Transcript
Available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/watres Kinetics of the oxidation of cylindrospermopsin andanatoxin-a with chlorine, monochloramine andpermanganate Eva Rodrı´ guez a , Ana Sordo a , James S. Metcalf b , Juan L. Acero a, a Departamento de Ingenieria Quimica y Energetica, Universidad de Extremadura, 06071 Badajoz, Spain b Division of Environmental and Applied Biology, School of Life Sciences, University of Dundee, Dundee DD1 4HN, Scotland, UK a r t i c l e i n f o Article history: Received 28 August 2006Received in revised form16 January 2007Accepted 26 January 2007Available online 13 March 2007 Keywords: CyanobacteriaCylindrospermopsinAnatoxin-aChlorinePermanganateRate constantsDrinking water oxidation a b s t r a c t Cyanobacteria produce toxins that may contaminate drinking water sources. Among others, the presence of the alkaloid toxins cylindrospermopsin (CYN) and anatoxin-a(ANTX) constitutes a considerable threat to human health due to the acute and chronictoxicity of these compounds. In the present study, not previously reported second-orderrate constants for the reactions of CYN and ANTX with chlorine and monochloramine andof CYN with potassium permanganate were determined and the influence of pH andtemperature was established for the most reactive cases. It was found that the reactivity of CYN with chlorine presents a maximum at pH 7 (rate constant of 1265M 1 s 1 ). However,the oxidation of CYN with chloramine and permanganate are rather slow processes, withrate constants o 1M 1 s 1 . The first chlorination product of CYN was found to be 5-chloro-CYN (5-Cl-CYN), which reacts with chlorine 10–20 times slower than the parent compound.The reactivity of ANTX with chlorine and chloramines is also very low (k o 1M 1 s 1 ).The elimination of CYN and ANTX in surface water was also investigated. A chlorinedose of 1.5mgl 1 was enough to oxidize CYN almost completely. However, 3mgl 1 of chlorine was able to remove only 8% of ANTX, leading to a total formation of trihalomethanes (TTHM) at a concentration of 150 m gl 1 . Therefore, chlorination is afeasible option for CYN degradation during oxidation and disinfection processes but not forANTX removal. The permanganate dose required for CYN oxidation is very high and notapplicable in waterworks. & 2007 Elsevier Ltd. All rights reserved. 1. Introduction Thepresence of toxiccyanobacterial blooms in naturalwatersused for drinking or for recreational purposes may presentserious health risks to the human population. Cylindrosper-mopsin (CYN), one of the most commonly occurring cyano-toxins, is known to be produced by the cyanobacterial genera Cylindrospermopsis , Anabaena, Umezakia and Aphanizomenon (Falconer, 2005). Due to its hepatotoxicity, cytotoxicity andgenotoxicity, there is a currently ongoing debate at the WorldHealth Organization (WHO) about whether or not theinformation published for CYN is sufficient to derive aguideline value. The uracil moiety is partly responsible forthe toxicity of CYN, presumably due to competitive orinhibitory binding of the toxin at a catalytic site (Banker etal., 2001). CYN has been found recently in Europe produced by Aphanizomenon flos-aquae in two German lakes (Preussel et al.,2006) and by Aphanizomenon ovalisporum in a Spanish waterreservoir (Quesada et al., 2006). Anabaena species, as well as Oscillatoria , and Aphanizomenon have been documented as ARTICLE IN PRESS 0043-1354/$-see front matter & 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.watres.2007.01.033 Corresponding author. Tel./fax: +34924289385.E-mail address:
