Environ. Sci. Technol. 2004, 38, 4134-4139
Odorous Products of the Chlorination of Phenylalanine in Water: Formation, Evolution, and Quantification I N G R I D F R E U Z E , * ,† S T EÄ P H A N B R O S I L L O N , † DORINE HERMAN,† ALAIN LAPLANCHE,† C H R I S T I A N D EÄ M O C R A T E , ‡ A N D JACQUES CAVARD§ Laboratoire de Chimie des Nuisances et Ge´nie de l’Environnement, Ecole Nationale Supe´rieure de Chimie de Rennes, Av. du Ge´ne´ral Leclerq, 35700 Rennes, France, Veolia Water, Compagnie Ge´ne´rale des Eaux, Immeuble Le Carillon, 6 Esplanade Charles de Gaulle, 92751 Nanterre, France, and Syndicat des Eaux d’Ile de France (SEDIF), 14 rue Saint Benoıˆt, 75006 Paris, France
To explain some of the possible origins of an odor episode which took place in a drinking water supply in the region of Paris (France), the chlorination reaction in water of phenylalanine was studied. This amino acid was chosen for first experiments because of its physical and chemical particular properties. Changes in the different byproducts formed were followed by high-performance liquid chromatography (HPLC) over a period of time. N-chlorophenylalanine (mono-N-chlorinated amino acid) and then phenylacetaldehyde were the major products formed for the lower chlorine to nitrogen molar ratios. For Cl/N molar ratios of 1 and beyond, phenylacetonitrile and N-chlorophenylacetaldimine appeared and increased with the chlorination level. N-chlorophenylacetaldimine was quantified by using its difference of stability in various organic solvents. Our attention was first directed to the monochlorinated derivative but further examination indicated that it could not be responsible for odor troubles: it dissociated before reaching the consumer’s tap and it was produced at consistently low yields under conditions relevant to drinking water treatment. On the contrary, chloroaldimine appeared to be a very odorous and water-stable product: it strongly smells of swimming pool with a floral background. The odor detection threshold is about 3 µg‚L-1 and it can persist for more than one week at 18 °C. It is now suspected of being a source of off-flavor concerns among consumers.
Introduction Drinking water disinfection is necessary for eliminating pathogenic organisms. Indeed, it must be without adverse effect on any form of life. Chlorine is one of the most important products used for chemical disinfection of drinking water because it is effective, cheap, and readily available. Moreover, it is one of the only disinfectants to have retentive * Corresponding author phone: (33) 223238048; fax: (33) 223238120; e-mail:
[email protected]. † Ecole Nationale Supe ´ rieure de Chimie de Rennes. ‡ Veolia Water. § Syndicat des Eaux d’Ile de France (SEDIF). 4134
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 15, 2004
power. But as an oxidizing disinfectant, chlorine can also react with natural organic matter (NOM) present in drinking water (1). The organic compounds composing NOM (polysaccharides, amino sugars and amino acids, proteins, and polyhydroxyaromatic compounds) can produce undesirable disinfection byproducts. Great concern has been focused on the health risks of drinking water because some of the disinfection byproducts have been linked to cancer in animals or are suspected to have possible reproductive and development effects (2), but little attention has been paid to offflavor properties of treated water. However, drinking water also needs to have good organoleptic properties. It should not present any odor, color, or taste because the consumer often uses sensory evaluation for judging water quality, assuming, most often wrongly, that water with an unpleasant smell is unsafe for consumption (3). A better understanding of the chemical causes of drinking water odor is by this fact necessary for water operators. The composition of water before chlorine addition has a great incidence on the formation of disinfection byproducts. Our study was financially supported by the Mery-sur-Oise treatment plant, located in the north suburbs of Paris. The study was based on an isolated odor episode following the implantation of a new membrane process. Part of the water (70%) from this plant has been nanofiltered since 1999 (4). So only very little organic matter is able to pass the membrane barrier. Only low-molecular-weight compounds such as amino acids (5) and short peptides can pass through the membrane. Now odorous chlorinated products, for example, are formed especially with the amino acids encountered at low concentration in water, under a dissolved free form or even more as little peptides. The major amino acids present in treated water are alanine, glycine, valine, phenylalanine, serine, threonine, isoleucine, aspartic acid, tyrosine, proline, glutamic