Environ. Sci. Technol. 2004, 38, 1807-1812
Applying the Nernst Equation To Simulate Redox Potential Variations for Biological Nitrification and Denitrification Processes C H E N G - N A N C H A N G , * ,† HONG-BANG CHENG,† AND ALLEN C. CHAO‡ Department of Environmental Science, Tunghai University, Taichung City, 407 Taiwan, and Department of Civil Engineering, North Carolina State University, Raleigh, North Carolina 27695
In this paper, various forms of Nernst equations have been developed based on the real stoichiometric relationship of biological nitrification and denitrification reactions. Instead of using the Nernst equation based on a one-toone stoichiometric relation for the oxidizing and the reducing species, the basic Nernst equation is modified into slightly different forms. Each is suitable for simulating the redox potential (ORP) variation of a specific biological nitrification or denitrification process. Using the data published in the literature, the validity of these developed Nernst equations has been verified by close fits of the measured ORP data with the calculated ORP curve. The simulation results also indicate that if the biological process is simulated using an incorrect form of Nernst equation, the calculated ORP curve will not fit the measured data. Using these Nernst equations, the ORP value that corresponds to a predetermined degree of completion for the biochemical reaction can be calculated. Thus, these Nernst equations will enable a more efficient on-line control of the biological process.
Introduction Conventional activated sludge systems have been used for many decades for removing carbonaceous organic substances from wastewaters. In recent years, systems that combine anoxic and oxic units (A/O systems) for accomplishing removal of both carbonaceous and nitrogenous organics are becoming popular. Nitrogen removal in the biological A/O system is achieved through oxic nitrification followed by anoxic denitrification. Although large wastewater treatment plants use continuous-flow processes, medium or small wastewater treatment plants often use batch or sequencing batch reactors. Unlike the continuous-flow reactor, a steady state is never established in these reactors. The reaction time is the most important factor to affect the design, operation, and cost-effectiveness of batch or sequencing batch reactors. Currently, most plants use a fix-time strategy to operate the reactor at a predetermined reaction or hydraulic detention time. With expected short- or long-term variations of the incoming wastewater characteristics and/or other conditions, * Corresponding author phone: +886 4 2359 3660; fax: +886 4 2359 0876; e-mail:
[email protected]. † Tunghai University. ‡ North Carolina State University. 10.1021/es021088e CCC: $27.50 Published on Web 02/06/2004
2004 American Chemical Society
this fix-time method may not yield the optimum results. Dynamically adjusting the reaction time in response to changes of wastewater characteristics and other conditions for the intended biochemical reaction is more efficient and cost-effective (1-3). This real-time control strategy will result in great savings in the design and operation of batch and sequencing batch reactors. Using system-monitoring parameters (e.g., oxidationreducation potential (ORP), pH, and DO) as indicators of reaction progress has been investigated for the real-time control strategy. Andreottola et al. (4) operated a sequencing batch reactor (SBR) system with on-line control of ORP, pH, and DO for treating the high ammonia wastewater from a wood factory. A nearly complete removal of ammonia (99%) was reported at 20 °C. The on-line control allows the applied load to be doubled without causing an obvious decrease in the ammonia removal efficiency. Lee et al. (5) used the measured on-line ORP, pH, and DO as real-time control parameters to adjust the duration of each operation phase in an A/O SBR. The SBR that is operated with real-time control shows a better performance in the removal of phosphorus and nitrogen than the SBR with fixed-time operation. The on-line monitored values of ORP, pH, and DO were reported to be indicative of the dynamic nutrient concentration variations in the SBR studied. Using the ORP measurement as the on-line control of the biological nitrification/denitrification process depends on the detection of an obvious change of the measured system ORP. Under normal loading conditions, the system ORP of a nitrification process continues to rise and then approaches a plateau when the biochemical process is completed. This trend of ORP variation offers an end point, referred to as the RORP,RO2 by Paul et al. (6), that corresponds to the sudden change of the measured system ORP. During the denitrification process, the system ORP decreases initially at a slower rate but rapidly drops when the