Navigating the design space of inorganic materials synthesis using statistical methods and machine learning

Accelerating the design of lattice structures using machine learning

Lattices remain an attractive class of structures due to their design versatility; however, rapidly designing lattice structures with tailored or optimal mechanical properties remains a significant challenge. With each added design variable, the design space quickly becomes intractable. To address this challenge, research efforts have sought to combine computational approaches with machine learning (ML)-based approaches to reduce the computational cost of the design process and accelerate mechanical design. While these efforts have made substantial progress, significant challenges remain in (1) building and interpreting the ML-based surrogate models and (2) iteratively and efficiently curating training datasets for optimization tasks. Here, we address the first challenge by combining ML-based surrogate modeling and Shapley additive explanation (SHAP) analysis to interpret the impact of each design variable. We find that our ML-based surrogate models achieve excellent prediction capabilities (R2 > 0.95) and SHAP values aid in uncovering design variables influencing performance. We address the second challenge by utilizing active learning-based methods, such as Bayesian optimization, to explore the design space and report a 5 × reduction in simulations relative to grid-based search. Collectively, these results underscore the value of building intelligent design systems that leverage ML-based methods for uncovering key design variables and accelerating design.

A Data-Driven Approach to Molten Salt Synthesis of N-Rich Carbon Adsorbents for Selective CO2 Capture

Applying a design of experiments methodology to the molten salt synthesis of nanoporous carbons enables inverse design and optimization of nitrogen (N)-rich carbon adsorbents with excellent CO2 /N2 selectivity and appreciable CO2 capacity for carbon capture via swing adsorption from dilute gas mixtures such as natural gas combined cycle flue gas. This data-driven study reveals fundamental structure-function relationships between the synthesis conditions, physicochemical properties, and achievable selective adsorption performance of N-rich nanoporous carbons derived from molten salt synthesis for CO2 capture. Taking advantage of size-sieving separation of CO2 (3.30 Å) from N2 (3.64 Å) within the turbostratic nanostructure of these N-rich carbons, while limiting deleterious N2 adsorption in a weaker adsorption site that harms selectivity, enables a large CO2 capacity (0.73 mmol g-1 at 30.4 Torr and 30 °C) with noteworthy concurrent CO2 /N2 selectivity as predicted by the ideal adsorbed solution theory (SIAST = 246) with an adsorbed phase purity of 91% from a simulated gas stream containing only 4% CO2 . Optimized N-rich porous carbons, with good physicochemical stability, low cost, and moderate regeneration energy, can achieve performance for selective CO2 adsorption that competes with other classes of advanced porous materials such as chemisorbing zeolites and functionalized metal-organic frameworks.

Using Data-Driven Learning to Predict and Control the Outcomes of Inorganic Materials Synthesis

The design of inorganic materials for various applications critically depends on our ability to manipulate their synthesis in a rational, robust, and controllable fashion. Different from the conventional trial-and-error approach, data-driven techniques such as the design of experiments (DoE) and machine learning are an effective and more efficient way to predictably control materials synthesis. Here, we present a Viewpoint on recent progress in leveraging such techniques for predicting and controlling the outcomes of inorganic materials synthesis. We first compare how the design choice (statistical DoE vs machine learning) affects the type of control it can offer over the resulting product attributes, information elucidated, and experimental cost. These attributes are supported by discussing select case studies from the recent literature that highlight the power of these techniques for materials synthesis. The influence of experimental bias is next discussed, followed finally by our perspectives on the major challenges in the widespread implementation of predictable and controllable materials synthesis using data-driven techniques.

Predictive Synthesis of Copper Selenides Using a Multidimensional Phase Map Constructed with a Data-Driven Classifier

Copper selenides are an important family of materials with applications in catalysis, plasmonics, photovoltaics, and thermoelectrics. Despite being a binary material system, the Cu-Se phase diagram is complex and contains multiple crystal structures in addition to several metastable structures that are not found on the thermodynamic phase diagram. Consequently, the ability to synthetically navigate this complex phase space poses a significant challenge. We demonstrate that data-driven learning can successfully map this phase space in a minimal number of experiments. We combine soft chemistry (chimie douce) synthetic methods with multivariate analyses via classification techniques to enable predictive phase determination. A surrogate model was constructed with experimental data derived from a design matrix of four experimental variables: C-Se bond strength of the selenium precursor, time, temperature, and solvent composition. The reactions in the surrogate model resulted in 11 distinct phase combinations of copper selenide. These data were used to train a classification model that predicts the phase with 95.7% accuracy. The resulting decision tree enabled conclusions to be drawn about how the experimental variables affect the phase and provided prescriptive synthetic conditions for specific phase isolation. This guided the accelerated phase targeting in a minimum number of experiments of klockmannite CuSe, which could not be isolated in any of the reactions used to construct the surrogate model. The reaction conditions that the model predicted to synthesize klockmannite CuSe were experimentally validated, highlighting the utility of this approach.

