Mixtures Recomposition by Neural Nets: A Multidisciplinary Overview.

Artificial Neural Networks (ANNs) are transforming how we understand chemical mixtures, providing an expressive view of the chemical space and multiscale processes. Their hybridization with physical knowledge can bridge the gap between predictivity and understanding of the underlying processes. This overview explores recent progress in ANNs, particularly their potential in the 'recomposition' of chemical mixtures. Graph-based representations reveal patterns among mixture components, and deep learning models excel in capturing complexity and symmetries when compared to traditional Quantitative Structure-Property Relationship models. Key components, such as Hamiltonian networks and convolution operations, play a central role in representing multiscale mixtures. The integration of ANNs with Chemical Reaction Networks and Physics-Informed Neural Networks for inverse chemical kinetic problems is also examined. The combination of sensors with ANNs shows promise in optical and biomimetic applications. A common ground is identified in the context of statistical physics, where ANN-based methods iteratively adapt their models by blending their initial states with training data. The concept of mixture recomposition unveils a reciprocal inspiration between ANNs and reactive mixtures, highlighting learning behaviors influenced by the training environment.

Ultrafast materials synthesis and manufacturing techniques for emerging energy and environmental applications.

Energy and environmental issues have attracted increasing attention globally, where sustainability and low-carbon emissions are seriously considered and widely accepted by government officials. In response to this situation, the development of renewable energy and environmental technologies is urgently needed to complement the usage of traditional fossil fuels. While a big part of advancement in these technologies relies on materials innovations, new materials discovery is limited by sluggish conventional materials synthesis methods, greatly hindering the advancement of related technologies. To address this issue, this review introduces and comprehensively summarizes emerging ultrafast materials synthesis methods that could synthesize materials in times as short as nanoseconds, significantly improving research efficiency. We discuss the unique advantages of these methods, followed by how they benefit individual applications for renewable energy and the environment. We also highlight the scalability of ultrafast manufacturing towards their potential industrial utilization. Finally, we provide our perspectives on challenges and opportunities for the future development of ultrafast synthesis and manufacturing technologies. We anticipate that fertile opportunities exist not only for energy and the environment but also for many other applications.

Bayesian chemical reaction neural network for autonomous kinetic uncertainty quantification.

Chemical reaction neural network (CRNN), a recently developed tool for autonomous discovery of reaction models, has been successfully demonstrated on a variety of chemical engineering and biochemical systems. It leverages the extraordinary data-fitting capacity of modern deep neural networks (DNNs) while preserving high interpretability and robustness by embedding widely applicable physical laws such as the law of mass action and the Arrhenius law. In this paper, we further developed Bayesian CRNN to not only reconstruct but also quantify the uncertainty of chemical kinetic models from data. Two methods, the Markov chain Monte Carlo algorithm and variational inference, were used to perform the Bayesian CRNN, with the latter mainly adopted for its speed. We demonstrated the capability of Bayesian CRNN in the kinetic uncertainty quantification of different types of chemical systems and discussed the importance of embedding physical laws in data-driven modeling. Finally, we discussed the adaptation of Bayesian CRNN for incomplete measurements and model mixing for global uncertainty quantification.

Autonomous Discovery of Unknown Reaction Pathways from Data by Chemical Reaction Neural Network.

Chemical reactions occur in energy, environmental, biological, and many other natural systems, and the inference of the reaction networks is essential to understand and design the chemical processes in engineering and life sciences. Yet, revealing the reaction pathways for complex systems and processes is still challenging because of the lack of knowledge of the involved species and reactions. Here, we present a neural network approach that autonomously discovers reaction pathways from the time-resolved species concentration data. The proposed chemical reaction neural network (CRNN), by design, satisfies the fundamental physics laws, including the law of mass action and the Arrhenius law. Consequently, the CRNN is physically interpretable such that the reaction pathways can be interpreted, and the kinetic parameters can be quantified simultaneously from the weights of the neural network. The inference of the chemical pathways is accomplished by training the CRNN with species concentration data via stochastic gradient descent. We demonstrate the successful implementations and the robustness of the approach in elucidating the chemical reaction pathways of several chemical engineering and biochemical systems. The autonomous inference by the CRNN approach precludes the need for expert knowledge in proposing candidate networks and addresses the curse of dimensionality in complex systems. The physical interpretability also makes the CRNN capable of not only fitting the data for a given system but also developing knowledge of unknown pathways that could be generalized to similar chemical systems.

Stiff-PINN: Physics-Informed Neural Network for Stiff Chemical Kinetics.

The recently developed physics-informed neural network (PINN) has achieved success in many science and engineering disciplines by encoding physics laws into the loss functions of the neural network such that the network not only conforms to the measurements and initial and boundary conditions but also satisfies the governing equations. This work first investigates the performance of the PINN in solving stiff chemical kinetic problems with governing equations of stiff ordinary differential equations (ODEs). The results elucidate the challenges of utilizing the PINN in stiff ODE systems. Consequently, we employ quasi-steady-state assumption (QSSA) to reduce the stiffness of the ODE systems, and the PINN then can be successfully applied to the converted non-/mild-stiff systems. Therefore, the results suggest that stiffness could be the major reason for the failure of the regular PINN in the studied stiff chemical kinetic systems. The developed stiff-PINN approach that utilizes QSSA to enable the PINN to solve stiff chemical kinetics shall open the possibility of applying the PINN to various reaction-diffusion systems involving stiff dynamics.

