemically tailored through scalable modification processes to further enhance properties such as oxidation resistance or compatibility with the substrate's coefficient of thermal expansion [15–18]. Furthermore, as the tailored precursor remains liquid or dissolvable, it is compatible with widely used industrial deposition methods, including spin coating, spray coating, and dip coating [19]. These layers can then be converted into ceramics, typically via thermal treatment. However, several challenges remain for both the processing route and the material system to be employed. 1. Processing: In the field of coating applications, thermal treatment is known to limit the achievable thickness of coatings. This limitation arises from the constrained shrinkage Introduction 3 of the polymer coated on a substrate during the thermal treatment. Exceeding a critical coating thickness under these conditions results in coating failure, manifesting as cracking, delamination, or spallation. To address this issue, two primary approaches have been explored: the first involves repeated coating, which is time-consuming, energy-intensive, and labor-demanding, yet never led to coating thicknesses beyond a few micrometers; the second approach uses fillers, which frequently introduce inhomogeneities and porosity into the coating [20,21]. 2. Performance: The complex interfaces and their development, especially at high temperatures, have not yet been studied. Additionally, there is a lack of research focusing on the oxidation resistance of filler-free PDC coatings. As a result, the high- temperature performance of the entire system requires thorough investigation, followed by adjustments to optimize its functionality. Thus, the main goals of this project are summarized as follows: 1. Develop a suitable processing method to deposit dense, crack-free, and homogeneous filler-free PDC coatings with thicknesses exceeding 10 μm for use as EBCs. 2. Develop a method to deposit mostly crack-free and porous PDC coatings with thicknesses exceeding 100 μm for use as TBCs. 3. Investigate the high-temperature behavior of these systems to identify and address potential issues. State of The Art 4 2 State of The Art 2.1 Thermal and Environmental Barrier Coatings 2.1.1 Background The Carnot cycle, an ideal process for converting heat into work, provides us with the ideal efficiency for combustion processes η, given in Equation (2-1). From this, it becomes apparent that the ratio of the process’s low and high temperature TC/TH determines the process’s efficiency and that increasing the high temperature TH leads to an uplift thereof. 𝜂 = 1 − 𝑇𝐶 𝑇𝐻 (2-1) This observation holds in the real-world application of turbine engines. Thus, there has been an ongoing effort to raise the maximum operating temperatures of turbines. Currently, the most advanced systems use actively cooled blades that are composed of superalloys in combination with thermal barrier coatings. Active cooling enables higher operating temperatures, thereby increasing efficiency, but it is also counteracted by the required power consumption of cooling. A turbine capable of operating at approximately 1300 °C without active cooling of the turbine blades could increase output power by nearly 50% [4]. For turbine operation, turbine blades must convert forces arising from the pressure gradient in the combustion chamber into rotational motion. To achieve this, they require sufficient mechanical strength. At such extreme temperatures, however, most materials do not offer sufficient mechanical or chemical stability to do so. The prime material class suited for the operation in such rotating blades is metals, as they offer ductility. Ductility allows non-critical deformation that could be caused by collisions and secondly allows relaxation of stresses. This means metals can often withstand what would cause critical failure in ceramic materials, e.g. by crack propagation [22–24]. However, the high operation temperatures mean that the melting points of metals and metal alloys are commonly approached or even surpassed, which not only reduces their strength but also strongly accelerates creep – both of which can lead to failure. Furthermore, the materials may not experience significant corrosion damage. If there is no protective scale hindering the corrosion of the bulk at these temperatures, the materials are bound to fail. Particularly noteworthy are corrosion processes that occur below the maximum operation temperature. Typical issues here are correlated to pesting, type I, and type II hot corrosion [25,26]. In general, the formation of a protective scale is substantial for the protection of such a system, State of The Art 5 and to provide sufficient protection, this scale must have a low diffusivity of the corrosive species over the whole temperature range. A key requirement for this is the formation of a dense layer, the kinetics of the scale growth play an important role, which can lead to a given scale only forming within a specific temperature range. Common examples of such a scale are alumina, silica, or chromia. An interesting example of these considerations is the development of Mo-Si-B alloys. The refractory metal molybdenum is chosen due to its high melting point of 2623 °C, providing high-temperature stability and the needed high-temperature mechanical properties [27]. Silicon is then introduced to provide a protective scale as pure molybdenum suffers from detrimental oxidation behavior due to its oxide MoO3 being volatile. However, Mo-Si alloys suffer from pesting, as no protective scale is formed at temperatures below 1000 °C [26,28,29]. Introducing boron into this system, however, allows the formation of a faster-growing continuous borosilicate scale, improving the overall corrosion resistance and making Mo-Si-B an interesting candidate for high-temperature applications – prompting further developments in this material field [30–32]. One of these more recent developments is the eutectic Mo-20Si-52.8Ti alloy which is of interest in this study. The introduction of titanium causes the enhanced pesting resistance with a SiO2/TiO2 scale already forming at 800 °C and protecting the alloy up to 1200 °C [11,12,33,34]. At higher temperatures, however, Mo-20Si-52.8Ti shows strong oxidation. Thus, to operate at temperatures of 1300 °C or higher, it would be necessary to either introduce cooling and a barrier that reduces the metal’s surface temperature or to reduce the concentration of corrosive species at the metal’s surface. 2.1.2 Function and Requirements of Barrier Coatings The purpose of thermal barrier coatings (TBCs) is to reduce the substrate’s surface temperature while that of environmental barrier coatings (EBCs) is to reduce the concentration of corrosive species at the substrate’s surface. The underlying physical principles governing these two purposes are the transport of heat and the diffusion of corrosive species. As there is little control over the influence of radiation, the primary factor for heat transfer in a solid material is the conduction of heat. According to Fourier’s law, heat flux q is proportional to the temperature gradient ∇𝑇, flowing from hot to cold volumes and is quantified by the material property k, the thermal conductivity, as shown in Equation (2-2). 𝑞 = −𝑘∇𝑇 (2-2) State of The Art 6 For our simplified model, composed of a coating and a substrate, illustrated in Figure 2–1, this yields Equation (2-3), demonstrating that the heat per area passing through the coating is proportional to the differences in temperature ∆𝑇 and time ∆𝑡. Most importantly for coating design, it is inversely proportional to the thickness of the coating Lc and proportional to its thermal conductivity 𝑘𝑐. Figure 2–1: Schematic drawing of a bilayer system composed of coating (light yellow) and substrate (light blue) with thicknesses Lc and Ls, depth l, thermal conductivities kc, and ks, and diffusion coefficients Dc and Ds of the coating and the substrate respectively. 𝑞 = 𝑘𝑐 −∆𝑇𝑐𝑜𝑎𝑡𝑖𝑛𝑔 𝐿𝑐 (2-3) From this relation, it becomes clear that the thicker the coating and the lower its thermal conductivity, the lower the heat flux into the substrate. While calculating T in real systems requires solving the heat equation, which for three-dimensional parts usually requires numerical approximations, simplified steady-state one-dimensional models can typically be solved analytically [35]. For the case mentioned above, consisting of a coating and a substrate (where the coating has a thickness of Lc and a thermal conductivity of kc, the substrate respectively of Ls and 𝑘𝑠), the equilibrium heat flux in a static temperature field (e.g., achieved through continuous combustion and cooling) must be constant. Therefore, the temperature drop over the coating ∆𝑇𝑐𝑜𝑎𝑡𝑖𝑛𝑔 and the temperature drop over the substrate ∆𝑇𝑠𝑢𝑏𝑠𝑡𝑟𝑎𝑡𝑒 are described by Equation (2-4). −𝑞 = ∆𝑇𝑐𝑜𝑎𝑡𝑖𝑛𝑔 𝑘𝑐 𝐿𝑐 = ∆𝑇𝑠𝑢𝑏𝑠𝑡𝑟𝑎𝑡𝑒 𝑘𝑠 𝐿𝑠 ⇒ ∆𝑇𝑐𝑜𝑎𝑡𝑖𝑛𝑔 ∆𝑇𝑠𝑢𝑏𝑠𝑡𝑟𝑎𝑡𝑒 = 𝐿𝑐 𝑘𝑐⁄ 𝐿𝑠 𝑘𝑠⁄ (2-4) This means that the ratios of the thermal resistances, Li/ki, determine the contribution to the total temperature drop. To illustrate this relationship, let us assume the coating surface temperature is 1400 °C and the substrate surface temperature is 1000 °C, the thicknesses are Lc = 100 μm and Ls = 10 mm, and the thermal conductivities are kc = 1 W m-1 K-1 and ks = 25 W m-1 K-1. This then implies that the addition of a coating with one hundredth the thickness of the substrate reduces the heat flux from 1000 kW m-2 to 800 kW m-2, and the temperature at the surface of the substrate is reduced from 1400 °C to 1320 °C. State of The Art 7 For an EBC, the diffusion of species is governed by Fick’s second law, Equation (2-5). The change of concentration c of a species over time t is proportional to the second derivative of the concentration over space and quantified by a diffusivity D that depends on the material, the species, and temperature. 