We report a method to convert discrete representations of molecules to and from a multidimensional continuous representation. This model allows us to generate new molecules for efficient exploration and optimization through open-ended spaces of chemical compounds. A deep neural network was trained on hundreds of thousands of existing chemical structures to construct three coupled functions: an encoder, a decoder, and a predictor. The encoder converts the discrete representation of a molecule into a real-valued continuous vector, and the decoder converts these continuous vectors back to discrete molecular representations. The predictor estimates chemical properties from the latent continuous vector representation of the molecule. Continuous representations of molecules allow us to automatically generate novel chemical structures by performing simple operations in the latent space, such as decoding random vectors, perturbing known chemical structures, or interpolating between molecules. Continuous representations also allow the use of powerful gradient-based optimization to efficiently guide the search for optimized functional compounds. We demonstrate our method in the domain of drug-like molecules and also in a set of molecules with fewer that nine heavy atoms.
Virtual screening is becoming a ground-breaking tool for molecular discovery due to the exponential growth of available computer time and constant improvement of simulation and machine learning techniques. We report an integrated organic functional material design process that incorporates theoretical insight, quantum chemistry, cheminformatics, machine learning, industrial expertise, organic synthesis, molecular characterization, device fabrication and optoelectronic testing. After exploring a search space of 1.6 million molecules and screening over 400,000 of them using time-dependent density functional theory, we identified thousands of promising novel organic light-emitting diode molecules across the visible spectrum. Our team collaboratively selected the best candidates from this set. The experimentally determined external quantum efficiencies for these synthesized candidates were as large as 22%.
Rational Design of Electrolyte Material 84Designing an appropriate organic molecule as electrolyte material starts from 85 identifying redox-active cores followed by functionalization of the core structure to 86 achieve a practical reduction potential and solubility. We observed that riboflavin 5' (Fig. 1c). The larger separation between its oxidation 102 and reduction peaks than those of FMN and lumichrome is likely due to slower kinetics. 103From our rotating disk electrode (RDE) measurement, the reduction rate constant was 104 measured to be 1.2±0.2 × 10 −5 cm s −1 ( Supplementary Fig. 3). Nevertheless, this value is 105 still an order of magnitude higher than that of the slower side of all-vanadium RFBs. 6 106Besides the large shift in reduction potential moving from isoalloxazine to 107 alloxazine, we also observed a significant increase in chemical stability in alkaline 10% to 90% SOC (Fig 2a). Polarization studies conducted at room temperature showed a 128 peak power density of 0.35 W cm −2 at a current density of 0.58 A cm −2 . The linearity of 129 the polarization curves allows us to derive a polarization area-specific resistance (ASR), 130 which is 1.03 Ω cm 2 at 50% SOC. About 70% of this cell ASR is contributed by the 131 membrane ( Supplementary Fig. 6), similar to our previous observation. 10 Note that ACA ACA was evaluated based on an extended charge-discharge study over 400 cycles (Fig. 136 2c). The current efficiency exceeded 99.7% at 0.1 A cm -2 , which is indicative of Fig. 7). From this result, the measured capacity retention from 400 cycles was 95%, i.e. 144the loss rate was 0.013% per cycle. We believe the discrepancy between this 145 measurement and the capacity retention observed during constant-current cycling was 146 due to an increase of system resistance (which we infer from decreasing energy 147 efficiency with cycle number); this effect moved the charging and discharging curves 148 closer to the cutoff voltages, resulting in less complete charging and discharging with 149 increasing cycle count ( Supplementary Fig. 8). We expect further cell development, 150including variations in pH, membrane and sealing method, to lead to further improvement 151 of capacity retention. By increasing the concentration of ACA to 1 M, we increased the 152 electrolyte charge density by almost two-fold ( Supplementary Fig. 9a). Together with 153 adjusted cell compression and higher ACA concentration, we were able to improve 154 round-trip energy efficiency to 74%, while retaining the same level of current efficiency 155 (99.7%) and capacity retention per cycle (99.95%) (Supplementary Fig. 9b). 156 Theoretical Modeling and Screening 157One useful feature of organic electrolyte materials is the ability to optimize their 158 properties through chemical modification, a process that can be accelerated by virtual 159 testing with computational methods. 8,20 We assayed the chemical landscape around the (Fig 3c and d). (Fig. 3b) 210Chemical synthesis and characterization 3,4-diaminophenol was purcha...
