Environ. Sci. Technol. 2004, 38, 1221-1227
Metribuzin Degradation by Membrane Anodic Fenton Treatment and Its Interaction with Ferric Ion QIQUAN WANG, EMILY M. SCHERER, AND ANN T. LEMLEY* Graduate Field of Environmental Toxicology, TXA, MVR Hall, Cornell University, Ithaca, New York 14853-4401
Metribuzin, a widely used herbicide and a frequently detected pollutant in the environment, was studied as a target compound for membrane anodic Fenton treatment (AFT), a Fenton technology with application potential for onsite treatment of pesticide wastewater. It was found that the degradation kinetics of metribuzin do not obey the AFT model, a previously developed model that fit AFT degradation kinetics of all previously investigated pesticides. The lack of fit for metribuzin data was determined to result from a weak interaction between metribuzin and the ferric ion, resulting in a significant reduction in availability of metribuzin for reaction with hydroxyl radicals during AFT, thus slowing degradation. A revised kinetic model was developed based on the original AFT model with the addition of this interaction. Results demonstrate that the new kinetic model fits metribuzin degradation data quite well at different delivery rates of Fenton reagent and at different temperatures. This weak interaction is also found to exist between ferric ion and several other triazinone/ triazine herbicides during membrane AFT. The interaction intensity correlates with the electron-withdrawing/donating property of substituents on the triazine/triazinone ring. The stronger the electron-donating ability of substituents, the stronger the interaction.
Introduction The intensive application of pesticides has become a notable characteristic of modern agriculture. As much as 912 million pounds of conventional pesticides were used in the U.S. in 1 year (1999) as estimated by the United State Environmental Protection Agency (1). As a result of this heavy application, pollution of water resources by pesticides has received wide attention (2-5). There has been recent emphasis on controlling pesticide contamination resulting from nonpoint source pollution, such as leaching and runoff from fields, but at the same time studies are revealing that considerable environmental contamination is caused by point-source pollution (6). The recycling of unwanted pesticides and the disposal of rinsates from pesticide application equipment have become major issues in pesticide waste management (7, 8). Technologies for farmers and commercial applicators for on-site management of small-scale wastewater containing high concentrations of pesticides are needed. Among available treatment technologies, classic Fenton treatment (9, 10), photo-Fenton treatment (11-13), electrochemical Fenton treatment (14), cathodic Fenton treatment (15, 16), and anodic Fenton treatment (AFT) (17, 18) * Corresponding author phone: (607) 255-3151; fax: (607) 2551093; e-mail:
[email protected]. 10.1021/es0345827 CCC: $27.50 Published on Web 01/16/2004
2004 American Chemical Society
are of great interest due to the strong oxidation ability of hydroxyl radicals, the broad spectrum of target pollutants, and the simple equipment and operation procedures. The substitution of an ion exchange membrane for the salt bridge in the original AFT and the advantages of high treatment efficiency, partially neutralized effluent, no need of handling hygroscopic ferrous salts or easily oxidized ferrous solutions, no addition of acid, no need for pure oxygen, and fast treatment rate make membrane AFT one of the Fenton technologies best suited for on-site treatment of pesticide wastewater. A kinetic model for AFT, based on 2,4-D degradation kinetics, was previously developed (19). This model was found to fit the AFT degradation kinetics of pure and formulated diazinon (20), carbaryl (18), carbofuran (21), and several other carbamate insecticides (22). Metribuzin (4-amino-6-(1,1-dimethylethyl)-3-(methylthio)1,2,4-triazin-5(4H)-one) is a 1,2,4-triazinone herbicide used worldwide for pre- and post-emergence control of many grasses and broad-leaved weeds in the cultivation of soya beans, potatoes, grains, beets, and sugarcane (23, 24). It has been frequently detected with other heavily used pesticides in surface water (2, 4, 25). A comparative study showed that metribuzin is significantly more toxic to five species of macrophytes and six species of algae than atrazine, a popular triazine herbicide, and much more toxic than alachor and metolachlor, two widely used chloroacetanilide herbicides (26). An early mammal test showed that metribuzin at 200 mg/kg or greater can cause a large increase in serum glutamicpyruvic transaminase (GPT) activity of mice within 24 h (27). On the basis of the potential hazardous effects of metribuzin on the environment and on human health, a study of its degradation by membrane AFT and microbial methods was recently carried out in our laboratory (28). Unlike previous pesticides that we have investigated, metribuzin does not obey the AFT kinetic model. The work described herein investigates this lack of fit and proposes that it is due to a weak interaction between metribuzin and ferric ion, which can significantly reduce the availability of metribuzin for reaction with hydroxyl radicals in the AFT system. The objectives of this paper are (i) to verify the existence of a weak interaction between metribuzin and ferric ion in the AFT system, (ii) to develop a revised AFT kinetic model for metribuzin by adding this weak interaction to the original AFT model, (iii) to investigate the temperature effect on degradation of metribuzin by AFT, and (iv) to examine the existence of this weak interaction between ferric ion and other pesticides with structures similar to that of metribuzin.
