Non-targeted analysis with high-resolution mass spectrometry for investigation of riverbank filtration processes

Abstract

A fully non-targeted analytical workflow for the investigation of a riverbank filtration site located at the river Danube has been developed and applied. Variations of compound intensities at different sampling locations of the riverbank filtration site and, for a single production well, over a monitoring period of one year have been investigated using liquid chromatography combined with time-of-flight-mass spectrometry followed by evaluation via non-targeted data analysis. Internal standardization and appropriate quality control strategies have been implemented into the workflow for reduction of possible methodological biases influencing data interpretation. Emphasis was placed on the assessment of different blank elimination steps and the final blank elimination strategy is reported. The spatial study of the selected riverbank filtration site revealed a homogenous composition of the filtered water sampled at 11 different locations across the 32,000 m² site, except for one sampling location in a zone of the aquifer, which was only weakly connected to the well field in terms of hydrogeological conditions. The examination of time-dependent changes of the composition of surface and groundwater obtained at the riverbank filtration system revealed that the non-targeted workflow is fit-for-purpose regarding the assessment the stability of filtration efficiency and compound residence time in the riverbank filtration compartment. In total, 677 compounds were selected for the investigation of the time-dependent variations of the filtration process. Analysis of the signal intensities of these compounds revealed that the riverbank filtration is significantly reducing the intensity and number of compounds present in surface water over a wide polarity range. In addition, the method enabled the determination of compound residence times in the riverbank filtration system ranging from 5 to 7 days.

Keywords: High-resolution mass spectrometry; Non-targeted analysis; Riverbank filtration; Solid-phase extraction; Surface water/groundwater interactions.

Fig. 1

Map of the investigated riverbank filtration site

Fig. 2

Effects of compound filtering steps in NTA of water samples. a Percentage and number of compounds classified according to the applied filtering step; b number of single detections from every sample classified according to the applied filtering step; (A) spatial study, positive ionization mode; (B) spatial study, negative ionization mode; (C) temporal study, positive ionization mode; (D) temporal study negative ionization mode. It is noteworthy that the low compound number in spatial analysis a and b is due to the sample collection from 11 sampling locations on a single day, while in temporal analysis, 25 samples were collected in one year

Fig. 3

Number of substantiated compounds detected at different sampling locations in spatial study of the riverbank filtration site

Fig. 4

Physicochemical parameters determined in-situ in the investigated groundwater samples. The precision of the conductivity, oxygen, and redox-potential measurement are ± 0.5% RSD, ± 0.5% RSD, and ± 0.2 (mV), respectively

Fig. 5

Substantiated compounds detected in the temporal study of the riverbank filtration site (positive ionization mode)

Fig. 6

Heat map of 112 substantiated compounds remaining in the dataset after the prioritization step described above (detection in groundwater in at least 60% of the samples). Each column represents a sampling day. Surface water samples obtained from the river Danube (a) were compared to the corresponding groundwater samples obtained from the investigated riverbank filtration site (b). The color range represents the abundance of the substantiated compounds (blue = low abundant, red = high abundant)

Fig. 7

100% stacked column chart of the detected compounds categorized according to relative fold changes between surface water (SW) and groundwater (GW). For interpretation of the legend, see the “Assessment of riverbank filtration efficiency” section