𝜕𝑐 𝜕𝑡 = 𝐷𝛻2𝑐 (2-5) For the simplest system that can provide us with insight, we reduce the problem to one dimension, with depth l, and assume a material with isotropic and constant diffusivity D, located at l > 0. For the atmosphere, we let the concentration of corrosive species c(l ≤ 0, t) be c0 for all times. The concentration in the material at the beginning c(l > 0, t = 0) is 0. Under these conditions, the solution of Fick’s second law takes the form of the error function Equation (2-6) [36]. 𝑐(𝑥, 𝑡) = 𝑐0 erfc ( 𝑙 2√𝐷𝑡 ) (2-6) From this, we can already see that with increasing product of D and t, the species penetrate deeper into the material. To better illustrate these processes, Figure 2–2 shows the concentration of a corrosive species along with the temperature over the depth l of a bilayer system at different times. The simulation assumed the following parameters: Ls/Lc = 100, Ds/Dc = 10, ks/ks = 25 (with identical density and heat capacity), c(l ≤ 0, t) = c0, c(l > 0, t=0) = 0, c(l = Lc + Ls) = 0, T(l = 0) = const and T(l = Lc + Ls) = const. From this simulation, we can see that very quickly, the temperature distribution reaches its equilibrium state with linear temperature profiles over depth, whereas the diffusion process is a lot slower. After penetrating the coating, the higher diffusivity of the substrate accelerates diffusion at the coating-substrate interface, shifting the concentration gradient within the coating from an initially Gaussian profile to a constant. From this point onwards, the concentration profile gradually approaches equilibrium, where, similarly to the temperature profile, the ratios of thickness over diffusivity determine the contribution to the concentration drop. This simple simulation demonstrates that the role of thermal conductivity and diffusion coefficient is paramount. Even relatively thin coatings can significantly reduce the temperature and amount of corrosive species the substrate is exposed to, provided these factors are sufficiently low. State of The Art 8 Figure 2–2: Functionality of TBCs and EBCs illustrated by cross-sections at different times of temperature T and concentration c for a modeled exemplary bilayer system such as discussed above, composed of coating (light yellow) and substrate (light blue), where Ls/Lc = 100, ks/kc = 25 (with identical density and heat capacity) and Ds/Dc = 10. All quantities have been normalized. Hot combustion gasses from the left heat the coating and the substrate and introduce corrosive species. To reflect this, the boundary conditions have been set to show a fixed temperature and concentration difference between hot and cooling air. The simulation was performed by Y. Yang. Thus, the predominant property characterizing a suitable material for a TBC is a low thermal conductivity k, especially at high temperatures, while for an EBC, it is low diffusivity D of the corrosive species across the entire temperature range – with properties such as porosity significantly influencing these values. For example, to reduce thermal conductivity, a certain degree of porosity is desirable in TBCs, whereas for EBCs, porosity should be as low as possible to minimize diffusivity [37–39]. Similarly, minor cracks in TBCs are not detrimental, whereas, in an EBC, they could result in pit corrosion. Furthermore, there are eight essential properties that both types of protective coatings must fulfill: Temperature stability, oxidation and corrosion resistance, chemical compatibility, good State of The Art 9 adhesion, well-matching coefficients of thermal expansion (CTEs, preferably with no or little anisotropy), no phase transitions over the operating temperatures (especially those with substantial volume change), and thermal shock stability [40–42]. A lack of temperature stability or oxidation resistance could cause the coating to volatilize or lose critical properties, ultimately leading to system failure. Poor chemical compatibility arises when a reaction or diffusion process between at least two layers in a system causes one of the other requirements to no longer be met. Poor adhesion facilitates delamination or spallation, particularly degrading thermal shock and thermal cycling performance. A strong mismatch in the CTE 𝛼, given in Equation (2-7), would result in a difference in strain 𝜀 between the different phases at an interface, causing delamination or spallation if a critical difference in temperature 𝑇 is exceeded. 𝛼 = 𝜕𝜀 𝜕𝑇 (2-7) Phase transitions can lead to undesirable material properties or cause failure akin to a CTE mismatch if accompanied by a volume change, which generates local stresses due to strain differences. As coatings are exposed to many thermal cycles, thermal shock stability is also a critical factor. Larger temperature gradients, as well as the suppression of kinetically hindered processes, may alter material behavior compared to the above-mentioned quasi-static properties, potentially triggering phenomena such as martensitic transformations. 2.1.3 State of the Art and Current Developments Currently, the most established application of barrier coatings in high-temperature applications is thermal barrier coatings. Since their introduction in the 1970s, zirconia-based coatings have been further developed to become the state of the art TBC [43]. While pure zirconia undergoes a phase transition from monoclinic to tetragonal over the operational temperature range, which is accompanied by a volume change of approximately 5%, the introduction of yttria stabilizes the t’ phase, eliminating these phase transitions and enabling the use of yttria-stabilized zirconia (YSZ) as a TBC material [7,44]. YSZ offers low thermal conductivity – about 0.8–1.1 W m-1 K-1 for atmospheric plasma sprayed (APS) YSZ and about 1.5–1.9 W m-1 K-1 for electron beam physical vapor deposited (EB-PVD) YSZ – and a low CTE mismatch with the nickel- or cobalt-based superalloys with its CTE of 11–13 10-6 K-1 [45,46]. The introduced porosity of the YSZ further reduces stresses, while its high hardness, corrosion resistance, and high melting point complete its set of advantageous material properties [47]. State of The Art 10 The two most common deposition techniques of YSZ, APS and EB-PVD, grant access to distinct microstructures, with efforts being made to combine their advantages. While APS produces lamellar microstructures with high porosity, the more expensive EB-PVD process yields columnar microstructures, as illustrated in Figure 2–3, which offer greater longevity but also increased thermal conductivity [48,49]. Figure 2–3: Schematic drawings of current TBC (in the columnar microstructure typical for EB-PVD, left) and EBC systems (right). Adhesion to the substrate is achieved through the implementation of a metallic bond coat between the top coat and superalloy that simultaneously functions as an oxidation barrier and reduces the thermally induced stresses [50]. During operation, the bond coat’s surface oxidizes to form a thermally grown oxide (TGO). This oxide layer is designed to form a protective scale with low oxygen diffusivity. By combining these coatings with active cooling of the substrate, the metal surface temperature can be reduced by 100 K or more, thereby allowing the operation of the turbines at gas temperatures well above the metal’s melting point [47]. However, common TBC materials already exhibit a limited lifetime at maximum surface temperatures of 1200–1300 °C [51]. In these systems, a further increase in the surface temperature would amplify existing problems and introduce new ones. One limiting factor is the vulnerability to calcia-magnesia-aluminosilicate (CMAS) attack at temperatures above ca. 1240 °C. At these temperatures, CMAS salts melt and adhere to the coating surface. Through the infiltration of pores, grain-boundary penetration, or the formation of reaction layers with incompatible material properties, the coatings are subsequently damaged [52–54]. For YSZ TBCs, especially the infiltration of columnar gaps and the reaction with stabilizers can cause severe damage to the coating by cold shock delamination [53,55]. Further problems arise due to YSZ sintering and phase instability when surpassing 1200 °C. First, accelerated sintering leads to increased thermal conductivity and stiffness, promoting spallation [56,57]. Second, the non-transformable t’ phase of YSZ decomposes into tetragonal State of The Art 11 and cubic phases, which then transform into cubic and monoclinic phases, accompanied by a volume shrinkage of about 5% [9,58,59]. This means that, to obtain TBCs suitable for the targeted surface temperatures of 1300–1500 °C, new material systems need to be developed. Some promising alternatives include hexaaluminates, pyrochlores, and multilayer approaches [8,51,59,60]. Since the limiting factor for the maximum operation temperature of superalloys is their melting point, there has been a focus on the development of thermal barrier coatings. However, due to recent advances in ceramic matrix composites (CMCs), environmental barrier coatings have also gained increasing interest, as shown in Figure 2–4 [61,62]. Figure 2–4: Annual publications for EBCs and TBCs (search results for the respective term in Web of Science on March 11, 2024). A promising representative of CMCs are the SiCf/SiC composites where silicon carbide fibers, due to their low defect density offering outstanding mechanical properties, reinforce a silicon carbide matrix. The main advantages of these CMCs are their low density and excellent thermal stability. However, a significant drawback is their susceptibility to corrosion in water vapor, with the oxidation of the fiber/matrix interface, often composed of boron nitride, playing a large role in material failure [63–65]. The underlying reactions creating porosity in hot steam environments are given in Equation (2-8) and Equation (2-9). 