Global sensitivity analysis of a model for silicon nanoparticle synthesis

• We model formation of silicon aerosols using a detailed population balance model.
• The model response is studied across 19 different experiments.
• A High Dimensional Model Representation (HDMR) is generated for each of the responses.
• The HDMR reveals information about the model׳s sensitivity to its adjusted parameters.

This paper presents a global sensitivity analysis of the detailed population balance model for silicon nanoparticle synthesis of Menz & Kraft [2013a, A new model for silicon nanoparticle synthesis, Combustion & Flame, 160:947–958]. The model consists of a gas-phase kinetic model, fully coupled with a particle population balance. The sensitivity of the model to its seven adjusted parameters was analysed in this work using a High Dimensional Model Representation (HDMR). An algorithm is implemented to generate response surface polynomials with automatically selected order based on their coefficient of determination. A response surface is generated for 19 different experimental cases across a range of process conditions and reactor configurations. This enables the sensitivity of individual experiments to certain parameters to be assessed. The HDMR reveals that particle size was most sensitive to the heterogeneous growth process, while the particle size distribution width is also strongly dependent on the rate of nucleation.