Article pubs.acs.org/est
Molecular Characterization of Dissolved Organic Matter through a Desalination Process by High Resolution Mass Spectrometry Nuria Cortés-Francisco and Josep Caixach* Mass Spectrometry Laboratory/Organic Pollutants, IDAEA-CSIC, Jordi Girona 18, 08034 Barcelona, Spain S Supporting Information *
ABSTRACT: The effect of different water treatments such as ultrafiltration (UF) and reverse osmosis (RO) on dissolved organic matter (DOM) is still unknown. Electrospray ionization Fourier transform orbitrap mass spectrometry has been used to provide valuable information of marine DOM evolution through a desalination process on a molecular scale. In the present manuscript, the characterization of four real composite water samples from a desalination pilot plant installed in the coast of Barcelona (Spain) has been carried out. The sampling was performed on each point of the pilot plant: raw seawater (RSW), UF effluent, brine RO and permeate RO. The mass spectra of the different samples show several thousand peaks, however for the present screening study, only the mass range m/z 200−500 and the main signals in this mass range (relative intensities ≥1%) have been considered. The analysis of RSW and UF samples reveal that there is little effect on DOM by the UF pilot. However, when the water is treated on the RO an important change on DOM has been observed. The recurring periodical patterns found in RSW and UF are lost in Permeate RO sample. Compounds with more aliphatic character, with higher H/C ratio (H/Cav 1.72) are present in the Permeate and some of them have been tentatively identified as fatty acids.
■
INTRODUCTION
In the last years, a desalination treatment plant which takes the water from the Mediterranean Sea and supplies fresh water to the city of Barcelona has been built.9 In parallel to the construction of the desalination plant there was a desalination pilot plant installed, in order to investigate in the pretreatments and the reverse osmosis processes with the same real water intake. Our goal was to characterize the molecular composition of water-soluble organic matter from Mediterranean seawater and see the evolution of DOM through a desalination pilot plant. In this study, DOM has been studied by use of electrospray ionization high-resolution Fourier transform orbitrap mass spectrometry (ESI LTQ-FT-Orbitrap-MS), due to the fact that high resolution mass spectrometry (HRMS) is one of the techniques that has given more information of the molecular species, exact masses and molecular formulas of DOM.8 Due to the difficulty of data interpretation and limitations of the technique (lacking ability for identification of isomers, limited resolution) a strict protocol has been applied to ensure reliable results. To the best of our knowledge, no studies have evaluated this evolution to understand the changes in DOM through a desalination process, which is a critical aspect to better describe pretreatment and reverse osmosis processes.
In the last years, there has been a shortage of fresh water, due to overuse and misuse. Seawater and saline aquifers account for 97.5% of the Earth’s water resources and represent a potential source of drinking water. Thus, desalination techniques involving seawater reverse osmosis (RO) have emerged as important candidates.1 The properties and amount of dissolved organic matter (DOM) in natural waters can significantly affect the production of drinking water.2−4 It may cause adverse aesthetic qualities such as color, taste, and odor, but also it can affect the performance of water treatment processes, such as activated carbon adsorption, ozonation, or membrane treatments. Moreover, DOM can serve as main precursor of disinfection byproducts (DBPs) during chlorination.5,6 DOM is an extremely complex mixture of organic compounds and the amount and characteristics of it depends on climate, geology, and topography.7 Marine DOM is found at very low concentrations in comparison to huge amounts of inorganic salts, so the removal of DOM and salts to obtain fresh water from seawater represents a real challenge. In order to improve and optimize removal processes of this organic matter, the characterization of DOM at different purification and treatment processes stages is important.2 In treatment lines, the characterization usually consists on bulk parameters such as total organic carbon (TOC), dissolved organic carbon, and spectroscopy techniques such as, ultraviolet and visible absorption spectroscopy, specific UV-absorbance, and excitation emission matrix fluorescence spectroscopy.8 © 2013 American Chemical Society
