Environ. Sci. Technol. 2001, 35, 4252-4259
MTBE Oxidation by Conventional Ozonation and the Combination Ozone/Hydrogen Peroxide: Efficiency of the Processes and Bromate Formation JUAN L. ACERO,† STEFAN B. HADERLEIN, TORSTEN C. SCHMIDT, MARC J.-F. SUTER, AND URS VON GUNTEN* Swiss Federal Institute for Environmental Science and Technology (EAWAG), Ueberlandstrasse 133, CH-8600 Du ¨ bendorf, Switzerland
The present study investigates the oxidation of methyl tert-butyl ether (MTBE) by conventional ozonation and the advanced oxidation process (AOP) ozone/hydrogen peroxide under drinking water treatment conditions. The major degradation products identified were tert-butyl formate (TBF), tert-butyl alcohol (TBA), 2-methoxy-2-methyl propionaldehyde (MMP), acetone (AC), methyl acetate (MA), hydroxyisobutyraldehyde (HiBA), and formaldehyde (FA). The rate constants of the reaction of ozone and OH radicals with MTBE were found to be 0.14 and 1.9 × 109 M-1 s-1, respectively. The rate constants for the same oxidation processes were also measured for the degradation products TBF, MMP, MA, and HiBA (kO3-TBF ) 0.78 M-1 s-1; kOH-TBF ) 7.0 × 108 M-1 s-1; kO3-MMP ) 5 M-1 s-1; kOH-MMP ) 3 × 109 M-1 s-1, kO3-MA ) 0.09 M-1 s-1, kO3-1 s-1; k 9 -1 s-1). Since all HiBA ) 5 M OH-HiBA ) 3 × 10 M compounds reacted slowly with molecular ozone, only the degradation pathway of MTBE with OH radicals has been determined, including the formation of primary degradation products. In experiments performed with several natural waters, the efficiency of MTBE elimination and the formation of bromate as disinfection byproduct have been measured. With a bromide level of 50 µg/L, only 35-50% of MTBE could be eliminated by the AOP O3/ H2O2 without exceeding the current drinking water standard of bromate (10 µg/L). The transient concentrations of MTBE and its primary degradation products were modeled using a combination of kinetic parameters (degradation product distribution and rate constants) together with the ozone and OH radical concentration and were in good agreement with the experimental results.
Introduction The widespread occurrence of methyl tert-butyl ether (MTBE), a major gasoline component, in drinking water resources such as groundwater and surface water (1) poses significant problems to the drinking water industry. MTBE is poorly biodegradable (especially under anaerobic condi* Corresponding author telephone: +41-1-8235270; fax: +41-18235210; e-mail:
[email protected]. † Present address: Departamento de Ingenieria Quimica, Universidad de Extremadura, 06071 Badajoz, Spain. 4252
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 21, 2001
tions), has a high water solubility (50 g/L), and is mobile in the subsurface (2). There is an ongoing scientific and public debate on safe drinking water standards for MTBE. Currently, there is no regulated value in place. However, the U.S. EPA has issued a consumer acceptability advisory on MTBE with a maximum tolerable concentration range in drinking water of 20-40 µg/L (on the basis of taste and odor considerations). Recently, the California Department of Health Services adopted a secondary maximum contaminant level of 5 µg/L for MTBE (3). Appropriate treatment options for MTBE-contaminated raw water at waterworks are therefore necessary. The use of air stripping, advanced oxidation processes (AOPs), and adsorption to granular activated carbon or synthetic resins has recently been compared in a comprehensive study with emphasis on cost-effectiveness (4). The performance of all methods depends on site-specific conditions such as water flow, observed influent, and required effluent concentration. However, MTBE is more difficult to remove from contaminated waters than other gasoline components when using the more traditional treatment techniques (4). AOPs, which are characterized by the production of hydroxyl radicals (•OH), are considered a promising not yet fully established treatment option for MTBE-contaminated water (4). Several studies have been performed considering ozone/hydrogen peroxide (5, 6), UV/hydrogen peroxide (7-9), Fenton’s reagent (10), UV/TiO2 slurries (11), and sonolytic destruction in the presence of ozone (12, 13). The objectives of these studies were to investigate the efficiency of MTBE degradation by different AOPs and to identify the degradation products formed. The major degradation products of MTBE are tertbutyl formate (TBF), tert-butyl alcohol (TBA), acetone (AC), and formaldehyde (FA) (5, 8, 10-12). In addition, 2-methoxy2-methyl propionaldehyde (MMP), hydroxyisobutyraldehyde (HiBA), and methyl acetate (MA) were identified as primary degradation products in some investigations (8). Although a general reaction mechanism of MTBE degradation by OH radicals has been elucidated, there is still a lack of information about the relative contribution of the different degradation pathways and, therefore, the distribution of intermediates during a given AOP treatment. Moreover, although some investigations have included the degradation of MTBE in natural waters (6, 9), the influence of drinking water parameters such as pH, alkalinity, natural organic matter (NOM), and concentration of other