Dottorato in SCIENZE FARMACEUTICHE

Titolo della tesi: THALLIUM INTERACTION WITH ORGANIC NATURAL MATTER IN THE DRINKING WATER

Thallium is an element widely distributed in nature even though its abundance is very low. Although its presence in drinking water distribution systems (WDSs) is usually negligeable, extremely rare cases of contamination, resulting from its occurrence in raw water, has aroused concern in consumers due to its toxicity. Despite the growing attention of the scientific community on its environmental fate and behavior, the extent of the role played by natural organic matter (NOM) on its mobility and migration has been poorly explored. NOM could play an important role in Tl bioavailability, chemical reactivity and transporting in aquatic systems. The aim of this study was the analysis of possible interactions between Tl and NOM extracted, purified and pre-concentrated from surface water samples by an innovative procedure developed during the PhD thesis. In September 2014, a severe contamination was detected in two WDSs of Pietrasanta (Lucca, Tuscany, Italy) due to the presence of Tl in a water supply source located near a dismissed mining site. After the hydraulic disconnection of the contaminated source, an unexpected increase in concentration (up to 60 µg/L) was detected due to thallium migration into water from the network pipelines where Tl had been accumulated over 50 years. To understand this phenomenon, it was necessary to characterize all the possible Tl fractions characterized by different migration kinetics. In this regard, a multi-sequential ultrasonic-assisted extraction was proposed by our working group to determinate different thallium fractions along the inner surface of the pipelines. Inorganic fractions release (i.e., Tl(I) and Tl2O3) was obtained by the first two phases of the procedure while organic fractions were supposed to be correlated to the release of thallium during the remaining part of the procedure. In the first phase of this work, the focus was on the optimization of the third step of the multiple sequential extraction procedure, which resulted in the release of thallium from organic matter in oxidizing conditions. These preliminary findings supported the possibility of interactions between thallium and natural organic substances in the distribution network. In the second phase, a procedure for the extraction and preconcentration of humic acids by non-ionic macroporous absorbing resins and freeze-drying treatment was applied. The concentrated humic acids were then placed in contact with thallium oxide and analyzed by size exclusion chromatography coupled with inductively-coupled plasma mass spectrometry (SEC/ICP-MS), which revealed the presence of two forms of thallium (that is, free Tl (I) and an organic complex of Tl (III)). This result provided a first strong evidence supporting the hypothesis formulated in the previous phase. In the third phase, a procedure to isolate and preconcentrate most of the dissolved organic component from surface water while preserving its chemical composition was developed and optimized using pooled techniques. Indeed, a volume of water was microfiltrated to retain suspended solids, sand and bacteria. Then, a series of a complexing resin and a cation exchange resin was used to remove multivalent cations to prevent their competitive effects in thallium complexation as well as fouling phenomena in the following nanofiltration (NF). Finally, the latter technique was firstly used to clean-up and extract NOM and, subsequently, to separate Tl-NOM complexes obtained by sonicating Tl2O3 with the extracted NOM. The application of the third phase to real samples was preceded by an extensive study of the phenomena involved during the nanofiltration process and of the effects produced by the experimental conditions. Before assembling the NF device, its components (i.e., pipes, fittings, washers, and NF membrane) were subjected to a prolonged cleaning procedure consisting in Soxhlet and ultrasonic extractions to significantly reduce organic release from organic polymers and rubbers. The procedure was optimized by trial and errors using UV spectrophotometry to check partial and final results. A short daily cleaning procedure was also developed to guarantee the absence of residuals and interferences immediately before the use of the NF device. The NF device mainly consisted of a membrane pump that made the aqueous sample to flow from a beaker to the inlet of the NF membrane holder at a controlled flow rate. The outlet of the holder corresponding to the retentate was throttled with a needle valve to fine-tune the flow rate and to set a counter-pressure at the NF membrane. Water flow coming from the retentate outlet was re-introduced into the beaker where distilled water was also continuously pumped inside to reintegrate the volume taken away at the permeate outlet of the NF device. During the NF process, aliquots of the permeate and feeding solution were collected for the determination of the monitored species (organics, Tl, anions and cations) by UV spectra, ICP-MS analysis, suppressed ion chromatography (SIC), and anodic stripping voltammetry (ASV). Temporal trends of the monitored species detected in the permeate and in the feeding solution were compared with a mathematical model developed to describe permeation kinetics of not-retained species. NF permeation studies and optimizations were carried out using aqueous solutions in the pH range of drinking water (6.5≤ pH ≤9.5) containing sodium and potassium salts of anions normally present in surface water, TlNO3, organic Tl(III) complexant agents (i.e., DTPA and commercial NOM extract), as well as their Tl(III) complexes. Tl(III) complexes were obtained by sonicating the complexing agents in the presence of Tl2O3. Since Tl2O3 is insoluble but prone to be reduced into Tl(I), the effect of operating conditions on Tl speciation was accurately controlled. The results obtained from the NF tests showed that this technique is able to isolate NOM from other electrolytes, which permeate through the NF membrane without interacting with it. Permeation kinetics seemed to be pH independent in the range allowed for the drinking waters and followed the mathematical model. As expected, Tl(III) complexes with DTPA or NOM were not able to permeate due to their molecular weight (MW) greater than the molecular weight cut-off (MWCO) of the NF membrane. Moreover, these complexes remained absorbed on the membrane surface as an effect of the action of dispersive forces. Anionic free complexant agents with MW lower than or comparable to MWCO (i.e., DTPA and small NOM components) were partially retained by the membrane by Colombian attraction since this kind of NF membrane is positively charged at the operating pH. At the same time, anionic free complexant agents with MW greater than MWCO (i.e., most of the NOM components) were completely retained by the NF membrane as an effect of size exclusion and Colombian attraction. Tl (III) organic complexes were quantitatively recovered from the NF membrane by pumping 2 L of demineralized water in counter-flow. Subsequently, anionic free complexant agents were desorbed by the membrane with 2 L of H2SO4 (at pH 3) in counter-flow. The whole procedure was applied to surface water samples collected from Bolsena lake, whose NOM was able to form stable complexes with Tl(III). These findings confirm that unstable species, such as Tl (III), may be solubilized by water-soluble NOM and carried inside a drinking-water distribution network. Their presence shall be investigated thoroughly to get information useful in the risk assessment of water contaminated by such elements.