process approaches the end point. Thus, a change of the ORP curve seen in the first or the second derivative of the ORP curve has been proposed as an indication of the end point for both the nitrification and the denitrification processes. However, for biological systems under high or low C/N loadings, the measured system ORP may not show an obvious end point even though the chemical process is completed. Kim and Hao (7) have reported that, for low influent ratios of COD to total Kjeldahl nitrogen (TKN) concentration (COD:TKN ratios), the predicted reaction time for the anoxic stage is much longer than the actual time needed to complete the denitrification process while at high COD:TKN ratios the measured ORP curve shows no obvious end point known as the knee point. Paul et al. (6) also reported similar experience in determining the end point based on ORP measurements for the oxic and anoxic processes under various high and low loading conditions. The interference from the residual oxygen has been cited to cause the difficulty in using the ORP measurement as an on-line control of the biological system. The nitrification and denitrification processes are oxidation-reduction reactions in which electrons are transferred from the reducing to oxidizing agents until the reaction reaches an equilibrium. The electrochemical potential between the reducing and the oxidizing agents is known as the ORP that measures the net potential of the system (8). The measured ORP values in an aqueous system are related to the changing concentrations of the reducing and the oxidizing species with the Nernst equation (9):
E ) E0 + (RT/nF) ln([Oxi]/[Red]) VOL. 38, NO. 6, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
(1) 9
1807
where E is the ORP (mV), E0 is the standard ORP for the given oxidation-reduction process (mV), R is the gas constant (8.314 J mol-1 K-1), T is the absolute temperature (K), n is the number of electrons transferred in the reaction, F is the Faraday constant (96 500 C mol-1), [Oxi] is the concentration of oxidation agent, and [Red] is the concentration of reduction agent. This form of Nernst equation based on a 1:1 stoichimetric ratio has been used as bases for developing empirical formulas to delineate the reduction of a target chemical species versus the measured solution ORPs in an oxidationreduction system. In this approach, key oxidizing and reducing agents are identified (e.g., NH3 and NO3-), and their concentrations are substituted in the above form of Nernst equation. Since a one-to-one stoichiometric relationship is assumed for all chemical species involved in reaction, this form of Nernst equation may not fit laboratory and/or field data for complicate real-world nitrification/denitrification reactions. Using ORP Based on Stoichiometric Relationship of the Species Involved in the Reaction. The measured ORP is a quantitative indicator of the degree of completion of the chemical reaction in question. It is undoubtedly a valuable parameter for real-time on-line control of the nitrification and denitrification processes. However, there are two major flaws for the previously proposed use of the ORP measurement. First, if the Nernst equation is based on a 1:1 ratio stoichiometric relationship of the species involved in the chemical equation, it will not truly relate the measured ORP values to indicate the degree of chemical reaction. Second, a sudden change of the system ORP curve (e.g., the bending point, RORP,RO2) obtained by a sign change of the second derivative of the ORP curve is empirical (10, 11). It is not be a direct indication of the completion of the biological reaction. This may be the reason that the current ORP control strategy does not consistently yield satisfactory results for all nitrification and denitrification processes. Instead of using the knee point to detect the completion of the biological reaction, the authors have proposed a new control strategy using the Nernst equation to calculate the ORP value for any degree of reaction completion. The biochemical process will be terminated when the measured system ORP value approaches a predetermined ORP value. Using this strategy, any degree of reaction can be achieved for the nitrification and denitrification processes. Thus, the Nernst equation based on a 1:1 stoichiometric ratio of the chemical species is not appropriate. A more useful Nernst equation should be developed based on considerations of the stoichiometric relationship unique to each chemical reaction in question. This paper presents the assumptions, procedures, and simplifications for deducing different forms of Nernst equation for the nitrification and denitrification processes with the ultimate goal of using ORP as a quantitative real-time process control parameter. Verifying the validity of the developed Nernst equation is included in the paper.