Materials and Methods Chemicals. Metribuzin (99.5%), ametryn (99.1%), simetryn (99.5%), prometon (99%), terbumeton (99%), atrazine (99%), cyanazine (99%), metamitron (99%), and carbaryl (99%) were purchased from Chem Service (West Chester, PA). Hydrogen peroxide (AR), acetonitrile (HPLC grade), water (HPLC grade), potassium permanganate (AR), and hydroxylamine (AR) were purchased from Mallinckrodt (Paris, KY). Sodium chloride (certified), phosphoric acid (AR), and ammonium acetate (certified) were purchased from Fisher Scientific (Fair Lawn, NJ). Gallic acid (97%), 2,4-D (98%), and 1,10-phenathroline (99%) were purchased from Aldrich Chem (Milwaukee, WI). Acetic acid (GR) and hydrochloric acid (GR) were purchased from EM Science (Gibbstown, NJ). Ferrous ammonium sulfate (certified) was purchased from GFS Chemicals (Columbus, OH). Methylene chloride (AR) was purchased from J.T. Baker (Phillipsburg, NJ). Deaminated metribuzin (metribuzin-DA) (99%) was donated from Bayer Co. (Stillwell, KS). VOL. 38, NO. 4, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1221
Anion exchange membrane (ESC-7001) with electrical resistance of 8 Ω cm-2 in 1 M NaCl solution at 25 °C was purchased from Electrosynthesis (Lancaster, NY). Membrane AFT Treatment. All degradation experiments were carried out in an H-shaped apparatus, which was illustrated in our previous work (18). Two 300-mL glass cells separated by anion exchange membrane were used as anodic and cathodic half-cells. Two hundred milliliters of 100 µM metribuzin or other pesticide solution with 0.02 M NaCl was added into the anodic half-cell and an equivalent volume of 0.08 M NaCl solution was added into the cathodic half-cell. Each half-cell was stirred using a magnetic stirring bar. An iron plate (2 × 10 × 0.2 cm) and a graphite rod (1(i.d.) × 10 cm) were used as the anode and cathode, respectively. Electrolysis current was controlled at 0.050 A by a BK Precision dc power supply 1610. Hydrogen peroxide at 0.311 M was pumped into the anodic half-cell at 0.50 mL min-1 using a Fisher peristaltic pump. The ratio of H2O2:Fe2+ was 10:1. At different treatment times, 1 mL of anodic solution was taken into a GC-vial containing 0.10 mL of methanol for HPLC analysis. The treatment temperature was controlled typically at 24.0 ( 0.1 °C by a HAAKE K20 water circulator. Treatments were repeated for a total of three replicates. For the treatment at different delivery rates of Fenton reagent, the electrolysis current was controlled at 0.010, 0.020, 0.030, 0.050, and 0.070 A. Hydrogen peroxide concentration was correspondingly 0.062, 0.124, 0.187, 0.311, and 0.435 M to keep the same ratio of H2O2:Fe2+. To investigate temperature dependency, treatments were performed additionally at 16 and 32 °C. Concentration Analysis. During a separate batch of AFT treatment with metribuzin, samples from the anodic halfcell were promptly taken out and immediately added into flasks containing acetate buffer and 1,10-phenathroline for instantaneous ferrous ion concentration determination. The concentration of total iron ions was analyzed using the same method after reduction of ferric (29). Hydrogen peroxide concentration was analyzed by titration using standard potassium permanganate solution (30). Concentration of pesticides was determined by a HP 1090 HPLC equipped with a diode array detector. Acetonitrile and water (pH was adjusted to 3.0 using H3PO4) at different ratios were used as the mobile phase and a C18-5µm-250 mm (L) × 4.6 mm (i.d.) PRISM RP column was used for separation. For detection of 2,4-D and carbaryl, the working wavelength of DAD was set at 280 ( 20 nm with the reference at 450 ( 80 nm. For detection of other pesticides, the wavelength was set at 225 ( 20 nm with the reference at 450 ( 80 nm. For