Predicting Indium Phosphide Quantum Dot Properties from Synthetic Procedures Using Machine Learning

Abstract

The prediction of chemical reaction outcomes using machine learning (ML) has emerged as a powerful tool for advancing materials synthesis. However, this approach requires large and diverse datasets which are extremely limited in the field of nanomaterials synthesis, due to inconsistent and nonstandardized reporting in the literature, and a lack of understanding of synthetic mechanisms. In this study, we extracted parameters of InP quantum dot (QD) syntheses as our inputs, and resultant properties (absorption, emission, diameter) as our outputs from 72 publications. We "filled in" missing outputs using a data imputation method to prepare a complete dataset containing 216 entries for training and testing predictive ML models. We defined the descriptor space in two ways (condensed and extended) based on the chemical identity or role of reagents to explore the best approach for categorizing input features. We achieved mean absolute errors (MAEs) as low as 20.29, 11.46, and 0.33 nm for absorption, emission, and diameter respectively with our best ML model. We used these models to deploy an accessible and interactive webapp for designing syntheses of InP (https://share.streamlit.io/cossairt-lab/indiumphosphide/Hot_injection/hot_injection_prediction.py). Using this webapp, we investigated the power of ML to uncover chemical trends in InP syntheses, such as the effects of common additives. We also designed and conducted new experiments based on extensions of literature procedures and compared our experimentally measured properties to predictions, thus evaluating the "real-life" accuracy of our models. Conversely, we designed an experiment to obtain InP QDs with specific properties. Finally, we applied the same approach to train, test, and launch predictive models for CdSe QDs by expanding a previously published dataset. Altogether, our data pre-processing method and ML implementations in this study show the ability to design materials with targeted properties and explore underlying reaction mechanisms despite limited data resources.

Synthesis, Surface Chemistry, and Fluorescent Properties of InP Quantum Dots

“…Nguyen et al recently used machine learning models comparing over 200 published InP QD syntheses to investigate trends in InP syntheses and found that annealing temperature, annealing time, and the presence of zinc additives were the most influential parameters studied when predicting optical peaks and nanocrystal size . The first two make sense considering the significant role annealing temperature and time play in nucleation and growth kinetics.…”

Colloidal III–V quantum dots: a synthetic perspective

“…Nonetheless, combining the In and P source into a single molecule often results in a stable solid under ambient conditions, a desirable trait for industrial applications. 17 In addition, the preformed bonds between the metal and chalcogenide ensure fewer defects are present within the crystal matrix. Finally, single-source precursors simplify the nucleation process by slowing down the conversion rate of the group V precursor, perhaps the greatest limitation of binary reactions.…”

“…In its current form, this protocol has been labelled as the heat-up method and alongside the hot-injection protocol (discussed in more detail in Section 3) represents over 80% of the reported literature. 38 Coordinating solvents, such as the ubiquitous trioctylphosphine (TOP) and trioctylphosphine oxide (TOPO), have been probed as both reaction mediums and stabilising/solvating agents. Notable examples include the synthesis of InP QDs in either a mixture of TOP/TOPO or pure TOP using chloroindium oxalate and P(SiMe 3 ) 3 .…”

“…These challenges have served as the impetus for extensive research into in-situ optimising of III-V QD formation, in particular, InP (Figure 1). 17 This has regrettably resulted in a vast array of isolated research results with little cohesion between publications. Navigating this synthetic landscape can be a daunting task for the uninitiated.…”

Electronic Lab Notebooks for Materials Synthesis

“…Thanks to ELNs, new researchers can easily access current synthesis protocols and also learn what has already been tried within the lab. Last but not least, ELNs represent a huge resource for data-driven discovery, which many materials chemists are pursuing in collaboration with computational scientists. , Using reliable data generated within the same laboratory and under standardized conditions is invaluable for comparison with larger data sets extracted via data mining from the literature, which is challenged by the lack of completeness and reporting standardization across different laboratories . Because ELNs provide the primary interface to research data, an opportunity exists to directly integrate computational approaches in order to automatically derive insights into the data and suggest the next step in the synthesis development.…”