Preliminary Classification of Black Writing Pen Inks Using Optical-probing Photoacoustic Measurement by Pulsed Laser Excitation Combined with Machine Learning for Forensic Applications

Abstract We successfully measured the photoacoustic (PA) signals of black inks in solution from marking pens using a combination of optical-probing photoacoustic spectroscopy (OPPAS) with nanosecond pulsed laser excitation. OPPAS is a highly sensitive technique that detects heat emitted by molecules through non-radiative transitions after photoexcitation. It has never been applied to the measurement of writing pen inks. The amplitudes and waveforms of the PA signals largely depended on the type of the ink. The PA signal amplitudes of the pigment-based inks were generally higher than those of the dye-based inks. The PA signals were evaluated using four machine learning methods to classify the inks. The best Balanced Accuracy (0.94) was achieved using a Support Vector Machine classifier. The score of the pigment-based inks was better than that of the dye-based inks. However, there was no difference in the conventional absorption spectra of the pigment-based inks. These results suggest that OPPAS combined with machine learning could be useful for the classification of writing inks for forensic applications in the future.