4BN +3O2(g),−2N2(g) → 2B2O3 +2H2O(g) → 4HBO2(g) (2-8) SiC +2O2(g),−CO2(g) → SiO2 +2H2O(g) → Si(OH)4(g) (2-9) Thus, to allow their operation at temperatures above the current limits of superalloy-based turbines, it is necessary to protect them against corrosion. Therefore, there has been a large interest in the development of EBCs for SiCf/SiC CMCs. After mullite-based coatings proved to State of The Art 12 be prone to hot steam corrosion, the focus shifted towards rare earth silicates, especially ytterbium and yttrium silicate [40]. While monosilicates offer better corrosion resistance, they also show higher CTEs than SiC, which is why disilicates are often favored [66,67]. Currently, there is growing interest in airtight EBCs that offer crack-healing capabilities [40]. Based on this approach, a new idea was proposed: a rare earth silicate that could be applied as a dense layer and feature low thermal conductivity could combine the functions of both EBCs and TBCs – withstanding the commonly limiting factors of hot steam corrosion, CMAS attack, and phase transformations, YxYb2-xSi2O7 would be a prime candidate [68]. An alternative solution to this problem is the multilayer approach, combining several protective layers – with both EBCs and TBCs being used as top coats, depending on the material system [51,59,69–71]. In doing so, weaknesses of one system can be mitigated. Figure 2–5 shows two examples of such setups. Figure 2–5: Multilayer approaches to enhance TBC performance by protecting them from CMAS attack and erosion (left) and multilayer approach to protect EBCs from CMAS attack and reduce EBC surface temperature (right). CMCs however have another major drawback: their lack of ductility. This constitutes a large hurdle towards their application in moving parts and therefore, there remains strong interest in the development of metal-based substrates to replace superalloys [22–24]. To increase the maximum operation temperature, refractory metals or alloys with very high melting points of around or above 2000 °C are often chosen. The limiting factor for their application often becomes corrosion resistance. Hence, the interest in the development of suitable EBCs in this field. For instance, the corrosion resistance of Mo-Si-B alloys has been increased by the addition of protective layers, such as aluminum pack cementation or silica coatings [72,73]. In a review study by Gatzen et al. [74], several EBCs for the protection of Mo- based alloys were compared. They indicated that the use of metallic Si and B, which form borosilicate glasses, was superior in terms of long-term oxidation resistance compared to the State of The Art 13 deposition of ceramic silica as the ceramic layers were not dense [74]. Combining the concept of a ceramic matrix with metallic Si and B fillers, however, was deemed very promising [74]. This was achieved by implementing metallic Si and B in a slurry with a silicon-based polymer precursor that was then pyrolyzed to obtain a ceramic matrix capable of achieving very low oxidation rates [75]. However, while highly promising for the protection of Mo-alloys in their pesting-regime around 800 °C, this specific polymer-derived ceramic-based coating would not be able to grant protection in the targeted hostile environments of gas turbines at temperatures above 1300 °C. Not only does silicon melt at 1414 °C but, as described by Equation (2-8) and Equation (2-9), both boria and silica form gaseous hydroxides in the presence of water vapor. 2.2 Polymer Derived Ceramics 2.2.1 The Polymer Derived Ceramics Route The polymer-derived ceramics (PDC) route grants access to a wide range of silicon-based ceramics, which can be attributed to their distinctive fabrication process. This process can commonly be categorized into up to five steps: synthesis, shaping, crosslinking, pyrolysis, and crystallization. The base for every PDC lies in the chemistry of its precursor. It is the foundation of both the composition and structure of the later derived ceramic, which together define the material properties. Aside from the oxygen-based polysiloxanes, there are further preceramic precursors such as carbon-based polycarbosilanes, nitrogen-based polysilazanes, and boron-based polyborosilanes, as well as combinations thereof [19]. While this difference in the polymer backbone already grants access to several highly interesting materials, there are plentiful options to further modify and tune the precursor structure to obtain the desired phase composition and microstructure. Analogous to carbon-based polymer chemistry, this can be done by the molecular approach of copolymers or by blending polymers. Further, chemical modification of the precursors is possible through the moieties. There have been numerous studies on the chemical modification of SiOC, SiC, SiN, and SiCN precursors, e.g., by the introduction of B, Al, Ti, Zr, Hf, or Ta, to enhance the ceramics’ phase stability and oxidation behavior [18,76–82]. Common chemical structures are displayed in Figure 2–6. These polymers can be processed similarly to carbon-based polymers, which is another major strength of this process, as shaping methods such as injection molding, warm pressing, ink State of The Art 14 jetting, infiltration, dip-coating and many more can be used to generate a large variety of shapes and dimensions – be it powders, fibers, coatings, monoliths or complex structures [19]. Figure 2–6: Common preceramic polymers. Redrawn with permission from [83]. Before transforming the polymeric precursors into ceramics, they first are crosslinked. Crosslinking is the formation of bonds between individual polymer chains. Thereby, a more rigid and interconnected polymer structure is obtained and the precursor transforms from a viscous liquid into a thermoset. Typical reactions underlying this step are condensation, dehydrocoupling, hydrosilylation, and transamination [16]. They are introduced in Table 2-1. Table 2-1: Typical reaction mechanisms for crosslinking [16]. Condensation R1–OH + R2–OH → R1–O–R2 + H2O Dehydrocoupling R1–H + R2–H → R1–R2 + H2 Hydrosylilation R1–CH=CH2 + R2–H → R1–CH2–CH2–R2 or R1–CH(CH3)–R2 Transamination R1–NH–R2 + R3–NH–R4 → R1–NR3–R2 + R4NH2 These reactions are usually provoked through heat, but methods such as ultraviolet light curing and catalytic curing are promising alternatives [84–87]. For some applications, such as anti- graffiti-coatings, this will already be the final processing step. In this case, the increased chemical stability, hardness, and elastic modulus obtained by the intermolecular bonds are of State of The Art 15 primary interest. For the application as ceramics, the most relevant feature of this processing step, however, is the reduction of the number of short chains that otherwise would evaporate during the pyrolysis. This reduction in mass loss allows for the high ceramic yields that are required in many PDC applications. The release of volatile species during crosslinking already leads to a distinctive mass loss that causes either shrinkage or porosity during this step, which occurs at temperatures of up to ca. 400 °C. This is representatively shown in the thermal gravimetric analysis (TGA) of several commercially available precursors in Figure 2–7. Once the polymers are no longer viscous, in thick samples, the forming gases can lead to the build-up of pressure in the material, which, in some cases, leads to fracture – to avoid this, pathways for gas release during thermal treatment play a crucial role [88]. During pyrolysis, the polymer-to-ceramic (PTC) transformation takes place. At temperatures typically ranging from 400 °C to 1100 °C, the remaining volatile short and side chains, as well as light elements, are released, and the remaining highly networked structure forms an amorphous ceramic typically composed mostly of Si and a combination of C, N, O, and B. One prominent reaction mechanism here is the formation and recombination of radicals [16]. Figure 2–7: Thermal gravimetric analysis of several commercially available precursors in argon indicating their temperature