A Data Mining and Machine Learning Software for Accelerated High-Temperature Corrosion Resistant Materials Design

Deposit-induced degradation of alloys and ceramic coatings in turbine engines is regarded as one of the core challenges preventing further gains in performance and efficiency. Historically, hot corrosion (HC, i.e., accelerated oxidation) of alloys in the presence of chloride and sulfate deposits has been treated separately from the degradation arising when molten silicate deposits (CMAS)1,2 infiltrate and react with thermal and environmental barrier coatings (TBC and EBC). This distinction arose in part based on the temperatures at which each predominates. Hot corrosion caused by Na-based sulfates is most pronounced near 700ºC (due to high pSO3) and 850ºC (via scale fluxing) and subsides above 950ºC as the Na2SO4 evaporates. Conversely, silicate melts form most readily above 1150ºC and their ability to infiltrate (and thus stiffen) porous or segmented TBCs increases as the melt viscosity drops at higher temperatures.    Specifically, deposits containing CaSO4 and mixed oxides persist to higher temperatures, expanding the range for HC phenomena to at least 1150ºC (and likely up to the design limits dictated by alloy strength). Likewise, transient chlorides and sulfates (and their associated oxides) lower the melting temperature of CM(F)AS-type deposits2,5 leading to infiltration of TBCs well below the oxide eutectics around 1150ºC. Finally, there is evidence of in-service vapor-phase infiltration of CaSO4 and mixed oxides into TBCs absent melt formation.6,7 These phenomena extend beyond the existing mechanistic understanding of the degradation processes. A new materials design paradigm capturing the effects of complex deposit chemistries (including mixed chlorides, sulfates, and oxides) and the gaseous environment (e.g., pSO3, pH2O) is urgently needed to accelerate the implementation of next-generation materials.   In the Phase II program, QuesTek will continue leading the efforts on developing/applying computational tools and ML technology for rapid design of new materials with superior high temperature corrosion resistance for naval applications. Thermomechanical factors in materials degradation will be included in the Phase II effort, in addition to the thermochemical processes. Prof. David Poerschke at UMN will continue providing domain knowledge related to CMAS-HC degradation and carry out comprehensive, systematically designed experiments to generate new data for ML model training and validation. Prof. Carlos Levi and Prof. Brian Gleeson, who have decades of experience related to CMAS-HC, will continue serving as the consultants in Phase II to provide technical guidance and data. Partners at GE Aviation and Pratt & Whitney will provide industrially relevant guidance in Phase II and will act as voice-of-the-customer to facilitate future implementation and commercialization of the developed technology.