Journal of Computational Chemistry

Staemmler, Volker; Franke, Robert

doi: 10.1002/jcc.70014pmid: 40153476

The origin of rotational barriers around CC single bonds is still vividly discussed and often referred to concepts like steric repulsion or hyperconjugation. In 1990, a paper was published in which the physical causes for the rotational barrier in ethane, that is, the well‐known finding that the potential energy in the eclipsed form is slightly higher than in the staggered form, appears as a consequence of a lowering of the nuclear‐electron attraction and not as a greater electronic repulsion in the eclipsed form. Surprisingly, this finding has practically not found its way neither into the textbook literature nor into the scientific discourse. Here we will show, by a careful analysis of the components to the total energy and their dependence on the geometry of the molecule, that the kinetic energy of the electrons and the virial theorem play the decisive role. This is very similar to their role for the origin of the chemical bond.

Frenking, Gernot

doi: 10.1002/jcc.70068pmid: 40155197

Schwarz, W. H. Eugen; Frenking, Gernot; Pan, Sudip

doi: 10.1002/jcc.70085pmid: 40152636

The molecular virial theorem relates kinetic and potential energies (T & V) to total energy and forces (E & R·∂E/∂R); it is a useful tool for analyzing the data, but does not provide clues on the origin of the stability of the “bonded” state. A strict conceptual distinction between cause and effect is recommended. Depending on the physical relationships, the induced change of one variable of the system leads to a resulting change of another variable; relaxation or response of the system can either moderate this change (in the sense of Le Chatelier's principle), enhance it, or even reverse it. Such unexpected, paradoxical behavior is common in reality and in daily life. As two examples of conceptual mix‐up in molecular chemistry, we discuss details of the origin of the steric pair‐pair repulsion and of the internal rotation barrier in ethane.

Nanba, Yusuke; Koyama, Michihisa

doi: 10.1002/jcc.70086pmid: 40119565

The band center of d orbitals (d‐band center) has been widely used as an effective descriptor for analyzing material properties. However, in high‐entropy nanoalloys, the diverse atomic environments present challenges in systematically exploring all possible combinations. Due to computational resource limitations, generating a sufficient number of samples is infeasible. Consequently, the d‐band center should be treated as a response variable in machine‐learning models. We calculated the d‐band center for individual atoms and applied supervised learning techniques to identify key factors influencing its behavior. While several factors were identified, their physical significance in predicting d‐band centers remained unclear. To address this issue, we incorporated various interatomic distance terms as descriptors, along with element‐based coordination numbers (ECN). The resulting model closely resembled the overlap integral of the Slater‐type orbital, and the regression coefficients of the ECN exhibited sensitivity to the effective principal quantum number and nuclear charge. Understanding the physical significance of these descriptors is crucial for improving property predictions and facilitating data collection on novel materials.

Jiang, Minghong; Wang, Zhanfeng; Chen, Yicheng; Zhang, Wenhao; Zhu, Zhenyu; Yan, Wenjie; Wu, Jianming; Xu, Xin

doi: 10.1002/jcc.70081pmid: 40099806

With the development of artificial neural networks (ANNs), its applications in chemistry have become increasingly widespread, especially in the prediction of various molecular properties. This work introduces the X2‐PEC method, that is, the second generalization of the X1 series of ANN methods developed in our group, utilizing pair energy correction (PEC). The essence of the X2 model lies in its feature vector construction, using overlap integrals and core Hamiltonian integrals to incorporate physical and chemical information into the feature vectors to describe atomic interactions. It aims to enhance the accuracy of low‐rung density functional theory (DFT) calculations, such as those from the widely used BLYP/6‐31G(d) or B3LYP/6‐31G(2df,p) methods, to the level of top‐rung DFT calculations, such as those from the highly accurate doubly hybrid XYGJ‐OS/GTLarge method. Trained on the QM9 dataset, X2‐PEC excels in predicting the atomization energies of isomers such as C6H8 and C4H4N2O with varying bonding structures. The performance of the X2‐PEC model on standard enthalpies of formation for datasets such as G2‐HCNOF, PSH36, ALKANE28, BIGMOL20, and HEDM45, as well as a HCNOF subset of BH9 for reaction barriers, is equally commendable, demonstrating its good generalization ability and predictive accuracy, as well as its potential for further development to achieve greater accuracy. These outcomes highlight the practical significance of the X2‐PEC model in elevating the results from lower‐rung DFT calculations to the level of higher‐rung DFT calculations through deep learning.