metribuzin, metribuzin-DA, atrazine, cyanazine, and carbaryl, the ratio of acetonitrile to water was 50:50 and their retention times were 5.8, 4.1, 8.0, 6.8, and 6.9 min, respectively. For simetryn and metamitron, the ratio was 45:55 and their retention times were 3.8 and 3.9 min, respectively. For prometon and terbumeton, the ratio was 40:60 and their retention times were 4.1 and 3.9 min, respectively. For ametryn and 2,4-D, the ratio was 55:45 and their retention times were 4.8 and 10.25 min, respectively. Previously Developed AFT Kinetic Model. A detailed derivation of the AFT kinetic model was previously reported (19). For better understanding the results discussed in this paper, a brief description of the AFT model is presented here. During AFT treatment, the ferrous ion is generated at v0 (µM min-1) and is consumed by continuous reaction with hydrogen peroxide. It is assumed that the concentration of ferrous is constant,
[Fe2+] ) v0π
(1)
where [Fe2+] is the instantaneous concentration of ferrous 1222
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 4, 2004
ion and π is the average life of ferrous ion in the reaction system (min). To achieve optimal treatment efficiency, hydrogen peroxide, which is continuously added into the AFT system at a constant rate, is controlled to be in excess of ferrous ion; thus, hydrogen peroxide can be gradually accumulated. Then, we assume that the hydrogen peroxide concentration increases linearly with treatment time and can be described as
[H2O2] ) ωv0t
(2)
where ω is a constant related to the delivery ratio of hydrogen peroxide to ferrous ion and to the consumption ratio of hydrogen peroxide; t is treatment time (min). Since the Fenton reaction obeys second-order kinetics, the generation rate of hydroxyl radical can be written as
) k [Fe (d[‚OH] dt )
2+
1
g
][H2O2] ) k1πωv02t
(3)
where k1 is the rate constant of the Fenton reaction (µM-1 min-1). The reaction of hydroxyl radical with the target compound also obeys second-order kinetics. The degradation rate of the target compound can be described as
-
d[D] ) k[‚OH][D] dt
(4)
where [D] and [‚OH] are the concentrations of target compound and hydroxyl radical (µM), respectively; k is the reaction rate constant for this reaction (µM-1 min-1). Since hydroxyl radicals are very reactive with other species as well as themselves, their lives in the reaction system are very short. Thus, it is assumed that the instantaneous concentration of hydroxyl radical is proportional to its generation rate. Then eq 4 can be written as follows,
-
(
)
d[D] d[‚OH] [D] ) kk1λπωv02t[D] ) kλ dt dt g
(5)
where λ is the average life of the hydroxyl radical (min). After integration of eq 5, the AFT kinetic model was obtained describing the concentration changes of the target compound during AFT.
( )
ln
[D]t
[D]0
1 ) - Kλπωv02t2 2
(6)
where K ) kk1 (µM-2 min-2) and [D]t and [D]0 are the target concentrations at t and 0 min.
Results and Discussion Metribuzin Degradation Kinetics. The AFT kinetic model was previously developed based on the degradation of 2,4-D by AFT; it was further found to fit several other pesticides well (18-22). Recent work in our laboratory combining AFT and microbial degradation of metribuzin (28) suggested that the degradation kinetics of metribuzin by AFT do not obey the AFT model well. A broad-scale study of AFT metribuzin degradation was undertaken to understand this deviation from the model and is reported in this paper. The full-scale fitting results of metribuzin degradation using the AFT model (dashed line in Figure 1) show a slower degradation than the experimental points in the first 2 min of treatment and a faster degradation than experimental points after 2 min, demonstrating a significant difference between experimental data and the model.
