A versatile method for accurately predicting electronic absorption spectra of tetrapyrrole macrocycles.

Understanding the UV-Vis-NIR electronic spectra of tetrapyrrole macrocycles (porphyrins, chlorins, and bacteriochlorins) is crucial to rational designs of novel functional molecules and materials for energy conversion, catalysis, bio-imaging, and medical applications. Despite this, no methods have been defined to enable high throughput screening for a large body of tetrapyrrole molecules ([Formula: see text]) in a short-time period and accurately predict their electronic absorption spectra. Traditional time dependent-density functional theory (TD-DFT) method is cheap for single molecule simulation, but it can be expensive and time-consuming for massive numbers of molecules . Using a combination of computational chemistry, Gouterman four-orbital model, and machine learning methods, we show accurate predictions of the electronic absorption spectra of tetrapyrroles bearing various peripheral substituents across the UV-Vis-NIR spectral regions. Optimized AutoGluon pipelines trained on the full data set achieve high predictive fidelity (R