Machine learning enables computers to address problems by learning from data. Deep learning is a type of machine learning that uses hierarchical recombination of features to extract pertinent information, and then learn the patterns represented in the data. Over the last eight years, its abilities have increasingly been applied to a wide variety of chemical challenges, from improving computational chemistry, to drug and materials design, and even synthesis planning. This review aims to explain the concepts of deep learning to chemists from any background and will follow this with an overview of the diverse applications demonstrated in the literature. We hope that this will empower the broader chemical community to engage with this burgeoning field and foster the growing movement of deep learning accelerated chemistry.

In the present work, we examine the effects of zinc chloride (ZnCl2) and aluminium chloride (AlCl3) complexation on the photochemistry of two well-known Type I photoinitiators, methyl-4ʹ-(methylthio)-2-morpholinopropiophenone (MMMP) and 2,2-dimethoxy-2-phenylacetophenone (DMPA). High-level ab initio calculations and experimental results demonstrate that Lewis acid complexation has a significant effect on the individual processes that comprise radical photoinitiation. Theoretical calculations predict that ZnCl2 coordinates to MMMP and DMPA to form thermodynamically stable bidentate ketone-amine and ketone-ether chelates, respectively. Meanwhile, the AlCl2 + cation coordinates to MMMP and DMPA to form a tridentate ether-amine-ketone chelate and a bidentate ketone-ether chelate, respectively. We found that addition of ZnCl2 and AlCl3 to solutions containing MMMP significantly increase its molar extinction coefficient (ε) between 350-360 nm. In contrast, the complexation of either ZnCl2 or AlCl3 to DMPA slightly reduces the value of ε in the 350-360 nm range. Time dependent density functional theory (TD-DFT) calculations demonstrate that Lewis acid complexation blue shifts the nπ* excitations of both DMPA and MMMP, while concurrently red shifting the ππ* transitions. Complexation also significantly alters the stability and reactivity of the photoinitiator fragment radicals. Lewis acid complexation localizes and destabilizes acyl radicals, resulting in significantly increased reactivity towards methyl methacrylate (MMA). In contrast, complexation of Lewis acids dramatically reduces the reactivity of the morpholine substituted isopropyl radical and the dimethoxyphenyl radical towards MMA. Alternative complexation at the methyl ester side-chain of MMA has a beneficial effect on the reactivity of all fragments, increasing addition rate coefficients by 2-4 orders of magnitude. We discuss some of the important implications of these findings for pulsed laser polymerization (PLP), and acetophenone photochemistry more generally.

The oxidation potential of a test set of 21 nitroxide radicals, including a number of novel compounds, has been studied experimentally in acetonitrile and correlated with theoretical calculations. It was found that both Hammett constants (σ) of the substituents on the nitroxide radicals and hyperfine splitting constants of the respective nitrogen atoms (α) were well correlated to their experimental oxidation potentials. Theoretical calculations, carried out at the G3(MP2,CC)(+)//M06-2X/6-31+G(d,p) level of theory with PCM solvation corrections, were shown to reproduce experiments to within a mean absolute deviation of 33 mV, with a maximum deviation of 64 mV. The oxidation potentials of the nitroxides examined varied over 400 mV, depending on ring size and substitution. This considerable variation can be rationalised by the ability of various substituents to electrostatically stabilize the oxidised oxoammonium cation. Importantly, this can be quantified by a simple predictive relationship involving the distance scaled dipole and quadrupole moments of the analogous cyclohexyl ring. This highlights the often-overlooked role of through-space electrostatic substituent effects, even in formally neutral compounds.