Our research in predictive microbiology is dedicated to establishing the fundamental principles that govern microbial inactivation during food processing. We move beyond empirical correlations to develop mechanistic models grounded in physics and chemistry. For thermal processing of low-moisture foods, our work has demonstrated that temperature-dependent water activity (aw) is a primary determinant of bacterial thermal resistance. By modeling the thermodynamic properties of food matrices like oils, we have successfully predicted the sharp decrease in aw at elevated temperatures and explained the resulting protective effect on pathogens like Salmonella. Our continuing research on the thermal processing of low-moisture food has obtained thermal resistance data at 120 °C and 0% RH. We have also demonstrated that organic acids significantly reduce Salmonella’s thermal resistance in dry heating environment. We are now extending this first-principles approach to non-thermal technologies. For instance, in our study of cold plasma-activated water (PAW) treatment, our preliminary results reveal that microbial inactivation is not merely a function of dose, but is strongly correlated with the dynamic history and concentration of specific reactive species, such as ozone, in the water. This allows us to model the antimicrobial effect based on quantifiable chemical kinetics, paving the way for predictable and controllable non-thermal sanitation.

This pillar focuses on engineering and validating the next generation of non-chemical, energy-based food processing technologies to enhance safety and sustainability. A cornerstone of our research is the development of pulsed laser sanitation. Our extensive investigation using a 5-watt Q-switched 355 nm UV laser has yielded compelling proof-of-concept results. We have achieved greater than a 6-log reduction of Salmonella on various surfaces, a critical benchmark for effective sanitation. Importantly, this high level of inactivation was accomplished with remarkable energy efficiency and, crucially, without inducing any observable thermal or physical damage to the treated surfaces. These findings establish pulsed UV lasers as a potent and precise tool for surface disinfection, overcoming the efficacy and material compatibility limitations of conventional methods. Building on this success, we are also engineering scalable cold plasma reactors for agricultural water treatment and advancing validation protocols for UVC sanitation by integrating novel sensing and modeling techniques.

Our vision is to create "smart" food manufacturing systems by seamlessly integrating advanced sensing, predictive modeling, and data-driven control. We are developing predictive digital twins of food processes, which serve as high-fidelity virtual counterparts to physical operations. These are not static simulations but dynamic models powered by Physics-Informed Neural Networks (PINNs), a hybrid approach that merges fundamental physical laws (from Pillar 1) with the adaptive power of machine learning. To inform these digital twins, we are pioneering advanced process sensors capable of operating in harsh industrial environments (like drying and roasting). Furthermore, we are innovating new ways to visualize and quantify process efficacy of UV light treatment. By using LiDAR and advanced image sensors, we create quantitative 3D models of food and contact surfaces. This allows us to understand, for the first time, how complex surface topography influences the performance of line-of-sight technologies like UV light. This data provides a crucial feedback loop for optimizing our predictive models and can be visualized through augmented reality (AR) to provide operators with intuitive, actionable insights, ultimately enabling a new paradigm of intelligent process control.