Model Methodology (A) Model Developing of the Various Forms of the Nernst Equation. (i) Nernst Equation for a General OxidationReduction Reaction. A generic oxidation-reduction reaction can be represented as eq 2 (12):
aA + bB S cC + dD
(2)
where A and B are reactants; C and D are products; and a-d are the stoichiometric coefficients for A-D, respectively. All species involved in the chemical reaction are assumed to be dissolved solutes. If solids are involved in the reaction, their concentrations are assumed unchanged during the reaction. Similarly, if the reagent is a pure substance rather than the component of a solution, the partial molar free energy equals 1808
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 6, 2004
to the free energy of 1 mol of the substance. The Nernst equation for a generalized oxidation-reduction reaction is obtained:
E ) E0 -
c d RT (aC) (aD) ln nF (a )a(a )b A
B
or
E ) E0 +
a b RT (aA) (aB) ln c nF (a ) (a )d C D
(3)
(ii) Nernst Equation for the Nitrification Process. The process in which ammonia nitrogen is converted to nitrate is an autotrophic process in which the energy for bacterial growth is derived by the oxidation of nitrogen compounds, primarily in ammonia form. It consists of two steps involving two genera of microorganisms. For the conversion of ammonia nitrogen into nitrite by Nitrosomonas, the following chemical reaction is assumed:
NH4+ + 3/2O2 f NO2- + 2H+ + H2O
(4)
Nitrite is subsequently converted to nitrate by Nitrobactor, as in the following equation:
NO2- + 1/2O2 f NO3-
(5)
For all practical purposes, the overall conversion of NH4+ to NO3- is represented by the following chemical equation:
NH4+ + 2O2 f NO3- + 2H+ + H2O
(6)
Applying the generalized Nernst equation to eq 6, eq 7 is obtained:
(
[NH4+](PO2)2 RT E)E + ln nF [NO -][H+]2[H O] 3 2 0
)
(7)
The above equation could be further modified as
E ) E0 +
(
) ( )
1 RT 2RT RT ln([NH4+]) + ln(PO2) + ln + nF nF nF [NO -] 3 1 1 RT 2RT ln ln + (8) nF [H2O] nF [H+]
( )
Both atmospheric oxygen partial pressure (PO2) and temperature in eq 8 are assumed unchanged during the nitrification process and are replaced by constants. Replacing ln(1/[H+]) for 2.3026 × (pH), eq 9 is obtained:
[NH4+] E ) a′ + b′pH + c′ log [NO3-]
(9)
The three constants a′, b′, and c′ are defined as follows:
a′ ) E0 +
( )
1 2RT RT ln PO2 + ln nF nF [H2O]
b′ )
2.3026 × 2RT nF
c′ )
2.3026 × RT nF
Equation 9 shows that the Nernst equation based on a 1:1
stoichiometric ratio of the chemical species involved in the reaction is applicable to the overall nitrification process. (iii) Nernst Equation for the Denitrification Process. The main biological denitrification to convert nitrate to nitrogen gas under anoxic conditions is heterotrophic and is expressed by the following generic stoichiometric equation (13):
If the quantity of organic substrates that donate electrons in the denitrification process is limited, the variation in the solution pH is insignificant and the pH term in eq 14 can be assumed constant. The Nernst equation can be expressed as
E ) a′′ + c′′ log([NO3-]) + d′′ log([CxHyOz])
(15)
-
5CxHyOz + (4x + y - 2z)NO3 f 5xCO2 + (2y - 2x +
z)H2O + (4x + y - 2z)OH- + (2x + 1/2y - z)N2 (10)
Applying the generalized Nernst equation, the Nernst equation for the denitrification process is 0
(
)
E)E +
[CxHyOz]5[NO3-](4x+y-2z) RT (11) ln 5x nF (P )(P2x+1/2y-z)[H O](2y-2x+z)[OH-](4x+y-2z) CO2 N2 2
or
E ) E0 +
( ) ( )
( )
(5x)RT (2x + 1/2y - z)RT 1 1 ln ln + + nF PCO2 nF PN2
(2y - 2x + z)RT 1 5RT ln ln([CxHyOz]) + + nF nF [H2O] (4x + y - 2z)RT (4x + y - 2z)RT 1 ln([NO3-]) + ln nF nF [OH-]
( ) (12)
A simplified form is obtained with constants i′′, j′′, k′′, and m′′ as defined in the following equations:
i′′ ) E0 +
( )
( ) ( )
(5x)RT (2x + 1/2y - z)RT 1 1 + + ln ln nF PCO2 nF PN 2
j′′ )
(2y - 2x + z)RT 1 ln nF [H2O] 5RT nF
k′′ )
(4x + y - 2z)RT nF
m′′ )
(4x + y - 2z)RT nF
m′′ ln
