Over the past decade, science and technology have achieved great advancements driven by the synergy between materials and manufacturing processes; coupled with the growth of informatics, which offers powerful tools to process and interpret data, new analytical devices have been developed. This work describes a modular 3D printed instrument that utilizes the AS7262 light sensor coupled with a LED to perform absorbance and reflectance measurements. The mode of operation can be switched by conveniently attaching different 3D printed parts. An Arduino Nano is used for operating the electronics, and a python-based software is employed for data handling. The device, beside spectra acquisition, allows rapid identification and quantification of samples through a database and machine learning (ML) algorithms. A recursive methodology for regression specifically designed allowed sample quantification in a range spanning around 2.5 orders of magnitude with errors generally below 10%. PySpectro was used on homogeneous solution and on PADs (Paper-based Analytical Devices) for food dyes and phosphomolybdic assay for phosphate. The device may find applications in any colorimetric detection also outside the laboratory environment and can be a time-saving tool for fast preliminary determinations or educational purposes.
PySpectro: A modular 3D printed, machine learning assisted optical device for recognition and quantification of samples
Abate M.;Dossi N.
2025-01-01
Abstract
Over the past decade, science and technology have achieved great advancements driven by the synergy between materials and manufacturing processes; coupled with the growth of informatics, which offers powerful tools to process and interpret data, new analytical devices have been developed. This work describes a modular 3D printed instrument that utilizes the AS7262 light sensor coupled with a LED to perform absorbance and reflectance measurements. The mode of operation can be switched by conveniently attaching different 3D printed parts. An Arduino Nano is used for operating the electronics, and a python-based software is employed for data handling. The device, beside spectra acquisition, allows rapid identification and quantification of samples through a database and machine learning (ML) algorithms. A recursive methodology for regression specifically designed allowed sample quantification in a range spanning around 2.5 orders of magnitude with errors generally below 10%. PySpectro was used on homogeneous solution and on PADs (Paper-based Analytical Devices) for food dyes and phosphomolybdic assay for phosphate. The device may find applications in any colorimetric detection also outside the laboratory environment and can be a time-saving tool for fast preliminary determinations or educational purposes.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
