Photothermal microscopy beyond intensity detection: exploiting spatial signal distributions for enhanced sensitivity

This paper reports a substantial enhancement in the sensitivity of photothermal microscopy by leveraging the inevitable spatial mismatch of pump and probe beams, which has conventionally been regarded as an experimental imperfection to be minimized. For aqueous solutions, the limit of detection (LOD) of conventional photothermal microscopy has been restricted to an absorbance equivalent of approximately 10^-3 for a 1 cm optical path length, corresponding to performance two to three times worse than that of standard commercial dual-beam spectrophotometers. We overcome this limitation by exploiting the spatial distribution of the photothermal signal, which is captured by imaging the probe beam with a camera. Using photothermal reflectance microscopy as a representative implementation, pump-induced reflectivity changes at the silica-aqueous interface generate a two-dimensional asymmetric intensity distribution in the probe beam owing to imperfect spatial overlap of the pump and probe beams at the focus. This asymmetric beam pattern is distinguishable from noise-induced intensity variations, enabling highly sensitive photothermal detection. Because quantitative analysis of these two-dimensional data is challenging due to their complexity, a deep-learning-based analysis was employed as a practical means to extract concentration-dependent information encoded in the spatial distribution. As a result, we achieved an LOD equivalent to an absorbance of 2 × 10^-4 for a 1 cm optical path length, representing a 3-5-fold improvement over previous photothermal microscopy limits and surpassing the performance of conventional spectrophotometers while preserving microscopic spatial resolution.