Machine learning–based inverse design for electrochemically controlled microscopic gradients of O ₂ and H ₂ O ₂

Abstract

A fundamental understanding of extracellular microenvironments of O 2 and reactive oxygen species (ROS) such as H 2 O 2 , ubiquitous in microbiology, demands high-throughput methods of mimicking, controlling, and perturbing gradients of O 2 and H 2 O 2 at microscopic scale with high spatiotemporal precision. However, there is a paucity of high-throughput strategies of microenvironment design, and it remains challenging to achieve O 2 and H 2 O 2 heterogeneities with microbiologically desirable spatiotemporal resolutions. Here, we report the inverse design, based on machine learning (ML), of electrochemically generated microscopic O 2 and H 2 O 2 profiles relevant for microbiology. Microwire arrays with suitably designed electrochemical catalysts enable the independent control of O 2 and H 2 O 2 profiles with spatial resolution of ∼10 1 μm and temporal resolution of ∼10° s. Neural networks aided by data augmentation inversely design the experimental conditions needed for targeted O 2 and H 2 O 2 microenvironments while being two orders of magnitude faster than experimental explorations. Interfacing ML-based inverse design with electrochemically controlled concentration heterogeneity creates a viable fast-response platform toward better understanding the extracellular space with desirable spatiotemporal control.

Electrochemical platform of controllable pH gradients in microbial systems

“…[22][23][24][25] The properties of gradients generated by electrochemical technique can be modulated in several ways, such as by the arrangement of electrodes in combination with diffusion or by spatiotemporal variation of the applied potential. 7,[26][27][28][29] However, design spatial and temporal control over electrochemical gradients is still in its infancy.…”

Inverse design of intermediate band solar cell via a joint drift-diffusion simulator and deep reinforcement learning scheme

“…Machine learning or reinforcement learning-based inverse design attempts to invert the problem to come up with a design that renders desired properties, generally amounting to too many nonlinear problems that are difficult to solve, although various approaches have been proposed. [5][6][7][8][9][10][11][12] By far, most of the effective inverse design problems based on surrogate machine learning and reinforcement learning model have been dominantly focused on the field of nano-photonics 4,[13][14][15][16][17][18][19][20][21][22] as well as the field of quantum control. [23][24][25][26][27] Few attempts have been done in the field of energy conversion devices such as photovoltaic devices.…”