Artificial Neural Networks Applied to Colorimetric Nanosensors: An Undergraduate Experience Tailorable from Gold Nanoparticles Synthesis to Optical Spectroscopy and Machine Learning

Abstract

Nowadays, technologies involving nanoparticles, colloids, sensors, and artificial intelligence are widespread in society, media, and industry. It is thus mandatory to integrate them into the curricula of students enrolled in chemistry and materials science. To this purpose, we designed a simple assay for the detection of glutathione (GSH) using surface-clean gold nanoparticles (Au NPs). The alteration of the electric double layer of the Au NPs with increasing GSH concentration causes the particles to aggregate, producing a measurable change in color. This behavior, which is widely exploited for optical sensing, has been introduced in an undergraduate course to familiarize the students with the concepts of nanoparticles, colloids, colloidal stability, and sensor features (selectivity, sensitivity, detection range). Nonetheless, there are no analytical models to quantitatively relate the absorption of Au NP colorimetric sensors to analyte concentration, which is the ideal condition for resorting to machine learning (ML). Hence, an artificial neural network was instructed in a students' collective data-sharing experiment about machine learning. Overall, the laboratory experience is safe and highly tailorable to students' background, course duration, available instruments, and teacher's didactic objectives. For instance, it can be lifted to the Master's or Ph.D. level by improving the spectroscopic and ML contents or shifted toward the industrial ground by focusing on the nanoparticle synthesis. We propose the integration of this laboratory experience in the undergraduate and Master's academic programs to stimulate the students with a collection of hot topics that at the same time can consolidate their preparation on arguments of great relevance for their professional life.

Implementation of an R Shiny App for Instructors: An Automated Text Analysis Formative Assessment Tool for Evaluating Lewis Acid–Base Model Use

“…Technologies help facilitate learners’ access to and participation in learning resources that provide formative feedback . The development of this Shiny app based on our previous machine learning-based tool is a small start, adding to other machine learning-based resources to support the learning of chemistry topics. − We have made our R code freely available so that others might use our code to translate other assessment instruments and prompts into practical tools for educators more easily . We should note that other researchers have also begun to develop tools such as dashboards to provide students and instructors with automated feedback; , this is yet another example of how such technologies can be used to transform and promote learning.…”

Prácticas de Materiales y Nanomateriales para Estudiantes de Primer Curso de Ingeniería Física

“…En este sentido, la educación universitaria de la química hace uso de las prácticas de laboratorio como un contexto privilegiado de enseñanza (Sánchez, 2017). La importancia de las prácticas se pone de manifiesto en la bibliografía, donde se han reportado experiencias de laboratorio puesto que permiten a los estudiantes aplicar y conectar diferentes conceptos químicos del área de química de materiales además de usar diversas técnicas de laboratorio e instrumentación (Todd, 2022), y el uso de nanopartículas como sensores (Revignas, 2022). Por tanto, la idea desarrollada en este proyecto para que el alumnado pueda preparar materiales porosos tipo MOF y nanopartículados basados en los diferentes conceptos del temario de la asignatura, se ha cumplido con éxito y como resultado se han podido elaborar dos prácticas de laboratorio innovadoras adecuadas a las habilidades y los conocimientos que se les exige a los estudiantes.…”

Exploring Machine Learning in Chemistry through the Classification of Spectra: An Undergraduate Project

“…13 A number of authors have approached the subject of teaching machine learning via artificial neural networks. 14,15 However, the data sets that are required for training these networks need to be very reproducible to avoid the ANNs finding patterns in the data that do not exist. We therefore avoided the use of ANNs in this study, given the complex nature of the substrates that we planned to sample.…”