Photoswitches, molecules which undergo reversible structural changes under photo-irradiation, have in the past decades emerged as a crucial component in photo-responsive systems. For their wide applications in the fields of biomedical and material science, it is often highly desirable to trigger the photoisomerization processes with long-wavelength light, especially red- and NIR-light, due to its lower propensity to cause damage as well as superior ability to penetrate the biological tissues. To this end, indigo as one of the most historical and abundant dyes utilized in human history has recently attracted much attention. The unique structure of this blue dye grants its natural absorbance in the red-light region (and hence its blue color), thus making it an ideal candidate for building red-light responsive photoswitches.
In this present work, we developed a copper-catalyzed indigo N-arylation to efficiently convert this blue dye into functional red-light photoswitches. N-aryl substituents of indigos have previously been implemented to control the thermal stability of the switched state (i.e. thermal half-lives) without sacrificing the red-light addressability. In this method, we for the first time achieved the installation of indigo N-aryl groups containing synthetically useful functional motifs, representing an important step towards their real applications. The surprising selectivity for mono-N-arylation is explained by theoretical calculations, where a key bis-copper-indigo intermediate is proposed to lower the energy barrier of the most challenging step. Furthermore, through a data-science workflow involving N-aryl group representation*1, DFT featurization*2, and multivariate linear regression*3, models correlating the thermal half-lives and structures of these indigo photoswitches were constructed, which is expected to aid future rational design of this attractive class of compounds. This work combining experimental, computational, and data-science approaches thus serves as an important landmark on the avenue leading to widely applicable indigo photoswitches.
- *1. Representation: a method to extract simplified but crucial structural components from the original complex form of the molecules.
- *2. Featurization: a process to convert the data in its crude form to quantifiable parameters for model construction
- *3. Multivariate linear regression: linear regression is a tool to correlate or predict continuous outcomes (outcome variables) based on given data (explanatory variables). When there are multiple explanatory variables, it is called multivariate linear regression.