[email protected] (J.L. Acero). WATER RESEARCH 41 (2007) 2048– 2056 producing anatoxin-a (ANTX), a neurotoxin which can affectthe central nervous system (Edwards et al., 1992). ANTXsimulates acetylcholine, containing both amine and carbonylmoieties, to overstimulate muscle cells and cause paralysis(Carmichael et al., 1992). Concentrations of CYN and ANTX innatural water samples have been reported to be up to 9.4 and51.3 m gl 1 , respectively (Quesada et al., 2006; Park et al., 1998). CYN can be eliminated from natural waters with some of the oxidants and disinfectants typically applied in water-works. Disinfection by chlorination has been evaluated as afeasible process to remove CYN from water at pH values of 6–9 (Senogles et al., 2000). Oxidation of CYN with chlorinemodified the uracil moiety to produce a chlorinated product,5-chloro-cylindrospermopsin (5-Cl-CYN), or a truncated car-boxylic acid derivative of CYN, cylindrospermic acid (Bankeret al., 2001). These two derivatives were found to be non-toxicin mice, suggesting that the uracil moiety is partly respon-sible for the toxicity of CYN. The formation of trihalo-methanes (THMs) and haloacetic acids (HAAs) fromchlorination of CYN were well below guideline values(Senogles-Derham et al., 2003). Photocatalytic degradation of CYN using titanium dioxide and UV irradiation have beendescribed as effective (Senogles et al., 2001). However,chloramine and chlorine dioxide were found to be less potentwith respect to CYN oxidation (Bankeret al., 2001). Ozone alsoreacts with the deprotonated amine moieties of CYN (Onstadet al., 2006) and is effective for the destruction of residualdissolved CYN (Newcombe and Nicholson, 2004). Howeverfew studies have been carried out reporting ANTX oxidationin natural waters. Ozone and OH radicals are powerfuloxidants for ANTX removal (Onstad et al., 2006). Hall et al. (2000) found that ANTX can be oxidized by permanganate andozone during drinking water treatment, but no discernibleremoval was observed by chlorination.Mostofthe above-mentionedstudies mainly focusedon thepercentage toxin oxidation by a given oxidant dose or theresidual oxidant concentration required for complete toxinoxidation. However, there has not been a thorough investiga-tion to determine rate constants for the oxidation of CYNwith chlorine, chloramines and permanganate or ANTX withchlorine and chloramine. Therefore, the objective of thepresent study was the calculation of such second-order rateconstants, which would be useful for predicting degradationefficiencies during full-scale treatment processes. The influ-ence of some operating variables such as pH and temperatureon the most reactive processes was also established. FinallyCYN and ANTX oxidation in natural water was investigated tocorroborate the reactivity of the studied oxidants with bothtoxins, to determine the required oxidant dose for toxinoxidation and to measure the total formation of THMs(TTHM) in the case of chlorination. 2. Materials and methods 2.1. Materials All reagents (buffers, oxidants, etc.) were of at least reagentgrade quality. Solutions of analytical reagents, chlorine,permanganate and phosphate buffers were prepared withultra-pure water produced by a MilliQ (Millipore) waterpurification system.Stock solutions of chlorinewere preparedby diluting a commercial solution of sodium hypochlorite (4%active chlorine, Aldrich) and standardized spectrophotome-trically in the presence of excess of iodide to form triiodine ( e at 351nm ¼ 25,700M 1 cm 1 ) (Bichsel and von Gunten, 1999). Permanganate stock solutions were also standardized spec-trophotometrically ( e at 526nm ¼ 2460M 1 cm 1 ) (Stewart,1973). Monochloramine solution was prepared daily by adding HOCl to ammonium nitrate, pH 8 at a molar ratio of 1:1. Sucha solution was standardized by the ABTS method (Pinkernellet al., 2000).CYN was isolated and purified from Cylindrospermopsisraciborskii strain CR3 (Metcalf et al., 2002) and ANTX from Phormidium strain DUN903 (Edwards et al., 1992). Standardiza-tion and handling followed a manual of standard operating procedures within the European Union project ‘‘TOXIC’’(Meriluoto and Codd, 2005). 