Non-Targeted and Suspect-Screening of Reuse Water by Large-Volume Injection Liquid Chromatography and High Resolution Mass Spectrometry.

Advances in High Resolution Mass Spectrometry and its Emerging Environmental Applications
Oral Presentation

Prepared by W. Backe
Minnesota Department of Health, 601 Robert Street North, Saint Paul, Minnesota, 55155, United States


Contact Information: Will.Backe@state.mn.us; 651-201-4864


ABSTRACT

Seven source-water samples for water reuse applications and one city-supply water sample were collected to determine the presence of chemical contamination and to determine bacterial and virus activity. The source-water included stormwater, rooftop runoff, wastewater, drinking water treatment back-flush water, and mixed sources. The source-water was reused mainly for irrigation and in one instance cooling water for a power plant. To determine chemical contamination, non-targeted and suspect screening analysis of the samples was performed using large-volume (700uL) direct-injection high-performance liquid chromatography (LVI-HPLC) and quadrupole time-of-flight mass spectrometry. Virus and bacterial activity was determined using microfluidic quantitative polymerase chain reaction to look for genetic markers for human pathogens.

Two LVI-HPLC methods were developed for the separation of positively and negatively ionizing compounds, respectively. The LVI-HPLC method for positively ionizing compounds was successfully tested on 59 suspect analytes ranging in polarity from log Kow -0.78 to 5.0. Similarly, the chromatographic method for negatively ionizing compounds was tested on 56 suspect analytes ranging in polarity from log Kow -2.4 to 5.8. LVI-HPLC facilitates non-targeted analysis of samples without the need for multiple solid-phase extraction techniques which are typically used to ensure the representative extraction of cationic, anionic, polar, and non-polar compounds. Retention time models were developed using multiple linear regression for the two separation methods as an additional metric of confidence when identifying unknown compounds. The two models have predictive R2-- which indicates how well the models predicts new observations—of 0.88 and 0.87, respectively.

After data reduction, the number of chemical features detected in the samples ranged from 74 to 945 in positive ionization and from 149 to 567 in negative ionization. A principle component analysis of the chemical features shows that sample tend to cluster by source-water type. Of the 59 suspect-screening compounds analyzed by positive ionization mode, 12 were positively identified. Of the 56 suspect-screening compounds analyzed by negative ionization mode, 13 were positively identified. Benzotriazole derivatives and PFAS were the most commonly detected compounds during suspect screening. Other compounds detected by suspect screening included pharmaceuticals, drug metabolites, x-ray contrast media, and wastewater indicators.

In positive ionization mode, 59 non-target compounds were tentatively identified. Of those 59 compounds, 42 were confirmed by standards, 7 were rejected by standards, and for 10 compounds standards were not able to be acquired. N,N-dibenzylformamide (n=6), metalachlor (n=4) and 2-hydroxy atrazine (n=4) were the most commonly detected. In addition, 14 pesticides or pesticide transformation products and 15 pharmaceuticals or pharmaceutical metabolites were identified.

In negative ionization mode, 31 non-target compounds were tentatively identified. Of those 31 compounds, 18 were confirmed by standards, 7 were rejected by standards, and 6 could not be assessed. Of the tentatively identified compounds, N,N’-ethylene distearylamide (n=7), dodecyl phosphate (n=6), and (3,4-Dichlorophenoxy)acetic acid (n=5) were the most commonly detected. Six pesticides or pesticide transformation products and four pharmaceuticals were also identified. Collaborators are currently working on the bacterial and virus data, which will be presented alongside the chemical data if available.