interfering compounds (e.g., bromide) has not been investigated. In the present study, we focus on ozone-based oxidation processes for MTBE degradation in natural waters. The impact of water quality parameters on MTBE oxidation efficiency and the formation of degradation products and bromate were evaluated. Ozonation can lead to oxidation of micropollutants either by a direct reaction of the compounds with molecular ozone or by OH radicals that are produced during the decay of ozone. The concentration of both oxidants must be known to define and calibrate an ozonation process with respect to its oxidation capacity. An experimental approach to determine the concentration of ozone and OH radicals during conventional ozonation or the AOP O3/H2O2 has been developed in previous studies (14-16). As described by Elovitz and von Gunten (14), a Rct value can be defined as the ratio of the exposures of OH radicals and ozone (i.e., concentration of oxidant integrated over the reaction time). The Rct value can be calculated from the measurement of the decrease of a probe compound, which reacts fast with •OH but not with ozone, and a simultaneous determination of the ozone 10.1021/es010044n CCC: $20.00
2001 American Chemical Society Published on Web 09/22/2001
TABLE 1. Water Quality Parameters lake water
well water
Zurich (LZ water)
Murten (LM water)
Porrentruy (WP water)
1.4 2.5 7.8 11.3 28.4
2.7 3.8 7.8 17.6 16.0
0.8 5.0 7.2 16.2 18.0
DOC (mg/L) alkalinity (mM HCO3-) pH bromide (µg/L) ammonia (µg/L of N)
concentration. Once the Rct value is known, the elimination of a compound (M), which reacts with both oxidants, can be calculated by second-order kinetics and expressed as a function of Rct, kO3, k•OH, and the ozone exposure (∫[O3] dt) (16):
ln
( )
∫
[M] ) - ( [O3] dt)(kOHRct + kO3) [M]0
(1)
where k•OH and kO3 are the second-order rate constants for the reactions of a micropollutant (M) with •OH and O3, respectively. Equation 1 can be used to estimate the elimination of degradation products as well. In a previous study, the degradation of atrazine and the formation of its degradation products could be predicted successfully by measuring the Rct and the rate constants for the oxidation of atrazine and its degradation products with ozone and OH radicals (16). This quantitative kinetic approach, which has also been applied to asses the oxidation of metal-diethylenetriaminepentaacetate (DTPA) complexes during drinking water ozonation (17), will be used to predict the behavior of MTBE and its primary degradation products during ozonation and the AOP O3/H2O2 for various raw waters.
Materials and Methods Standards and Reagents. MTBE and its main degradation products (TBF, TBA, AC, MA, and FA) where obtained from Aldrich and Sigma, each with a purity above 98%. All chemicals used for solutions (buffer, eluent, etc.) were reagent grade and were used without further purification. Stock solutions were prepared in bidistilled water. Concentrated ozone solutions were produced by passing O3-containing oxygen through bidistilled water that was cooled in an ice bath (18). Natural Water Systems. To evaluate the influence of the water quality parameters on the degradation of MTBE together with the formation of degradation products, we studied the processes in the following Swiss natural waters (for water quality, see Table 1): (i) raw water of Lake Zurich collected from a depth of 30 m was obtained from a drinking water plant in Zurich (LZ water) and is characterized by a low alkalinity and low DOC; (ii) raw water from Lake Murten sampled at a depth of 20 m (LM water) is characterized by a relatively high alkalinity and DOC content; and (iii) raw well water collected at the drinking water plant in Porrentruy (WP water). It originates from a karstic system and is characterized by a high alkalinity and low DOC. All waters were filtered (0.45 µm cellulose nitrate) within 24 h after sampling and stored at 4 °C until use. Analytical Methods. MTBE and most of its degradation products (TBF, TBA, AC, and MA) were analyzed by direct aqueous injection GC/MS. Positive electron impact (EI+) mass spectra were obtained with a Fisons MD800 quadrupole mass spectrometer interfaced to a Fisons GC8000 gas chromatograph. Spectra were acquired at an electron energy of 70 eV. The source temperature was set at 200 °C, and the interface temperature was set at 250 °C. The mass spectra were scanned from 25 to 150 (m/z) at a scan time of 0.5 s for