ranges for shaping, crosslinking, pyrolysis, and crystallization. Wacker Belsil MK, Silres MK, and Silres 604 are SiOC precursors with increasing amounts of carbon, Starfire SMP 10 is a SiC(N) precursor, AZ PHPS is a SiN precursor, and Merck Durazane 1800 is a SiCN precursor. Thermal gravimetric analysis was performed by C. Fasel. While this step is usually performed in an inert atmosphere, reactive atmospheres such as air or ammonia can also be used [89,90]. The former leads to increased oxygen content in the obtained ceramic, and the latter leads to increased nitrogen content. Notably, due to State of The Art 16 comparably low temperatures and holding times, these pyrolyzes consume significantly less energy than common sintering processes. To showcase such an amorphous structure, Figure 2–8 shows the difference between the α-cristobalite structure and amorphous silica. Both structures have the composition of SiO2, where each silicon atom is coordinated by four oxygen atoms in a tetrahedral structure, and each oxygen atom is coordinated by two silicon atoms in a bent structure. With the angles and distances of the bonds quite tightly fixed, both structures show a distinct and similar short- range order. For these systems, their pair distribution functions for first neighbors are very similar, with narrow distributions. However, one remaining variable for the amorphous silica remains: rotation. With the tetrahedra being arranged in varying, non-repeating rotation angles, the pair distribution function becomes more and more broad with increasing distance, and thus there is no long-range order. Figure 2–8: Left: SiO4-tetrahedra in α-cristobalite. Right: Amorphous silica structure, redrawn with permission from [91]. This non-crystalline structure however is only metastable and upon further thermal treatment, the tetrahedra can rearrange and find a local energy minimum, leading to ordered orientation and thereby creating repeated structures with long-range order. This process is called crystallization and for PDCs, it typically takes place at temperatures above 1000 °C – with some PDCs showing significantly higher crystallization temperatures. The access to tunable chemical modifications on a polymer chemistry basis, together with the availability of many processing techniques with adjustable parameters, allows for the combination of unique phases, specific phase compositions, and adjustable microstructures for ceramics in many shapes. This variety in chemical composition, as well as the structure – from State of The Art 17 the nanometer to the millimeter scale – results in highly tailorable material properties specifically designed for the needs of the application [19,20,92]. This flexibility also fuels the constantly increasing interest in this field of research, which is reflected in the steadily increasing number of publications on PDCs over the years, as shown in Figure 2–9. Figure 2–9: Annual publications on PDCs, violet for “polymer derived ceramic” and light violet for “precursor derived ceramic” or “polymer derived ceramic” (search results for the respective terms in Web of Science on March 11, 2024). This is achieved partly by advances in high-temperature or high hardness PDCs but also by opening up new areas of research, such as functional materials. For example, in the fields of supercapacitors, catalysts, or gas separators, a high specific surface area with distinct pore size distribution and structure is desired. This can be achieved by chemical design, solvent evaporation, sacrificial fillers, or templates [93]. Another example is piezo resistivity, which can be adjusted by the amount and structure of carbon phases in the ceramics, as well as by the choice of precursor and thermal treatment [94]. One outstanding achievement in the field of PDCs is the establishment of the Yajima process in the 1970s where a polysilane (Si backbone) underwent the Kumada rearrangement into a polycarbosilane (Si–C backbone) that could be easily spun into fibers [95]. Since then, ceramic fibers have become one of the main applications for polymer-derived ceramics due to their outstanding tensile strength of about 3 GPa, and there remains strong interest in further development of ceramic fibers, e.g., for use in ceramic matrix composites [96]. Another highly interesting rearrangement in PDCs is that of the carbon in SiOC and SiCN ceramics upon thermal treatment. In the first step, within the randomly structured matrix, some carbon segregates. Further increasing the temperature, the segregations can grow and either form amorphous or crystalline carbon. This carbon phase is not only crucial for some functional State of The Art 18 applications but also enhances phase stability by embedding the silicon-containing phases, thereby hindering crystallization and decomposition [92,97,98]. For SiCN, not only the carbon content but also the type of precursor plays an important role in their structural development. Polysilazanes tend to form single-phase amorphous SiCN with mixed bonds while polysilyscarbodiimides tend to form nanocomposites of C/SiC/Si3N4 [19,83]. This is illustrated in Figure 2–10. While the latter, therefore, show enhanced phase stability and crystallization resistance, they are also less homogeneous, and the free carbon phase can serve as a site for oxygen attack [99]. Figure 2–10: Comparison of the microstructures of polysilazane and polysilylcarbodiimide-derived SiCN. Redrawn with permission from [83]. 2.2.2 Polymer Derived Ultra-High Temperature Ceramic Nanocomposites To be a suitable preceramic precursor for use as an EBC or TBC, the requirements discussed in 2.1.1 need to be fulfilled. Some key properties are good adhesion, high phase stability, and well-matching CTEs, as well as oxidation and corrosion persistence. For EBCs, low diffusivity of corrosive species is a further specification, whereas, for TBCs, low thermal conductivity is vital. In addition, the top layer should exhibit suitable mechanical properties, i.e., high hardness, to persist erosion. Chemical bonding or interdiffusion is required to show good adhesion. Here, PDCs are generally highly suitable materials due to the inherent reactivity of the preceramic precursors, which is a prerequisite for cross-linking. When applied on common metallic substrates, they form chemical bonds with the metal (Me) hydroxide surface groups (Me–OH). While for nitrogen-containing precursors, this reaction is driven by amine groups or dehydrocoupling, for oxygen-containing precursors, this is driven by condensation of the hydroxy-group – a functionality commonly utilized for grafting [100,101]. The formed Me–O–Si bonds connect the metal to the polymeric State of The Art 19 backbone, thereby granting excellent adhesion already in the polymeric state. Depending on the substrate material, there also is the possibility of the formation of interdiffusion layers, e.g., silicides, that can further enhance adhesion upon thermal treatment. For SiO-based PDCs, pure silica has a relatively low melting point of 1710 °C and undergoes several phase transitions between room temperature and 1500 °C [102]. SiN-based PDCs suffer from the decomposition of silicon nitride into liquid silicon and gaseous nitrogen [103]. SiC-based PDCs show the highest resistance to decomposition in inert atmospheres since, even after phase separation, both SiC and C are highly stable phases [104]. For high silicon contents, however, excess silicon already melts at 1414 °C. While the ternary PDCs SiOC and SiCN benefit from increased crystallization resistance, they introduce a further decomposition mechanism: the free carbon leads to carbothermal reduction of silica or silicon nitride starting at temperatures of about 1500 °C, following Equation (2-10) and Equation (2-11) [97,105]. SiO2 + 2C→ SiC + CO2(g) (2-10) Si3N4 + 3C → 3SiC + 2N2(g) (2-11) All silicon-based PDCs are either composed of a silica matrix or form silica upon oxidation. Thus, for SiN- and SiC-based PDCs, the oxidation of the bulk is only kinetically hindered, and the ability to form a dense protective scale is essential to obtain low oxidation rates. While silica is already fully oxidized, it forms volatile silicon hydroxide in the presence of water [106]. Due to this, silica and SiOC show the lowest resistance to steam corrosion of the discussed systems, and while Si3N4 shows improved resistance, SiC is the most stable [107]. On the other hand, in a pure oxygen atmosphere, the oxidation kinetics of Si3N4 have been reported to be slower than those of SiC [108]. For SiCN, boron is commonly used for modification to yield a PDC with a very high crystallization temperature [19,92,109]. Contrasting the experience with SiOC, the implementation of boron into the already highly crystallization-resistant SiCN matrix has resulted in amorphous ceramics that are crystallization-resistant up to about 1800 °C, which has been attributed to a kinetic hindrance; e.g., low mobility of grain boundaries that are being constrained by turbostratic BN(C), and increased structural disorder [92,110–112]. A promising method to further enhance the material properties is the chemical modification of PDCs by incorporating elements that form ultra-high temperature ceramics (UHTCs). This modification not only