E ) a′′ + b′′pH + c′′ log([NO3-])
(16)
Both eqs 15 and 16 show different forms of the Nernst equation that is different from the equation derived on the basis of a one-to-one stoichiometic relationship for the denitrification process. In the presence of an excessive carbon source, the Nernst equation includes an extra pH term, while under conditions of limited carbon source the organic carbon is included in the Nernst equation. (B) Model Verification. Relevant experimental results published by other researchers in the literature on nitrification and denitrification processes under different conditions are used to verify the validity of the various Nernst equation forms derived in this paper. All data used in this verification process must be the results of continuously monitoring the reaction over the entire study period. Additionally, the data for all relevant parameters used in the Nernst equation must be available. For example, if the organic substrate for denitrification is excessive, the pH data must be presented in the publication. On the other hand, if the organic substrate for denitrification is limited, concentrations of both nitrate and organic substrate must be available. Among the numerous publications, four sets of data that are sufficiently distinct and meet these requirements have been selected for the subsequent validity verification. One set of data concerns the nitrification process, and the other three sets cover the denitrification process. Since raw data are not available in the literature, all data are digitized based on the figures published. Multiple regression analyses (14) are used to calculate the constants in the various forms of Nernst equation. A good fit between the published ORP data and the calculated ORP variation curve indicates that the proposed ORP equation is valid in predicating the variation trend of the measure ORP values.
Results and Discussion
then eq 12 could be simplified as eq 13:
E ) i′′ + j′′ ln([CxHyOz]) + k′′ ln([NO3-]) +
With excessive carbon substrate in the denitrification process, the substrate concentration can be assumed constant. But the change in pH is too significant to be ignored. Thus, the Nernst equation takes the following form:
( )
1 (13) [OH-]
Replacing 2.3026 × (14 - pH) for ln(1/[OH-]), one obtains
E ) a′′ + b′′pH + c′′ log(NO3-) + d′′ log(CxHyOz) (14) where a′′, b′′, c′′, and d′′ are defined as
a′′ ) i′′ + 2.3026 × 14 × m′′ b′′ ) - 2.3026 × m′′ c′′ ) 2.3026 × k′′ d′′ ) 2.3026 × j′′ Since the denitrification process can be carried out under condition of either excessive carbon source or limited carbon source, different forms of the Nernst equation will be derived.
Nernst Equation for the Nitrification Process. The data published by Yu et al. (15) on continuous-flow SBR to remove organic carbon nitrogen and phosphorus in a single tank are used for verifying the validity of eq 9. Each cycle of the SBR consists of four stages: anaerobic (1 h), aerobic (5.5 h), settling (1 h), and drawing (0.5 h). The nitrification process occurs during the aerobic stage. The resulting ORP equation is expressed as
E ) -153.43 + 20.32pH - 59.88 log([NH4+]/[NO3-]) R 2 ) 0.95 (17) where E is the redox potential (mV). Figure 1 shows the comparison of the original published ORP data and the ORP curve calculated using eq 17. The variation trend of the measured ORP curve is well-predicted in most parts by the calculated ORP curve but with a slight deviation toward the latter part of the ORP curve. The slight deviation may be caused by the errors introduced during the process of digitizing the ORP data from the published data. Additionally, the reported conditions for the nitrification VOL. 38, NO. 6, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1809
FIGURE 1. Simulation of the measured ORP for the biological nitrification in a continuous-flow SBR. Original data were obtained from ref 15.
