Detection of pollutants in water represents a challenge for environmental monitoring. Even at ultratrace levels (defined as a concentration below 1 nmol/L or 1 µg/L depending on the unit), these pollutants can be harmful to living organisms in the short or long term. Among detection techniques, chromatography coupled with mass spectrometry stands out for its sensitivity on the order of ng/L. However, its high cost and the complexity of its implementation limit its use for the field. In this context, Surface-Enhanced Raman Spectroscopy (SERS) emerges as an alternative. The SERS technique relies on the amplification of the Raman signal of molecules near metallic nanostructures, acting as plasmonic resonators. This less expensive technique can acquire a SERS spectrum in a few seconds and is deployable for field applications. However, its main drawback remains its relatively low limit of detection compared to mass spectrometry (on the order of µg/L). In this doctoral project, we will use this technique to develop a sensor combining a superhydrophobic surface and plasmonic resonators, capable of analyzing organic compounds present at ultratrace levels within a micro-volume of liquid after the evaporation of a water droplet containing the compound to be detected. In an environmental context, we aim to apply this technology to water analysis to detect the presence of organic pollutants. The study was initially conducted on ultrapure water and then on drinking water. Work was carried out on the cleanroom fabrication of this sensor using silicon technology and on its reproducibility. We propose a methodology that allows the optimization of nanofabrication steps and ensures a high density of SERS signal enhancement hotspots on the analytical zone of the sensor. Experimental and spectral analysis methods were then developed to guarantee detection reliability and minimize interpretation errors. These methods include careful sample preparation, the use of multiple controls mimicking serial dilution steps, and the analysis of SERS spectra using statistical, probabilistic, and spectral classification methods based on a clustering algorithm. We evaluated the sensor’s performance for detecting two pharmaceutical molecules (paracetamol and aspirin), a dye (Rhodamine B), and a phthalate (phthalic acid). Phthalates are chemical compounds derived from phthalic acid, some of which are recognized as endocrine disruptors. The developed technology, coupled with the SERS spectral processing method, allowed the detection of pharmaceutical pollutants and phthalate at femtomolar concentrations, and the detection of Rhodamine B at attomolar concentrations in ultrapure water. Our method 5 was also used to detect the release of phthalic acid from polypropylene tubes, demonstrating its potential for use in laboratory quality control studies. Finally, although our methods were validated with laboratory-prepared solutions in which the target molecules were serially diluted in ultrapure water, initial field tests conducted on drinking water highlighted the difficulty of detecting pollutants such as Rhodamine B and aspirin at micromolar concentrations. These initial tests emphasize the importance of a preliminary purification step to mitigate matrix effects, which on our sensor induces a masking of the plasmonic resonators by salt crystals formed during evaporation.