Received: Revised: Accepted: Published: 9619
January 3, 2013 July 17, 2013 July 24, 2013 July 24, 2013 dx.doi.org/10.1021/es4000388 | Environ. Sci. Technol. 2013, 47, 9619−9627
Environmental Science & Technology
■
Article
EXPERIMENTAL SECTION Chemicals. All reagents were of analytical or highperformance liquid chromatographic grade. Dichloromethane, methanol, and hydrochloric acid were purchased from Merck (Darmstadt, Germany). Isopropyl alcohol was from Carlo Erba (Milan, Italy), and formic acid was from Panreac (Barcelona, Spain). Highpurity water produced with a Milli-Q Organex-Q System Millipore (Millipore Corp., Bedford, MA) was used. The internal standards sodium dodecyl-d25 sulfate was purchased from CDN Isotopes (Quebec, Canada) and the dioctyl sodium sulfosuccinate was purchased from SigmaAldrich (Schnelldorf, Germany). Suwannee River Fulvic Acid (SRFA) (1S101F) and Suwannee River Natural Organic Matter (SRNOM) (1R101N) standards from International Humic Substance Society (Minnesota, United States) have been diluted in methanol and analyzed to find common components between the real samples and the standards. Sample collection and preparation. Four real composite water samples from a pilot desalination plant installed in the coast of Barcelona (Spain) were collected in Pyrex borosilicate amber glass bottles. The seawater intake system is located 2200 m from the shoreline and 31 m below sea level in the Mediterranean Sea.9 Raw seawater (RSW) is passed through an out/in ultrafiltration (UF) hollow fiber membrane (polyvinylidene fluoride; 0.02 μm nominal pore size). The UF effluent is then passed through 5 μm security cartridge filters and fed through a RO module (thin film composite membrane operating at 14 Lm −2 h −1 and 45% of recovery). The sampling was performed on each point of the pilot treatment plant: RSW, UF effluent, brine RO, and permeate RO. Samples were stored at 4 °C and analyzed within 48 h. We acidified 1.5 L from each sample with 10% hydrochloric acid to pH 2 and extracted with 2 × 100 mL of dichloromethane/isopropyl alcohol (90:10 v/v) and the extracts were concentrated down to 250 μL at 40 °C under nitrogen. 100 fold dilution and reanalysis was performed to the brine RO sample. Sodium dodecyl-d25 sulfate and dioctyl sodium sulfosuccinate were added as internal standards used as lock mass for internal calibration of the spectra. This extraction protocol was chosen in order to manipulate as less as possible the real nature of the water, and analyze the whole water without prior filtration, as well as to eliminate any remaining salts which would suppress the ion generation within electrospray. LTQ-Orbitrap Mass Spectrometry. Flow injection analysis (FIA)10 of 10 μL of the standards and the samples has been carried out with a LTQ-Orbitrap XL (Thermo Fisher Scientific, Bremen, Germany) equipped with an ESI source. The LC system consists of a Surveyor MS Plus pump and a Micro As autosampler (Thermo Fisher Scientific, San Jose, California). The mobil phase was methanol:water (80:20) at 50 μL/min. All mass spectra were acquired in negative ionization mode produced by an ESI source voltage of 3.00 kV, capillary voltage −35 V and tube lens −90 V, after ion source parameters optimization. The sheath gas flow was set at 20 and aux gas flow 5, both arbitrary units and capillary temperature of 300 °C. Negative ionization was used, because it has been reported to produce more ions from marine DOM samples11,12 and we were not specifically searching for compounds containing nitrogen, which then positive mode would have been more appropriated. The mass range was m/z 50−1200, because prior analysis showed no signals in a higher m/z range.
A protocol has been developed to better evaluate the uncertainty of the accurate mass measurements and to build the van Krevelen plots in a systematic way, so that the different data from each sample can be comparable. The first step was to evaluate the uncertainty of the accurate mass measurements of the mass spectrometer, based on prior experience.13 Internal and external calibration experiments were performed taking into account the internal standards added (data not shown). Finally, the spectra were internally calibrated using as lock masses sodium dodecyl-d25 sulfate (m/z 290.3048) and dioctyl sodium sulfosuccinate (m/z 421.2265) added during the sample treatment. This way, the accuracy and precision in the relevant mass range (m/z 200−500) was always
