Abstract
Calcium magnesium alumino-silicate (CMAS) attack is a pressing issue in the development of environmental barrier coatings (EBCs) for ceramic matrix composites (CMCs). CMAS originates as siliceous debris such as sand or volcanic ash, which can be ingested into aircraft turbine engines during flight. At temperatures greater than ̴1200°C, CMAS melts and can penetrate EBC materials, causing premature coating failure.
There is debate in the literature as to whether EBC materials sought for CMAS mitigation should be reactive or nonreactive with molten deposits. Due to the presence of coating defects such as grain boundaries, cracks, and pores, which are susceptible to CMAS penetration, it is hypothesized that an EBC material that reacts rapidly with CMAS to form a protective crystallized layer is better able to resist CMAS infiltration compared to one that is inert. This hypothesis was confirmed in the current work through investigation of the interaction behavior between CMAS and either rare earth (RE) titanates, RE silicates, or standalone apatite (Ca2RE8(SiO4)6O2; a commonly observed reaction product between RE coatings and CMAS) at 1300°C. The results of this work show that inducing rapid crystallization at the coating/glass interface reduces the ability for CMAS to penetrate defects and leads to slower infiltration. Of the EBC materials studied, Yb2SiO5 (YbMS) was best able to resist CMAS infiltration. YbMS reacted quickly with the glass to form either apatite or Yb2Si2O7 (YbDS), depending on the initial CMAS composition. Infiltration in YbMS was drastically slower than in the current standard EBC material, YbDS, which did not react with CMAS to form new crystalline phases and was instead penetrated via grain boundaries.
The effect of EBC microstructure on CMAS infiltration behavior was explored in model materials containing controlled amounts of YbMS within a YbDS matrix. YbMS was introduced either as “splats” (to model an air plasma spray (APS)-deposited EBC) or as a “fine dispersion.” The addition of YbMS to YbDS improves overall material resistance to CMAS infiltration. It was determined that including ≥ 20 vol% YbMS was beneficial in reducing glass penetration, as compared to phase pure YbDS. Model APS coatings exhibited a combination of grain boundary attack (of the YbDS matrix) and reactive crystallization of YbMS granules (most notably to form apatite). The formation of apatite slowed the incoming CMAS front. Fine dispersion samples were not penetrated as deeply as model APS materials due to glass spreading on their surfaces.
Experimental viscosity data was obtained for several CMAS-related melts. The effect of Ca/Si ratio and Al2O3/MgO content was assessed. Increasing the Ca/Si ratio (from 0.37 to 0.73) and MgO content (to 20 mol%) resulted in decreased viscosity, while increasing the Al2O3 content (to 30 mol%) resulted in increased viscosity. Results were compared to three viscosity models commonly cited in the literature (by FactSage, Fluegel, and Giordano et al.). The FactSage model was unequivocally the best at describing experimental viscosity data. CMAS viscosity was related to coating infiltration using melt infiltration models available in the literature. Infiltration cannot be described without considering the formation of crystalline reaction products between coating and glass.
The findings from this work address critical questions related to EBC design for CMAS mitigation. EBCs deposited by APS will contain coating defects such as grain boundaries and pores. Thus, material selection should seek to maximize reactivity between the coating and glass to form favorable phases that can slow incoming CMAS. The addition of YbMS to YbDS drastically improves the ability for model materials to resist infiltration. This is an important insight, as actual APS-deposited YbDS coatings will contain some YbMS. The results of this study show that a critical amount of YbMS is needed to induce improved behavior. This suggests that APS processing techniques can be tailored to optimize CMAS mitigation while maintaining other critical coating requirements. Finally, experimental viscosities for CMAS glasses were obtained and it was found that the FactSage model is best suited for future applications needing to describe glass viscosity. Current coating infiltration models indicate that EBC porosity is the most important coating-related parameter and glass viscosity is the most important CMAS-related parameter in predicting CMAS infiltration rates. Additionally, EBC phase constitution is critical, as reaction product formation between the coating and CMAS can block pores/penetration pathways. However, none of the available models accurately described experimental coating infiltration results.
