Which molecular properties determine the impact sensitivity of an explosive? A machine learning quantitative investigation of nitroaromatic explosives DOI Creative Commons
Itamar Borges, Júlio César Duarte,

Romulo Dias da Rocha

et al.

Published: Nov. 15, 2022

We decomposed density functional theory charge densities of 53 nitroaromatic molecules into atom-centered electric multipoles using the distributed multipole analysis that provides a detailed picture molecular electronic structure. Three multipoles, ∑▒〖Q_0 (NO_2)〗 (the nitro groups), ∑▒〖Q_1 total dipole, i.e., polarization, ∑▒〖Q_2 (C) 〗 electron delocalization C ring atoms), and number explosophore groups (#NO_2) were selected as features for comprehensive machine learning (ML) investigation. The target property was impact sensitivity h_50 (cm) values quantified by drop-weight measurements. After preliminary screening 42 ML algorithms, four based on lowest root mean square errors: Extra Trees, Random Forests, Gradient Boosting, AdaBoost. predicted having very different sensitivities algorithms are in range 19% - 28% compared to experimental data. most important properties predicting atoms polarization with averaged weights 39% 35%, followed (16%) (10%) groups. A significant result is how contribution these depends its sensitivities: sensitive explosives (h_50 up ~ 50 cm), contribute reducing h_50, intermediate ones (~ cm ≲ 100 cm) #NO_2 increasing it other two it. For highly insensitive (h_50≳ 200 all essentially These results furnish consistent basis known also can be used developing safer new ones.

Language: Английский

Which molecular properties determine the impact sensitivity of an explosive? A machine learning quantitative investigation of nitroaromatic explosives DOI
Júlio César Duarte,

Romulo Dias da Rocha,

Itamar Borges

et al.

Physical Chemistry Chemical Physics, Journal Year: 2023, Volume and Issue: 25(9), P. 6877 - 6890

Published: Jan. 1, 2023

Machine learning was used to rationalize the molecular origin of impact sensitivity nitroaromatic explosives.

Language: Английский

Citations

24
Calibration-free reaction yield quantification by HPLC with a machine-learning model of extinction coefficients DOI Creative Commons
Matthew A. McDonald, Brent A. Koscher, Richard B. Canty

et al.

Chemical Science, Journal Year: 2024, Volume and Issue: 15(26), P. 10092 - 10100

Published: Jan. 1, 2024

Reaction optimization and characterization depend on reliable measures of reaction yield, often measured by high-performance liquid chromatography (HPLC). Peak areas in HPLC chromatograms are correlated to analyte concentrations way calibration standards, typically pure samples known concentration. Preparing the material required for runs can be tedious low-yielding reactions technically challenging at small scales. Herein, we present a method quantify yield without needing isolate product(s) combining machine learning model molar extinction coefficient estimation, both UV-vis absorption mass spectra. We demonstrate variety important medicinal process chemistry, including amide couplings, palladium catalyzed cross-couplings, nucleophilic aromatic substitutions, aminations, heterocycle syntheses. The were all performed using an automated synthesis isolation platform. Calibration-free methods such as presented approach necessary platforms able discover, characterize, optimize automatically.

Language: Английский

Citations

9
A comparative study of methods for estimating model-agnostic Shapley value explanations DOI Creative Commons
Lars Henry Berge Olsen, Ingrid K. Glad, Martin Jullum

et al.

Data Mining and Knowledge Discovery, Journal Year: 2024, Volume and Issue: 38(4), P. 1782 - 1829

Published: March 29, 2024

Language: Английский

Citations

8
Predicting the enthalpy of formation of energetic molecules via conventional machine learning and GNN DOI
Di Zhang, Qingzhao Chu, Dongping Chen

et al.

Physical Chemistry Chemical Physics, Journal Year: 2024, Volume and Issue: 26(8), P. 7029 - 7041

Published: Jan. 1, 2024

Different ML models are used to map the enthalpy of formation from molecular structure, and impact different feature representation methods on results is explored. Among them, GNN achieve impressive results.

Language: Английский

Citations

6
Advancements in methodologies and techniques for the synthesis of energetic materials: A review DOI Creative Commons
Wei Du, Lei Yang, Jing Feng

et al.

Energetic Materials Frontiers, Journal Year: 2024, Volume and Issue: unknown

Published: June 1, 2024

Recent years have witnessed significant advancements in methodologies and techniques for the synthesis of energetic materials, which are expected to shape future manufacturing applications. Techniques including continuous flow chemistry, electrochemical synthesis, microwave-assisted biosynthesis been extensively employed pharmaceutical fine chemical industries and, gratifyingly, found broader This review comprehensively introduces recent utilization these emerging techniques, aiming provide a catalyst development novel green methods synthesizing materials.

