Accelerating the Flame Retardant Design: From Small-scale to Bench-scale

Abstract

This research first investigated small-scale flammability and thermal stability for novel nanofillers in polymeric nanocomposites. The fire tests were then scaled up to the bench scale and addressed the flame retardancy with a cone calorimeter. Finally, an accelerated design of flame-retardant polymeric nanocomposites via a machine learning prediction model was performed to associate the observations in the previous tests.

In detail, a two-dimensional nanomaterial MXene-organic hybrid (O-Ti₃C₂) was applied to polystyrene (PS) as a nanofiller. The resulting PS nanocomposite (PS/O-Ti₃C₂) showed good thermal stability and lower flammability, evidenced by small-scale thermogravimetric analysis (TGA) and pyrolysis-combustion flow calorimetry (PCFC). Furthermore, by comparing the TGA and PCFC results between different samples, the thermal stability and 2D thermal- and mass-transfer barrier effect of MXene-organic hybrid nanosheets were revealed to play essential roles in delaying polymer degradation.

Subsequently, MIL-125 is proposed as a novel type of flame-retardant nanofiller to improve further the ignitability, thermal stability, and flame retardancy of acrylonitrile butadiene styrene (ABS) engineering plastics. Through the “booster” effect with intumescent flame retardants (IFR) and MIL-125, the resulting ABS composite received promising flame retardancy and smoke suppression performance in multiple bench-scale fire tests. Furthermore, from the residue analysis, the compact carbon char indicates that an effective heat-insulating protective layer is formed, proving that the ABS/IFR/MIL-125 system is a highly efficient flame-retardant system.

Finally, a flame retardancy database was built, including information on polymer flammability, thermal stability, and nanofiller properties. Five machine-learning algorithms were then applied to predict the flame retardancy index for different polymeric nanocomposites. Among them, extreme gradient boosting regression gives the best prediction with a coefficient of determination (R²) of 0.94 and a root-mean-square-error (RMSE) of 0.17, which in turn was used to guide the design of polymeric nanocomposites for flame retardant applications. Following the guidelines, a high-performance flame-retardant polymeric nanocomposite was designed and synthesized, of which the experimental result was well-matched with the machine learning prediction (up to 95% accuracy). This result demonstrated a fast identification of flame retardancy of polymeric nanocomposite without bench-scale fire tests, which could accelerate the design of functional polymeric nanocomposites in the flame retardant field.