Optimizing the sensitivity of a non-targeted LVDI-LC-HRMS platform for environmental applications by Design of Experiment

Non-targeted analytical approaches based on liquid chromatography coupled with high-resolution mass spectrometry (LC-HRMS) are gaining attention for the comprehensive detection and identification of chemicals of environmental concern (CECs) occurring in environmental samples, alleviating the need for reference standards [1]. LC-HRMS non-targeted approaches are playing a relevant role in monitoring CEC occurrence in surface waters and effluents from wastewater treatment plants, in both of which transformation products (TPs) can occur after chemical, photochemical, and biological degradation processes. Consequently, non-targeted approaches tend to lower selectivity by reducing sample handling, without drastically affecting sensitivity often performing large-volume direct injection (LVDI) [2]. On this basis, a Design-of-Experiment (DoE) optimization workflow is here presented for increasing the sensitivity of a LVDI-LC-HRMS method, performed using a training set of 42 CECs (e.g., PFASs and estrogens) and TPs from 2013-2020 regulations and watch lists, covering a wide range of physicochemical properties, being thus characterized by different chromatographic and ionization behaviours. Chromatographic analysis was performed in reversed phase mode on a F5 Kinetex analytical column coupled with an Orbitrap Exploris 120 mass spectrometer by an Ion Max Source operating in ESI mode. DoE was implemented evaluating 4 critical factors for the in-source ionization process: (i) acidity of aqueous eluent, expressed as percentage of formic acid, (ii) eluent modifier, expressed as mM concentration of ammonium formate, (iii) source ion spray voltage (kV), and (iv) temperature (°C), each one investigated at 3 levels through a D-optimal design (n=3 center points), considering mixed and two- factors interactions. The dependent variables, i.e. the averaged chromatographic areas of critical CECs in the investigated training set, were fitted using PLS algorithm, and the obtained models transformed and refined where needed. For the optimization step, only a reduced set of responses was kept. In fact, chromatographic areas that exhibited a non-significant and non-valid model were removed to reduce the uncertainty in prediction. The identified optimal conditions for the LVDI-LC- HRMS (aqueous eluent = 0.057%, eluent modifier = 9,7 mM, ion spray voltage = 2,5 kV, and ion spray temperature = 348 °C) method will be applied, as a further perspective, to the non-targeted analysis of surface and ground water samples related to the potabilization and depuration management of several urban areas in Tuscany (Italy), in order to identify known and unknown CECs and TPs that can potentially impact drinking water safety.