Kalamatianos, K. G.; Flenga, Olga N.

doi: 10.1002/jcc.70083pmid: 40110645

In this study, a hybrid machine learning (ML) approach is presented for accurately predicting activation energies (Ea) of gas‐phase elementary reactions involving organic compounds with up to seven non‐hydrogen atoms. Given the importance of activation energies in reaction studies and modeling, ML composite models were created that effectively integrate molecular descriptors with semi‐empirical and single energy density functional theory (DFT) calculations. The dataset, containing 300 randomly selected elementary gas‐phase reactions, was assembled using accurate DFT (ωB97X‐D3/def2‐TZVP) values for activation energies Ea from a database alongside semi‐empirical computations. For accurate predictions, this approach required the inclusion of both physical organic and geometric/empirical descriptors in the training procedure. The best two ML models demonstrated efficient Ea prediction capability, achieving a mean absolute error (MAE) of 1.314 kcal mol−1 and R2 of 0.992 (Model 3) and (MAE) of 1.949 kcal mol−1 and R2 of 0.979 (Model 2) in validation tests. Notably, this performance approaches the threshold of “chemical accuracy” of 1 kcal mol−1. Model's 3 robustness was tested across the reaction types present in the dataset, demonstrating its ability in properly predicting activation energies, which is critical for the study and optimization of chemical processes.

Alves, Rodolpho L. R.; Leitão, Ezequiel F. V.; Ventura, E.; Monte, S. A.

doi: 10.1002/jcc.70079pmid: 40099792

Normally, carbocations and carbanions of hydrocarbons react to form a CC σ bond. However, if the ions are very stable with a large steric repulsion between them, one can also have the formation of an ion pair, generally much less stable than the covalent form. For an extremely large steric repulsion in the covalent form, the ion pair can become the most stable form, even in the gas phase. DFT and CASSCF calculations indicate that the tert‐butylfulleride anion and the tris[1‐(5‐isopropyl‐3,8‐dimethylazulenyl)]‐cyclopropenylium cation form an ion pair. This is the first study of this ion pair, although it is the building block of a salt that has already been synthesized. DFT results indicate that this ion pair is considerably more stable than its covalent counterpart. Nevertheless, several properties of the ionic and covalent forms are similar. An energy decomposition analysis indicates the polarization, electrostatic, and dispersion terms as the most important attractive terms between the ions.

Siegbahn, Per E. M.

doi: 10.1002/jcc.70088pmid: 40127040

Cytochrome c nitrite reductase (CcNiR) activates nitrite and produces ammonia. It is one of several enzymes that use a redox‐active cofactor to perform its reaction. In this case, the cofactor has a heme with a lysine as the proximal ligand and a charged nearby arginine. The role of a tyrosine, which is also close, has been less clear. There are also four bis‐histidine‐ligated hemes involved in the electron transfers. CcNiR has been studied before, using essentially the same methods as here. However, the mechanism is very complicated, involving six reductions, and quite different results for the mechanism have been obtained here. For example, the tyrosine has here been found to be redox active in the final step when ammonia is produced. Also, the arginine has here been found to stay protonated throughout the mechanism, which is different from what was found in the previous study. The present results are in very good agreement with experimental findings and are, therefore, another case where the methodology has been shown to work very well. Previous examples include Photosystem II and Nitrogenase, normally considered to be the most important enzymes in nature for the development of life.

doi: 10.1002/jcc.70123pmid: N/A

Odonkor, Gideon; Odoh, Samuel O.

doi: 10.1002/jcc.70090pmid: 40135997

Computational investigations of Inverted Singlet‐Triplet (INVEST) emitters often rely on ADC(2) and TD‐DFT excitation energies (EEs) obtained with the vertical approximation. Here, we first considered several cyclazine derivatives and examine the sensitivity of vertical EEs (VEEs) as well as singlet‐triplet gaps, ΔES1T1 gaps, to the level at which the ground state (S0) structure was optimized. For cyclazine, VEEs and vertical gaps from ADC(2) or TD‐DFT are spread over a narrow range (< 0.064 eV) whether the S0 structure is optimized with various DFT, CCSD, and RI‐MP2 methods. However, for asymmetric cyclazines, depending on the protocol for optimizing S0 structures, not only are VEEs spread over a substantially wider range (up to 0.75 eV) but so are vertical ΔES1T1 gaps (up to 0.30 eV), leading to cases where, with different S0 structures, one obtains positive vertical ΔES1T1 gaps or significantly negative gaps. We relate this behavior to the introduction of significant asymmetry and bond‐length variations in the cyclazine derivatives, formed by ligand functionalization or modification of the cyclazine core. On a more positive note, adiabatic EEs (AEEs) and adiabatic ΔES1T1 gaps display significantly lower sensitivity (7–30× less) to the geometry optimization protocols than their vertical analogs. Crucially, for cyclazine, the M06‐HF functional with 100% non‐local exchange provides the closest S0 geometry to available CCSD(T) data. We show that this effect exists also for other frameworks (e.g., azulene, pentaazaphenalene, and non‐alternant polycyclic hydrocarbons) that have been considered for the INVEST property, with VEEs spread over a broader range of up to 1.19 eV and vertical ΔES1T1 gaps over a range of 0.62 eV. For INVEST emitters, it is therefore extremely important to judiciously choose the computational protocol for optimizing ground state geometries, in computing VEEs and vertical ΔES1T1 gaps.