Machine Learning Predicting Optimal Preparation of Silica-Coated Gold Nanorods for Photothermal Tumor Ablation

Gold nanorods (GNRs) coated with silica shells are excellent photothermal agents with high surface functionality and biocompatibility. Understanding the correlation of the coating process with both structure and property of silica-coated GNRs is crucial to their optimizing preparation and performance, as well as tailoring potential applications. Herein, we report a machine learning (ML) prediction of coating silica on GNR with various preparation parameters. A total of 306 sets of silica-coated GNRs altogether were prepared via a sol-gel method, and their structures were characterized to extract a dataset available for eight ML algorithms. Among these algorithms, the eXtreme gradient boosting (XGboost) classification model affords the highest prediction accuracy of over 91%. The derived feature importance scores and relevant decision trees are employed to address the optimal process to prepare well-structured silica-coated GNRs. The high-throughput predictions have been adopted to identify optimal process parameters for the successful preparation of dumbbell-structured silica-coated GNRs, which possess a superior performance to a conventional cylindrical core-shell counterpart. The dumbbell silica-coated GNRs demonstrate an efficient enhanced photothermal performance in vivo and in vitro, validated by both experiments and time domain finite difference calculations. This study epitomizes the potential of ML algorithms combined with experiments in predicting, optimizing, and accelerating the preparation of core-shell inorganic materials and can be extended to other nanomaterial research.

Creating ground truth for nanocrystal morphology: a fully automated pipeline for unbiased transmission electron microscopy analysis

Control over colloidal nanocrystal morphology (size, size distribution, and shape) is important for tailoring the functionality of individual nanocrystals and their ensemble behavior. Despite this, traditional methods to quantify nanocrystal morphology are laborious. New developments in automated morphology classification will accelerate these analyses but the assessment of machine learning models is limited by human accuracy for ground truth, causing even unsupervised machine learning models to have inherent bias. Herein, we introduce synthetic image rendering to solve the ground truth problem of nanocrystal morphology classification. By simulating 2D images of nanocrystal shapes via a function of high-dimensional parameter space, we trained a convolutional neural network to link unique morphologies to their simulated parameters, defining nanocrystal morphology quantitatively rather than qualitatively. An automated pipeline then processes, quantitatively defines, and classifies nanocrystal morphology from experimental transmission electron microscopy (TEM) images. Using improved computer vision techniques, 42 650 nanocrystals were identified, assessed, and labeled with quantitative parameters, offering a 600-fold improvement in efficiency over best-practice manual measurements. A classification algorithm was trained with a prediction accuracy of 99.5%, which can successfully analyze a range of concave, convex, and irregular nanocrystal shapes. The resulting pipeline was applied to differentiating two syntheses of nominally cuboidal CsPbBr₃ nanocrystals and uniquely classifying binary nickel sulfide nanocrystal phase based on morphology. This pipeline provides a simple, efficient, and unbiased method to quantify nanocrystal morphology and represents a practical route to construct large datasets with an absolute ground truth for training unbiased morphology-based machine learning algorithms.

Inverse Design of Materials by Machine Learning

It is safe to say that every invention that has changed the world has depended on materials. At present, the demand for the development of materials and the invention or design of new materials is becoming more and more urgent since peoples' current production and lifestyle needs must be changed to help mitigate the climate. Structure-property relationships are a vital paradigm in materials science. However, these relationships are often nonlinear, and the pattern is likely to change with length scales and time scales, posing a huge challenge. With the development of physics, statistics, computer science, etc., machine learning offers the opportunity to systematically find new materials. Especially by inverse design based on machine learning, one can make use of the existing knowledge without attempting mathematical inversion of the relevant integrated differential equation of the electronic structure but by using backpropagation to overcome local minimax traps and perform a fast calculation of the gradient information for a target function concerning the design variable to find the optimizations. The methodologies have been applied to various materials including polymers, photonics, inorganic materials, porous materials, 2-D materials, etc. Different types of design problems require different approaches, for which many algorithms and optimization approaches have been demonstrated in different scenarios. In this mini-review, we will not specifically sum up machine learning methodologies, but will provide a more material perspective and summarize some cut-edging studies.