identification of degradation products. For quantitative analysis, tert-amyl methyl ether (TAME) was selected as the internal standard, and only five masses were recorded in selective ion monitoring (SIM) mode, corresponding to the main mass signals of MTBE and its degradation products (73 for MTBE and TAME and 57, 58, 59, and 74 for TBF, AC, TBA, and MA, respectively). Aqueous samples (1 µL) without further pretreatment were injected on-column onto a Stabilwax (Crossbond Carbowax-PEG) column (60 m × 0.32 mm, 1 µm film thickness). The oven temperature was kept at 50 °C during the first 10 min and then increased to 100 °C at 10 °C/min. After 5 min at 100 °C, the temperature was increased to 200 °C at 30 °C/min. Helium was the carrier gas, and the column head pressure was set at 100 kPa. The aldehyde intermediates were analyzed by HPLC (Hewlett-Packard, 1050 series; column: 250 × 4 mm i.d. Zorbax ODS (5 µm particle size)) after derivatization with 2,4-dinitrophenylhydrazine (2,4-DNPH) following a method adapted from ref 19. To 5 mL of sample solution, 2 mL of an aqueous solution containing 800 µM 2,4-DNPH and 500 mM HCl were added and maintained at room temperature for 2 h before analysis. Separation was performed using gradient elution at a flow rate of 1 mL/min. Initial conditions were 45% acetonitrile and 55% water with a linear gradient to 51% of acetonitrile at 8 min and then increased linearly to 80% at 15 min. After that, the initial conditions were reached within 2 min, and the system equilibrated for another 8 min. The sample volume injected varied from 25 to 250 µL depending on the MTBE concentration. Absorbance was measured at 360 nm. MMP and HiBA were quantified by using trimethylacetaldehyde (TMA) and FA as reference compounds since these aldehydes are not commercially available. The identification of the aldehyde intermediates was performed by LC/MS according to the method described above. The spectra were obtained on a Platform LC single quadrupole mass spectrometer equipped with an electrospray interface (ESI) from Micromass UK Ltd. (Manchester, U.K.). The spectra were acquired in negative ion mode over the mass range of 50-500 Da at 1 scan/s. The ESI interface temperature was set to 150 °C, and the nitrogen gas flow was set to 500 L/h. The cone voltage was 50 V. Parent ions at m/z 281 for MMP and 267 for HiBA were found, which corresponds to (M - H)- ions as observed for other dinitrophenylhydrazones. Characterization of Ozonation and AOP. Dissolved ozone was analyzed by the Indigo method (20). OH radicals were indirectly measured by a probe compound, which reacts slowly with ozone and fast with OH radicals. The selected probe compound was p-chlorobenzoate (pCBA), which reacts with •OH with a second-order rate constant of k ) 5 × 109 M-1 s-1 (21). pCBA was determined by HPLC (column: Merck Lichrospher 100, RP18 5 µm, 125 × 3 mm i.d.) with an eluent containing 45% 10 mM H3PO4 and 55% methanol at 1 mL/ min and detected at 234 nm (14). Hydrogen peroxide was determined by the peroxidase-DPD method (22). Bromide and bromate were measured by ion chromatography with postcolumn reaction (23). Mechanistic Investigations. Oxidation of MTBE and its degradation products by molecular ozone is too slow to be used for drinking water treatment. Therefore, AOP O3/H2O2 experiments with MTBE, TBF, and TBA in bidistilled water at 20 °C could be used to determine the degradation pathway with OH radicals. These experiments were carried out by adding H2O2 (0.1 mM) to the initial solution of these compounds (40-80 µM in all experiments) followed by the addition of O3. Under these conditions, •OH was the predominant oxidant species. The pH was kept constant at 7 (with 10 mM phosphate buffer). Residual degradation product concentrations were analyzed after total O3 consumption in experiments with varying O3 dosages (from 0.1 to 10 mg/L). VOL. 35, NO. 21, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4253
Experiments in Natural Waters. Natural water samples were buffered to pH 7 or 8 by adding 10 mM borate buffer and adjusting the pH with HCl. MTBE was added at relatively low initial concentrations (2 µM or 176 µg/L) to minimize competition reactions of the oxidants with the natural water matrix. Furthermore, pCBA was added as a probe compound (0.5 µM). Bromide was spiked to the samples to reach an initial concentration of 50 µg/L. H2O2 was used only in AOP experiments, with a ratio of 0.34 mg of H2O2/mg of O3 (0.5 M/M). After addition of ozone (from 2 to 4 mg/L), samples were withdrawn with a dispenser after certain time intervals (18). The reaction was stopped with indigo for residual ozone and bromate measurements. For MTBE and pCBA analysis, residual ozone was quenched by adding a small volume of a concentrated (1 M) sodium thiosulfate solution. Determination of Rate Constants of MTBE and Its Degradation Products with Ozone and •OH. The rate constants of the reaction of ozone with MTBE and its degradation products were measured by an absolute method involving the measurement of the decrease of ozone in the presence of an excess of MTBE to fulfill pseudo-first-order conditions (24). The temperature was varied from 5 to 20 °C, and the pH was kept constant at 2 (50 mM phosphate buffer). Ozone (2 × 10-5 M) was added to buffered solutions of MTBE and its degradation products (1 × 10-3 M). The ozone concentration was monitored by measuring its absorbance at 258 nm ( ) 3150 M-1 cm-1; 25). Rate constants of MTBE and some degradation products with •OH were determined by competition kinetics (26), using pCBA as a reference compound. OH radicals were produced by adding different O3 doses (from 0.25 to 5 mg/L) to solutions containing MTBE or degradation products (10 µM), pCBA (10 µM), and H2O2 (0.1 mM) at pH 7 (50 mM phosphate buffer). The relative decrease of pCBA and MTBE or its degradation products was measured after complete depletion of ozone.
