Project Portugal 2030

Summary

Surface plasmons are charge density oscillations coupled to an intrinsic electromagnetic field. Under certain conditions, they can be resonantly excited by external electromagnetic waves. The resonance frequencies are sensitive to changes in the local dielectric environment, e.g. caused by analyte molecules. Yet, are they sensitive enough? How can the detected changes be made analyte-specific? Are there other elementary excitations that can be used instead, or combined with surface plasmons to improve sensing devices or make them cheaper? These are some of the pertinent questions which the research team aims to answer. The studies will be focused on systems with surface plasmons where the charge density oscillations occur in metal nanoparticles (NPs). They can interact with target molecules and possibly with other elementary excitations present in nearby 2D materials such as Graphene and TMDCs, and these materials will be used together with plasmonic films containing Au-Ag NPs embedded in a dielectric matrix to optimize their sensing properties. Machine Learning (ML) Algorithms will use as input optical (either spectroscopy or polarimetry) data to tackle the challenge of identifying different analytes in contaminated water samples. The main objectives of the project are: 1) Development of optical sensors with enhanced sensitivity and selectivity, based on new hybrid nanomaterials The hybrid materials will result from combining plasmonic substrates (thin films containing plasmonic NPs embedded in an oxide matrix) with 2D materials, that are functionalized with recognition elements to achieve the chemical selectivity of pollutants. The target molecules will be water pesticide contaminants, especially two recently added molecules to the 4th Watch List of the EU (Azoxystrobin and Diflufenican). 2) Search for alternative (physical) mechanisms for specific detection of analyte molecules. The second objective of the project will address alternative mechanisms, which may be intrinsically specific. In this way, the alternative or enhanced sensing mechanisms will include: (i) charge transfer events between graphene and nanoparticles in the presence of the analyte; (ii) exciton photoluminescence (PL) from TMDCs, coupled to plasmons in the NPs, spectrally modified (Fano-type resonance), enhanced or quenched in the presence of the analyte. All these possibilities will be analysed theoretically and modelled. 3) Development of Machine Learning algorithms to identify the presence of different types of analytes in complex water samples. Machine Learning (ML) algorithms based on the Decision Trees/ Random Forest approach will be developed to solve classification problems, for identifying the targeted analyte molecules (Azoxystrobin and Diflufenican) in complex environments as contaminated waters. The inputs will be LSPR spectroscopy data (transmittance spectra) and polarimetry data (Mueller matrix measurements). 4) Validation of the developed technology and paving the way for on-site identification of water contaminants using machine learning algorithms. The developed technology will be validated against standard techniques (chromatography-mass spectrometry methods) and compared to competing optical methods (SERS). Envisaging miniaturization and simplification of the hardware, a portable device prototype will be designed to enable on-site detection of targeted water contaminants (Azoxystrobin and Diflufenican), using the ML identification algorithm.