FIGURE 1. Degradation kinetics of metribuzin by membrane AFT and attempted fit to the AFT model. To elucidate the cause of this poor fit, the first four experimental data points (open circles in Figure 1) were regressed using the AFT model and the fitting result was extended to 0 µM metribuzin concentration (shown as a dotted line in Figure 1). The AFT model appears to fit the first four experimental points well with a regression coefficient (r) of 0.99. An extension of the model using this partial fitting produces metribuzin concentrations below those of the experimental data. The concentration difference becomes increasingly significant with treatment time from 2 to 4 min and begins to be less significant when the metribuzin concentration becomes very low after 4 min of treatment, implying that some species is gradually generated and accumulated during the metribuzin AFT. This species gradually slows down metribuzin degradation, causing the metribuzin degradation kinetics to digress from the AFT model. On the basis of the chemical species present in the metribuzin AFT system, we believe that there are only three possible explanations for the AFT kinetics of metribuzin. The first is a possible combination of gradually accumulated metribuzin degradation product(s) with ferrous ion, decreasing the generation rate of hydroxyl radicals, and thus gradually decreasing the degradation rate of metribuzin. The second possible explanation is the protonation of the weakly alkaline metribuzin, making metribuzin difficult to degrade by hydroxyl radicals in the pH-decreasing environment. It is observed that, during the AFT treatment of metribuzin at 0.050 A, the anodic pH decreases from 5.9 at 0 min to 3.3 at 10 min. The third potential explanation is a possible weak interaction between metribuzin and ferric ion, causing a decrease in metribuzin availability to hydroxyl radicals. If this weak complex of metribuzin with ferric ion is dissociated in the column during HPLC analysis, the analytical results will show total nondegraded metribuzin concentration for each sample. Figure 2 shows the concentration of total ferrous ion as well as total iron ion during the AFT treatment of metribuzin. The low and constant concentration of ferrous ion illustrates that it does not accumulate because of a possible combination with metribuzin or its degradation product(s). Thus, the first possibility, that is, the competition for ferrous ion, either does not happen or is too weak to affect the Fenton reaction and the generation rate of hydroxyl radical. The second possible explanation of the AFT kinetics of metribuzin was examined by investigating the degradation of metribuzin by AFT with different initial pH values (shown in Figure 3). The degradation kinetics do not obey the AFT model at either pH. The degradation of metribuzin with an initial pH at 3.3 is significantly faster than that at 5.9,
FIGURE 2. Concentration changes of total ferrous ion and total iron ions during membrane AFT of 100 µM metribuzin at 0.050 A.
FIGURE 3. Degradation of metribuzin by membrane AFT with different initial pHs. Points are experimental data and dotted lines are fitting results using the AFT model. confirming that the protonation of metribuzin does not decrease its reaction with the hydroxyl radical, and the poor fit of metribuzin kinetics is not caused by the decreasing pH during AFT treatment. If the poor fit of metribuzin degradation kinetics to the AFT model is caused by the third possibility, namely, a weak interaction of metribuzin with ferric ion, the metribuzin kinetics should fit the AFT model when a ligand is added that chelates the ferric ions generated from the Fenton reaction, thus preventing the combination of metribuzin with ferric ion. The ideal ligand should meet the following four criteria: (i) possess an appropriate ability to chelate ferric ion; (ii) be inert to ferrous ion or form a complex with ferrous ion that is too unstable to affect the Fenton reaction; (iii) exert no pH-increasing effect that can precipitate ferrous ion and greatly slow the Fenton reaction; (iv) be nonreactive to hydroxyl radicals. However, it is very difficult to find a ligand that meets all four criteria. We investigated inorganic ligands such as fluoride ion and phosphate ions; they meet criteria (i) and (iv), but fail to meet the others (experimental data not shown). Gallic acid was found to be an effective ligand chelating ferric ion while exerting no significant hindrance on the Fenton reaction (31, 32), and was chosen as the appropriate ligand to test our hypothesis, although it is reactive to hydroxyl radicals. The degradation of 100 µM metribuzin by membrane AFT with and without the presence of 300 µM gallic acid is shown in Figure 4a,b. The initial pH of the pure metribuzin solution was adjusted to 3.9, the same as the solution with gallic acid. VOL. 38, NO. 4, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1223