Results and Discussion Rate Constants of MTBE and Its Degradation Products with Ozone and OH Radicals. The rate constants of the reaction of MTBE and of each of its degradation products with both ozone and OH radicals have to be known to design and interpret a comprehensive kinetic study on MTBE degradation in drinking water treatment. Since some of these rate constants are well-known from the literature (21, 24), only unknown rate constants or those which are controversial in the literature were determined in the present study. Our experimental data are compiled in Table 2 together with published values. MMP and HiBA were not commercially available. Therefore, we obtained an estimate of their reaction rate constants with ozone and OH radicals from the average of the respective rate constants of structurally related aldehydes TMA and isobutyraldehyde. It can be seen that the degradation of MTBE and most of its intermediates by direct attack of ozone is very slow and therefore negligible when the AOP O3/H2O2 is applied. The oxidation by ozone might be important only for some aldehyde intermediates. Rate constants for the reaction of OH radicals (kOH) with MTBE and some aldehydes (MMP, HiBA, and FA) exceed 109 M-1 s-1, corroborrating previous results of fast degradation of these compounds by OH radicals. The rate constants of the other intermediates of MTBE degradation for the reaction with OH radicals are significantly lower and range from 108 to 7 × 108 M-1 s-1, indicating a potential accumulation of these compounds during AOPs. FA exhibits a higher reactivity than AC and MA. The rate constant for the reactions of MTBE and TBF with OH radicals have been measured before by competition kinetics methods (7, 21, 27). Our experimentally determined 4254
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 21, 2001
TABLE 2. Rate Constants for Reaction of MTBE and Its Degradation Products with Ozone and OH Radicals compd
kO3 (M-1 s-1)
ref
MTBE TBF TBA MMP MA AC HiBA FA TMA isobutyraldehyde
0.14 0.78 0.003 5a 0.09 0.032 5a 0.1 5.9 4.2
this study this study 24 this study this study 24 this study 24 this study this study
a
kOH (M-1 s-1) 1.9 × 109 7.0 × 108 6.0 × 108 3.0 × 109 2.3 × 108 1.3 × 108 3.0 × 109 1.0 × 109 3.2 × 109 2.8 × 109
a
a
ref this study this study 21 this study 21 21 this study 21 this study this study
Estimate from the values obtained for TMA and isobutyraldehyde.