2.2. Analytical methods Concentrations of CYN and ANTX were determined by HPLC(Waters Alliance 2695 Separation Module and Waters 996Photodiode Array Detector) using a C18 end-capped MerckPurospher STAR RP-18e, 3 m m particles, LiChroCART55 4mm I.D column (temperature 40 1 C). The photodiodearray (PDA) detector scanned from 200 and 300nm, withwavelengths of 262 and 227nm selected for the quantificationof CYN and ANTX, respectively. The concentration of 5-Cl-CYN was determined by using the CYN calibration curvetaking into account the ratio between the absorption coeffi-cients of both compounds at 262nm. The absorption coeffi-cient of CYN at 262nm is 6100 (Harada et al., 1994). Theabsorption coefficient of 5-Cl-CYN at 262nm was determinedfrom its coefficient at 277nm (4400M 1 cm 1 , Banker et al.,2001) and its PDA response at both wavelengths. A pre-concentration step was not needed due to the high concen-tration of toxins used in the experiments. Gradient elution(flow rate 1mlmin 1 ) of water with 0.05% trifluoroacetic acid(A) and acetonitrile with 0.05% trifluoroacetic acid (B) wasused by varying the percentage of A from 99% to 98% over5min, for the analysis of CYN, and by varying the percentageof A from 99% to 93% over 5min, for the analysis of ANTX. Inall cases the injection volume was 25 m l.Total THMs (TTHM) were measured by HS-GC-ECD (Golfi-nopoulos et al., 2001). The permanganate concentration innatural water samples was analyzed with DPD (Clesceri et al.,1999) after filtering the sample through 0.22 m m nylon filtersto remove manganese oxide. The chlorine concentration insurface water experiments was measured by the ABTSmethod (Pinkernell et al., 2000). 2.3. Rate constants of chlorine, monochloramine and permanganate with toxins Oxidation experiments were conducted under pseudo-first-order conditions where the oxidant was at least 8 fold inexcess. Ultra-pure water produced by a MilliQ system wasused to prepare the reaction samples. Initial concentrationsof oxidants and toxins are detailed in the Section 3. For the ARTICLE IN PRESS WATER RESEARCH 41 (2007) 2048– 2056 2049 most reactive oxidants, the temperature was varied in therange 10–30 1 C and the pH between 4 and 9 with 10mMphosphate buffer and adjusted with NaOH. In a typicalexperiment, a 5ml volume of buffered toxin solution wasprepared in a batch reactor (serum vial of 5.5ml), which wascapped with a PTFE-faced silica septum and located in athermostatic bath. The experiments were started afteraddition of an aliquot of the oxidant stock solution into thereactor while stirring. At fixed time intervals, 0.3ml of samplewas rapidly transferred with a syringe into a HPLC vialcontaining 2 m l of thiosulfate (0.1M) to stop the reaction. Theconcentration of residual toxin was analyzed directly byHPLC. No replicates were considered necessary based on thelow standard deviation obtained. 2.4. Experiments with natural water To investigatethe oxidationofCYN and ANTXby chlorine andpermanganate in natural water, some experiments wereperformed with surface water from Lake ‘‘Villar del Rey’’,located in the South-West of Spain. The main water qualityparameters were: pH ¼ 7.3, DOC ¼ 6.7mgl 1 , [bicarbona-te] ¼ 0.6mM, [ammonia] ¼ 2.3 m M. The initial dose of chlorineand potassium permanganate were varied from 0.2 to 3 andfrom 0.2 to 1.5mgl 1 , respectively, typical of a pre-oxidationstep during drinking water treatment. High initial concentra-tions of CYN and ANTX (1 m M) were spiked into the surfacewater in order to avoid solid phase extraction (SPE) beforeHPLC analysis. These experiments were started by adding different amounts of the oxidant stock solution to aliquots of 10ml of buffered toxin solutions at pH 7. The residual toxinconcentrations and TTHM formation (in chlorinationexperiments) were measured after 24 and 3h for chlorineand permanganate, respectively, time enough to reachcomplete oxidant consumption (determined by analyzing periodically the residual concentrations of chlorine orpermanganate). 