A Data-Rich Self-Driving Fluidic Lab for Accelerated Development of Colloidal Quantum Dots

Self-driving labs (SDLs) are emerging as a powerful tool to speed up the research, development, and manufacturing of advanced functional materials. By integrating lab automation, robotics, and artificial intelligence (AI), SDLs can autonomously explore complex synthesis-property parameter space of advanced materials based on human-defined objectives [1]. The SDLs' developments over the past few years have been driven by process intensification (PI) principles, to achieve faster, safer, and more efficient processes toward Self-Driving Fluidic Labs (SDFLs) that leverage flow synthesis technologies and real-time characterization techniques that could enable more efficient and sustainable data generation schemes.
SDFLs equipped with microscale fluidic reactors benefit from enhanced heat and mass transfer rates, short start-up and equilibration times, and reduced chemical consumption. When interfaced with high-throughput in-situ characterization techniques, SDFL platforms can enable accurate mapping between material properties and reaction conditions. However, current data generation schemes in SDFLs focus only on steady-state measurements, overlooking observations during transient states. Properly leveraging the transient data of flow chemistry platforms can offer deep insights into the material synthesis mechanisms, potentially speeding up experiments, providing valuable information to the AI agents of SDFLs [2,3]. This study introduces dynamic flow experiments as a data-rich method for mapping quantum dot (QD) synthesis space to material properties operating at least 100× faster than traditional steady-state flow synthesis strategies.
The dynamic flow experiments of the developed SDFL utilize in-situ characterization techniques to generate time-series data from controlled variation of reaction parameters. By correlating instantaneous flow conditions to steady-state equivalent residence times, we showcase the significant data intensification potential of continuous flow reactors to provide high quantity of high-quality experimental data to the AI agent of SDFLs. The dynamic flow experiment strategy increases experimental data throughput of SDFLs over 100-fold while reducing chemical consumption by at least 3 times, all in half the time compared to conventional steady-state experiments.
Specifically, we use cadmium selenide (CdSe) QDs as a case study to demonstrate accelerated parameter space mapping capabilities of the developed data-rich SDFL technology. Moreover, the resulting large experimental dataset paves the way for creating a robust digital twin of the reactive system, enabling successful autonomous closed-loop experimentation of multiple target emission wavelengths.
References:
[1] M. Abolhasani, E. Kumacheva, The rise of self-driving labs in chemical and materials sciences, Nat. Synth. 2 (2023) 483–492. https://doi.org/10.1038/s44160-022-00231-0.
[2] F. Florit, A.M.K. Nambiar, C.P. Breen, T.F. Jamison, K.F. Jensen, Design of dynamic trajectories for efficient and data-rich exploration of flow reaction design spaces, React. Chem. Eng. 6 (2021) 2306–2314. https://doi.org/10.1039/D1RE00350J.
[3] K.C. Aroh, K.F. Jensen, Efficient kinetic experiments in continuous flow microreactors, React. Chem. Eng. 3 (2018) 94–101. https://doi.org/10.1039/C7RE00163K.