TABLE 1. Results of Regressive Analyses on the Nernst Equation for Biological Denitrification with Excessive Carbon Sourcea Nernst equation constants substrate mixed substances glycerin ethanol glucose a
a′′
b′′
c′′
R2
-2434.04 -4194.99 826.69 -2135.86
336.61 575.75 -79.72 288.27
98.60 27.73 315.10 9.88
0.99 0.99 0.99 0.99
Original data were obtained from ref 16.
FIGURE 2. Simulation of the measure ORP values for the biological denitrification with excessive carbon source using four different carbon substrates. Original data were obtained from ref 16.
TABLE 2. Nernst Equations in Different Operation Strategies and Reaction Conditionsa Nernst equation constants
phase I phase II shock test a
a′′
b′′
c′′
R2
632.71 32.31 414.96
-49.60 30.75 -56.33
191.65 1651.38 292.98
0.98 0.94 0.99
Original data were obtained from ref 17.
reaction temperature are slightly different from the standard condition of 20 °C and 1 atm. Thus, the calculated ORP curve may not closely fit the measured ORP data as shown in Figure 1. However, the trend of ORP variation follows the pattern as predicted using eq 17. Nernst Equation for the Denitrification Process with Excessive Organic Carbon Sources. Case 1: For the denitrification process under conditions of excessive organic carbon source, the Nernst equation (eq 15) that includes both the pH and the nitrate terms in the following form is used:
E ) a′′ + b′′pH + c′′ log([NO3-])
(18)
The first multiple regression analysis useing the data published by Drtil et al. (16) on the denitrification process in a sequencing batch rector are used. Four types of substrates including mixed substances, glycerin, ethanol, and glucose have been reported in the denitrification study. After linear regression analyses, values of the coefficients are listed in Table 1. Comparisons of the reported lab ORP data as reported by Drtil et al. and the calculated ORP curves using eq 18 for biological denitrification with excessive carbon source are shown in Figure 2. Analyzed results fitted experimental data well for the various substrates used as the carbon sources in the denitrification process. Case 2: Chang and Hao (17) applied pH and ORP as the process parameters in a two-phase nitrification and denitrification. Phase I was operated with a cyclic duration of 8 h (three cycles a day) for completing a series operations of feed, mixed-react, react, settle, draw, and idle. In phase II, a shorter duration (6 h) was used with the same series operations. After phase II, nitrogen shock loading (additional 20 mg/L NH4+-N) was imposed on four consecutive cycles to observe the applicability of using pH/ORP as process control parameters. 1810
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 6, 2004
FIGURE 3. ORP course of denitrification and simulated Nernst equation in different operation strategies and shock loading. Original data were obtained from ref 17. The Nernst equation developed for denitrification with excess carbon source (eq 18) is used to simulate the observed ORP for the denitrification process. The results of multiple regression analyses are listed in Table 2. Comparison of the reported ORP data and the simulated ORP curve is shown in Figure 3. It is obvious from the results that simulated ORP can fit the experiment data for the denitrification process undergoing COD shock loadings. Nernst Equation for the Denitrification Process with Limiting Organic Carbon. Most wetland treatment systems are effective in removing nitrogen because they provide diverse physical, chemical, and biological environments as well as relatively long detention times. Biological denitrification reaction occurs at anoxic microsites adjacent to aerobic sites. In addition to anoxic conditions, a sufficient carbon source is necessary to complete the denitrification reaction. The carbon from decaying vegetation can serve as partial carbon sources, and the critical carbon/nitrogen ratio appears to be 5 when dried plant material was used as the carbon source (18). The wetland system exhibits a strong “buffer”
TABLE 3. Simulated Model of Nernst Equation in Denitrification by Unplanted Cell and Cells with Four Species of Aquatic Planta Nernst equation constants species
a′′
c′′
d′′
R2
unplanted canary reed bulrush
-646.23 -983.97 -2140.19 102.19
-1174.10 -287.08 -210.97 -91.13
1919.11 987.87 3286.44 444.15
0.99 0.98 0.99 0.99
a
Original data were obtained from ref 20.