3. Results and discussion 3.1. Kinetics of the reaction of chlorine withcylindrospermopsin Second-order rate constants for the oxidation of CYN withchlorine have been determined under pseudo-first-orderconditions (at least 8-fold excess of chlorine with respect toCYN) for the pH range 4–9. The reactions of chlorine withdifferent types of organic compounds are of second orderoverall, first order in active oxidant (HOCl+OCl ) and firstorder in organic compound (Rebenne et al., 1996; Acero et al., 2005a). Particularly, the chlorination reaction of microcystins(MC) was found to be of second order, first order in chlorineand first order in MC (Acero et al., 2005b). Therefore, thechlorination of CYN can be assumed to follow second-orderkinetics, first order with respect to each reactant (Eq. (1)), anaspect that will be confirmed later.d ½ CYN d t ¼ k app ½ CYN ½ chlorine t , (1)where k app is the apparent second-order rate constant at aspecific pH and [chlorine] t is the total concentration of chlorine species (HOCl+OCl ), which can be consideredconstant during the experiment. Thus, Eq. (1) can besimplified to Eq. (2)d ½ CYN d t ¼ k obs ½ CYN (2)which can be integrated to Eq. (3)ln ½ CYN ½ CYN 0 ¼ k obs t . (3)In order to confirmthe assumptionthat the observed loss of toxin in the presence of free chlorine is of first order withrespect to toxin concentration, the plot of Eq. (3) was used.Table 1 shows the initial concentrations of CYN and chlorineand pH applied in the experiments, being free chlorinepresent in excess with respect to CYN (pseudo-first-orderconditions). Fig. 1 shows the plot of Eq. (3) for experimentsperformed with similar initial concentrations of chlorine andCYN at different pHs. As can be observed, points liesatisfactorily around straight lines, whose slopes are theobserved first-order rate constants. After linear regressionanalysis ( r 2 4 0.99), the calculated values of k obs for eachexperiment are summarized in Table 1. These results pointout that CYN exhibitsa first-orderdependency with respect toitself in the presence of excess of chlorine, and therefore, thereaction of CYN with chlorine is of first order with respect tothe toxin.The apparent second-order rate constant can be calculatedfrom Eq. (4). Thus, if the reaction is of first order with respectto chlorine, apparent second-order rate constants for experi-ments performed at a specific pH with different initialchlorine concentrations must be similar. Effectively, thevalues calculated for the apparent second-order rate constantat pH 6.4 using Eq. (4) were 739.1 and 790.9M 1 s 1 forexperiments performed with initial chlorine concentrationsof 5.6 10 5 and 9.18 10 5 M, respectively. Therefore, anaverage value of 765.0 7 25.9M 1 s 1 can be proposed for theapparent rate constant at pH 6.4. Similar calculations (Eq. (4))were performed for the remaining pHs, being the valuesobtained for the apparent second-order rate constantsincluded in Table 1. k app ¼ k obs ½ chlorine t . (4)Fig. 2a shows the pH profile of the apparent second-orderrate constants. The maximum degradation rate of CYN wasobserved at a pH around 7, with lower reactivity at higher andlower pHs. This pH effect is similar to that obtained for thechlorination of dissociating compounds such as phenoliccompounds (Acero et al., 2005a) and certain pharmaceuticals(Pinkston and Sedlak, 2004).The pH dependence of the apparent second-order