Modeling of Complex Material Systems in Extreme Environments for Space Technology

The original proposal for this fellowship aimed to employ computational models to better inform studies of pyrolytic chemistry in phenolic resins. The current state of the art in phenolic-impregnated carbon ablative (PICA) thermal protection systems (TPS) employs a phenolic matrix processed for low density and low thermal conductivity with a carbon fiber internal structure. Under the extreme conditions of atmospheric entry, the phenolic matrix undergoes pyrolysis towards a state of amorphous carbon char, with much of the non-carbon byproducts blown off in gaseous form. The char provides high thermal emissivity from the surface, and the intermediate remaining PICA impedes the transmission of heat to the payload with a combination of endothermic chemical reactions and a low-conductivity microstructure. A potential path forward for PICA ablators is to engineer the evolution of microstructure and chemical composition to optimize the shedding or blocking of heat for a particular atmospheric entry event. Phenolic pyrolysis is, however, an exceedingly complex chemical process, and detailed studies of its chemical pathways are recent.1,2 Infrared3 and optical probes (Pugel et al, NASA Goddard 2006) have been employed to attempt detailed analysis of pyrolysis. While such probes have had some success, they leave room for improvement in terms of the exact chemical information needed for microstructural and chemical engineering of heat shields. We proposed to employ a relatively recent and versatile methods in computational electronic structure (time dependent density functional theory, TD-DFT) to attempt interpretation of optical, infrared, and UV probes in the context of phenolic pyrolysis. According to TABS 9.1.1.1, advances are needed in the state of the art of ablative TPS for milestones as early as 2020-30, targeting high-velocity EDL on Mars and Earth and eventually exploration of Venus. Woven and conformal TPS are highlighted as potential candidates to meet these milestones, as well as multi-TPS designs. An optical diagnostic utility could provide useful feedback for research and development in these areas. One of our research goals is to inform the development of such a probe. Our approach to this goal has led us through a range of topics. Tangentially, we performed a study of tailored ionic response with THz/IR electric fields, and a benchmark study of TD-DFT computed absorption spectra. Our best attempt to address the proposed optical/IR/UV diagnostic modeling was made in the fourth year, by assembling our computational results for a simulated pyrolyzed phenolic and computing a regression based model to determine the material’s composition from features of the computed spectra. A year by year summary of our research (and conclusions) over the course of this fellowship follows: 1. First year (continuing from the Fellow’s pre-NSTRF research): • We completed and published a TD-DFT based study of optical signatures of shock-induced chemistry in nitromethane, another chemical reaction whose precise details could be used for improved engineering.4 This work set much of the foundation for our proposed work on pyrolysis in phenolic resin. • Conclusion 1: Infrared absorption strength appears to be correlated with the disappearance of transient CNO and NO molecules as the backbone of the nitromethane molecule dissociates. This result has a nuanced relationship with contemporary experiments performed at Los Alamos National Laboratory.5 • We demonstrated TD-DFT based computed spectra for a variety of molecules that are likely to be represented in the chemical pathway of pyrolysis. • We began to attempt DFT and TD-DFT computations of optical spectra of a simulated pyrolyzing phenolic using atomistic coordinates produced in a previous DFT-based molecular dynamics (MD) study1. 2. Second year: • We began to compare computed absorption spectra against any available experiments, launching an attempt to benchmark TD-DFT computed spectra for the sorts of molecules present in pyrolyzing phenolic. • We began to study coherent ionic control with electric fields, initially targeting chemical isomerization as a proof of principle. Potential space technology targets include ultrafast switches, nonvolatile memory, photonic catalysis for chemical synthesis, and field-driven actuators in nano-electromechanical systems (NEMS). • We contributed computations to a collaborative study of second-harmonic generation (SHG) signatures under intense photoexcitation for molybdenum disulfide.6 • Conclusion 2: We found that electronic degrees of freedom could react strongly to radiative excitation without allowing damage to the atomistic structure. 3. Third year: • We completed our study of coherent control over chemical isomerization with electric fields, demonstrating traversal of energy barriers of ~0.1eV with residual heating of about 10 degrees celsius.7 • Conclusion 3: Coherent navigation of a potential surface may be possible in cases where specific vibrations can be targeted with an electric field that does not couple to electronic excitations. • We continued our work in TD-DFT based absorption spectra by benchmarking against measurements for molecules representing probable pyrolysis products, yielding statistics for the expected error in absorption peaks for 30 molecules and 6 different exchange and correlation (XC) formulations. • Conclusion 4: Absorption spectrum peaks tend to be computed to within about 0.5eV by the first generation of XC approximations, while popular approximations of the next generation systematically undershoot the absorption peaks by about 1eV. The latest formulations fall within about 0.3eV. Without careful attention to the choice of XC formulation, absorption peak errors of more than 1.5eV are not unexpected. 4. Fourth year: • The Fellow attempted to refine the aforementioned TD-DFT benchmark study to gain an objective and systematic approach for using measured data to inform the computation of an absorption spectrum from vertical excited states. • The Fellow assembled all computational work to date on phenolic pyrolysis and investigated the potential for a regression-based model to determine composition from spectral features. • Conclusion 5: A regularized regression model can be trained specifically to identify bonding motifs in pyrolyzing phenolic to within a few atomic percent from spectral band intensities. The most useful spectral features are the aromatic pi-pi* excitation and low-frequency vibrations. The most difficult species to predict are alkane groups and less populous species (in this case, ether and alkyne groups). • The Fellow completed his defense and dissertation, graduating in June 2016.