kOH values for MTBE (1.9 × 109 M-1 s-1) and TBF (7 × 108 M-1 s-1) fall within the range of published data of 1.6 × 109 (21) and 3.9 × 109 M-1 s-1 (7) for MTBE and of 4.1 × 108 (27) and 1.2 × 109 M-1 s-1 (7) for TBF. Oxidation of MTBE and Its Degradation Products by OH Radicals. To determine the degradation pathway of MTBE by OH radical attack, experiments were conducted with high MTBE concentration (1 × 10-4 M) in the presence of H2O2 (2 × 10-4 M) at pH 7 and varying the ozone dosage (from 2 to 10 mg/L). The identification of the reaction intermediates was performed by GC/MS and LC/MS as described above. The following degradation products were identified: TBF, TBA, MMP, AC, MA, HiBA, and FA. These degradation products have also been identified in previous studies when other OH radical producing AOPs were applied (UV/H2O2 (8), sonolysis (12)). Once the intermediates were identified, the AOP O3/H2O2 was applied individually to MTBE and its most important degradation products (TBF and TBA) to quantify the kinetics of these reactions and to obtain the complete reaction sequence. These experiments were carried out with an initial concentration of MTBE or intermediates of 40-80 µM at pH 7 in the presence of H2O2 (1 × 10-4 M) and varying the ozone dosage (from 0.10 to 10 mg/L). Under these experimental conditions, OH radicals are the main oxidant species. The degradation of MA and AC was not the subject of our study because of their low reactivity with OH radicals (see Table 2). In addition, the degradation of AC was studied in detail before (28, 29). Figure 1 shows the degradation of MTBE and the evolution of its major degradation products as a function of the ozone dosage for the AOP system. The primary degradation products were TBF, TBA, MMP, AC, MA, and FA. A detailed scheme of MTBE degradation in the UV/H2O2 system has been proposed by Stefan et al. (8), including the mineralization of secondary transformation products. However, some important kinetic data including rate constants and percentage of formation of some intermediates are not reported. From the available data, it can be estimated that at low ozone doses, OH radicals react primarily with MTBE while significant degradation of MTBE transformation products only takes place after an important fraction of MTBE has been degraded. Therefore, the percentage of each degradation product formed from MTBE can be calculated during the initial phase of a transformation experiment conducted at such conditions (see Figure 2). The percentages given indicate the fraction of MTBE oxidized to a particular product. FA has to be considered separately because it is formed simultaneously from several degradation products. The total mass balance has been calculated considering the compounds with high carbon atom number formed directly from MTBE (sum of MTBE and degradation products concentration except FA). The mass balance was close to 100% at very low ozone dose (see Figure 1) and decreased with increasing ozone dosage
FIGURE 1. Concentration profiles of MTBE (left axis) and its degradation products (right axis) as a function of ozone dose during the AOP O3/H2O2. Experimental conditions: pH 7, T ) 20 °C, [H2O2]0 ) 1 × 10-4 M, [MTBE]0 ) 40 µM. The data shown represent final concentrations after complete ozone depletion.
FIGURE 2. MTBE degradation and formation of degradation products. Percentages given refer to the relative importance of a pathway. (75% after 90% of MTBE degradation), probably due to the formation of unidentified intermediates. Initial OH radical attack on MTBE by H-abstraction can occur at either the methoxy group or any of the three methyl groups. The carbon-centered radical generated after addition of oxygen forming the peroxyl radical decomposes through bimolecular reactions to different intermediates (30). Reaction at the methoxy group is more likely to happen since H-abstraction occurs most easily at the position R to the ether function (30). This reaction pathway generates TBF, TBA, AC, and FA (8). The second possible pathway (OH radical attack on the methyl groups) leads to the formation of MMP, MA, AC, and FA. The contribution to the overall transformation of MTBE of each pathway can be calculated from the product distribution data shown in Figure 2. Considering that MA and AC are generated in equimolecular amounts from the OH radical attack on a methyl group, the quantity of AC formed from the OH radical attack on the methoxy group can be calculated by subtracting the amount of MA from the total amount of AC. Thus, around 44% AC originate from OH radical attack on the three methyl groups, and 56% are generated from OH radical attack on the methoxy group. This result confirms that the presence of oxygen in the molecule facilitates the electrophilic attack of OH radicals on the methoxy group (56% of the total process). This ratio agrees well with the data from Stefan et al. (8), who found 40% MTBE decay as a result of OH radical attack at the tert-
