Photoreactions are often unpredictable and unwanted reactions are hard to control because of the large driving force of light. Modern supercomputers allow chemists to predict the photochemistry of molecules. We explore multidimensional potential energy surfaces of the photoreactivity of molecular catalysts and photoswitches using quantum mechanics to build mechanistic understanding. In particular, we have developed a new computational pump-probe absorption method that produces transient spectra that can be directly compared to experimental ultrafast transient absorption spectroscopy (TAS). TAS measures the complex landscape of relaxation paths of photoexcited states which are difficult to untangle in general. This new method uses the same standard linear response time-dependent density functional theory (LR-TDDFT) of more traditional stead-state approaches. Through simple post-processing of a typical LR-TDDFT output, the energy differences between various excited states are approximated and their corresponding oscillator strengths are resolved from the transition dipole moments between the LR-TDDFT wavefunctions. By coupling multiple excited state features, computational difference spectra can be directly compared to experimental TAS at various time delays. Pump-probe TDDFT captures all relevant excited state absorption features of photoisomerization of azobenzene.