has shown the potential to increase the PDCs’ crystallization resistance but also their exposure to high temperatures leads to the precipitation of the UHTC phases, State of The Art 20 thereby creating UHTC nanocomposites (NCs) [14]. Figure 2–11 gives an overview of the precipitating phases in a SiOC matrix upon modification of the precursor with the respective element. For nitride, carbide, and carbonitride PDCs, the precipitating carbides and nitrides are known for their high hardness and melting points [14]. Especially noteworthy regarding their melting temperature are Hf(C,N) and (Ta,Hf)C [113,114]. Figure 2–11: Precipitating phases within the SiOC matrix upon thermal treatment of the PDC NC precursors modified with the respective elements: Dark blue elements form carbide precipitates, light blue elements form metal precipitates, green elements form silicide precipitates, yellow elements form silicate precipitates, light red elements form oxide precipitates at temperatures below 1100 °C and silicates at higher temperatures, red elements form oxide precipitates. Redrawn with permission after [14]. For SiOC, the addition of metals such as Zr or Hf to SiOC has been shown to increase its phase stability and corrosion resistance [78,107]. HfO2/SiOC was oxidation-stable until 1200 °C; above that temperature, it decomposed into SiO2 and SiC [79]. At 1500 °C, a cristobalite scale formed but was unstable [79]. Modification of a SiCN precursor with B and Hf has shown remarkable oxidation resistance up to temperatures of 1500 °C [17,115]. Importantly, the derived precursor remained soluble in toluene, and coatings derived from it proved to protect the underlying chromium substrate against oxidation at 1050 °C more effectively than reactive element alloying [15]. Similarly, SiC precursors have been chemically modified, e.g., with Hf and Ta [116]. The derived ceramics have shown strongly improved oxidation resistance due to (Hf,Ta)C solid solutions that form Hf6Ta2O17, which has no phase transitions up to well above 1500 °C, suitable CTE, and also features low thermal conductivity, slow oxidation kinetics, and persistence to CMAS attack [20,117–121]. However, the modified precursors are no longer dissolvable in common solvents, impeding their application as homogeneous and dense EBCs. Further, due to the high thermal conductivity of SiC, their use as a TBC material is also not feasible [122]. Therefore, the two most promising candidate systems for coating applications are modified SiOC and SiCN. Amorphous SiOC generally shows thermal conductivities comparable to fused State of The Art 21 silica, depending on the carbon content: Low carbon content SiOC can show thermal conductivities lower than that of fused silica, while values well above 2 W m-1 K-1 result from high carbon contents. Upon introduction of transition metals, the nanocomposites HfO2/SiOC and ZrO2/SiOC, obtained after heat treatment at 1600 °C, have also shown similar thermal conductivities to fused silica in the range of 1.5 W m-1 K-1 [123,124]. For SiCN and SiBCN, thermal conductivities close to or even below 1 W m-1 K-1 have been reported [125,126]. Further, SiCN and SiOC both show remarkable creep resistance [19]. Similar to Si3N4 and SiO2, SiCN generally shows higher hardness and elastic modulus than SiOC [110,127–129]. In amorphous PDCs, this has been ascribed to the higher crosslinking degree of the covalent network in SiCN [19]. However, the exact values depend on the amount of carbon [130]. Further, mechanical properties are highly dependent on porosity. For example, fully dense SiCN shows significantly increased hardness, elastic modulus, and bending strength when compared to SiCN with 6–10% open porosity [131,132]. Again, the incorporation of HfO2 into the SiCN(O) matrix resulted in enhanced properties, i.e., increased hardness, as well as elastic modulus [133]. 2.2.3 Preparation of Polymer Derived Ceramic Coatings For the preparation of polymer-derived ceramic coatings, again, their excellent processability comes into play, as facile preparation techniques known from carbon-based polymers can be used to apply the preceramic polymers, and comparatively low processing temperatures are required to transform them into ceramic coatings. The access to a wide variety of compositions allows a cost-effective deposition of coating compositions otherwise only available via more expensive and complex methods such as magnetron sputtering or plasma-enhanced chemical vapor deposition, e.g., the aforementioned SiHfBCN that has been shown to effectively protect against oxidation and corrosion at temperatures above 1000 °C [15,134–136]. The prerequisites to obtaining PDC coatings have been summarized by Barroso et al.: “First, precursors must be liquids or have an appropriate solubility in common solvents to enable coating deposition in liquid state. Another requirement, important both for polymeric and ceramic coatings, is a latent reactivity, provided by the presence of specific chemical groups, which induce cross-linking upon exposure to thermal, radiative, or chemical stimuli. […] Finally, a sufficiently high molecular weight is required to prevent volatilization of compounds with low molecular weight during cross-linking and conversion into ceramic“ [20]. While there is a large variety of coating techniques and some of them are quite advanced and highly interesting for the field of PDC coatings, e.g. the additive manufacturing methods of State of The Art 22 aerosol-jet printing or laser-engineered net shaping, the most commonly applied and most easily accessible methods to apply the polymers on substrates on a lab scale are spray coating, dip coating, spin coating, and tape casting. Spray coating is a prevalent method for depositing polymeric coatings, e.g., in the automotive industry. A strong gas flow atomizes a low-viscosity lacquer, a solution or dispersion of the material to be deposited on the substrate, and transports it to the target, where most of the lacquer (up to about 80%) deposits. Film thickness is dependent on many parameters such as gas pressure and flow rate, viscosity, concentration, nozzle size and shape, distance, and angle between nozzle and target, wetting behavior, surface preparation, topography, as well as time and number of coating layers [137]. Thus, high curvature or poor accessibility tend to result in inhomogeneous layer thickness. Still, this method commonly results in even and homogeneous polymeric coatings, even for complex sample geometries but requires a skilled individual or automatization to perform it reproducibly. There also have been several advanced spray coating methods developed such as thermal and cold spraying. For dip coating, the substrate is immersed in the lacquer and then commonly removed at a set speed, as shown in Figure 2–12. During the withdrawal, two main processes happen simultaneously: firstly, the lacquer is dragged up with the substrate, and secondly, gravity draws the lacquer back into the reservoir. Thus, not only higher viscosities 𝜇 but also faster withdrawal speeds 𝑣 generally increase coating thickness 𝐿𝑐. For the relatively low speeds and viscosities typical for PDCs, evaporation also influences film thickness, which then follows Equation (2-12), as described by Landau and Levich [138,139]. 𝐿𝑐 ∝ (𝜇𝑣) 2/3 (2-12) Dip coating requires a sufficient amount of lacquer to immerse the sample in. This means that for a single sample, the amount of waste is extremely high, whereas, for mass production, it is nearly negligible. The lacquer also needs to be able to get in contact with all the parts to be coated and be able to drain off. Therefore, this is a method commonly used for simple geometries. Also, while this method is highly reproducible, the thickness of the coating varies depending on geometry and position – in general, coatings are thinnest at the top and thickest at the bottom [140]. For spin coating, lacquer is applied on the substrate, either before spinning (static dispense technique) or while spinning (dynamic dispense technique). Following this, rotation of the substrate drives the lacquer outwards as soon as the centripetal force and interface energies surpass liquid-liquid interactions. After the solvent is distributed, excess lacquer is spun off, State of The Art 23 followed by solvent evaporation. The different stages of this process are illustrated in Figure 2–12. Figure 2–12: Schematics of the dip coating (left) and spin coating (right) processes. Stages of Dip coating: immersion, holding, and withdrawal. Stages of spin coating: wetting, distribution, thinning, and drying. Meyerhofer has described the resulting layer thickness: Without evaporation, film thickness 𝐿𝑐 is given by Equation (2-13), with evaporation 𝐿𝑐 is given by Equation (2-14) [141]. In both cases, it increases with viscosity μ and concentration 𝑐 but decreases with frequency 𝑓. 