Language: Английский

Citations

6
Bionic inspired multifunctional modular energetic materials: an exploration of new generation of application-oriented energetic materials DOI
Yujia Wen, Linyuan Wen, Bojun Tan

et al.

Journal of Materials Chemistry A, Journal Year: 2024, Volume and Issue: 12(16), P. 9427 - 9437

Published: Jan. 1, 2024

Aiming to balance the pertinence and universality of energetic materials, this study proposes a new concept bionic inspired multifunctional modular materials seeks out potential monomers via high-throughput screening strategy.

Language: Английский

Citations

4
Prediction and Construction of Energetic Materials Based on Machine Learning Methods DOI Creative Commons
Xiaowei Zang, Xiang Zhou, Haitao Bian

et al.

Molecules, Journal Year: 2022, Volume and Issue: 28(1), P. 322 - 322

Published: Dec. 31, 2022

Energetic materials (EMs) are the core of weapons and equipment. Achieving precise molecular design efficient green synthesis EMs has long been one primary concerns researchers around world. Traditionally, advanced were discovered through a trial-and-error processes, which required research development (R&D) cycles high costs. In recent years, machine learning (ML) method matured into tool that compliments aids experimental studies for predicting designing EMs. This paper reviews critical process ML methods to discover predict EMs, including data preparation, feature extraction, model construction, performance evaluation. The main ideas basic steps applying analyzed outlined. state-of-the-art about applications in property prediction inverse material is further summarized. Finally, existing challenges strategies coping with proposed.

Language: Английский

Citations

19
Artificial Intelligence Approaches for Energetic Materials by Design: State of the Art, Challenges, and Future Directions DOI Creative Commons
Joseph B. Choi, Phong Nguyen, Oishik Sen

et al.

Propellants Explosives Pyrotechnics, Journal Year: 2023, Volume and Issue: 48(4)

Published: Feb. 18, 2023

Artificial intelligence (AI) is rapidly emerging as an enabling tool for solving various complex materials design problems. This paper aims to review recent advances in AI-driven materials-by-design and their applications energetic (EM). Trained with data from numerical simulations and/or physical experiments, AI models can assimilate trends patterns within the parameter space, identify optimal material designs (micro-morphologies, combinations of composites, etc.), point superior/targeted property performance metrics. We approaches focusing on such capabilities respect three main stages materials-by-design, namely representation learning microstructure morphology (i.e., shape descriptors), structure-property-performance (S-P-P) linkage estimation, optimization/design exploration. provide a perspective view these methods terms potential, practicality, efficacy towards realization materials-by-design. Specifically, literature are evaluated capacity learn small/limited number data, computational complexity, generalizability/scalability other species operating conditions, interpretability model predictions, burden supervision/data annotation. Finally, we suggest few promising future research directions EM meta-learning, active learning, Bayesian semi-/weakly-supervised bridge gap between machine research.

Language: Английский

Citations

11
Force field-inspired transformer network assisted crystal density prediction for energetic materials DOI Creative Commons
Jun-Xuan Jin,

Gao‐Peng Ren,

Jianjian Hu

et al.

Journal of Cheminformatics, Journal Year: 2023, Volume and Issue: 15(1)

Published: July 19, 2023

Abstract Machine learning has great potential in predicting chemical information with greater precision than traditional methods. Graph neural networks (GNNs) have become increasingly popular recent years, as they can automatically learn the features of molecule from graph, significantly reducing time needed to find and build molecular descriptors. However, application machine energetic materials property prediction is still initial stage due insufficient data. In this work, we first curated a dataset 12,072 compounds containing CHON elements, which are traditionally regarded main composition elements materials, Cambridge Structural Database, then implemented refinement our force field-inspired network (FFiNet), through adoption Transformer encoder, resulting (FFiTrNet). After improvement, model outperforms other learning-based GNNs-based models shows its powerful predictive capabilities especially for high-density materials. Our also capability crystal density (i.e. Huang & Massa dataset), will be helpful practical high-throughput screening

Language: Английский

Citations

10
Assisted Energetic Material Property Prediction through Advanced Transfer Learning with Graph Neural Networks DOI

Jianjian Hu,

Jun-Xuan Jin,

Xiao‐Jing Hou

et al.

Industrial & Engineering Chemistry Research, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 16, 2025

In this study, we explore the use of transfer learning to predict properties energetic materials using a force-field-inspired transformer graph neural network (FFiTrNet). We began by pretraining model on large data set CHNOF compounds and then fine-tuning it smaller experimental enthalpy formation for materials. Our results show that significantly enhances accuracy predicting formation, with reduction in mean absolute error root-mean-square compared direct training set. Furthermore, demonstrate effectiveness other materials, highlighting its potential improve predictive capabilities machine models range properties. The result is most accurate among state-of-the-art material used enriches database materials' properties, making valuable publicly available future research.

Language: Английский

Citations

0