Fabrication of SERS Substrates Using Silver-Coated Gold Nanostars for Chemical Sensing: A Multiobjective Bayesian Optimization Approach

Field detection of trace analytes remains a challenge in many sectors, including national security, public health, and environmental monitoring. This challenge arises from the scarcity of field-deployable technologies capable of identifying trace analytes in the complex matrices of real-world samples. Surface-enhanced Raman scattering (SERS) is a promising technique for chemical sensing because it offers sensitive and specific analyte detection with commercial portable Raman systems. Despite the immense potential of SERS, commercially available SERS technologies that are accurate, reliable, cost-effective, and compatible with portable spectrometers remain scarce. This scarcity is due to the challenges associated with the complex design and resource-intensive optimization of the manufactured substrates that can achieve both high SERS intensity and signal uniformity. Machine learning (ML) tools are ideal for addressing complex optimization problems, but their application in SERS substrate fabrication remains largely unexplored. We present a pioneering example of applying ML to optimize SERS substrate fabrication using silver-coated gold nanostars for chemical sensing. In this study, we used multiobjective Bayesian optimization to achieve both high SERS intensity and signal uniformity across the SERS substrate. Optimal fabrication parameters were identified for spin and flow coating of silver-coated gold nanostars. The optimized spin coated substrate achieved an experimental limit of detection of 100 nM with an application relevant compound (ammonium nitrate), suggesting its potential for trace-level sensing. This work underscores the potential for ML to accelerate the development of field-deployable SERS technologies for trace analyte detection. Furthermore, we provide a general framework for optimized fabrication of nanoparticles on surfaces for other applications such as catalysis, electronics, and medicine.