Robot-Accelerated Perovskite Investigation and Discovery

Li, Zhi; Najeeb, Mansoor Ani; Alves, Liana; Sherman, Alyssa Z.; Shekar, Venkateswaran; Parrilla, Peter Cruz; Pendleton, Ian M.; Wang, Wesley; Nega, Philip W.; Zeller, Mat­thias; Schrier, Joshua; Norquist, Alexander J.; Chan, Emory M.

doi:10.1021/acs.chemmater.0c01153

Cited by 136 publications

(147 citation statements)

References 88 publications

Supporting

Mentioning

143

Contrasting

Unclassified

Order By: Relevance

“…[54] Realizing the opportunities along the way toward materials by design, several studies were recently published in the field of halide perovskites combining high throughput synthesis and machinelearning (ML)-assisted data analysis to identify novel functional materials. [55][56][57][58] Moving to the next level in accelerating synthesis and fully exploiting the advancements of automated and autonomous synthesis it can be advantageous to implement in situ diagnostics in ML-assisted high throughput experiments that will allow for real-time monitoring and data-driven stirring of the experiment. The advancement of characterization techniques and computational frameworks together with implementation of automated workflows can change the way we are doing synthesis at the moment.…”

Section: Perspective and Outlookmentioning

confidence: 99%

The Value of Watching How Materials Grow: A Multimodal Case Study on Halide Perovskites

Sutter‐Fella

2021

Advanced Energy Materials

View full text Add to dashboard Cite

natorial parameter spaces and real time control can change the way we are used to do material synthesis. Realizing this paradigm would not only speed up new materials discovery, shorten the development and optimization time of deploying new materials in commercial products by 10 times, [5] but also make better use of research resources. Toward this goal, further developments spanning new computational tools, materials databases, and synthesis and characterization techniques, need to converge. One approach to rationally synthesize materials, is to leverage in depth understanding of how complex materials assemble, how their properties evolve over time, and how this affects the overall physiochemical macroscopic properties. In fact, the mission innovation report recently identified the lack of structure-property relationships as one significant gap toward accelerated materials discovery using data-driven approaches. [6] Much of today's experiments are conducted by measuring single properties one by one ex situ, mostly near equilibrium. Ex situ measurements (referred to as "black box" synthesis) describe materials or device characterization post fabrication and do for example not allow for insights into crystallization pathways, chemical transformations, as well as the evolution or reciprocal correlation of functional properties that are happening dynamically over time during synthesis and device operation (Figure 1). Some experiments are conducted as break-off experiments where the synthesis is interrupted at different points during the reaction to conduct ex situ measurements and to capture snapshots of the reaction. More recently, in situ measurements are performed in real time as the material is synthesized, i.e., measurements are conducted during synthesis and one can monitor the possibly intricate interplay between thermodynamic and kinetically driven formation pathways. As an example, in situ diffraction measurements during synthesis that span several processing steps can reveal metastable crystalline phases that form as intermediates and then disappear at the end of the reaction. [7][8][9] With the beginning of modern science there has been a relationship between scientific discovery and innovative tools. Fast data collection was enabled by detector developments with integration times on the order of ≈ms and increasing X-ray fluxes available at synchrotron facilities. [10] These developments allow now to follow growth kinetics, structural transformations, and Material synthesis is one of the most important aspects in humankinds' endeavor to discover and create new materials for energy applications. One strategy to tailor materials with desired functions in a rational way is by knowing how functions relate to structure, synthetic variables, arrangement of atoms and molecules, and how functions evolve during synthesis. In order to accelerate materials synthesis, discovery, and optimization by 10 times it is the right time now to integrate computational tools, synthesis, and characterization. One particular barr...

show abstract

Section: Perspective and Outlookmentioning

confidence: 99%

The Value of Watching How Materials Grow: A Multimodal Case Study on Halide Perovskites

Sutter‐Fella

2021

Advanced Energy Materials

View full text Add to dashboard Cite

show abstract

“…However, these past studies only advised researchers on the next experiment to perform, leaving experiment planning, execution, and analysis to the researcher. Recent advances in robotics have shifted the burden of materials synthesis from human experts to automated systems, accelerating materials discovery 15 , 16 . Concurrently, active learning has been demonstrated to accelerate property optimization by guiding simulations of known phases 17 .…”

Section: Introductionmentioning

confidence: 99%

On-the-fly closed-loop materials discovery via Bayesian active learning

et al. 2020

View full text Add to dashboard Cite

Active learning—the field of machine learning (ML) dedicated to optimal experiment design—has played a part in science as far back as the 18th century when Laplace used it to guide his discovery of celestial mechanics. In this work, we focus a closed-loop, active learning-driven autonomous system on another major challenge, the discovery of advanced materials against the exceedingly complex synthesis-processes-structure-property landscape. We demonstrate an autonomous materials discovery methodology for functional inorganic compounds which allow scientists to fail smarter, learn faster, and spend less resources in their studies, while simultaneously improving trust in scientific results and machine learning tools. This robot science enables science-over-the-network, reducing the economic impact of scientists being physically separated from their labs. The real-time closed-loop, autonomous system for materials exploration and optimization (CAMEO) is implemented at the synchrotron beamline to accelerate the interconnected tasks of phase mapping and property optimization, with each cycle taking seconds to minutes. We also demonstrate an embodiment of human-machine interaction, where human-in-the-loop is called to play a contributing role within each cycle. This work has resulted in the discovery of a novel epitaxial nanocomposite phase-change memory material.

show abstract

“…Ref. 27 is another recent example of successful optimization for the synthesis of metal halide perovskite materials in which a large initial experimental data set of more than 8000 conditions obtained by RS is used to train support vector machines and neural networks. How much experimental data is necessary to accurately train a DNN or any other regressor remains an open question in ML.…”

Section: Discussion On ML Initializationmentioning

confidence: 99%

Two-step machine learning enables optimized nanoparticle synthesis

et al. 2021

View full text Add to dashboard Cite

In materials science, the discovery of recipes that yield nanomaterials with defined optical properties is costly and time-consuming. In this study, we present a two-step framework for a machine learning-driven high-throughput microfluidic platform to rapidly produce silver nanoparticles with the desired absorbance spectrum. Combining a Gaussian process-based Bayesian optimization (BO) with a deep neural network (DNN), the algorithmic framework is able to converge towards the target spectrum after sampling 120 conditions. Once the dataset is large enough to train the DNN with sufficient accuracy in the region of the target spectrum, the DNN is used to predict the colour palette accessible with the reaction synthesis. While remaining interpretable by humans, the proposed framework efficiently optimizes the nanomaterial synthesis and can extract fundamental knowledge of the relationship between chemical composition and optical properties, such as the role of each reactant on the shape and amplitude of the absorbance spectrum.

show abstract

Robot-Accelerated Perovskite Investigation and Discovery

Cited by 136 publications

References 88 publications

The Value of Watching How Materials Grow: A Multimodal Case Study on Halide Perovskites

The Value of Watching How Materials Grow: A Multimodal Case Study on Halide Perovskites

On-the-fly closed-loop materials discovery via Bayesian active learning

Two-step machine learning enables optimized nanoparticle synthesis

Contact Info

Product

Resources

About