A philosophy for defining what constitutes a virtual high-throughput screen is discussed, and the choices which influence decisions at each stage of the 'computational funnel' are investigated, including an indepth discussion of the generation of molecular libraries. Additionally advice on the storing, analysis and visualization of data is given, based upon extensive experience in the group.
Molecular dynamics simulations provide theoretical insight into the microscopic behavior of materials in condensed phase and, as a predictive tool, enable computational design of new compounds. However, because of the large temporal and spatial scales involved in thermodynamic and kinetic phenomena in materials, atomistic simulations are often computationally unfeasible. Coarsegraining methods allow simulating larger systems, by reducing the dimensionality of the simulation, and propagating longer timesteps, by averaging out fast motions. Coarse-graining involves two coupled learning problems; defining the mapping from an all-atom to a reduced representation, and the parametrization of a Hamiltonian over coarse-grained coordinates. Multiple statistical mechanics approaches have addressed the latter, but the former is generally a hand-tuned process based on chemical intuition. Here we present Autograin, an optimization framework based on autoencoders to learn both tasks simultaneously. Autograin is trained to learn the optimal mapping between all-atom and reduced representation, using the reconstruction loss to facilitate the learning of coarse-grained variables. In addition, a force-matching method is applied to variationally determine the coarse-grained potential energy function. This procedure is tested on a number of model systems including single-molecule and bulk-phase periodic simulations.
The authors investigate four anthraquinone derivatives as negative electrolyte candidates for an aqueous quinone-bromide redox flow battery: anthraquinone-2-sulfonic acid (AQS), 1,8-dihydroxyanthraquinone-2,7-disulfonic acid (DHAQDS), alizarin red S (ARS), and 1,4-dihydroxyanthraquinone-2,3-dimethylsulfonic acid (DHAQDMS). The standard reduction potentials are all lower than that of anthraquinone-2,7-disulfonic acid (AQDS), the molecule used in previous quinone-bromide batteries. DHAQDS and ARS undergo irreversible reactions on contact with bromine, which precludes their use against bromine but not necessarily against other electrolytes. DHAQDMS is apparently unreactive with bromine but cannot be reversibly reduced, whereas AQS is stable against bromine and stable upon reduction. The authors demonstrate an AQS-bromide flow cell with higher open circuit potential and peak galvanic power density than the equivalent AQDS-bromide cell. This study demonstrates the use of chemical synthesis to tailor organic molecules for improving flow battery performance.
Zeolites are versatile catalysts and molecular sieves with large topological diversity, but managing phase competition in zeolite synthesis is an empirical, labor-intensive task. Here, we controlled phase selectivity in templated zeolite synthesis from first principles by combining high-throughput atomistic simulations, literature mining, human-computer interaction, synthesis, and characterization. Proposed binding metrics distilled from over 586,000 zeolite-molecule simulations reproduced the extracted literature and rationalize framework competition in the design of organic structure-directing agents. Energetic, geometric, and electrostatic descriptors of template molecules were found to regulate synthetic accessibility windows and aluminum distributions in pure-phase zeolites. Furthermore, these parameters allowed realizing an intergrowth zeolite through a single bi-selective template. The computation-first approach enabled controlling both zeolite synthesis and structure composition using a priori theoretical descriptors.
The stability limits of quinones, molecules that show promise as redox-active electrolytes in aqueous flow batteries, are explored for a range of backbone and substituent combinations with high-throughput virtual screening.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.