butyl side and 60% at the methoxy group. Taking into account that 42% of MTBE decays to form TBF, this implies that 75% of MTBE degradation by the OH radical attack on the methoxy group leads to TBF. This value is somewhat higher than the 66% reported by Stefan et al. (8). We also investigated the degradation pathway of the reaction of the main intermediates (TBF and TBA) with OH radicals as shown in Figure 2. The degradation of TBA leads to the formation of AC, HiBA, and FA. The fraction of unidentified primary transformation products was very low since the mass balance was close to 100%. The same degradation products were identified during the oxidation of TBF. However, the mass balance was only 75% after 50% degradation of TBF, which indicates that degradation products are formed that were not detectable by our analytical methods. These results disagree with those proposed by Stefan et al. (8), who reported AC and FA as major primary intermediates of TBF degradation. In addition, TBA was not found during the degradation of TBF in accordance with the relatively slow hydrolysis of TBF to TBA at neutral pH (halflife of 5 days; 31). Degradation of MTBE in Natural Waters. In addition to MTBE, a small amount of a probe compound (pCBA) was added to the natural waters to quantify the concentration of OH radicals through the Rct value concept (14). During conventional ozonation, a biphasic behavior was observed. An initial phase with high k (first-order rate constant for ozone decomposition) and Rct values was followed by a second phase with significantly lower values. For the AOP O3/H2O2, only one phase with constant k and Rct values was noted (see Table 3; only the values of the second phase are included for conventional ozonation). Figure 3 shows the course of MTBE concentration versus reaction time in experiments performed with LZ water for different oxidant concentrations and pH values. A relatively slow MTBE degradation occurred in conventional ozonation experiments due to the low reactivity of MTBE with molecular ozone. Addition of hydrogen peroxide (0.34 mg of H2O2/mg of O3) accelerated the ozone decomposition into OH radicals (32), which in turn oxidized MTBE (cf. k and Rct values in Table 3). The application of an AOP therefore leads to a faster MTBE degradation. However, even the AOP is not very effective in removing MTBE from raw waters at oxidant doses typically applied in drinking water treatment. For instant, a dose of 2 mg/L of ozone and a ratio of 0.34 mg of H2O2/mg of O3 resulted in a final MTBE conversion of less than 50% in LZ water. To increase the MTBE degradation, a higher oxidant concentration must be applied. An O3 dose of 4 mg/L yielded 71% of MTBE elimination in LZ water (Figure 3 and Table 3). An increasing pH leads to a faster ozone decomposition according to the higher values of k and Rct obtained at pH 8 (Table 3). This faster ozone decomposition into OH radicals leads to a faster MTBE degradation but not to a higher degree of MTBE elimination (Figure 3). The influence of various water quality parameters on the O3 decay and OH radicals formation can be evaluated by comparing the values of k and Rct given in Table 3 for each natural water in experiments referring to similar experimental conditions. An increase of NOM concentration accelerates ozone decomposition into OH radicals as indicated by the increasing values of k and Rct at higher DOC concentrations. This is due to the promoting effect of NOM, which results in an acceleration of the radical-type chain reactions for ozone decomposition. The influence of the water quality parameters on k and Rct values is more pronounced in conventional ozonation because in the AOP O3/H2O2 the initiation of O3 decomposition by HO2- is an important step and NOM promotion becomes less relevant in the whole process. The influence of the water quality parameters on MTBE degraVOL. 35, NO. 21, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4255
TABLE 3. MTBE Removal, Bromate Formation, Ozone Decomposition Rate Constant, and Rct Values for Natural Waters Treatment pH 7 [O3] ) 2 mg/L ozone
ozone/H2O2
pH 8 [O3] ) 4 mg/L ozone/H2O2
[O3] ) 2 mg/L ozone
a
ozone/H2O2
s-1)
% elim MTBE bromatea (µg/L) k (s-1) Rct
LZ Water (DOC: 1.4 mg/L; Alkalinity: 2.5 mM; Scavenging Capacity: 5.6 × 39 46 71 35 15.1 8.8 16.8 12.5 1.1 × 10-4 5.3 × 10-3 6.8 × 10-3 4.7 × 10-4 6.4 × 10-10 4.4 × 10-8 6.1 × 10-8 2.8 × 10-9
% elim MTBE bromatea (µg/L) k (s-1) Rct
WP Water (DOC: 0.8 mg/L; Alkalinity: 5 mM; Scavenging Capacity: 6.2 × 104 s-1) 28 37 65 22 20.7 12.5 33.5 12.5 -5 -3 -3 6.8 × 10 3.8 × 10 5.3 × 10 2 × 10-4 2.2 × 10-10 1.9 × 10-8 3.6 × 10-8 4 × 10-10
% elim MTBE bromatea (µg/L) k (s-1) Rct
LM Water (DOC: 2.7 mg/L; Alkalinity: 3.8 mM; Scavenging Capacity: 10 × 104 s-1) 33 34 46 12 5.6 8.7 7.7 × 10-4 8.3 × 10-3 1 × 10-2 4 × 10-9 4.2 × 10-8 4.4 × 10-8
104
36 4.2 1.2 × 10-2 7.4 × 10-8
Initial bromide concentration in the experiments: 50 µg/L.