Stiff neural ordinary differential equations.

Neural Ordinary Differential Equations (ODEs) are a promising approach to learn dynamical models from time-series data in science and engineering applications. This work aims at learning neural ODEs for stiff systems, which are usually raised from chemical kinetic modeling in chemical and biological systems. We first show the challenges of learning neural ODEs in the classical stiff ODE systems of Robertson's problem and propose techniques to mitigate the challenges associated with scale separations in stiff systems. We then present successful demonstrations in stiff systems of Robertson's problem and an air pollution problem. The demonstrations show that the usage of deep networks with rectified activations, proper scaling of the network outputs as well as loss functions, and stabilized gradient calculations are the key techniques enabling the learning of stiff neural ODEs. The success of learning stiff neural ODEs opens up possibilities of using neural ODEs in applications with widely varying time-scales, such as chemical dynamics in energy conversion, environmental engineering, and life sciences.

Facile Thermal and Optical Ignition of Silicon Nanoparticles and Micron Particles.

Silicon (Si) particles are widely utilized as high-capacity electrodes for Li-ion batteries, elements for thermoelectric devices, agents for bioimaging and therapy, and many other applications. However, Si particles can ignite and burn in air at elevated temperatures or under intense illumination. This poses potential safety hazards when handling, storing, and utilizing these particles for those applications. In order to avoid the problem of accidental ignition, it is critical to quantify the ignition properties of Si particles such as their sizes and porosities. To do so, we first used differential scanning calorimetry to experimentally determine the reaction onset temperature of Si particles under slow heating rates (∼0.33 K/s). We found that the reaction onset temperature of Si particles increased with the particle diameter from 805 °C at 20-30 nm to 935 °C at 1-5 µm. Then, we used a xenon (Xe) flash lamp to ignite Si particles under fast heating rates (∼103 to 106 K/s) and measured the minimum ignition radiant fluence (i.e., the radiant energy per unit surface area of Si particle beds required for ignition). We found that the measured minimum ignition radiant fluence decreased with decreasing Si particle size and was most sensitive to the porosity of the Si particle bed. These trends for the Xe flash ignition experiments were also confirmed by our one-dimensional unsteady simulation to model the heat transfer process. The quantitative information on Si particle ignition included in this Letter will guide the safe handling, storage, and utilization of Si particles for diverse applications and prevent unwanted fire hazards.

Effects of composition, thermal, and theological properties of rice raw material on rice noodle quality.

The study aims to evaluate the relationships between characteristics of regional rice raw material and resulting quality of rice noodles. Four of most commonly used rice cultivars in Guangxi for noodles production were investigated. The results showed that compositions of rice flour primarily affected gelatinization and retrogradation, which then influenced the textural and sensory properties of rice noodles. Amylose content had strong positive correlation with peak viscosity (PV) and trough viscosity (TV) of rice flour (P < 0.01). PV and TV had strong negative correlations with adhesive strength (P < 0.01) and positive correlations with chewiness (P < 0.05), hardness, peak load and deformation at peak of rice noodles (P < 0.01). Protein content had positive correlation with the Setback of rice flour (P < 0.05), which is known to have influences on retrogradation. In addition, solubility had positive correlations with cooking loss (P < 0.01) and broken rate (P < 0.05) of rice noodles and strong negative correlation with its springiness (P < 0.01). Swelling power had negative correlation with broken rate (P < 0.05). As sensory score of rice noodles was negatively correlated with broken rate (P < 0.05) and cooking loss (P < 0.01) and positively correlated with springiness (P < 0.01), solubility and swelling power of rice flours were presumed to be useful for predicting consumer acceptability of rice noodles.

Acid-in-Clay Electrolyte for Wide-Temperature-Range and Long-Cycle Proton Batteries.

Proton conduction underlies many important electrochemical technologies. A family of new proton electrolytes is reported: acid-in-clay electrolyte (AiCE) prepared by integrating fast proton carriers in a natural phyllosilicate clay network, which can be made into thin-film (tens of micrometers) fluid-impervious membranes. The chosen example systems (sepiolite-phosphoric acid) rank top among the solid proton conductors in terms of proton conductivities (15 mS cm-1 at 25 °C, 0.023 mS cm-1 at -82 °C), electrochemical stability window (3.35 V), and reduced chemical reactivity. A proton battery is assembled using AiCE as the solid electrolyte membrane. Benefitting from the wider electrochemical stability window, reduced corrosivity, and excellent ionic selectivity of AiCE, the two main problems (gassing and cyclability) of proton batteries are successfully solved. This work draws attention to the element cross-over problem in proton batteries and the generic "acid-in-clay" solid electrolyte approach with superfast proton transport, outstanding selectivity, and improved stability for room- to cryogenic-temperature protonic applications.

