Laboratory of computer assistance to chemical researches (№50)
Modeling of chemical objects and processes and high-performance computing technologies development for solving chemical problems:
- mathematical modeling of the dynamics of the states of a pair of molecules in the reaction center of a bimolecular secondary photoreaction and the spectro-temporal characteristics of the corresponding fluorescence;
- the mechanisms and kinetics of processes initiated by electron transfer;
- methods for accelerating resource-intensive quantum chemical calculations of bioorganic molecules;
- studying the nature of non-covalent interactions using quantum chemical methods;
- research on the performance of computer systems for computational chemistry problems.
✓ Mathematical modeling of the dynamics of the states of a pair of molecules in the reaction center of a bimolecular secondary photoreaction and the spectro-temporal characteristics of the corresponding fluorescence
Mathematical modeling of the dynamics of secondary photoreaction of two-level molecules emitting spontaneous or resonant selective fluorescence upon excitation of a reagent molecule by an ultra-short or long-term pulse, respectively, monochromatic light is carried out. A system of equations for dynamics is obtained based on the use of solutions of a modified system of Schrodinger equations for the amplitudes of states of a composite system consisting of a pair of molecules of a reaction center (RC) and a quantized radiation field. The modification used was that the equations for the amplitudes were written in such a way that their solutions reflect the dynamics of the population of the ground state of each RC molecule when it emits a corresponding fluorescence photon, while determining their relative amount, thus fully and in detail characterizing the efficiency of the reaction (including its quantum yield as the population of the ground state of the product molecule). This makes the resulting system of equations different from the well-known equations for the dynamics of bimolecular secondary photoreactions (for example, the well-known system of Bloch optical equations), which non-selectively describe the fluorescence of a pair of RC molecules, determining only the population of the “collective” ground state of this pair, without providing information about the efficiency of the reaction.
Reaction efficiency F(w, Δ) as a function of the parameters of the reaction center molecules (interaction of molecules w and the frequency difference of their own transitions Δ).
✓ Mechanisms and kinetics of processes initiated by electron transfer
A integrated study of the thiocyanate anion electrooxidation by experimental (cyclic voltammetry, chronoamperometry, electrolysis) and theoretical methods (numerical modeling and quantum chemical calculations) revealed that, in addition to the dimerization of the thiocyanate radical, the anodic process also involves its coupling with an anion to form the thiocyanogen anion radical and a reversible coupling reaction of thiocyanogen with thiocyanate anion. The latter reaction is also observed during the reduction of thiocyanogen. The established mechanisms were successfully used both to correctly interpret results of electroanalytical studies on the thiocyanation of organic compounds and to optimize the corresponding electrosynthesis processes.
✓ The electrochemical oxidation of (iso)thiocyanic acid was studied for the first time by complex of experimental (cyclic voltammetry, controlled potential electrolysis) and theoretical (digital simulations, quantum chemical calculations) methods. We unexpectedly discovered that electrooxidation of (iso)thiocyanic acid leads to isothiocyanogen (and not to thiocyanogen, as in the oxidation of thiocyanate salts). It is the first case where isothiocyanogen was synthesized in solution under mild conditions. The reactivity of isothiocyanogen has been evaluated using indole thiocyanation as a model reaction. The high yield of 3-thiocyanato-1H-indole (93 %) obtained in this process underscores its practical interest.
✓ Studying the nature of non-covalent interactions using quantum chemical methods
Based on quantum chemical analysis of anion complexes with hydrogen halide molecules, it was demonstrated that the nature of the halogen bond depends on the energy of the anion's lone pair orbital, which can be considered an indicator of the orbital-orbital interaction of the monomers. It was shown that the high energy of the alkyl carbanion's lone pair orbital determines the significant contribution of the charge transfer component and the partially covalent nature of the intermolecular halogen bond. A linear correlation was found between the halogen bond energy and the potential energy density at the critical point of intermolecular contact.
The influence of the donor's lone electron pair on the hybrid orbital of the Cl atom
and the energy ENBO of orbital-orbital interaction
✓ Calculations of binary complexes with the C(sp3)…Cl-Li intermolecular bond formed by alkyl carbanions with lithium chloride were performed using the MP2/aug-cc-pVTZ method. The disappearance of the σ-hole on the Cl atom upon replacement of the hydrogen atom in the HCl molecule with lithium leads to a change in the nature of the intermolecular interaction. According to calculations, the increase in the binding energy Ebind in a series of carbanion complexes with lithium chloride is determined not by an increase in the electron density rBCP at the critical point of the C…Cl intermolecular contact, but by an increase in the dispersion interaction energy, which correlates with the binding energy.