FIGURE 5. Comparison of the calculated ORP curve using the correct and the incorrect forms of Nernst equations. The dotted data points are measured ORP values of denitrification by canary (20). Curve A shows a perfect fit if the correct form of Nernst equation (eq 16) is used; while curve B shows that, if an incorrect form of the Nernst equation is used (eq 15), the calculated ORP curve does not fit the measured data.
FIGURE 4. ORP curves for denitrification and simulated data by Nernst equation in constructed wetlands. Original data were obtained from ref 20. capacity with respect to pH fluctuation. The humic substances generated within a wetland via growth, death, and decomposition cycles would precipitate under acidic conditions, and their protonated form tends to be less solubility. Consequently, the wetland system is buffered against incoming wastewater (19). The ORP variation is simulated using the form of Nernst equation (eq 16) for relative constant pH and limiting carbon source. The results published by Zhu and Sikora (20) on the removal of nitrogen in a constructed wetland sheltered in a greenhouse are used in the following ORP simulation. Their studies covered four types of vegetation species including canarygrass (Phalaris arundinacea), reed (Phragmites communis Trin.), bulrush (Scirpus atrovirens georgianus), and typha (Typha latifolia). After linear regression analyses, values of the coefficients are listed in Table 3. Graphic comparisons of the simulated ORP variations and the ORP data are shown in Figure 4. For the four types of predominant vegetation, the simulated ORP curve fit the experimental data. Appropriate Functional Form of Nernst Equation. In this study, several forms of Nernst equation have been developed for the different biological nitrification and denitrification processes based on the stoichiometric relationship of the biochemical reactions. Use of the appropriate functional form of Nernst equation is a key to the success of fitting lab data with Nernst equation. If the appropriate functional form of Nernst is used to simulate the variation of ORP at different reaction times, a good fit of the measured ORP with calculated ORP can be obtained (Figures 1-4). However, if an inappropriate functional form of the Nernst equation is used, the measured ORP cannot be fitted by the calculated ORP. Using the denitrification process with limiting organic matter as an example, its ORP variation should be simulated using eq 16. If the form of Nernst equation developed for the denitrification process with excess organic carbon is used for wetland, the regression analysis results show a R 2 value of only 0.25. This is much lower than the R 2 values of 0.99-0.94 usually obtained when the appropriate functional form of the Nernst equation is used. As shown in
Figure 5, the measured ORP data cannot be fitted by the calculated ORP curve (curve B). Bending Point. The main objective of measuring and simulating variations of the various system parameter (e.g., ORP, temperature, pH, etc.) is to use the resulting model for achieving on-line control of the biological chemical reaction. Several researchers have developed the technique to predict a “bending point”, which corresponds to the point at which the target chemical reaction is completed and the operation should be terminated. For the nitrification process, the bending point is obtained by selecting the point at which the value of the first derivative of the pH curve becomes 0 (i.e., dpH/dt ) 0). At this point, the conversion of ammonia nitrogen to nitrate nitrogen is almost completed (21). For the denitrification process, the bending point is obtained by a zero second derivative of the ORP curve or d2 ORP/dt2 ) 0 (7). However, the bending point thus obtained is somewhat empirical because it has not been developed based on sound theoretical considerations. Additionally, not all bending points match the end point of the biological reactions (22, 23). In this study, appropriate functional forms of Nernst equation have been developed based on actual stoichiometic chemical equations for various biochemical reactions. The ORP value for any degree conversion of reactants to products can be precisely calculated. Hence, once the degree of conversion is decided, the ORP value corresponding to the end point can be calculated so that on-line control of the biological system can be effectively practiced.
