Feasibility of Activation Energy Prediction of Gas-Phase Reactions by Machine Learning

Sunghwan Choi; Yeonjoon Kim; Jin Woo Kim; Zeehyo Kim; Woo Youn Kim

doi:10.1002/chem.201800345

The formation of urea in space

Astronomy and Astrophysics ◽

10.1051/0004-6361/201731610 ◽

2018 ◽

Vol 610 ◽

pp. A26 ◽

Cited By ~ 7

Author(s):

Flavio Siro Brigiano ◽

Yannick Jeanvoine ◽

Antonio Largo ◽

Riccardo Spezia

Keyword(s):

Activation Energy ◽

Interstellar Medium ◽

Gas Phase ◽

Organic Molecules ◽

Peptide Bond ◽

Biological Molecules ◽

Gas Phase Reactions ◽

Molecule Reaction ◽

Quantum Chemistry Calculations ◽

Highly Correlated

Context. Many organic molecules have been observed in the interstellar medium thanks to advances in radioastronomy, and very recently the presence of urea was also suggested. While those molecules were observed, it is not clear what the mechanisms responsible to their formation are. In fact, if gas-phase reactions are responsible, they should occur through barrierless mechanisms (or with very low barriers). In the past, mechanisms for the formation of different organic molecules were studied, providing only in a few cases energetic conditions favorable to a synthesis at very low temperature. A particularly intriguing class of such molecules are those containing one N–C–O peptide bond, which could be a building block for the formation of biological molecules. Urea is a particular case because two nitrogen atoms are linked to the C–O moiety. Thus, motivated also by the recent tentative observation of urea, we have considered the synthetic pathways responsible to its formation. Aims. We have studied the possibility of forming urea in the gas phase via different kinds of bi-molecular reactions: ion-molecule, neutral, and radical. In particular we have focused on the activation energy of these reactions in order to find possible reactants that could be responsible for to barrierless (or very low energy) pathways. Methods. We have used very accurate, highly correlated quantum chemistry calculations to locate and characterize the reaction pathways in terms of minima and transition states connecting reactants to products. Results. Most of the reactions considered have an activation energy that is too high; but the ion-molecule reaction between NH2OHNH2OH2+ and formamide is not too high. These reactants could be responsible not only for the formation of urea but also of isocyanic acid, which is an organic molecule also observed in the interstellar medium.

Download Full-text

Soot Activation Energy for a Xylene-Fueled CVD Reactor

International Journal of Chemical Reactor Engineering ◽

10.2202/1542-6580.1388 ◽

2007 ◽

Vol 5 (1) ◽

Author(s):

Wahed Wasel ◽

Kazunori Kuwana ◽

Kozo Saito

Keyword(s):

Activation Energy ◽

Carbon Nanotube ◽

Species Composition ◽

Gas Phase ◽

Inverse Method ◽

Soot Formation ◽

Effect Of Temperature ◽

Gas Phase Reactions ◽

Concentration Data ◽

The Past

The past attempts of Arrhenius plots of carbon nanotube (CNT) yield (or CNT length) to obtain the overall activation energy for CNT formation under changing reaction temperature conditions have created substantial errors. Here we propose an inverse method to accurately account for the effect of temperature on gas-phase reactions and species composition. The overall activation energy of soot formation for a xylene-based CVD reactor was calculated and successfully compared with the measured gas-phase concentration data. Our proposed inverse method is expected to help improve the performance of CVD reactors and optimize its design.

Download Full-text

Gas Phase Adduct Reactions in MOCVD Growth of GaN

MRS Proceedings ◽

10.1557/proc-395-97 ◽

1995 ◽

Vol 395 ◽

Cited By ~ 2

Author(s):

A. Thon ◽

T.F. Kuech

Keyword(s):

Activation Energy ◽

Gas Phase ◽

Decomposition Temperature ◽

Adduct Formation ◽

Gas Phase Reactions ◽

Flow Tube ◽

Exponential Factor ◽

First Order ◽

Mocvd Growth ◽

Tube Reactor

ABSTRACTGas phase reactions between trimethylgallium (TMG) and ammonia were studied at high temperatures, characteristic to MOCVD of GaN reactors, by means of insitu mass spectroscopy in a flow tube reactor. It is shown, that a very fast adduct formation followed by elimination of methane occurs. The decomposition of TMG and the adduct - derived compounds are both first order and have similar apparent activation energy. The pre-exponential factor of the adduct decomposition is smaller, and hence is responsible for the higher full decomposition temperature of the adduct relative to that of TMG.