acid, and leucine (6). Amino acid concentrations in drinking water range from 0.33 to 1.05 µg‚L-1 (7). According to Dossier Berne (8), free amino acid concentrations determined in several drinking water factories of western France would vary between 0 and 30 µg‚L-1. It is well-known that the reaction of aqueous chlorine with amino acids leads to the formation of N-chloramines, nitriles, and aldehydes (9), which are most often odorous products and can cause some odor problems in distributed water (10). Indeed, amino acid chlorination is always a subject of great concern and has received attention for different reasons: chloramines are believed to be toxic to some fishes (11, 12) and are also strongly suspected of interfering with the disinfectant concentration measurements in water (13-15). Several studies (16-20) presented a new scheme for the pathway of amino acids chlorination. They demonstrated that N-chloroaldimines can be produced during this reaction but no information was available for their odor properties. The aim of this work was to identify the byproducts formed during the chlorination reaction of amino acids, which are suspected of being odorous. It has been known for a long time that aldehydes have weak odor detection thresholds, closed to µg‚L-1 concentration (21). The kind of odor presented by these aldehydes is related to their molecular weight: low-molecular-weight aldehydes present unpleasant odors and high-molecular-weight aldehydes present floral or fruity odors (22). Our study focused on the chloroaldimine, for which no odor description was available in the literature but which rapidly appeared odorous at low concentrations. In this way, the amino acid chlorination reaction was studied to determine the odor-producing potential of this product in the drinking water industry. 10.1021/es035021i CCC: $27.50
2004 American Chemical Society Published on Web 06/22/2004
FIGURE 1. Simplified scheme for chlorination of phenylalanine, adaptedfrom Conyers and Scully (18). Principal byproducts formed. Even if it is not the most abundant amino acid in water, phenylalanine was chosen for its UV-absorbing properties. This enabled us to follow the chlorination reactions and the appearance of byproducts easily. Moreover, this product has only one nitrogen atom, which restricts its reactivity with chlorine and thus the complexity of the reaction mixtures obtained. Various Cl/N molar ratios and various reaction times were studied (Figure 1). Other amino acids will be studied to extend the results.
Materials and Methods Reagents and Apparatus. D,L-Phenylalanine (99%) was obtained from Jansen Chimica, phenylacetonitrile (98-99%) was from EGA Chemie, and phenylacetaldehyde (90%) was purchased from Aldrich. The sodium hypochlorite solution (2 mol‚L-1 minimum) was purchased from Prolabo and its concentration in active chlorine was regularly checked by thiosulfate titration (23) to record the decay rate of the solution in consideration. All the buffers and solutions were prepared in demineralized Millipore water with no chlorine demand (24). The free and combined chlorine were determined by reaction with N,N-diethyl-p-phenylenediamine (DPD) in absence or presence of iodide (24). Indeed, DPD gives a specific pink color in the presence of chlorine, the intensity of which can be measured by spectrophotometry (λ ) 515 nm). Addition of potassium iodide crystals is necessary to reveal combined chlorine but this method cannot differentiate organic and mineral chloramines. Dechlorination was realized by addition of 3 equiv (with regards to the quantity of chlorine introduced) of sodium thiosulfate 0.1 mol‚L-1 and 15 min of good stirring. Measurements of the optical density for the determination of free and combined chlorine with DPD titration were made with a spectrofluorimeter Seconam S.750. This titration was available for chlorine concentrations between 0.05 and 4 mg Cl2‚L-1. The equation of the curve was OD (optical density) ) 0.1806C (C was the chlorine concentration in mg Cl2‚L-1) with a correlation coefficient of 0.9939. Nuclear magnetic resonance (NMR) data were obtained with a Brucker FT ARX (1H 400.13 MHz and 13C 100.61 MHz). HPLC was performed using an Alliance Waters 2690 separation module with a Waters 2487 UV detector (λ ) 254 nm). A Waters fraction collector II was added at the outlet of the HPLC apparatus. The majority of the analyses were first performed after 30 min of contact time and then at various intervals during the week with an injection volume of 250 µL. Separations were carried out on a C18 Symmetry column (4.6 × 250 mm) maintained at 30 °C, with a dual solvent system. Solvent A was 90% water (pH 4 adjusted with acetic acid) and 10% acetonitrile and solvent B was 90% acetonitrile and 10% water (pH 4 adjusted with acetic acid). The solvent program (flow rate of 1 mL‚min-1) consisted of a 5-min isocratic elution with 85% A/15% B followed by a linear gradient to 55% A/45% B over 20 min. Then a second linear gradient was applied to
