Forecasting continuous carbon nanotube production in the floating catalyst environment

Abstract

We present validated statistical models and univariate correlations of carbon nanotube (CNT) textile properties (specific electrical conductivity, Raman G:D ratio and mass yield rate) extracted continuously from floating catalyst chemical vapour deposition (FC-CVD) reactors over a uniquely wide multivariate experimental space. This includes directly controlled reactor settings (e.g. precursor concentrations, gas flow rates, furnace temperatures and winding speeds), indirect parameters (e.g. ambient temperature and pressure), and time-dependent reactor influences such as reactor tube age. Two vertical FC-CVD reactors, with different precursor delivery architectures, were considered: 1) in which precursors were pre-mixed together as a liquid solution that was directly injected into reactor; 2) in which vaporised precursors were independently injected in the gas phase using Coriolis-based microfluidic mass flow controllers with concentrations monitored in-line using FTIR spectroscopy. Factors favouring highest electrical conductivity fibres include: lower hydrogen flows, lean fuel-to-gas mixtures, higher winding rates, higher argon flows, with many thiophene concentration interactions with other parameters; for highest Raman G:D ratios: leaner fuel-to-gas mixtures, lower thiophene concentrations, higher hydrogen flows, and greater external laboratory pressure; but for yield rate, systematic trends were harder to discern. This study demonstrates the degree of predictability in FC-CVD reactors, quantitatively ranks impact of FC-CVD parameters, and identifies regions of fibre “spinnability” which correspond well with a recent meta-analysis of experimental results in the literature.