𝐿𝑐 ∝ √𝜇 3 ⋅ 𝑓− 1 2 ⋅ 𝑐(1 − 𝑐)− 1 3 (2-13) 𝐿𝑐 ∝ √𝜇 3 ⋅ 𝑓− 2 3 ⋅ (1 − 𝑐)− 1 3 (2-14) During this process, most of the lacquer is spun off, resulting in very high amounts of waste. If the spun-off lacquer is recollected, the amount of waste, however, can be significantly reduced. While this method is restricted to flat substrates, in this use case it excels due to excellent homogeneity, evenness, and reproducibility [20,137]. While the aforementioned methods use rather low-viscosity lacquers, tape casting is a method used for the application of thick coatings from viscous slurries. Here, a blade pushes away excess slurry above a set height. Therefore, the thickness can be directly controlled and concentration primarily influences dry film thickness [142]. A comparison of characteristics for these four coating techniques is given in Table 2-2. Once shaping, i.e., film deposition, is completed, the processing route for PDCs is continued. First, the precursor is crosslinked, then it is pyrolyzed, and when desired, it is crystallized. State of The Art 24 Critically, since the film is required to remain attached, this means the coating is restrained in the two dimensions of the substrate, only allowing shrinkage in the third dimension, which, therefore, approximately represents the volumetric shrinkage. This means that a PDC that, as a bulk material, shows a linear shrinkage of 20%, in coatings shows a shrinkage of nearly 50%. Thus, usually, a large difference between dry film thickness and ceramic film thickness is observed. Table 2-2: Comparison of spray coating, dip coating, spin coating, and tape casting. Spray Coating Dip Coating Spin Coating Tape Casting Sample Geometry Complex Simple Flat Flat Lacquer Viscosity Low Low to Medium Low High Coating Evenness Medium to High Medium Very High High Reproducibility Medium High Very High High Inert Atmosphere Possible Possible Possible Possible Waste Amount Medium Low to High High Low Coating Thickness Low to High Low to Medium Very Low to Low Medium to High Furthermore, once the film is crosslinked, it no longer allows the accommodation of stresses by viscous flow. This means that starting from this point, upon further shrinkage, stresses build up between coating and substrate where the compressive stresses of the coating are responded with by tensile stresses of the substrate on the coating due to adhesion, thereby preventing it from shrinking in the plane of the substrate. This may lead to the alleviation of some of the stresses by deformation, which is in the form of a concave bend. Thereby, while retaining adhesion, the substrate is allowed to expand, and the coating is allowed to shrink. The radius of curvature R has been described to follow Equation (2-15), decreasing with residual stress and thickness of the coating 𝜎c and 𝐿𝑐 and increasing with the substrate’s thickness 𝐿𝑠, elastic modulus 𝐸𝑠 and Poisson’s ratio 𝜈𝑠 [143]. 𝑅 = 1 6 ⋅ 𝜎𝑐 ⋅ 𝐿𝑐 ⋅ 𝐸𝑠 ⋅ 𝐿𝑠 1 − 𝜈𝑠 (2-15) If, however, the stresses at the interface exceed the adhesion or the strength of one of the materials, typically the coating, this results in the formation of cracks, segmentation, spallation, or delamination. Since the stresses grow with increasing layer thickness, this results in a critical coating thickness where 50% of the coatings show such defects. Since the goal of this project is to obtain thick and at least mostly crack-free coatings, a high ceramic yield, resulting in low shrinkage, as well as good adhesion, are preferred, because they State of The Art 25 increase critical coating thickness. Still, the critical coating thickness of PDC coatings is restricted to about 1 μm [89,143,144]. 2.2.4 Overcoming Critical Coating Thickness To obtain effective environmental and thermal barrier coatings, in this project, the target thickness for EBCs has been set to above 10 μm and that of TBCs to above 100 μm, substantially exceeding the critical coating thickness of PDCs. To overcome the critical coating thickness, two routes have been found: The implementation of fillers, which reduces shrinkage or introduces porosity, and thereby alleviates stress, and repeated layer-by-layer deposition, resulting in multilayer coatings, which is known from carbon-based polymer coatings [145]. Several types of fillers have been studied in the field of PDCs, and the influence on the microstructure of the ones suitable for high temperatures, such as TBCs and EBCs, is shown in Figure 2–13 [19,20]. Since dense coatings are required for EBCs, the use of active fillers or a combination of active and passive fillers is most promising for them and has shown success in the past years where even zero-shrinkage and mostly dense PDC EBCs have been prepared [146–149]. Figure 2–13: Concepts of passive (left), active (middle), and sacrificial fillers (right) in polymer-derived ceramic coatings, increasing critical coating thickness (substrate in grey, filler in blue, PDC in red, pores in light red). Sacrificial fillers decompose during thermolysis, thereby introducing controlled porosity that allows the alleviation of stresses. Passive fillers are inert and thereby reduce the relative shrinkage of the coating. Active fillers undergo a reaction during thermolysis, commonly with the chosen reactive pyrolysis atmosphere, thereby increasing in volume and counteracting the shrinkage of the polymer during the PTC transformation [150]. When the filler growth and PDC shrinkage cancel each other out, zero-shrinkage coatings with strongly increased maximum coating thickness can be obtained. The alternative method to overcome the critical thickness, the multilayer approach, was explored to a lesser extent, mainly because the repeated application of coatings makes this process very slow, energy-intensive, and labor-intensive, therefore precluding industrial application of this method [21,151]. For the same reasons, until now, there have been no filler- free PDC coatings that reached the target thickness of 10 μm. So far, this approach is the only known route to obtain filler-free PDC coatings with a total coating thickness significantly exceeding critical coating thickness. Since the multilayer coatings closely resemble the single-layer coatings, they retain their advantages of high homogeneity State of The Art 26 both in chemistry and structure, especially when compared to filler-based coatings. This is exemplarily shown for a single source precursor-derived SiCN multilayer coating shown in Figure 2–14, prepared by Klausmann et al. [21]. Figure 2–14: Multilayer SiCN coating on a silicon substrate with a total layer thickness of about 900 nm and individual layer thickness of about 200 nm. Reprinted with permission from [21]. Experimental 27 3 Experimental 3.1 Precursor Synthesis The polymeric precursors Dura-b in this work were designed to form ceramics of the composition Six(HfaTa1-a)(Bb)CyNz with a = {0.7, 1} and b = {0, 0.5, 1}, where per gram of polyorganosilazane, 1.19 mMol of transition metals in a ratio equal to a and either no, half, or equal the amount of substance of boron is added. To do so, the commercially available preceramic polymer Durazane 1800 (Merck KGaA) is modified with varying amounts of tetrakis(diethylamino)hafnium (TDEAH, 99.99% trace metal bases, Sigma-Aldrich), pentakis(dimethylamino)tantalum (PDMAT, 99.99% trace metal basis, Sigma-Aldrich) and borane dimethyl sulfide complex (BMS, Sigma-Aldrich). In the following, an exemplary synthesis of the single-source precursor (SSP) Dur0.7-0.5 is described; other compositions or batch sizes were prepared with the same ratio of precursor to solvent. This process was performed in an argon atmosphere using gloveboxes and the Schlenk technique. First, 10.8 g of Durazane 1800 were dissolved in 20 mL of toluene (anhydrous, 99.8%, Sigma-Aldrich) in a 250 mL Schlenk flask. Then, under stirring, 4.20 g of TDEAH and 1.55 g of PDMAT in 15 mL of toluene were added dropwise. After stirring for two hours at room temperature, the flask was immersed in a dry ice and isopropanol cold bath at -78 °C where then 0.48 g of BMS dissolved in 12 ml of toluene was added dropwise. Subsequently, the mixture was allowed to reach room temperature. To remove the solvent and side products, it was vacuum-dried at 50 °C for about four hours, until the formation of highly viscous bubbles nearly came to a halt. While the tantalum-free compositions are nearly transparent, the obtained SSPs with tantalum show a shift in color from yellow to an intense red with increasing boron content, displayed in Figure 3–1. Figure 3–1: Modified single-source precursors Dura-b to yield ceramics of the compositions Six(HfaTa1-a)(Bb)CyNz, showing a shift in color from transparent (a = 1, b = 0.5) to yellow (a = 0.7, b = 0) to red (a = 0.7, b = 1). Adapted with permission from [152]. Precursor modification was performed in cooperation with J. Bernauer. All of the SSPs are highly viscous. Thus, dilution is necessary for most coating techniques, including spin and dip coating. Further, to prolong shelf-life and increase reproducibility, the precursors were promptly diluted to typically 35 wt% in toluene to kinetically hinder Experimental 28 crosslinking. The concentration of 35 wt% was chosen because it typically resulted in coatings around the critical coating thickness. When needed, this solution was then further diluted to the required concentrations. In addition, the solutions were stored in an opaque and air-tight glass container and used for a maximum of two weeks for coating applications. 