Abstract
Aerospace materials that can operate at higher temperatures lead directly to fuel and cost savings, provided they enable engines that are lighter and run hotter while producing the same power. The leading candidate is the SiC/SiC ceramic-matrix composite (CMC), with one third the density of the nickel superalloys it replaces and roughly 165 °C more temperature capability. So far only static, lightly loaded CMC parts have entered service. Rotating hardware and the full hot section have not, because CMCs recede in the water vapor produced by combustion. They require environmental barrier coatings (EBCs) to survive at all. The coating solves the recession problem and creates a new one. A multi-layer ceramic system fails in ways engines have not seen, and the models needed to predict those failures still rest on the most fundamental measured quantities, including modulus and residual stress. No continuous measurement of either exists in the literature for these coating systems through their processing and operating temperatures. A census conducted for this work found only 39 papers reporting a primary Yb-silicate modulus value, and zero at-temperature absolute modulus values for a spray-deposited Yb-disilicate coating.
This work describes the development of a pair of measurement systems that measure both at once on the same multilayer beam, using optical curvature for stress and impulse excitation for modulus, continuously to 1500 °C, beyond silicon degradation and the “2700 °F challenge” limits. The instruments, fixturing, and software were designed and built from scratch for this work, down to a casting process for the high-temperature parts for speedier iteration, and including furnace-survivable impulse excitation with automated multi-mode resonance tracking robust to the mode-crowding that appears at temperature. Layer properties are recovered in a deliberately conservative order: the substrate is characterized bare, the silicon bond coat on a freestanding reference, the pair validated on a bilayer, and only then is the top coat backed out from the tri-layer response and anchored to its own ex-situ measurement. This recovery is an inversion of measurements, not a simulation. Per-layer modulus and absolute through-thickness stress are obtained by imposing equilibrium on the measured curvature, an eigenstrain-closure inversion that reads the coating’s inelastic relaxation out of the data rather than assuming a creep or viscosity law, with every layer property checked against an independent number wherever one exists. The Yb-silicate modulus rises roughly sixfold through the first heat treatment, from 15–21 GPa as-sprayed to up to 124 GPa, and the as-sprayed value independently matches ex-situ measurement near 21 GPa. The coating finishes in 43–78 MPa of residual compression, with the silicon bond coat and CMC carrying the small opposite reaction, reproduced across four independently sprayed beams spanning two coating thicknesses and two heating rates.
These histories allow quantification of a mechanism that has, to the author’s knowledge, only ever been described qualitatively. The amorphous phase flows viscously until crystallization arrests it, and the stiffness the coating finally gains is related to how much flow occurred first. This is a measured self-healing law. A kinetics model expressing this mechanism reproduces both the curvature and modulus histories across heating rates from 1 to 5 °C/min, and the activation energy it yields (767 ± 67 kJ/mol by Kissinger analysis, R² = 0.99) falls within the 710–900 kJ/mol range of independent same-material DSC measurements. The findings reduce to practical heat-treatment design rules. Crystallization is temperature-gated near 1000–1030 °C, shifting about 30 °C across a fivefold change in heating rate. Ramps slower than about 3 °C/min may cost roughly 15 percent of the final coating modulus and buy no residual-stress benefit in return. Combining the per-layer stress and modulus histories gives the stored elastic strain energy, the quantity that drives delamination, resolved in time through processing. These records expose a class of events that endpoint characterization cannot see: stress and stored energy spike briefly during the metastable to stable phase change region, reaching the measured toughness of these interfaces. Damage in these systems may be decided in transients, not end states. Together these provide manufacturers with quantitative tools for designing pre-service heat treatments and a benchmark dataset against which process models of EBC systems can be tested.