Acknowledgments We thank the National Science Council, Republic of China, for financial support of this research (NSC 90-2211-E-029004).
Nomenclature E
oxidation reduction potential (ORP) (mV)
E0
standard ORP for the given oxidation reduction process (mV)
R
gas constant (8.314 J mol-1 K-1)
T
absolute temperature (K)
n
number of electrons transferred in the reaction
F
Faraday constant (96 500 C mol-1)
VOL. 38, NO. 6, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1811
aox
activity of oxidation agent
ared
activity of reduction agent
GA
free energy of reagent A (J) 0
GA
free energy of reagent A at standard conditions (J)
Literature Cited (1) Lee, P. G.; Lea, R. N.; Dohmann, E.; Prebilsky, W.; Turk, P. E.; Ying, H.; Whitson, J. L. Aquacult. Eng. 2000, 23, 37-59. (2) Wareham, D. G.; Hall, K. J.; Mavinic, D. S. Water Sci. Technol. 1993, 28, 273-282. (3) Zipper, T.; Fleischmann, N.; Haberl, R. Water Sci. Technol. 1998, 38, 307-314. (4) Andreottola, G.; Foladori, P.; Ragazzi, M. Water Sci. Technol. 2001, 43, 93-100. (5) Lee, D. S.; Jeon, C. O.; Park, J. M. Water Res. 2001, 35, 39683976. (6) Paul, E.; Plisson-Saune, S.; Mauret, M.; Cantet, J. Water Sci. Technol. 1998, 38, 299-306. (7) Kim, H.; Hao, O. J. Water Environ. Res. 2001, 73, 95-102. (8) Kim, Y. H.; Hensley, R. Water Environ. Res. 1997, 69, 10081014. (9) Lie, E.; Welander, T. Water Sci. Technol. 1994, 30, 91-100. (10) Vanrolleghem, P.; Coen, F. Water Sci. Technol. 1995, 31, 149160.
1812
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 6, 2004
(11) Pierson, J. A.; Pavlostathis, S. G. Water Environ. Res. 2000, 72, 585-592. (12) Barrow, G. M. Physical Chemistry; McGraw-Hill: New York, 1988. (13) Drtil, M.; Nemeth, P.; Kucman, K.; Bodik, I.; Kasperek, V. Water Res. 1995, 29, 1353-1360. (14) Chang, C.-N.; Lin, J.-G.; Chao, A. C.; Liu, C.-S. Water Sci. Technol. 1996, 34, 151-157. (15) Yu, R.-F.; Liaw, S.-L.; Chang, C.-N.; Lu, H.-J.; Cheng, W.-Y. Water Sci. Technol. 1997, 35, 57-66. (16) Drtil, M.; Nemeth, P.; Buday, J.; Bodik, I.; Hutnan, M. Chem. Pap. 1999, 53, 75-81. (17) Chang, C. H.; Hao, O. J. J. Chem. Technol. Biotechnol. 1996, 67, 27-38. (18) Ingersoll, T. L.; Baker, L. A. Water Res. 1998, 32, 677-684. (19) Kadlec, R. H.; Knight, R. L. Treatment Wetlands; CRC Press: Boca Raton, FL, 1996. (20) Zhu, T.; Sikora, F. J. Water Sci. Technol. 1995, 32, 219-228. (21) Al-Ghusain, I. A.; Huang, J.; Hao, O. J.; Lim, B. S. Water Sci. Technol. 1994, 30, 159-168. (22) Wareham, D. G.; Hall, K. J.; Mavinic, D. S. J. Environ. Eng. 1993, 119, 120-136. (23) Wouters-Wasiak, K.; He´duit, A.; Audic, J. M.; Lefe`vre, F. Water Sci. Technol. 1994, 30, 207-210.
Received for review December 26, 2002. Revised manuscript received November 21, 2003. Accepted November 24, 2003. ES021088E