10% A/90% B over 15 min and a 5-min isocratic elution with 10% A/90% B. These conditions were determined by Nweke and Scully (16). Chlorination and Identification of the Byproducts Formed. Chlorinations of phenylalanine were first conducted in both phosphate buffer (pH 7) and water for Cl/N molar ratios of 1 and 2. Results obtained were similar. The same final products were formed and reached the same levels of concentration. So all experiments were then carried out in water to simplify the procedure. These conditions were also chosen to ensure that the pH conditions were quite similar to industrial conditions. Model solutions of phenylalanine (50 mL, 10-3 mol‚L-1) were chlorinated to different chlorine to nitrogen molar ratios by addition of micro volumes of the hypochlorite solution, incubated in the dark at room temperature, and regularly analyzed by HPLC-UV. Cl/N molar ratios of 0.5, 1, 1.5, and 2 will be presented here. All the compounds formed were collected with the fraction collector. Once the compounds were isolated, their flavor characteristics were determined and their oxidizing ability was tested to determine if they were N-chlorinated products: fractions obtained were mixed with various quantities of DPD, phosphate buffer (0.5 mol‚L-1, pH 6.3), and potassium iodide (24). The N-chlorophenylacetaldimine was synthesized. A sodium hypochlorite solution (50 mL, Cl/N molar ratio of 2.4, providing the best conditions for chloroaldimine formation, as indicated below) was cooled in an ice bath, and phenylalanine (0.01 mol‚L-1, 1.65 g) was added while the mixture was stirred. After 10 min, the reaction mixture, which became cloudy, was extracted with 3 mL of CDCl3, dried over anhydrous Na2SO4, and analyzed by both NMR and HPLC. The stability of chloroaldimine in various solvents was studied to establish an extraction method for this product and furthermore to quantify it. A solution of phenylalanine (50 mL, 0.2 mol‚L-1) was chlorinated with a Cl/N molar ratio of 2.8 and fractionated. Solvent (100 mL ofwater, hexane, methanol, acetonitrile, or dichloromethane) was added to 10 mL of the reaction mixture. After 24 h, the solution was analyzed by HPLC (injection of the organic phase for hexane and dichloromethane). Because N-chlorophenylacetaldimine is not commercially available, a quantification method was investigated. Syntheses with different initial concentrations in phenylalanine (0.01, 0.02, 0.05, 0.1, and 0.2 mol‚L-1) were realized, with a Cl/N molar ratio of 2.4. Once the reaction stabilized, the solutions were shared out into two equal parts and each of them was percolated on a SepPak Cartridge C18, washed with equal quantities of water, and eluted with the same volume of solvent (5 mL approximately), hexane for one sample and acetonitrile for the other. To confirm the validity of the extraction protocol, the one in acetonitrile was injected immediately after concentration: it showed exactly the same quantity of aldehyde, nitrile, and chloroaldimine as the hexane extract. The synthesis of imine was realized by an organic way. Phenylacetaldehyde (46 mg, 0.38 × 10-3 mol‚L-1) was introduced into 50 mL of anhydrous methanol. This solution was added to a solution of 5 equiv of ammonium chloride (NH4Cl, 95 mg) in an additional 50 mL of methanol. Molecular sieves were used to absorb the water formed during the reaction. The solution was mixed with a magnetic stirrer for 1 h, then the reaction mixture was injected on HPLC. Olfaction Tests. The olfaction tests were realized with standard compounds for aldehyde and nitrile and with chlorinated amino acid solutions (Cl/N 2.4) for the chloroaldimine. In these conditions, N-chloroaldimine concentration was considered to be 35% of the initial amino acid concentration. The following laboratory-made protocol was employed. First, all the glass was washed with an odorless VOL. 38, NO. 15, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4135
FIGURE 2. High-performance liquid chromatogram on C18 column with UV detection of a solution of phenylalanine (1 mM) Cl/N ) 1.5 after 30 min chlorination. For chlorination rates close to 1 and for reaction times of several hours, another product was observed with a retention time of 39 min.
TABLE 1. Stable Levels Reached by Phenylalanine, Phenylacetaldehyde, and Phenylacetonitrile for Different Cl/N Molar Ratios of a Phenylalanine Solution (10-3 mol‚L-1) at 22 °C Cl/N ) 0.5 [Phe] (10-3mol‚L-1) 0.5 0.42 [Ald] (10-3mol‚L-1) [Nit] (10-3mol‚L-1) undetected
Cl/N ) 1 Cl/N ) 1.5 Cl/N ) 2
