Self‐Driven Multistep Quantum Dot Synthesis Enabled by Autonomous Robotic Experimentation in Flow

Abstract

Inorganic lead halide perovskite (LHP) quantum dots (QDs) have recently emerged as a promising class of semiconducting materials for next-generation, solution-processed optoelectronic devices. [1] For example, inorganic LHPs have surpassed the performance of conventional IV-VI QDs in photovoltaic devices. [2] The prominence of LHPs among other semiconductor nanocrystals is mainly attributed to their high photoluminescence quantum yield (PLQY), high defect tolerance, facile bandgap tunability, and narrow emission linewidth. The ease of peak emission bandgap tuning (1.7-3.1 eV) makes inorganic LHP QDs a versatile material for widespread applications ranging from solar cells (1.77 eV), [3-6] light-emitting diodes (blue 2.7 eV, green 2.39 eV, and red 1.88 eV), [7-9] and various photocatalytic reactions. [10-12] The peak emission energy of cesium lead halide QDs (CsPbX 3 , X ¼ Cl, Br, I) can be readily tuned by varying i) QD size using the quantum confinement effect, [13-16] ii) ligand composition, [17-19] iii) the chemical composition of the QD through anion, [20-22] and/or cation exchange, [23] and iv) the precursor halide content. [1,24] Despite producing high-quality monodispersed CsPbX 3 QDs, [1] flask-based hot-injection synthetic routes impose major challenges from large-scale manufacturing and reproducibility perspectives. Hot-injection colloidal synthesis requires operating at high temperatures (>150 C), which increases the overall energy costs and necessitates specific reactor design modifications to ensure homogenous, uniform heat distribution across the reactor. Furthermore, manual, flask-based colloidal syntheses are notorious for their lack of reproducibility (batch-to-batch variation and operator error), and difficulty of integration with material diagnostic probes. [13,24,25] Room-temperature colloidal synthesis (e.g., ligand-assisted reprecipitation strategy) [7,26,27] and post-synthesis halide exchange reactions [20-22,28] of CsPbBr 3 QDs are considered attractive alternatives to the hot-injection synthesis strategy for facile and precise bandgap engineering of LHP QDs. QD purification normally involves washing steps that consist of antisolvent addition followed by centrifugation, aliquot disposal, and fresh solvent addition. Moreover, washing and the subsequent redispersal of LHP QDs in fresh solvent disrupts the surface ligands, leading to ligand detachment, [29,30] surface defects (lowering the PLQY), and reduced colloidal stability of the LHP QDs. [30] Removal of the intermediate washing step of halide exchange reactions can enable end-to-end continuous manufacturing of inorganic LHP QDs and accelerate their adoption by chemical and energy technologies.

CsPbI3 Nanocrystals Go with the Flow: From Formation Mechanism to Continuous Nanomanufacturing

“…The use of PFO serves two purposes: i) it is used as a carrier phase to segment the reactive phase and induce an axisymmetric swirling pattern within the formed droplets to enhance mixing, and ii) it preferentially wets the PFA inner wall and forms a lubrication film to isolate the reactive droplets and prevent fouling. [4,[40][41][42] The Cs precursor stream is mixed with an octadecene (ODE)/oleic acid (OA) solvent stream at the room temperature, using a T-junction (IDEX Health and Sciences), before entering the flow segmentation module to achieve the desired Cs concentration (Figure 2A). The segmentation module is a custom-machined four-way junction (polyether ether ketone), specifically designed to achieve reliable and uniform flow segmentation with two reactive streams and a carrier fluid.…”

“…The flow cell is a machined aluminum block with three optical fiber (Ocean Insight, 600 µm diameter) ports and a fluorinated ethylene propylene tubing (OD: 1/16 in., ID: 0.03 in.). [40] The flow cell allows for real-time spectral characterization of the inflow synthesized CsPbI 3 NCs through UV-vis absorption and photoluminescence (PL) spectroscopy (Figure 2A). The optical fibers are connected to a broadband light source (Ocean Insight, DH-2000-BAL) for UV-vis absorption spectroscopy, a 450 nm light emitting diode (Thorlabs) as the PL excitation source, and a miniature fiber-coupled spectrometer (Ocean Insight, HDX).…”

Data-Centric Architecture for Self-Driving Laboratories with Autonomous Discovery of New Nanomaterials

“…However, there are examples of successful SDL projects and their applications for the discovery of new materials, including clean energy technologies [ 19 ], organic photovoltaics [ 67 ], discovery of thin film materials [ 17 ], autonomous synthesis of carbon nanotubes [ 68 ], and self-driving ”Artificial Chemist” [ 18 ] for producing inorganic perovskite and inorganic lead halide perovskite quantum dots [ 69 ]. It is worth noting that the functionality of each instance of SDL is limited and depends greatly on certain aims and areas of research and application.…”

“…In principle, operations that are simple, even for an inexperienced person, such as pressing a powder sample into a cell and connecting the necessary connectors in an X-ray absorption spectrometer, can cause extreme difficulties and be a serious obstacle for an automatic robotic platform. Due to this, the flow chemistry approach has more capabilities to be fully automated currently, for example [ 62 , 69 , 71 ]. However, regardless of whether the process is available or unavailable, the ability to obtain automated and autonomous chemistry synthesis processes followed by nanomaterial characterization the automated workflow is the main provider of new scientific research data at SDL architecture.…”

Self‐Driving Platform for Metal Nanoparticle Synthesis: Combining Microfluidics and Machine Learning

“…The ML algorithms may be thought of as an adaptive form of experimental design, which allows one to efficiently identify reaction conditions that yield NPs with desired properties. Self-driving MF platforms integrated with ML have been used for the synthesis of metal chalcogenide quantum dots, [32] perovskite, [33][34][35] and metal NPs. [36] The ML algorithms searched within a predefined parameter space when targeting for certain characteristics in the photoluminescence or absorption spectra of the NPs such as emission and absorption wavelength, as well as linewidth.…”