Spatial Temperature and Composition Inference from FTIR in an FCCVD CNT Reactor

Carbon nanotubes (CNTs) possess exceptional material properties with transformative potential across various fields. However, their widespread application is hindered by the need for more efficient and pure synthesis methods. Floating catalyst chemical vapor deposition (FCCVD) is a promising technique for mass-producing CNTs, but its efficiency is limited by the deactivation of catalyst nanoparticles due to carbon encapsulation. A lack of in-situ diagnostic tools for detailed internal chemistry and thermodynamics in FCCVD reactors further compounds this issue.<br/><br/>This study introduces a methodology to obtain synthetic Fourier Transform Infrared (FTIR) measurements from Computational Fluid Dynamics (CFD) simulations of an FCCVD CNT reactor. We validate these simulations against experimental data from the literature. Using OpenFOAM CFD software, we model detailed chemistry, including methane pyrolysis and CNT growth kinetics. The Radiant Monte-Carlo ray tracing software provides accurate solutions to the radiative transfer equation, enabling coupled radiation/chemistry modeling and generating incident spectra to an FTIR sampling device. Line-by-line accurate synthetic spectra are produced for gaseous species using the HITRAN and HITEMP databases. Spectral absorption coefficients are also introduced to account for emission and absorption by CNTs.<br/><br/>Our results, compared to literature for both emission-based and absorption-based FTIR, incorporate accurate radiative effects such as the spectral transmittance of quartz containers, boundary reflections, field-of-view restrictions, soot on the walls, and instrument line shape functions to mimic measurement uncertainties. This digital twin of an FCCVD CNT reactor FTIR measurements is then applied to a machine-learning assisted parameter inversion procedure to train chemical rate parameters in the CNT reactor. Additionally, a neural network is applied to obtain radial temperature and composition profiles using FTIR measurements. This approach has significant practical implications, enabling experiments for parameter inversion of chirality-dependent CNT growth kinetics, high throughput experimental testing and data acquisition, and optimal CNT reactor design.