In materials science, the discovery of recipes that yield nanomaterials with defined optical properties is costly and time-consuming. In this study, we present a two-step framework for a machine learning-driven high-throughput microfluidic platform to rapidly produce silver nanoparticles with the desired absorbance spectrum. Combining a Gaussian process-based Bayesian optimization (BO) with a deep neural network (DNN), the algorithmic framework is able to converge towards the target spectrum after sampling 120 conditions. Once the dataset is large enough to train the DNN with sufficient accuracy in the region of the target spectrum, the DNN is used to predict the colour palette accessible with the reaction synthesis. While remaining interpretable by humans, the proposed framework efficiently optimizes the nanomaterial synthesis and can extract fundamental knowledge of the relationship between chemical composition and optical properties, such as the role of each reactant on the shape and amplitude of the absorbance spectrum.

Bayesian optimization (BO) has been leveraged for guiding autonomous and high-throughput experiments in materials science. However, few have evaluated the efficiency of BO across a broad range of experimental materials domains. In this work, we quantify the performance of BO with a collection of surrogate model and acquisition function pairs across five diverse experimental materials systems. By defining acceleration and enhancement metrics for materials optimization objectives, we find that surrogate models such as Gaussian Process (GP) with anisotropic kernels and Random Forest (RF) have comparable performance in BO, and both outperform the commonly used GP with isotropic kernels. GP with anisotropic kernels has demonstrated the most robustness, yet RF is a close alternative and warrants more consideration because it is free from distribution assumptions, has smaller time complexity, and requires less effort in initial hyperparameter selection. We also raise awareness about the benefits of using GP with anisotropic kernels in future materials optimization campaigns.

The vibration of bubbles can produce intense microstreaming when excited by ultrasound near resonance. In order to study freely oscillating bubbles in steady conditions, we have confined bubbles between the two walls of a silicone microchannel and anchored them on micropits. We were thus able to analyse the microstreaming flow generated around an isolated bubble or a pair of interacting bubbles. In the case of an isolated bubble, a short-range microstreaming occurs in the channel gap, with additional in-plane vortices at high amplitude when Faraday waves are excited on the bubble periphery. For a pair of bubbles, we have observed long-range microstreaming and large recirculations describing a ‘butterfly’ pattern. We propose a model based on secondary acoustic Bjerknes forces mediated by Rayleigh waves on the silicone walls. These forces lead to attraction or repulsion of bubbles and thus to the excitation of a translational mode in addition to the breathing mode of the bubble. The mixed-mode streaming induced by the interaction of these two modes is shown to generate fountain or anti-fountain vortex pairs, depending on the relative distance between the bubbles.