Abstract: Marine aerosols have extensively been monitored in the framework of a research project to investigate the inorganic and organic nutrient fluxes of the atmospheric nitrogen input into the North Sea. As a part of this project bulk analysis was also performed by energy dispersive X-ray fluorescence (EDXRF) on a total of 366 samples in order to provide detailed elemental composition. Samples have been clustered per campaign and differences were related to backward trajectories. Elements as Cl, Si, and S appeared to be the best tracers for coastal aerosols. Therefore they have been chosen to represent each campaign and to map the seasonal variations. It is clear that Cl is an excellent tracer for oceanic trajectories, while a less clear correlation between the soil dust components and the air mass trajectories is found.

Abstract: Illicit drug consumption is a worldwide worrying phenomenon that troubles modern society. For this reason, law enforcement agencies (LEAs) are placing tremendous efforts into tackling the spreading of such substances among our community. New sensing technologies can facilitate the LEAs duties by providing portable and affordable analytical devices. Herein, we present for the first time a sensitive and low-cost electrochemical method, i.e. square-wave adsorptive stripping voltammetry on carbon screen-printed electrodes (SPE), for the detection of five illicit drugs (i.e. cocaine, heroin, 3,4-methylenedioxymethamphetamine, 4-chloro-alpha-pyrrolidinovalerophenone, and ketamine) in oral fluid by the aid of a surfactant. Particularly, the surfactant is adsorbed at the carbon electrode’s surface and yields the adsorption of illicit drug molecules, allowing for an enhanced electrochemical signal in comparison to surfactant-free media. First, the surfactant-mediated behavior is deeply explored at the SPE by cyclic voltammetry, electrochemical impedance spectroscopy, and Fourier-transform infrared spectroscopy. Subsequently, the electrochemical behavior of the five illicit drugs is studied and optimized to render optimal analytical performance. Accordingly, the analytical system exhibited a wide linear concentration range from 1 to 30 µM with sub-micromolar limits of detection and high sensitivity. This performance is similar to other reported electrochemical sensors, but with the advantage of using an unmodified SPE, thus avoiding costly and complex functionalization of the SPE. Finally, the methodology was evaluated in diluted oral fluid samples spiked with illicit drugs. Overall, this work describes a simple, rapid, portable, and sensitive method for the detection of illicit drugs aiming to provide oral fluid testing opportunities to LEAs.

Abstract: Shelf sediments of the southern North Sea, were studied with a microanalytical [electron probe X-ray microanalysis (EPXMA)] and two bulk [X-ray diffraction (XRD) and X-ray fluorescence (XRF)] techniques. The investigation proved that the promptness of the microanalytical method is combined with a reasonable analytical reliability. XRD studies of such a type of sediments with monotonous mineral composition are not able to provide mineralogical information beyond the main well-crystalline minerals and the mineralogical quantitative characteristic of the sediment based on XRD estimations are incorrect. The EPXMA mineralogical interpretations are based on the statistical evaluation of a huge data set (thousands of mineral particles) and provide a rather correct quantitative determination of the main minerals. The comparative EPXMAXRF study revealed that the Al, Si, K, Ca, Fe and to some extent Ti contents estimated by EPXMA are fairly reliable. In this respect the accuracy of the EPXMA-based mineral identification of the pure silicates, pure aluminosilicates, and Al-, Ca-, Fe- and Ti-containing minerals with simple composition is very high. Mg-calcite, augite and apatite determinations are assessed to be correct. The supposed accuracy of the clay mineral determinations is slightly lower (7080%) than that of the other main minerals due to the complex and varying composition of the clays. The identification of XRD-invisible accessory minerals and quantification of their presence in the sediments is an essential advantage of the EPXMA, which makes it a useful approach in tracing the origin of the sediments, the pathways of their transport and the geochemical processes they have undergone. However, the EPXMA has several flaws, which need to be solved in the future sediment investigations: (1) calibration with natural standards is needed in order to provide a higher accuracy of the mineral determinations; (2) any EPXMA study of sediments needs to be secured with XRF examinations of selected samples since EPXMA gives only semi-quantitative information about the abundance of the elements; (3) ultra-thin window EPXMA of low-Z elements has to be used since some of them (O, C) are always present in the main sediment components: silicates, aluminosilicates, carbonates and metal oxyhydroxides; (4) the interpretations of the clay fraction have to be supported with detailed XRD investigations of selected samples, while the mineralogy of the silt and sand fractions needs to be backed up with optical microscopy studies. The information from different analytical techniques (EPXMA with XRFXRD-optical microscopy of selected samples) combined with the knowledge about the most possible minerals in a given environment, would give the most reliable results in studying mineralogical composition of shelf sediments.

Abstract: The added value of supervised Machine Learning (ML) methods to determine the Absolute Configuration (AC) of compounds from their Vibrational Circular Dichroism (VCD) spectra was explored. Among all ML methods considered, Random Forest (RF) and Feedforward Neural Network (FNN) yield the best performance for identification of the AC. At its best, FNN allows near-perfect AC determination, with accuracy of prediction up to 0.995, while RF combines good predictive accuracy (up to 0.940) with the ability to identify the spectral areas important for the identification of the AC. No loss in performance of either model is observed as long as the spectral sampling interval used does not exceed the spectral bandwidth. Increasing the sampling interval proves to be the best method to lower the dimensionality of the input data, thereby decreasing the computational cost associated with the training of the models.