Global Uncertainty-Sensitivity Analysis on Mechanistic Kinetic Models: From Model Assessment to Theory-Driven Design of Nanoparticles

TL;DR

This paper demonstrates how global uncertainty and sensitivity analysis on mechanistic models can guide the rational design of nanoparticle synthesis, exemplified by calcium-silicate-hydrate precipitation, to optimize properties like specific surface area.

Contribution

It introduces a comprehensive framework combining mechanistic modeling with global UA/SA to inform nanoparticle synthesis protocols and understand model behavior.

Findings

UA/SA identifies key factors influencing particle surface area

Model analysis guides fine-tuning of synthesis conditions

Approach is applicable to various nanoparticle synthesis methods

Abstract

The optimal design of nanoparticle synthesis protocols is often achieved via one-at-a-time experimental designs. Aside from covering a limited space for the possible input conditions, these methods neglect possible interaction between different combinations of input factors. This is where mechanistic models embracing various possibilities find importance. By performing global uncertainty/sensitivity analysis (UA/SA), one can map out the various outcomes of the process vs. different combinations of operating conditions. Moreover, UA/SA allows for the assessment of the model behavior, an inevitable step in the theoretical understanding of a process. Recently, we developed a coupled thermodynamic-kinetic framework in the form of population balance modelling in order to describe the precipitation of calcium-silicate-hydrate. Besides its relevance in the construction industry, this inorganic nanomaterial offers potential applications in biomedicine, environmental remediation, and catalysis most notably due to ample specific surface area that can be achieved by carefully tuning the synthesis conditions. Here, we apply a global UA/SA to an improved version of our computational model in order to understand the effect of variations in the model parameters and experimental conditions (induced by either uncertainty or tunability) on the properties of the product. With the specific surface area of particles as an example, we show that UA/SA identifies the factors whose control would allow a fine-tuning of the desired properties. This way, we can rationalize the proper synthesis protocol before any further attempt to optimize the experimental procedure. This approach is general and can be transferred to other nanoparticle synthesis schemes as well.