Change in the stability of carbanion complexes when a hydrogen atom
in an HCl molecule is replaced by lithium
✓ The relative orientation of monomers in H2O...SHX complexes with chalcogen bonds and H2O...HSX complexes with hydrogen bonds is determined by the positions of the positive electrostatic potential maxima on the sulfur and hydrogen atoms of the SHX molecule (X is a halogen atom). According to calculations, the energies of chalcogen and hydrogen bonds have close values. Electrostatic interactions make the primary contribution to complex stabilization; however, in H-bonded complexes, the charge transfer component is also significant, contributing approximately 40% of the electrostatic component to the binding energy. Dispersion interactions are important for both types of complexes; in complexes with chalcogen bonds, the dispersion energy is comparable to the charge transfer component.
Structure of the transition state for the interconversion of complexes with a chalcogen S…O bond (A) and a hydrogen S-H…O bond (B). E1 and E2 show the activation barrier of the transitions A → B and B → A, respectively
✓ In binary complexes of methane and ethane with HCl and LiCl, the effect of monomers orientation on C...H-Cl/Li-Cl bonding was studied using the MP2/aug-cc-pVTZ method. Calculations predict the formation of ethane complexes with “parallel” and “perpendicular” orientation of HCl/LiCl molecules relative to the C-C covalent bond line of ethane. Decomposition of binding energy into components shows that in H-bonded complexes exchange repulsion and dispersion prevail over electrostatics and charge transfer. Stabilization of complexes with a Li-bond is mainly determined by the polarization component. Due to dispersion interactions, the perpendicular configuration of ethane complexes becomes more stable than the linear configuration
Molecular complexes formed by methane and ethane molecules with HCl (1-3) and LiCl
(4-6). Numbers indicate interatomic distances in Å.
Density difference maps for ethane complexes with HCl and LiCl molecules. Blue regions represent loss of electron density as a result of formation of the complex, relative to isolated monomers; purple regions indicate increased density. The contour shown is 0.0005 a.u., calculated at the MP2/aug-cc-pVTZ level.
✓ Methods for accelerating resource-intensive quantum chemical calculations of bioorganic molecules
Giant bioorganic molecules and their complexes, such as thousands of protein-ligand docking complexes with thousands of atoms, are complex (see Figure 1), but they are relevant, including in the development of future drugs. Quantum chemical calculations of such molecules are relevant and can significantly improve accuracy compared to faster non-quantum calculations, but they are extremely resource-intensive due to the extremely large size of the resulting arrays, as shown in the Table.
After our previous very powerful acceleration of faster, but less accurate semi-empirical quantum chemical calculations of giant bioorganic molecules (up to 120 thousand There are real protein atoms and more than 1 million. a random one-dimensional protein on a home PC back in 2008) laid the foundation for a complex set of methods to dramatically accelerate more accurate but ~ a thousand times slower modern nonempirical calculations of the DFT of giant bioorganic molecules, to begin with, see Article 1. Further, a complex set of methods was improved to dramatically accelerate calculations of the DFT of giant bioorganic molecules and accelerate mass calculations of the DFT of thousands of docking complexes of thousands of atoms.
Further, DFT methods were adapted to the specifics of the SSP to dramatically accelerate calculations of giant bioorganic molecules, which made it possible to accelerate both due to the convergence of the SCF (and a significant reduction in the increment of the density matrix to the next iteration of the SCF, especially in advanced iterations), and to take into account the specifics of various parts of giant bioorganic molecules and their complexes, For example, in thousands of docking complexes, take into account the specifics of each group of ligands near one of its parts of the giant protein cavity
Grouping of Ligands and localization of their interaction with the giant Protein
Table. Table of approximate array sizes for non-empirical quantum chemical calculations of the relatively small protein IMMUNOPHILIN FKBP-12, consisting of approximately 1700 atoms
|
Number of Basis Functions BF when using the 6-31G* basis, which is often used in DFT calculations |
∼ 14 000 |
| Number of Auxiliary Density Functions χp(r) |
∼ 30 000 |
| The number of non-negligible 4-center Coulomb integrals (fn|fm) from BF only | ∼ 2×1014 |
| The number of non-negligible 3-center Coulomb integrals from BF to VFP |
∼ 2×1011 |
✓ Research on the performance of computer systems for computational chemistry problems
An analysis of first-generation ARM server processors, including the Fujitsu A64FX, compared to Intel Xeon Skylake and Cascade Lake, and AMD EPYC Rome and Milan processors, while observing performance advantages of the A64FX over Cascade Lake for quantum chemistry problems taking electron correlation into account, did not reveal superior performance of the A64FX for molecular dynamics in the high-performance GROMACS software suite. ARM's advantages for molecular dynamics are related to its low power consumption.
An analysis of performance data for general-purpose AMD CDNA 2 GPUs and Nvidia Volta and Ampere GPUs shows that for quantum chemistry problems, the achieved efficiency is high for calculations that explicitly account for correlation, including using coupled cluster methods. Using these GPUs for DFT with a Gaussian basis can yield high performance gains when using specialized methods not available in most available quantum chemistry software packages.