3.2 Substrate Preparation The target substrate material Mo-20Si-52.8Ti is not yet commercially available and therefore its preparation is still labor-intense. Hence, only a limited number of both untreated and pack cemented substrates were available, the latter being even more labor-intensive. As a result, additional substrate materials have been used in this work. For developing the coating process, primarily silicon and molybdenum substrates were used as test substrates. The insights gained on these were then transferred to the intermetallic substrates. An overview of the commonly used substrates and their preparation is given in Figure 3–2. Figure 3–2: Overview of the used substrates. Silicon is one of the compounds of the target substrate. Additionally, it is a standard material for coating applications, and silicon wafers offer outstanding consistency. While being limited by its melting temperature of 1414 °C, it still allows heat treatment and exposure to dry and wet air at temperatures above 1000 °C. Single crystalline 2.54 cm diameter silicon wafers (Prime CZ-Si, 280 μm, 2-side polished, MicroChemicals) were used as received for the coating process. Molybdenum is another compound of the target substrate and also is commercially available. Its melting point of 2623 °C allows for high-temperature treatments under the prerequisite of inert atmospheres while also featuring a relatively similar coefficient of thermal expansion (CTE) to Mo-20Si-52.8Ti – 5.4–6.2 10-6 K-1 (manufacturer) and 7.0–9.1 10-6K-1 from 300–1200 °C [11], respectively. Molybdenum sheet metal of 0.4 mm thickness (99.97%, Plansee) was cut into discs and wet ground using SiC sandpaper until level. If not mentioned Experimental 29 to differ, once level, the discs were polished up to 1200 grit. In some cases, they were subsequently polished using diamond paste. They were then washed under strong water flow, followed by rinsing with deionized water, repeated sonication in organic solvents, and vacuum drying at approximately 80 °C. Mo-20Si-52.8Ti was prepared by Frauke Hinrichs at Karlsruher Institut für Technologie via arc melting of molybdenum, silicon, and titanium. More details can be found in [33]. The rod of the eutectic alloy was copper wire eroded into cylinders roughly 13 mm in diameter and 3 mm in height. The remainder of the copper was ground off, and then, shortly before coating, the substrates were polished to 1200 grit by hand with SiC sandpaper. The Al3Ti and Cr diffusion coatings were prepared via pack cementation by Katharina Beck at DECHEMA. The process is described in detail in [153,154]. If not indicated otherwise, Al3Ti pack cemented Mo-20Si-52.8Ti and Cr pack cemented Mo-20Si-52.8Ti were not further treated before coating and were stored under an argon atmosphere shortly after preparation. 3.3 Thermal Treatment For conventional crosslinking and pyrolysis, two tube furnaces (LOBA) were used with quartz glass tubes and the Schlenk technique under argon flow. For single-layer coatings, the heating and cooling rates were 100 K/h and the crosslinking temperature was set to 250 °C which was held for 2 h. On molybdenum, pyrolysis was then carried out at 900 °C for 3 h, on silicon samples at 1000 °C for 2 h, again with heating rates of 100 K/h. For multilayer coatings on molybdenum, after crosslinking at 250 °C with a 100 K/h heating rate, the heating rate was increased to 150 K/h up to the temperature of 900 °C which was held for 3 h before cooling at 300 K/h. When crosslinking was performed on a hot plate, the substrates were placed thereon at room temperature and then heated to a temperature of 300 °C within about 2 min, which then was held for 5 min, thereby avoiding inhomogeneities found when placed on a pre-heated hot plate. This process was carried out in an aluminum chamber in direct contact with the hot plate under a strong argon flow to avoid air contact. This argon flow was then also used to quickly cool down the samples afterward. The setup is schematically shown in Figure 3–3. Figure 3–3: Sketch of the fast crosslinking setup on a hot plate. Experimental 30 To access extremely high heating and cooling rates, an ultrafast furnace (UFF) was built according to the sketch in Figure 3–4, inspired by a similar setup by Wang, Ping, Bai, Cui, Hansleigh et al. [155]. In a chamber that can be operated under vacuum and argon, two strips of carbon paper are fixed on both ends between two copper blocks that are connected to a power supply, thereby applying a potential gradient to the carbon paper during operation, which leads to resistive heating. In the middle of these papers, a sample can be placed, and the local temperature is tracked by a pyrometer (KTRD 1485, MAURER, Germany, detection range of 700–3500 °C). The temperature is controlled by the power. In this work, the power is also given as the relative power Ψ, with respect to the maximum power. The change of relative power over time is further called ψ. Figure 3–4: Setup of the ultra-fast furnace. Two sheets of carbon paper are resistively heated in a vacuum/ inert gas chamber. In between them, a sample is placed, and the temperature is tracked using a pyrometer. Power can either be controlled manually or using the pyrometer as a thermostat. Adapted with permission from [152]. For annealing in nitrogen or argon for holding times up to 5 h, a carbon furnace (GT Advanced Technologies) was used. Heating rates of 20 K/min until 1200 °C and 10 K/min above 1200 °C were chosen. This temperature was then held for the selected dwelling time before subsequent cooling at 20 K/min. Annealing in nitrogen for holding times starting from 5 h was done in an alumina furnace (GERO) and SiC crucible with heating and cooling rates of 150 K/h. Experimental 31 Oxidation was performed in a muffle furnace (VMK 1600, Linn High Therm, Germany). For oxidation tests of 5-layer coatings on Mo-20Si-52.8Ti, the samples were heated at 20 K/min up to 1200 °C, then at 10 K/min up to 1400 °C, and cooled at 20 K/min. Oxidation tests up to 1200 °C used a heating and cooling rate of 20 K/min. 3.4 Coating Process 3.4.1 Coating of Filler-Free Coatings For filler-free coatings with the target application of environmental barrier coatings (EBCs), two processing routes were considered: dip coating and spin coating, both carried out in an argon-filled glovebox. For dip coating, a precursor solution of a concentration c was prepared. The substrate was cleaned by sonication in toluene and then vacuum dried at about 80 °C before fixing it to an in- house built dip-coater that then immersed the substrate at a speed of 5 mm/s, held it in position for 30 seconds and removed it at a withdrawal speed 𝑣. For spin coating, the substrates were cleaned by repeatedly wetting them with toluene and spinning them off. When the coating thickness was to be determined by profilometry, parts of the substrate were taped off, followed by additional cleaning as described above. Then, using the static dispense technique that proved to yield homogeneous coatings in previous studies on similar materials, a solution of concentration c was applied to the substrate through a syringe filter (0.2 μm mesh size, 25 mm hydrophobic polytetrafluoroethylene syringe filter, ROTILABO), to remove small particles that would lead to coating defects, then accelerated to frequency f by a spin coater (KL-SCI-20, Quantum Devices) and stopped after 40 s [84,94]. The remaining solution in the syringe was then disposed of in order not to affect the viscosity and molar mass distribution of the stock solution. To find suitable coating parameters, solutions of several concentrations were prepared. For each concentration, the lowest frequency that resulted in homogeneous coatings (in 1000 rpm steps) was selected to obtain thick coatings. The used frequencies were 4000 rpm for 14 wt% solutions and 6000 rpm for concentrations of 20 wt% or higher. An exemplary flow chart for spin coating on silicon is shown in Figure 3–5. Experimental 32 Figure 3–5: Exemplary flow chart for the preparation of coatings via spin coating on silicon. Adapted with permission from [152]. 