Available on-demand - *F.MT07.07.10
Accelerated Quantum Dot Development in Flow—Convergence of Flow Chemistry, Colloidal Synthesis and Machine Learning
North Carolina State University1
Metal halide perovskite quantum dots (QDs) have recently emerged as an exciting class of semiconducting materials, outperforming conventional II-VI, IV-VI, and III-V semiconductor nanocrystals in QD-based optoelectronic devices.1 Despite the substantial improvement in performance of perovskite QD-based optoelectronic devices over the past five years, two major obstacles hinder their development: (i) Edisonian (trial and error)-based QD synthesis, discovery, and optimization methods, and (ii) poor size dispersity of batch-synthesized colloidal nanocrystals, resulting in surface trap states. Existing material development strategies very often fail to overcome the demands of colloidal QDs’ vast colloidal synthesis and processing universe, resulting in time- and cost-intensive QD development efforts.
Recent advances in machine learning (ML),2 including deep neural networks (DNNs) and reinforcement learning algorithms, provide an exciting opportunity for reshaping the synthesis and development of perovskite QDs, through the machine-based direction of a high-throughput QD synthesis bot. Capitalizing on the recent progress of ML-based optimization algorithms, smart QD manufacturing strategies relying on DNNs trained on experimentally measured QD properties can be envisioned to accelerate the discovery, synthesis, and manufacturing of metal halide perovskite QDs with application-guided optoelectronic properties.
In this talk, I will present an Artificial Chemist,3 a modular intelligent fluidic microprocessor4 capable of autonomous synthesis, optimization, and end-to-end manufacturing of colloidal QDs for direct utilization in high-performance and low-cost optoelectronic devices. The Artificial Chemist can rapidly and efficiently (i) explore the massive parameter space of colloidal perovskite QDs, (ii) learn their synthesis and processing schemes, (iii) identify the composition and relevant synthesis route(s) of perovskite QDs to achieve specific optoelectronic properties, and (iv) continuously manufacture the rapidly optimized perovskite QDs at a fraction of time/cost of batch techniques. The reconfigurable Artificial Chemist technology utilizes a multimodal in-situ material diagnostic probe (absorption/photoluminescence (PL) spectroscopy) in conjunction with a real-time, ensemble DNN adaptive sampling algorithm to enable simultaneous optimization of PL quantum yield and emission linewidth of colloidal QDs for any desired peak emission energy. The Artificial Chemist technology can be readily adapted for accelerated development and autonomous end-to-end manufacturing of other solution-processed nanomaterials.
1. (a) Hazarika, A., et al., ACS Nano 2018, 12 (10), 10327-10337; (b) Sanehira, E. M., et al., Sci. Adv. 2017, 3 (10); (c) Swarnkar, A., et al., Science 2016, 354 (6308), 92-95; (d) Yoon, H. C., et al., ACS Applied Materials & Interfaces 2016, 8 (28), 18189-18200; (e) Lu, M., et al., Advanced Functional Materials 0 (0), 1902008.
2. (a) LeCun, Y., et al., Nature 2015, 521, 436; (b) Le, T. C.; Winkler, D. A., Chem. Rev. 2016, 116 (10), 6107-6132.
3. Epps, R. W., et al., Advanced Materials 2020, (in press), 2001626.
4. (a) Abdel-Latif, K., et al., Advanced Functional Materials 2019, 29 (23), 1900712; (b) Epps, R. W., et al., Lab Chip 2017, 17 (23), 4040-4047.