Download Full-text

Dynamics of the Gas-Phase Reactions of Chloride Ion with Fluoromethane: High Excess Translational Activation Energy for an Endothermic SN2 Reaction

Journal of the American Chemical Society ◽

10.1021/ja012031f ◽

2002 ◽

Vol 124 (2) ◽

pp. 336-345 ◽

Cited By ~ 20

Author(s):

Laurence A. Angel ◽

Sandra P. Garcia ◽

Kent M. Ervin

Keyword(s):

Activation Energy ◽

Gas Phase ◽

Chloride Ion ◽

Gas Phase Reactions ◽

Sn2 Reaction ◽

Translational Activation

Download Full-text

SiS formation via gas phase reactions between atomic silicon and sulphur-bearing species

Monthly Notices of the Royal Astronomical Society ◽

10.1093/mnras/staa269 ◽

2020 ◽

Vol 493 (1) ◽

pp. 299-304 ◽

Cited By ~ 1

Author(s):

Mateus A M Paiva ◽

Bertrand Lefloch ◽

Breno R L Galvão

Keyword(s):

Activation Energy ◽

Gas Phase ◽

Atomic Hydrogen ◽

Energy Surface ◽

Main Reaction ◽

Interaction Method ◽

Gas Phase Reactions ◽

Configuration Interaction Method ◽

Explicit Correlation ◽

No Activation

ABSTRACT The potential energy surface for the Si + SH and Si + SH2 reactions is explored using the highly accurate explicit correlation multireference configuration interaction method. For atomic silicon colliding with SH, SiS + H is predicted to be the main reaction channel with no activation energy. The reaction Si + SH2 → SiS + H2 is found to be largely thermodynamically favourable, but likely to be slow, due to its spin forbidden nature. Several details on possible mechanisms are evaluated, and implications for astrochemical models are discussed. Among other results, we show that SiS is stable towards collisions with H and H2, and that the HSiS molecule will quickly be converted to SiS in collisons with atomic hydrogen.

Download Full-text

BAND NN: A Deep Learning Framework For Energy Prediction and Geometry Optimization of Organic Small Molecules

10.26434/chemrxiv.9763094 ◽

2019 ◽

Author(s):

Siddhartha Laghuvarapu ◽

Yashaswi Pathak ◽

U. Deva Priyakumar

Keyword(s):

Machine Learning ◽

Density Functional ◽

Computational Cost ◽

Geometry Optimization ◽

Dft Methods ◽

Energy Prediction ◽

Machine Learning Model ◽

Equilibrium Structures ◽

High Level ◽

Non Equilibrium

Recent advances in artificial intelligence along with development of large datasets of energies calculated using quantum mechanical (QM)/density functional theory (DFT) methods have enabled prediction of accurate molecular energies at reasonably low computational cost. However, machine learning models that have been reported so far requires the atomic positions obtained from geometry optimizations using high level QM/DFT methods as input in order to predict the energies, and do not allow for geometry optimization. In this paper, a transferable and molecule-size independent machine learning model (BAND NN) based on a chemically intuitive representation inspired by molecular mechanics force fields is presented. The model predicts the atomization energies of equilibrium and non-equilibrium structures as sum of energy contributions from bonds (B), angles (A), nonbonds (N) and dihedrals (D) at remarkable accuracy. The robustness of the proposed model is further validated by calculations that span over the conformational, configurational and reaction space. The transferability of this model on systems larger than the ones in the dataset is demonstrated by performing calculations on select large molecules. Importantly, employing the BAND NN model, it is possible to perform geometry optimizations starting from non-equilibrium structures along with predicting their energies.

Download Full-text