Presentation Details
| Meniscus Integration for Thin film Nano Electronics Robot (MEITNER): Automated, Scalable Perovskite Manufacturing via Multi-Objective Bayesian Optimization Daniel Abdoue1, Ethan G.Schwartz2, Maimur Hossain2, Jake Kittell3, Kyle Sipe1, Danil Ryzhokhin1, Vivek Babu4, Chad Miller4, Tonio Buonassissi5, Devin Mackenzie2, 6, David Fenning1. 1University of California San Diego, San Diego, CA, USA.2University of Washington, Seattle, WA, USA.3Happiness Tech LLC, Morrisville, VT, USA.4Verde Technologies Inc., Burlington, VT, USA.5Massachusetts Institute of Technology, Cambridge, MA, USA.6Washington Clean Energy Testbeds, Seattle, WA, USA |
Abstract
Solution-processed metal halide perovskites are promising candidates for next-generation photovoltaics, yet scalable film formation remains difficult to optimize and reproduce due to the strong coupling between ink rheology, solvent evaporation, and crystallization kinetics. These challenges are particularly acute for meniscus-based deposition methods, whose processing physics differ fundamentally from spin coating, complicating recipe transfer and slowing scale-up. Here, we introduce MEniscus Integration for Thin Film Nano Electronics Robot (MEITNER), a fully robotic, meniscus-native platform that integrates gravity-fed meniscus coating, thermal processing, and multimodal characterization within a closed-loop experimental workflow guided by multi-objective Bayesian optimization. MEITNER enables rapid, automated exploration of manufacturing-relevant processing spaces while explicitly balancing optoelectronic performance, film uniformity, and crystalline phase purity. By coupling robotic execution with the Bayesian Optimization Shared Service (BOSS) framework, the platform minimizes experimental variability and reduces the total number of experiments required to identify Pareto-optimal processing windows. Importantly, optimization is performed directly on a scalable coating architecture with rapid chemistry exchange, allowing crystallization behavior and film formation to be evaluated under conditions relevant to large-area manufacturing rather than simplified laboratory deposition. This work demonstrates a pathway for autonomous, data-efficient perovskite process development that prioritizes scalable deposition physics from the outset. By integrating intelligent decision-making with manufacturing-relevant experimentation, MEITNER addresses key bottlenecks in perovskite materials research and provides a practical framework for accelerating translation from laboratory discovery to scalable photovoltaic manufacturing.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.