Raw biomass electroreforming coupled to green hydrogen generation.

Despite the tremendous progress of coupling organic electrooxidation with hydrogen generation in a hybrid electrolysis, electroreforming of raw biomass coupled to green hydrogen generation has not been reported yet due to the rigid polymeric structures of raw biomass. Herein, we electrooxidize the most abundant natural amino biopolymer chitin to acetate with over 90% yield in hybrid electrolysis. The overall energy consumption of electrolysis can be reduced by 15% due to the thermodynamically and kinetically more favorable chitin oxidation over water oxidation. In obvious contrast to small organics as the anodic reactant, the abundance of chitin endows the new oxidation reaction excellent scalability. A solar-driven electroreforming of chitin and chitin-containing shrimp shell waste is coupled to safe green hydrogen production thanks to the liquid anodic product and suppression of oxygen evolution. Our work thus demonstrates a scalable and safe process for resource upcycling and green hydrogen production for a sustainable energy future.

Erratum: Author Correction: Machine learning model to project the impact of COVID-19 on US motor gasoline demand.

[This corrects the article DOI: 10.1038/s41560-020-0662-1.].

Synergistically Chemical and Thermal Coupling between Graphene Oxide and Graphene Fluoride for Enhancing Aluminum Combustion.

Metal combustion reaction is highly exothermic and is used in energetic applications, such as propulsion, pyrotechnics, powering micro- and nano-devices, and nanomaterials synthesis. Aluminum (Al) is attracting great interest in those applications because of its high energy density, earth abundance, and low toxicity. Nevertheless, Al combustion is hard to initiate and progresses slowly and incompletely. On the other hand, ultrathin carbon nanomaterials, such as graphene, graphene oxide (GO), and graphene fluoride (GF), can also undergo exothermic reactions. Herein, we demonstrate that the mixture of GO and GF significantly improves the performance of Al combustion as interactions between GO and GF provide heat and radicals to accelerate Al oxidation. Our experiments and reactive molecular dynamics simulation reveal that GO and GF have strong chemical and thermal couplings through radical reactions and heat released from their oxidation reactions. GO facilitates the dissociation of GF, and GF accelerates the disproportionation and oxidation of GO. When the mixture of GO and GF is added to micron-sized Al particles, their synergistic couplings generate reactive oxidative species, such as CFx and CFxOy, and heat, which greatly accelerates Al combustion. This work demonstrates a new area of using synergistic couplings between ultrathin carbon nanomaterials to accelerate metal combustion and potentially oxidation reactions of other materials.

Energetic Performance of Optically Activated Aluminum/Graphene Oxide Composites.

Optical ignition of solid energetic materials, which can rapidly release heat, gas, and thrust, is still challenging due to the limited light absorption and high ignition energy of typical energetic materials ( e.g., aluminum, Al). Here, we demonstrated that the optical ignition and combustion properties of micron-sized Al particles were greatly enhanced by adding only 20 wt % of graphene oxide (GO). These enhancements are attributed to the optically activated disproportionation and oxidation reactions of GO, which release heat to initiate the oxidization of Al by air and generate gaseous products to reduce the agglomeration of the composites and promote the pressure rise during combustion. More importantly, compared to conventional additives such as metal oxides nanoparticles ( e.g., WO3 and Bi2O3), GO has much lower density and therefore could improve energetic properties without sacrificing Al content. The results from Xe flash ignition and laser-based excitation experiments demonstrate that GO is an efficient additive to improve the energetic performance of micron-sized Al particles, enabling micron-sized Al to be ignited by optical activation and promoting the combustion of Al in air.

Electroless Deposition and Ignition Properties of Si/Fe2O3 Core/Shell Nanothermites.

Thermite, a composite of metal and metal oxide, finds wide applications in power and thermal generation systems that require high-energy density. Most of the researches on thermites have focused on using aluminum (Al) particles as the fuel. However, Al particles are sensitive to electrostatic discharge, friction, and mechanical impact, imposing a challenge for the safe handling and storage of Al-based thermites. Silicon (Si) is another attractive fuel for thermites because of its high-energy content, thin native oxide layer, and facile surface functionality. Several studies showed that the combustion properties of Si-based thermites are comparable to those of Al-based thermites. However, little is known about the ignition properties of Si-based thermites. In this work, we determined the reaction onset temperatures of mechanically mixed (MM) Si/Fe2O3 nanothermites and Si/Fe2O3 core/shell (CS) nanothermites using differential scanning calorimetry. The Si/Fe2O3 CS nanothermites were prepared by an electroless deposition method. We found that the Si/Fe2O3 CS nanoparticles (NPs) had a lower reaction onset temperature (∼550 °C) than the MM Si/Fe2O3 nanothermites (>650 °C). The onset temperature of the Si/Fe2O3 CS nanothermites is also insensitive to the size of the Si core NP. These results indicate that the interfacial contact quality between Si and Fe2O3 is the dominant factor for determining the ignition properties of thermites. Finally, the reaction onset temperature of the Si/Fe2O3 CS NPs is comparable to that of the commonly used Al-based nanothermites, suggesting that Si is an attractive fuel for thermites.