A Materials Science Perspective of Midstream Challenges in the Utilization of Heavy Crude Oil

An increasing global population and a sharply upward trajectory of per capita energy consumption continue to drive the demand for fossil fuels, which remain integral to energy grids and the global transportation infrastructure. The oil and gas industry is increasingly reliant on unconventional deposits such as heavy crude oil and bitumen for reasons of accessibility, scale, and geopolitics. Unconventional deposits such as the Canadian Oil Sands in Northern Alberta contain more than one-third of the world's viscous oil reserves and are vital linchpins to meet the energy needs of rapidly industrializing populations. Heavy oil is typically recovered from subsurface deposits using thermal recovery approaches such as steam-assisted gravity drainage (SAGD). In this perspective article, we discuss several aspects of materials science challenges in the utilization of heavy crude oil with an emphasis on the needs of the Canadian Oil Sands. In particular, we discuss surface modification and materials' design approaches essential to operations under extreme environments of high temperatures and pressures and the presence of corrosive species. The demanding conditions for materials and surfaces are directly traceable to the high viscosity, low surface tension, and substantial sulfur content of heavy crude oil, which necessitates extensive energy-intensive thermal processes, warrants dilution/emulsification to ease the flow of rheologically challenging fluids, and engenders the need to protect corrodible components. Geopolitical reasons have further led to a considerable geographic separation between extraction sites and advanced refineries capable of processing heavy oils to a diverse slate of products, thus necessitating a massive midstream infrastructure for transportation of these rheologically challenging fluids. Innovations in fluid handling, bitumen processing, and midstream transportation are critical to the economic viability of heavy oil. Here, we discuss foundational principles, recent technological advancements, and unmet needs emphasizing candidate solutions for thermal insulation, membrane-assisted separations, corrosion protection, and midstream bitumen transportation. This perspective seeks to highlight illustrative materials' technology developments spanning the range from nanocomposite coatings and cement sheaths for thermal insulation to the utilization of orthogonal wettability to engender separation of water-oil emulsions stabilized by endogenous surfactants extracted during SAGD, size-exclusion membranes for fractionation of bitumen, omniphobic coatings for drag reduction in pipelines and to ease oil handling in containers, solid prills obtained from partial bitumen solidification to enable solid-state transport with reduced risk of damage from spills, and nanocomposite coatings incorporating multiple modes of corrosion inhibition. Future outlooks for onsite partial upgradation are also described, which could potentially bypass the use of refineries for some fractions, enable access to a broader cross-section of refineries, and enable a new distributed chemical manufacturing paradigm.

Prospects and challenges for autonomous catalyst discovery viewed from an experimental perspective

Autonomous catalysis research requires elaborate integration of operando experiments into automated workflows. Suitable experimental data for analysis by artificial intelligence can be measured more readily according to standard operating procedures.

Machine-Learning-Guided Discovery of 19F MRI Agents Enabled by Automated Copolymer Synthesis

Modern polymer science suffers from the curse of multidimensionality. The large chemical space imposed by including combinations of monomers into a statistical copolymer overwhelms polymer synthesis and characterization technology and limits the ability to systematically study structure-property relationships. To tackle this challenge in the context of 19F magnetic resonance imaging (MRI) agents, we pursued a computer-guided materials discovery approach that combines synergistic innovations in automated flow synthesis and machine learning (ML) method development. A software-controlled, continuous polymer synthesis platform was developed to enable iterative experimental-computational cycles that resulted in the synthesis of 397 unique copolymer compositions within a six-variable compositional space. The nonintuitive design criteria identified by ML, which were accomplished by exploring <0.9% of the overall compositional space, lead to the identification of >10 copolymer compositions that outperformed state-of-the-art materials.

