MSFig. 1. Summary of the procedural performance of the integrated screening met hod applied to the experiment with initial concentration of 500 g L−1. In step I, number srefer to the peaks remaining after each step; in step II, numbers refer to the peak intensity values for each entry; in step III, numbers before and after the slash refer to negative and positive ionizat ion mode, respectively, numbers in the square bracket srefer to the same compounds detected in the opposite ionization mode, numbe rsin round brackets refer to fragments and/or adducts, an example is shown in the figure data, as illustrated in Fig. 1 for the experiment with an initial concentration of 500 g L−1. In both ionizat ion modes, step I eliminated approximately 97% of the detected peaks, with a majority being class ified as noise (i.e. intensity below threshold). 80% of there maining peaks were discarded based on the time-trend criterion.Overall, less than 1% of the initial peaks were passed on to step III for further processing and structure assignment. Eleven promising components corre sponded to a UM-PPS prediction and another 5 corres ponded to a literature record. Only 4 candidate components in negative mode and 2 in positive mode (Fig. S2, Supple mentary Material) remain unknown. No prediction could be matched toone of these components, and no characteristic diagnostic struc-tural features (e.g., neutral CO2-loss) were observed in the mass spectra that could provide a link to a specific parent compound.For these potential TPs, elemental compositions were generated from their measured exact mass and isotopic patterns by the to olimplemented in the instrument software (Tab. S3, Supplementary Material). Incubation experiments with individual pharmaceuticals would be necessary for further exploring the identity of these TPs.