3.4.2 Coating of Filler-Based Coatings For the experiments on filler-based coatings with the target application of thermal barrier coatings (TBCs), first, a pre-pyrolyzed self-filler was prepared. Dur1-0.5 was prepared from 10.8 g Durazane 1800, vacuum dried at room temperature for 4 h (room temperature was chosen as this results in small amounts of toluene still dissolved in the polymer, therefore strongly reducing viscosity and facilitating removal from the flask), crosslinked in a tube furnace at 250 °C for 3 h with a 100 K/h heating and a 300 K/h cooling rate and then ground into a white powder by hand in an agate mortar and sieved through a 100 μm mesh in a glovebox. After pyrolysis at 1100 °C for 3 h in a tube furnace with 150 K/h heating and cooling rate, the powder was again ground by hand, sieved through a 32 μm mesh, and repeatedly ball milled (Retsch Mixer Mill MM400, zirconia balls) for one-hour periods until a particle size in the range of about 1 μm was confirmed in scanning electron microscopy. This particle size was chosen to limit the spacing between the particles in the final film, thereby reducing the build-up of local stresses during the polymer-to-ceramic transformation while still allowing porosity to reduce thermal conductivity and alleviate stresses. The required amounts were weighed into glass sample containers and heated at 80 °C for two hours before putting them into the glovebox. Experimental 33 The coating process is schematically drawn in Figure 3–6. In the first step, a thin Dur1-0.5 layer was applied as a bond coat through spin coating of a 25 wt% solution thereof at 6000 rpm, followed by crosslinking and pyrolysis in a tube furnace in argon with heating and cooling rates of 100 K/h and dwelling times of 2 h at 200 °C and 1000 °C. Figure 3–6: a) Deposition of bond coat via spin coating and b) tape casting of self-filled TBCs. A matrix of 12 slurry compositions was investigated. The compositions are listed in Table 3-1. Table 3-1: Test Matrix for TBC compositions with varying Precursor:Filler and Precursor:Toluene ratios, by mass. Polymer:Filler 3:1 Polymer:Filler 1:1 Polymer:Filler 1:4 Polymer:Filler 1:9 Polymer:Toluene 1:0 Slurry 1 → TBC 1 Slurry 4 → TBC 4 Slurry 7 → TBC 7 Slurry 10 → TBC 10 Polymer:Toluene 2:1 Slurry 2 → TBC 2 Slurry 5 → TBC 5 Slurry 8 → TBC 8 Slurry 11 → TBC 11 Polymer:Toluene 1:2 Slurry 3 → TBC 3 Slurry 6 → TBC 6 Slurry 9 → TBC 9 Slurry 12 → TBC 12 Names and order during the preparation of the coatings and slurries were randomized. In an inert atmosphere, 500 mg of each slurry was prepared by mixing the components for 5 min, followed by application through tape casting. To do so, the same day, directly before coating preparation, a new batch of Dur1-0.5 was prepared from 3.6 g Durazane 1800. Again, the polymer was vacuum-dried at room temperature for 3 h to obtain a spreadable liquid. After cleaning the substrates by repeatedly spinning off toluene, parts of the substrate were taped off using polyimide tape (3M, 5413) to provide a more rectangular surface area for the following spreading of the slurry. As this needed to be carried out in an inert atmosphere, it was not possible to use a complete tape casting setup, and thus, the applied thickness showed variance around the set-in thickness of 100 μm. The obtained coatings were then crosslinked at 250 °C for 2 h and pyrolyzed at 1000 °C for 2 h with heating and cooling rates of 100 K/h in a tube furnace. Out of these, several slurries could not be deposited: Slurry 3 was too thin, and the slurries 7, 10, and 11 were too thick. Slurries 6, 8, 9, and 12 were distributed evenly, whereas the coatings prepared from slurries 1, 2, 4, and 5 showed minor inhomogeneities, which are attributed to the manual operation. The as-coated state of these samples is shown in Figure 3–7. Experimental 34 Figure 3–7: Slurries 1–12, directly after tape casting on a Dur1-0.5-derived bond coat on Si. Thermal treatment thereafter led to strong delamination or spallation for all samples except TBC 6, TBC 9, and TBC 12. Tetrahydrofuran was tested as an alternative solvent with the aim of increased meso- and micro-porosity due to the sped-up solvent evaporation. However, the high volatility inhibited the preparation of spreadable slurries as they already dried up during their preparation and application. Attempts to use ultra-fast pyrolysis for the preparation of TBCs were not successful, as the obtained coating was strongly delaminated, exemplarily shown in Figure 3–8. Figure 3–8: Attempt to perform ultra-fast pyrolysis of Slurry 12, resulting in delamination of large parts of the coating. Experimental 35 3.5 Sample Preparation and Analysis Methods When silicon wafers needed to be disjoint, they were either diamond wire cut or broken. For Mo-20Si-52.8Ti, they were separated using a diamond blade set to a feed speed of 0.1 mm/s. For cross-sections on Si, the samples were either diamond wire cut or broken, whereas, for Mo and Mo-20Si-52.8Ti substrates, the samples were vacuum embedded in epoxy resin and then wet ground with SiC paper up to 4000 grit, always from the coated side to the uncoated side. In some cases, before grinding, a Cr layer in the micron range was applied with a sputter coater (Quorum Technologies, Q300TD) onto the coatings to avoid damage during sample preparation. Rheology was studied with a HAAKE MARS I Rheometer (Thermo Fisher Scientific, USA) at room temperature at 1 Hz. For gel-permeation chromatography (GPC), 1 mg vacuum-dried polymer was dissolved in 1 mL anhydrous tetrahydrofuran (Sigma Aldrich), filtered through a 0.2 μm syringe filter, and measured on an Agilent 1050 Series (Agilent, USA) with a reflective index detector, using a polystyrene reference. Fourier-transform infrared (FT-IR) spectra were detected from 4000 cm-1 to 550 cm-1 in attenuated total reflectance (ATR) using a Varian 670-IR (Agilent, USA). Thermal gravimetric analysis (TGA) of polymers was studied in an argon atmosphere with a STA449C Jupiter (Netzsch, Germany) at a heating rate of 5 K/min up to 1100 °C. For the plasma torch tests, the coated side of the sample was held 1.5 cm above the source of a radio frequency generated capacitance coupled plasma flame with a temperature of ca. 3000 K for 30 s. For profilometry, a DektakXT (Bruker, USA) was used with a stylus force of 3–5 mg. X-ray photoelectron spectroscopy (XPS) was carried out on a VG ESCALAB 250 (Al Kα 1486.68 eV, Thermo Fisher, USA). For reflective X-ray diffraction (XRD) of annealed (treated above 1200 °C) coatings on molybdenum, a D8 Advance (Bruker, USA, Cu Kα) was used. All other samples were characterized with a STADI MP (Mo Kα with Ge (111) monochromator, STOE, Germany), commonly in reflective mode. The same instrument was also used for grazing-incidence (GI) XRD at an incident angle of 1° for more surface-sensitive measurements. Experimental 36 Static water contact angle (WCA) measurements used an OCA-20 goniometer (DataPhysics Instruments GmbH, Germany) with the sessile drop method for seven 3 μL water droplets per sample. Micro-Raman spectra were collected on a LabRAM Horiba HR Raman spectroscope HR800 (Horiba Jobin Yvon GmbH) using either a green laser at 514.5 nm or a blue laser at 488 nm for more surface-sensitive measurements. Nanoindentation of as-prepared coatings was performed using a Nano Indenter G200 (Keysight Technologies, USA) with a Berkovich tip and a 1000 nm penetration depth for single-layer coatings and 2000 nm penetration depth for multilayer coatings at a strain rate of 0.05 s-1. Nanoindentation experiments after annealing and oxidation were conducted on a ZHN-S Nanoindenter (Zwick/Roell, Germany) with a Berkovich tip and a load rate of 22.2 mN s-1 up to 2 N load. Optical light microscopy (OLM) was performed on a Zeiss Axio Zoom.V16 (ZEISS, Germany). A ring light was chosen to illuminate the sample evenly from all sides while avoiding direct reflections that would otherwise dominate the picture – especially for polished metals and translucent coatings thereon –, thereby highlighting defects. To evaluate coating quality and structure, scanning electron microscopy (SEM) using both secondary electron (SE) and backscattered electron (BSE) detectors was used. Before investigation, some samples were coated with a thin film of carbon or gold to enhance electric conductivity to avoid charging. Sputtering with gold was performed in a sputter coater (Quorum Technologies, Q300TD, 30 mA for 30 s). Two different electron microscopes were used: The first is a PHILIPS XL30 FEG (Philips Electronics, Amsterdam, Netherlands), and the second one is a JEOL JSM-7600F (JEOL Ltd.) with an X-Max energy-dispersive X-ray spectroscope (Oxford Instruments). The acceleration voltage for energy-dispersive X-ray spectroscopy (EDS) measurements was 15 kV, for SE and BSE micrographs it was 10–15 kV. Electron probe microanalysis (EPMA) was carried out on a JEOL JXA-8100 (JEOL Ltd.) at 15 kV acceleration voltage. Transmission electron microscopy (TEM) was performed on a CM 200 (Philips, Schottky field emission gun, 200 kV). For investigating a coating-substrate cross-section, platinum was deposited on the sample surface and then a ca. 100 nm thick lamella was cut using a focused ion beam (FIB, DualBeam Strata 400S, FEI Company). The lamella was then applied on a TEM grid and only partially thinned to retain sufficient rigidity. Results and Discussion 37 4 Results and Discussion 4.1 Towards Thick Ceramic Coatings This chapter covers the route toward obtaining thick polymer-derived ceramic (PDC) coatings. It also touches on approaches that were not successful, both to avoid repetitions thereof and to gain valuable insights for further procedures from the identified causes of errors. While this chapter is meant to provide more general insights and benefit a broader range of precursors, within this dissertation, only the Dura-b precursors yielding Six(HfaTa1-a)(Bb)CyNz ceramics with a = {0.7, 1} and b = {0, 0.5, 1} are studied. 4.1.1 Synthesis and Properties of the Modified Precursor 4.1.1.1 Precursor Synthesis Studying the preceramic precursors is important because they determine processability as well as properties of the later derived ceramic coatings. For Dur1-0.5, this has previously been done by Yuan et al. [18]. Therefore, here only the, in cooperation with project 1 of the research training group, newly developed but chemically similar precursors Dur0.7-b are being discussed. The core idea behind this subtle change in composition is the highly desirable Hf6Ta2O17 phase that could be formed from this precursor, and that shows low thermal conductivity, no p