rateconstant can be explained by considering the speciation of the oxidant (reaction (5), pK a5 ¼ 7.5 (Albert and Serjeant,1984)) and CYN (reaction (6)), the reaction of HOCl with non-dissociated CYN (reaction (7)) and with dissociate CYN(reaction (8)) and the reaction of OCl with dissociate CYN(reaction (9)). These reactions describe the chlorination of CYN during water treatment at neutral or slightly alkalinepH. ARTICLE IN PRESS WATER R ESEARCH 41 (2007) 2048– 2056 2050 The reaction between OCl and non-dissociate CYN wasconsidered negligible because the concentration of non-dissociate CYN at high pH, when OCl oxidation would beimportant, must be very low.HOCl 3 OCl þ H þ K a5 ; (5)CYN 3 CYN þ H þ K a6 ; (6)CYN þ HOCl ! products k 7 ; (7)CYN þ HOCl ! products k 8 ; (8)CYN þ OCl ! products k 9 : (9)The degradation of CYN can be written from this reactionset as follows: d ½ CYN t d t ¼ k 7 ½ HOCl ½ CYN þð k 8 ½ HOCl þ k 9 ½ OCl Þ½ CYN , ð 10 Þ where [CYN] t is the total concentrations of toxin species(CYN+CYN ). Considering the speciation of chlorine q(reac-tion (5)) and CYN (reaction (6)), Eq. (10) is transformed ARTICLE IN PRESS Table 1 – Oxidation kinetics of CYN and 5-Cl-CYN with chlorine [CYN] 0 , m M ½ chlorine t 0 , m M T, 1 C pH k obs 10 3 ,s 1 R 2 k app , M 1 s 1 5.59 54.0 20 4.0 2.39 0.993 44.235.56 53.7 20 5.1 6.82 0.992 126.96.33 56.0 20 6.4 41.34 0.994 739.17.21 64.0 20 7.1 81.0 0.998 1265.16.92 55.8 20 8.0 29.9 0.994 536.86.27 57.0 20 9.0 3.03 0.993 53.186.76 54.5 10 6.4 25.7 0.992 470.96.72 54.5 30 6.4 64.1 0.999 1174.86.95 91.8 20 6.4 72.6 0.990 790.96.33 a 49.7 20 6.4 3.52 0.992 70.97.21 a 56.8 20 7.1 3.46 0.998 61.06.92 a 48.9 20 8.0 1.45 0.997 29.77.69 a 53.5 20 8.4 1.02 0.996 19.1 a Data corresponding to 5-Cl-CYN. 0 100 200 300 400-5-4-3-2-10 l n [ C Y N ] / [ C Y N ] 0 time, s Fig. 1 – Pseudo-first-order kinetic plot for the oxidation of CYN with chlorine at 20 1 C: Influence of pH on the pseudo-first-order rate constant (pH 4.0 ( ’ ); pH 5.1 ( & ); pH 6.4( K ); pH 7.1 ( J ); pH 8.0 ( m ); pH 9.0 ( W )). Initialconcentrations of CYN and chlorine are detailed in Table 1.The solid line is a linear least-squares regression of thedata. 1101001000100003 4 5 7 8 96 10 pH k a p p , M - 1 s - 1 k a p p , M - 1 s - 1 101006 6 98 pH ab Fig. 2 – pHdependenceof theapparentrateconstantsforthereaction of chlorine with CYN (a) and 5-Cl-CYN (b). The linesare calculated according to Eq. (12). WATER RESEARCH 41 (2007) 2048– 2056 2051 into Eq. (11) d ½ CYN t d t ¼ k 7 ½ H þ 2 þ k 8 ½ H þ K a6 þ k 9 K a5 K a6 ð K a5 þ½ H þ Þð K a6 þ½ H þ Þ ½ CYN t ½ Cl 2 t . ð 11 Þ Finally, the apparent second-order rate constant at each pH, k app , is given by Eq. (12) k app ¼ k 7 ½ H þ 2 þ k 8 ½ H þ K a6 þ k 9 K a5 K a6 ð K a5 þ½ H þ Þð K a6 þ½ H þ Þ . (12)Individual rate constants for the reactions between CYNand chlorine species (reactions (7)–(9)) and p K a values for CYN(reaction (6)) were evaluated from the values of the apparentsecond-order rate constant at different pHs and Eq. (12) bynon-linear least-squares regression analysis. The obtainedvalues were 38.1, 1.96 10 3 and o 0.01M 1 s 1 for k 7 , k 8 and k 9 ,respectively, and a p K a6 of 6.5. In the next step, the theoreticalpH profile of the apparent second-order rate constant wascalculated from Eq. (12) using the individual rate constantspreviously determined and represented in Fig. 2a (line). Thegood agreement between experimental values (symbols) andcalculated values (line) indicated that the proposed elemen-tary reactions are a good representation of the chlorination of CYN. Therefore, Eq. (7) is useful to determine the apparentsecond-order rate constant for the chlorination of CYN at anypH within the range typically applied in drinking watertreatment. The maximum in Fig. 2a corresponds