FIGURE 3. Influence of the oxidant concentration and pH on MTBE degradation. Experimental conditions: LZ water, T ) 12 °C, ratio of H2O2/O3 ) 0.34 w/w, [MTBE]0 ) 1.8 µM. Symbols represent measured data, and lines represent model calculations. Final reaction time for conventional ozonation experiments were 6 and 2 h at pH 7 and pH 8, respectively. dation is depicted in Figure 4 for experiments carried out under conditions yielding maximum MTBE degradation (pH 7, ozone dose of 4 mg/L, and presence of H2O2). The yield of MTBE degradation was highest in LZ water and lowest in LM water due to the differences in the overall scavenging capacities of the waters rather than due to the different ozone decomposition rates into OH radicals. The overall OH radical scavenging capacities are defined as follows: kOH-DOC[DOC] + kOH-HCO3-[HCO3-] + kOH-CO32-[CO32-]; kOH-DOC ) 2.5 × 104 mg-1 s-1L (25), kOH-HCO3- ) 8.5 × 106 M-1 s-1, kOH-CO32- ) 3.9 × 108 M-1 s-1 (21). At pH 7, the values for the scavenging capacities are 5.6 × 104, 6.2 × 104, and 1.0 × 105 s-1 for LZ water, WP water, and LM water, respectively. These values can explain the trend of the MTBE degradation yield observed in Figure 4 and Table 3 except those for the conventional ozonation of WP water and LM water (28% and 33% elimination, respectively). Oxidation of MTBE and Bromate Formation. In bromidecontaining waters, the ozone-induced oxidation of micropollutants is always accompanied by the formation of bromate. Because of the low drinking water standard for 4256
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 21, 2001
FIGURE 4. Influence of water source on MTBE degradation. Experimental conditions: T ) 12 °C, pH 7, [O3]0 ) 4 mg/L, ratio of H2O2/O3 ) 0.34 w/w, [MTBE]0 ) 1.8 µM. Symbols represent measured data, and lines represent model calculations. bromate in many countries (10 µg/L) (33, 34), a multiobjective optimization of the process is often required. Bromate formation occurs through a complicated mechanism involving both molecular ozone reactions and reactions with OH radicals (35). In hydrogen-peroxide-based AOPs, the reduction of hypobromous acid (HOBr) by hydrogen peroxide is a crucial process (36) that reduces bromate formation (37, 38). However, in the AOP O3/H2O2, bromate formation cannot be completely suppressed even at very high levels of hydrogen peroxide (37). In Table 3, MTBE oxidation and concomitant bromate formation are shown for three different waters under various oxidation conditions. All waters were spiked with bromide to yield an initial concentration of 50 µg/L. The same basic features occurred in all three of the waters and are discussed in detail below for LZ water. Since MTBE only reacts with OH radicals at significant rates, it can be taken as a measure for the OH radical exposure. For an ozone dose of 2 mg/L, addition of H2O2 increased the MTBE elimination at pH 7 only slightly as compared to the conventional ozonation test, indicating that the OH radical exposure remained almost unchanged. However, in the AOP O3/H2O2, the half-life of molecular ozone is much smaller and therefore the ozone exposure decreases. Thus the overall oxidation
capacity (OH radicals and ozone) decreases in the system. In addition, the presence of H2O2 leads to a partial reduction of HOBr, which is an important intermediate in bromate formation. These two factors reduced the bromate formation by approximately a factor of 2, from 15.1 to 8.8 µg/L. At higher ozone doses (4 mg/L), an increased bromate formation is found because of the increased overall oxidation capacity in the system. An increase of the ozone dose leads to an enhanced elimination of MTBE (by almost a factor of 2). This shows that the OH radical exposure is increased significantly with the higher ozone dose. In addition, the ozone exposure also increases. Both factors lead to an increased bromate formation by about a factor of 2 (from 8.8 to 16.8 µg/L). Therefore, MTBE oxidation in the LZ water is restricted by bromate formation at pH 7. An oxidation of about 50% can be achieved without exceeding the bromate standard of 10 µg/L when the AOP O3/H2O2 is applied with an initial bromide concentration of 50 µg/L. An increase in pH from 7 to 8 leads to a decrease of the bromate formation for both conventional ozonation and the AOP O3/H2O2. To compare both pH values, however, the elimination of MTBE has to be considered. In both cases, less MTBE is eliminated at the higher pH, which is the result of a lower OH radical exposure. In addition, the experiments show that the ozone exposure is lower (k increases) at the higher pH. Another reason for a reduced bromate formation is the enhanced reduction of HOBr by H2O2 at the higher pH value. The superposition of all three factors results in a lower bromate formation (36). Only when normalized to the same ozone exposure bromate formation increases with increasing pH (35). The two other waters (WP, LM) follow basically the same trend (see Table 3) as discussed for LZ water. However, the degree of bromate formation varied significantly. The comparison of LZ water and WP water for the AOP O3/H2O2 with an ozone dose of 4 mg/L shows that the MTBE oxidation was similar, whereas the bromate formation was almost twice in WP water. In the DOC-poor WP water, the half-life of ozone was substantially higher, which lead to a higher ozone exposure. In addition, due to the high carbonate content, carbonate radicals contributed to enhanced bromate formation (35). Therefore, in WP water only about 35% of MTBE can be oxidized by the AOP O3/H2O2 without exceeding the bromate standard with the initial bromide concentration used in these experiments (50 µg/L). In LM water, about 50% of MTBE can be oxidized without exceeding the bromate standard, but a higher ozone dose is required due to the high overall scavenging capacity of this water. In fact, the same degree of MTBE oxidation can be achieved for LM water with an ozone dose of 4 mg/L (AOP O3/H2O2) for LZ water with an ozone dose of 2 mg/L (AOP O3/H2O2). The bromate formation in both cases is identical at about 8.8 µg/L, and the ozone exposure is similar as well (1.4 × 10-4 M min-1). The factor of 2 in ozone dosage needed to obtain similar results is reflected in a 2-fold greater scavenging capacity for LM water as compared to LZ water. In conclusion, MTBE oxidation by conventional ozonation and the AOP O3/H2O2 has to be carefully optimized with regard to bromate formation. To achieve a similar degree of MTBE oxidation, the overall scavenging capacity has to be considered and the ozone dose adapted accordingly. Once a certain degree of MTBE oxidation has been set, the bromate formation is a function of the resulting ozone exposure. The oxidation of MTBE is an “unfortunate” case because of its relatively low reactivity toward OH radicals. For other compounds with higher rate constants with OH radical (e.g., aromatic compounds), better removal could be achieved, and the optimization pollutant elimination/bromate formation would be easier.
