Predicting retrosynthetic pathways using transformer-based models and a hyper-graph exploration strategy

We present an extension of our Molecular Transformer model combined with a hyper-graph exploration strategy for automatic retrosynthesis route planning without human intervention. The single-step retrosynthetic model sets a new state of the art for predicting reactants as well as reagents, solvents and catalysts for each retrosynthetic step. We introduce four metrics (coverage, class diversity, round-trip accuracy and Jensen–Shannon divergence) to evaluate the single-step retrosynthetic models, using the forward prediction and a reaction classification model always based on the transformer architecture. The hypergraph is constructed on the fly, and the nodes are filtered and further expanded based on a Bayesian-like probability. We critically assessed the end-to-end framework with several retrosynthesis examples from literature and academic exams. Overall, the frameworks have an excellent performance with few weaknesses related to the training data. The use of the introduced metrics opens up the possibility to optimize entire retrosynthetic frameworks by focusing on the performance of the single-step model only.

We present an extension of our Molecular Transformer model combined with a hyper-graph exploration strategy for automatic retrosynthesis route planning without human intervention.     

Diffusion of Gold Nanoparticles in Inverse Opals Probed by Heterodyne Dynamic Light Scattering

The diffusive behavior of nanoparticles inside porous materials is attracting a lot of interest in the context of understanding, modeling, and optimization of many technical processes. A very powerful technique for characterizing the diffusive behavior of particles in free media is dynamic light scattering (DLS). The applicability of the method in porous media is considered, however, to be rather difficult due to the presence of multiple sources of scattering. In contrast to most of the previous approaches, the DLS method was applied without ensuring matching refractive indices of solvent and porous matrix in the present study. To test the capabilities of the method, the diffusion of spherical gold nanoparticles within the interconnected, periodic nanopores of inverse opals was analyzed. Despite the complexity of this system, which involves many interfaces and different refractive indices, a clear signal related to the motion of particles inside the porous media was obtained. As expected, the diffusive process inside the porous sample slowed down compared to the particle diffusion in free media. The obtained effective diffusion coefficients were found to be wave vector-dependent. They increased linearly with increasing spatial extension of the probed particle concentration fluctuations. On average, the slowing-down factor measured in this work agrees within combined uncertainties with literature data.

     

Quenching of Premixed Flames at Cold Walls: Effects on the Local Flow Field

The presence of a turbulent premixed flame strongly influences the properties of the adjacent velocity boundary layer. This influence is studied here using a generic configuration where at atmospheric pressure turbulent premixed methane/air flames interact with a temperature stabilized wall. The experiment is optimized for well-defined boundary conditions and optical accessibility in the zone where the flame impinges at the wall. Laser based diagnostic methods are used to measure two components of the velocity field by particle image velocimetry simultaneously with the flame front position using laser induced fluorescence of the OH molecule. Two measurement planes are selected that are aligned perpendicularly to the surface of the wall. Based on this data, the flow field near the wall is analyzed by different methodologies using laboratory-fixed and flame-conditioned statistics, a quadrant splitting analysis of the Reynolds stresses and an evaluation of the production term of the turbulent kinetic energy. The results of chemically reactive cases are compared to their corresponding non-reactive flows for otherwise identical inflow conditions. In the zone of flame-wall interactions the boundary layer structure and its turbulence are dominated by the turbulent flame. Important features are that the flame compresses the boundary layer already upstream the location where the flame is finally quenched and that ejection and sweeps are no longer the dominant mechanisms as in non-reactive boundary layers. This experimental data may serve additionally as a database for model development for near wall reactive flows.     

Thiophenylazobenzene: An Alternative Photoisomerization Controlled by Lone-Pair⋅⋅⋅π Interaction

Azoheteroarene photoswitches have attracted attention due to their unique properties. We present the stationary photochromism and ultrafast photoisomerization mechanism of thiophenylazobenzene (TphAB). It demonstrates impressive fatigue resistance and photoisomerization efficiency, and shows favorably separated (E)- and (Z)-isomer absorption bands, allowing for highly selective photoconversion. The (Z)-isomer of TphAB adopts an unusual orthogonal geometry where the thiophenyl group is perfectly perpendicular to the phenyl group. This geometry is stabilized by a rare lone-pair⋅⋅⋅π interaction between the S atom and the phenyl group. The photoisomerization of TphAB occurs on the sub-ps to ps timescale and is governed by this interaction. Therefore, the adoption and disruption of the orthogonal geometry requires significant movement along the inversion reaction coordinates (CNN and NNC angles). Our results establish TphAB as an excellent photoswitch with versatile properties that expand the application possibilities of AB derivatives.