uncombined metribuzin (µM), respectively. The total metribuzin concentration in the AFT system is composed of the concentration of uncombined and combined metribuzin with ferric ion and can be expressed as
[D] ) [D]free + [D - Fe3+]
(8)
where [D - Fe3+] is the concentration of combined metribuzin with ferric ion. With substitution of eq 8 into eq 7, it can then be written as
-
d[D] ) kk1λπωv02t{[D] - [D - Fe3+]} dt
(10)
There is a possibility that the combination of metribuzin with ferric ion is a multiple-ligand complex. To simplify the model, we assume that the combination is a single-ligand complex which obeys the following equilibrium,
[D - Fe3+] [D]free[Fe3+]
) KD-Fe3+
(11)
where KD-Fe3+ is the equilibrium constant (µM-1). Equation 10 can then be written as
-
FIGURE 4. Degradation kinetics of 100 µM metribuzin by membrane AFT with (a) and without (b) 300 µM gallic acid. Initial pH of pure metribuzin solution was adjusted to 3.9, the same as the solution containing gallic acid. A significantly lower degradation of metribuzin in the first 2 min and a much faster degradation after 3 min were observed after the addition of gallic acid as compared with the degradation kinetics of pure metribuzin. A complete removal of metribuzin was achieved within a shorter time when gallic acid was added. It might well be that the relatively lower degradation rate of metribuzin at the beginning of the treatment was caused by the competition of gallic acid with metribuzin for hydroxyl radicals. Assuming that this competition does exist, the faster removal of metribuzin when gallic acid is present strongly implies that no metribuzinFe3+ interaction is taking place under these circumstances and that there is full availability of metribuzin for reaction with hydroxyl radicals. Though metribuzin degradation kinetics with gallic acid do not show a very good fit to the AFT model, it is reasonable to assume that the kinetics would follow the AFT model if gallic acid did not compete for hydroxyl radicals. Further Development of the AFT Kinetic Model. The above analysis and the experimental results strongly suggest a weak interaction between metribuzin and ferric ion that would cause some unavailability of metribuzin to hydroxyl radicals in the AFT system and thus result in a poor fit of metribuzin degradation kinetics to the AFT model. To better understand the degradation of metribuzin by AFT and the interaction of metribuzin with ferric ion, the original AFT kinetic model was further developed, taking into account the availability of metribuzin during AFT treatment. Since only uncombined metribuzin is available to hydroxyl radicals, eq 5 should be written as
-
d[D] ) kk1λπωv02t[D]free dt
(7)
where [D] and [D]free are the concentrations of total and 1224
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 4, 2004
d[D] [D] ) kk1λπωv02t dt 1 + KD-Fe3+[Fe3+]
(12)
The concentration of ferric ion in the AFT system depends not only on its generation from the Fenton reaction but also on the anodic solution pH. For simplification purposes, we assume that the ferric ion concentration is proportional to the product of the electrolysis current and treatment time. Thus,
[Fe3+] ) ηv0t
(13)
where η is the ratio of ferric ion concentration to total iron ion concentration during the AFT and is a constant less than 1. Then, eq 12 can be written as
-
d[D] [D] ) kk1λπωv02t dt 1 + KD-Fe3+ηv0t
(14)
After integration of eq 14, we obtain an equation describing the concentration changes of metribuzin during membrane AFT treatment as follows,
[D]t a a t ln ) - 2 ln a + bt b [D]0 b
(
)
(15)
where [D]0 and [D]t are metribuzin concentrations at 0 and t min, respectively; a ) 1/(Kλπωv02); b ) ηKD-Fe3+/(Kλπωv0); and K ) kk1. We call this kinetic model the revised AFT model. After regression of experimental data using eq 15, values of a and b can be obtained. Since the delivery rate of ferrous ion, v0, is known, the value of Kλπω can be calculated from a. The value of KD-Fe3+ η can then be derived by substituting the value of Kλπωv0 into b. Degradation of Metribuzin at Different Fenton Reagent Delivery Rates and Model Fitting. Degradation of metribuzin by membrane AFT at different delivery rates of Fenton reagent is shown in Figure 5. The revised AFT kinetic model appears to fit the experimental data quite well. Regression coefficients (r) are all above 0.999 (Table 1). Values of Kλπω and ηKD-Fe3+, as well as their P values, are also listed in Table 1. As previously shown with the degradation kinetics of 2,4-D (19) and carbaryl (18), an increase in Fenton reagent delivery rate can efficiently accelerate the degradation of metribuzin,
TABLE 1. Regression Results of Metribuzin Degradation Kinetics by Membrane AFT at Different Delivery Rates of Fenton Reagent Using the Revised AFT Model electrolysis current (A)
delivery rate of Fe2+ (µM min-1)
Kλπω (µM-2)
P value of Kλπω
ηKD-Fe3+ (µM-1)
P value of ηKD-Fe3+
regression coefficient, r
0.010 0.020 0.030 0.050 0.070
15.6 31.1 46.6 77.8 108.8
(3.69 ( 0.35) × 10-4 (1.70 ( 0.14) × 10-4 (1.66 ( 0.15) × 10-4 (1.31 ( 0.10) × 10-4 (1.12 ( 0.09) × 10-4