Kinetic Modeling of MTBE Degradation by the AOP O3/ H2O2. A kinetic model was applied to predict the evolution of MTBE and its degradation products in natural waters by conventional ozonation and the AOP O3/H2O2 and compared to experimental results. For this, in addition to the rate constants and the degradation product distribution, the concentration of both oxidants ozone and OH radicals were determined (k and Rct in Table 3). The contribution of several degradation pathways must be considered to estimate the transient formation of the degradation products. For that purpose, the overall rate constants of MTBE and of each of its degradation products with both oxidants have been split up according to the partial contribution of each pathway given in Figure 2. With the determined oxidant concentrations and the split rate constants, the concentrations of MTBE and its degradation products have been calculated from eq 1 with the help of the computer program ACUCHEM (39). The data shown in Figures 3 and 4 allow comparisons of the model calculations of MTBE degradation (lines) with measured data (symbols) for various experimental conditions and types of natural waters. It can be seen that MTBE oxidation is well predictable. Figure 5 shows measured data (symbols) and model calculations (lines) for MTBE removal and formation of its degradation products during treatment of LZ water (Figure 5a), WP water (Figure 5b), and LM water (Figure 5c) at the respective experimental conditions (for details, see figure captions). Experiments performed with natural waters resulted in measurable concentrations of MTBE, TBF, TBA, AC, and MA. Because of the low initial concentration of MTBE (2 µM) and the interference of thiosulfate in the analytical method, some aldehyde intermediates could not be quantified. FA could only be measured in the last reaction sample (after complete ozone depletion), and the obtained values were very similar to those predicted by model calculations. The concentrations of MMP (also a primary degradation product) as predicted by the kinetic model are also included in Figure 5 (dotted line), although we cannot compare these values to measured data. The dashed line in Figure 5 represents the mass balance including measured concentrations of the primary degradation products MTBE, TBF, TBA, AC, and MA and the modeled concentrations of MMP. Some control samples of natural waters before and after ozonation have been analyzed, and no MTBE degradation products were found (data not shown), which indicates that the measured mass balance is in fact due to MTBE degradation. The experimental mass balance was well above 90% of the initial MTBE concentration in all cases, indicating that the formation of the secondary degradation products is not relevant until MTBE degradation exceeds 50%. Note that measured concentrations of AC were often slightly higher than the model predictions (mainly in conventional ozonation experiments), which can be attributed to the formation of AC from oxidation of NOM or small interferences in the background signal of the GC/MS from the natural water. The otherwise good agreement between the model calculations and the experimental results validates the results of our study. Overall, the combination of experimental kinetic data with modeling enables to predict not only the time course of MTBE removal but also the concomitant formation of its degradation products during conventional ozonation and AOP treatment of natural waters. In addition, our results allow a quantitative prediction of the formation of primary degradation products based on the degree of MTBE elimination. This is shown in Figure 6 where the production of degradation products is correlated to the MTBE elimination (molar basis). The data shown in Figure 6 include both conventional ozonation and the AOP O3/H2O2 for the different water types investigated in the present study. The slopes of the straight lines are 0.35 (TBF and FA), 0.23 (AC), 0.11 (TBA), and 0.07 mol/mol (MA). VOL. 35, NO. 21, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4257
FIGURE 6. Formation of degradation products as a function of MTBE elimination. Intermediate and final amounts obtained in experiments shown in Table 3 are included. due to its higher reactivity toward ozone and OH radicals, which leads to its further degradation (Figure 6 inset). Our experiments at the lower level of MTBE (2 µM) showed a good mass balance for product formation even though they were extrapolated from high level MTBE experiments. A similar behavior can be expected for concentrations