Toward autonomous design and synthesis of novel inorganic materials

Autonomous experimentation driven by artificial intelligence (AI) provides an exciting opportunity to revolutionize inorganic materials discovery and development. Herein, we review recent progress in the design of self-driving laboratories, including robotics to automate materials synthesis and characterization, in conjunction with AI to interpret experimental outcomes and propose new experimental procedures. We focus on efforts to automate inorganic synthesis through solution-based routes, solid-state reactions, and thin film deposition. In each case, connections are made to relevant work in organic chemistry, where automation is more common. Characterization techniques are primarily discussed in the context of phase identification, as this task is critical to understand what products have formed during synthesis. The application of deep learning to analyze multivariate characterization data and perform phase identification is examined. To achieve "closed-loop" materials synthesis and design, we further provide a detailed overview of optimization algorithms that use active learning to rationally guide experimental iterations. Finally, we highlight several key opportunities and challenges for the future development of self-driving inorganic materials synthesis platforms.

Statistical Multiobjective Optimization of Thiospinel CoNi2S4 Nanocrystal Synthesis via Design of Experiments

Thiospinels, such as CoNi₂S₄, are showing promise for numerous applications, including as catalysts for the hydrogen evolution reaction, hydrodesulfurization, and oxygen evolution and reduction reactions; however, CoNi₂S₄ has not been synthesized as small, colloidal nanocrystals with high surface-area-to-volume ratios. Traditional optimization methods to control nanocrystal attributes such as size typically rely upon one variable at a time (OVAT) methods that are not only time and labor intensive but also lack the ability to identify higher-order interactions between experimental variables that affect target outcomes. Herein, we demonstrate that a statistical design of experiments (DoE) approach can optimize the synthesis of CoNi₂S₄ nanocrystals, allowing for control over the responses of nanocrystal size, size distribution, and isolated yield. After implementing a 2^5-2 fractional factorial design, the statistical screening of five different experimental variables identified temperature, Co:Ni precursor ratio, Co:thiol ratio, and their higher-order interactions as the most critical factors in influencing the aforementioned responses. Second-order design with a Doehlert matrix yielded polynomial functions used to predict the reaction parameters needed to individually optimize all three responses. A multiobjective optimization, allowing for the simultaneous optimization of size, size distribution, and isolated yield, predicted the synthetic conditions needed to achieve a minimum nanocrystal size of 6.1 nm, a minimum polydispersity (σ/d̅) of 10%, and a maximum isolated yield of 99%, with a desirability of 96%. The resulting model was experimentally verified by performing reactions under the specified conditions. Our work illustrates the advantage of multivariate experimental design as a powerful tool for accelerating control and optimization in nanocrystal syntheses.

A power-triggered preparation strategy of nano-structured inorganics: sonosynthesis

Ultrasound irradiation covers many chemical reactions crucially aiming to design and synthesize various structured materials as an enduring trend in frontier research studies. Here, we focus on the latest progress of ultrasound-assisted synthesis and present the basic principles or mechanisms of sonosynthesis (or sonochemical synthesis) from ultrasound irradiation in a brand new way, including primary sonosynthesis, secondary sonosynthesis, and synergetic sonosynthesis. This current review describes in detail the various sonochemical synthesis strategies for nano-structured inorganic materials and the unique aspects of products including the size, morphology, structure, and properties. In addition, the review points out the probable challenges and technological potential for future advancement. We hope that such a review can provide a comprehensive understanding of sonosynthesis and emphasize the great significance of structured materials synthesis as a power-induced strategy broadening the updated applications of ultrasound.

Using simulation to accelerate autonomous experimentation: A case study using mechanics

Autonomous experimentation (AE) accelerates research by combining automation and machine learning to perform experiments intelligently and rapidly in a sequential fashion. While AE systems are most needed to study properties that cannot be predicted analytically or computationally, even imperfect predictions can in principle be useful. Here, we investigate whether imperfect data from simulation can accelerate AE using a case study on the mechanics of additively manufactured structures. Initially, we study resilience, a property that is well-predicted by finite element analysis (FEA), and find that FEA can be used to build a Bayesian prior and experimental data can be integrated using discrepancy modeling to reduce the number of needed experiments ten-fold. Next, we study toughness, a property not well-predicted by FEA and find that FEA can still improve learning by transforming experimental data and guiding experiment selection. These results highlight multiple ways that simulation can improve AE through transfer learning.