A self-optimizing algorithm for precise nanocrystal synthesis

New system allows extremely rapid experimentation using a microfluidic reactor


A new autonomous system for nanoparticle synthesis and optimization has been developed by Léonard Bezinge and coworkers in the deMello group. By combining an advanced ‘goal seeking’ control algorithm with a microfluidic reactor system, the team demonstrated rapid and efficient targeted synthesis of fluorescent nanoparticles that vastly outperforms traditional batch-based approaches.


Léonard and the team used the system to synthesize lead halide perovskite nanocrystals, which show great promise for applications in solar cells, lighting and lasers. However, there is a need to improve their performance and stability before their promise can be realized. This is an extremely difficult task because of the immense variety of elemental compositions and reaction conditions that must be explored. Using traditional flask-based chemistry, where each reaction setup will yield only a single product, would be extremely costly, time consuming and inefficient.


The deMello group has previously developed a variety of segmented flow microfluidic reactors for rapid nanoparticle synthesis. Here the reaction solution is split into many thousands of droplets acting as isolated reaction vessels, using only small amounts of precious reagents. In line analysis of fluorescence emission then allows tuning of reaction conditions in real time. In their new work, Léonard and the team coupled such a microfluidic reactor with a new control algorithm. When told a desired outcome in terms of color of fluorescence, the system eventually returns an accurate model identifying reactions conditions that yield the desired nanoparticles. It does this by iteratively exploring the parameter space, studying previously selected experimental conditions, then selecting subsequent optimal sampling points with the highest probability of reaching the target product. Such automation is a vital asset in the study and optimization of complex chemical compounds.


Although demonstrated using lead halide perovskites, the new system may be applied to a variety of compounds, and could achieve high impact in rapid and efficient materials discovery.


Written by Philip Howes.

Read the full paper here.


The algorithm continually tunes reaction conditions, altering the color of light emitted from the individual reaction droplets

Links
ADDRESS

Inst. f. Chemie- u. Bioing.wiss.

HCI F 117

Vladimir-Prelog-Weg 1-5/10

8093 Zürich
Switzerland

2020 © by the contributing authors