Physically inspired deep learning of molecular excitations and photoemission spectra

被引:44
作者
Westermayr, Julia [1 ]
Maurer, Reinhard J. [1 ]
机构
[1] Univ Warwick, Dept Chem, Gibbet Hill Rd, Coventry CV4 7AL, W Midlands, England
基金
英国工程与自然科学研究理事会; 奥地利科学基金会;
关键词
ORBITAL ENERGIES; ACCURATE; ELECTRON; AZULENE; PREDICTIONS; DATABASE; STATES;
D O I
10.1039/d1sc01542g
中图分类号
O6 [化学];
学科分类号
0703 ;
摘要
Modern functional materials consist of large molecular building blocks with significant chemical complexity which limits spectroscopic property prediction with accurate first-principles methods. Consequently, a targeted design of materials with tailored optoelectronic properties by high-throughput screening is bound to fail without efficient methods to predict molecular excited-state properties across chemical space. In this work, we present a deep neural network that predicts charged quasiparticle excitations for large and complex organic molecules with a rich elemental diversity and a size well out of reach of accurate many body perturbation theory calculations. The model exploits the fundamental underlying physics of molecular resonances as eigenvalues of a latent Hamiltonian matrix and is thus able to accurately describe multiple resonances simultaneously. The performance of this model is demonstrated for a range of organic molecules across chemical composition space and configuration space. We further showcase the model capabilities by predicting photoemission spectra at the level of the GW approximation for previously unseen conjugated molecules.
引用
收藏
页码:10755 / 10764
页数:10
相关论文
共 93 条
[1]   Toward reliable density functional methods without adjustable parameters: The PBE0 model [J].
Adamo, C ;
Barone, V .
JOURNAL OF CHEMICAL PHYSICS, 1999, 110 (13) :6158-6170
[2]   The Cambridge Structural Database: a quarter of a million crystal structures and rising [J].
Allen, FH .
ACTA CRYSTALLOGRAPHICA SECTION B-STRUCTURAL SCIENCE, 2002, 58 (3 PART 1) :380-388
[3]   The GW method [J].
Aryasetiawan, F ;
Gunnarsson, O .
REPORTS ON PROGRESS IN PHYSICS, 1998, 61 (03) :237-312
[4]  
Batzner S., Nat Commun, V13, P2453
[5]   First Principles Neural Network Potentials for Reactive Simulations of Large Molecular and Condensed Systems [J].
Behler, Joerg .
ANGEWANDTE CHEMIE-INTERNATIONAL EDITION, 2017, 56 (42) :12828-12840
[6]   Learning the electronic density of states in condensed matter [J].
Ben Mahmoud, Chiheb ;
Anelli, Andrea ;
Csanyi, Gabor ;
Ceriotti, Michele .
PHYSICAL REVIEW B, 2020, 102 (23)
[7]   Quantum chemical accuracy from density functional approximations via machine learning [J].
Bogojeski, Mihail ;
Vogt-Maranto, Leslie ;
Tuckerman, Mark E. ;
Mueller, Klaus-Robert ;
Burke, Kieron .
NATURE COMMUNICATIONS, 2020, 11 (01)
[8]   Levels, sources and potential human health risks of organic pollutants in urban soils [J].
Cachada, A. ;
Pato, P. ;
Rocha-Santos, T. ;
Ferreira da Silva, E. ;
Duarte, A. C. .
SCIENCE OF THE TOTAL ENVIRONMENT, 2012, 430 :184-192
[9]   An Azulene-Containing Low Bandgap Small Molecule for Organic Photovoltaics with High Open-Circuit Voltage [J].
Chen, Yao ;
Zhu, Youqin ;
Yang, Daobin ;
Zhao, Suling ;
Zhang, Lei ;
Yang, Lin ;
Wu, Jianglin ;
Huang, Yan ;
Xu, Zheng ;
Lu, Zhiyun .
CHEMISTRY-A EUROPEAN JOURNAL, 2016, 22 (41) :14527-14530
[10]   A universal density matrix functional from molecular orbital-based machine learning: Transferability across organic molecules [J].
Cheng, Lixue ;
Welborn, Matthew ;
Christensen, Anders S. ;
Miller, Thomas F., III .
JOURNAL OF CHEMICAL PHYSICS, 2019, 150 (13)