approxi-mately to (p K a5 +p K a6 )/2, similar to the results found for thechlorination of other dissociating compounds such as chlor-ophenols (Acero et al., 2005a). These results can be explainedif HOCl is the majorelectrophilic species and indicate that themain reaction pathway at the circumneutral pH range is thereaction between HOCl and the dissociate form of CYN(reaction (8)). According to the values of the apparentsecond-order rate constant, CYN can be effectively oxidizedduring water chlorination in the pH range 6–8, typically foundin drinking waters.The temperature dependence of the apparent second-orderrate constant was determined at 10, 20 and 30 1 C in experi-ments performed with similar concentrations of CYN( E 6.5 m M) and chlorine ( E 55 m M) at pH 6.4 (Table 1). Thevalues obtained for the apparent rate constants following thecalculations above are shown in Table 1. These values wereused to calculate the activation energy by means of thelinearized Arrhenius expression (Eq. (13)).ln k app ¼ ln A E a = RT : (13)After linear regression analysis ( r 2 4 0.99), an activationenergy of 32.7kJmol 1 could be calculated. This activationenergy is higher than 20.1kJmol 1 obtained for Microcystin-LR (Acero et al., 2005b) and lower than 58.5kJmol 1 proposedby Lee (1967) for phenol. 3.2. Kinetics of the reaction of chlorine with 5-Cl-CYN Banker et al. (2001) identified 5-Cl-CYN as the first chlorina-tion product, which was found to be non-toxic. We identified5-Cl-CYN from chlorination of CYN by HPLC-PDA detectionaccording to the characteristics of its UV-spectrum describedby Banker et al. (2001). Since this compound is not commer-cially available, its concentration was determined using theCYN calibration curve as described in Section 2.2. Fig. 3 showsthe decay of CYN during a chlorination experiment (pH ¼ 7.1, T ¼ 20 1 C, [chlorine] 0 ¼ 64 m M, [CYN] 0 ¼ 7.21 m M) as well as theformation and subsequent decay of 5-Cl-CYN. It can beobserved that after 50s CYN has been chlorinated selectivelyto 5-Cl-CYN, its concentration being very close to the initialCYN concentration. After that, the chlorination of 5-Cl-CYNcontinues slowly. The apparent rate constant for the chlor-ination of 5-Cl-CYN can be determined from the decay of itsconcentration following the procedure described previouslyfor CYN. Thus, several experiments of CYN chlorination havebeen performed in which the formation and decay of 5-Cl-CYN has been measured over the reaction time (Table 1). Theobserved first-order rate constants were calculated by Eq. (3)leading to the values depicted in Table 1. Then the apparentsecond-order rate constant at each pH was calculatedaccording to Eq. (4). A negative effect of pH on the chlorina-tion rate of 5-Cl-CYN was observed (Table 1 and Fig. 2b). This pH effect can be explained if 5-Cl-CYN is a non-dissociating compound which can be attacked by HOCl and OCl (reac-tions (14) and (15)).5 Cl CYN þ HOCl ! products k 14 ; (14)5 Cl CYN þ OCl ! products k 15 . (15)With this set of reactions, the apparent second-order rateconstant at each pH, k app , is given by Eq. (16) k app ¼ k 14 ½ H þ þ k 15 K a5 ð K a5 þ½ H þ Þ . (16)Following a procedure similar to that of CYN, the valuescalculated for k 14 and k 15 were 77.4 and 12.8M 1 s 1 ,respectively. The theoretical pH profile of the apparentsecond-order rate constant was calculated from Eq. (16) using the individual rate constants and data in Fig. 2b (line). Again,the good agreement between experimental values (symbols) ARTICLE IN PRESS 0 50 100 150 200 250 300012345678 [ C ] , µ M time, s Fig. 3 – Formation of 5-Cl-CYN from CYN chlorination andsubsequent oxidation (CYN ( ’ ); 5-Cl-CYN ( & )).Experimental conditions: pH 7.1, T 20 1 C,[chlorine] 0 64 l M, [CYN] 0 7.21 l M. WATER R ESEARCH 41 (2007) 2048– 2056 2052
