This study examined the effect of initial manganous ion (Mn2+) concentration and oxidant dose on the oxidation of Mn2+ to Mn oxide during water treatment. The oxidants studied were chlorine dioxide (ClO 2 ), potassium permanganate (KMnO 4 ), and ozone. Initial Mn 2+ concentrations were 60, 200, and 1,000 µg/L, and the goal of the treatment was a final Mn2+ <10 µg/L. Bench-scale experiments were performed by applying different doses of each oxidant to a raw surface water and measuring Mn2+ residuals over time. For all experiments, the ambient raw water conditions were pH 7.0, 9oC, and total organic carbon = 3.4 mg/L. Oxidation kinetics for low initial Mn2+ concentrations (60 and 200 µg/L) were significantly slower than for high concentrations (1,000 µg/L). For the lower initial Mn2+ levels, ClO2 was the only oxidant that consistently produced final Mn2+ <10 µg/L, generally within 60–120 s. All oxidants produced final Mn2+ <10 µg/L when the initial Mn2+ concentration was 1,000 µg/L. Oxidation with ozone consistently resulted in soluble Mn2+ >20 µg/L after 5 min, with the formation of permanganate ion in many cases. For low Mn2+ concentrations, 90% oxidation by KMnO4 required 15–30 min. EFFECT OF soluble Mn concentration on oxidation kinetics BY DEAN GREGORY AND KENNETH CARLSON uring water treatment processes and in the distribution system, the manganous ion (Mn2+) oxidizes to Mn oxide (MnO2(s)). The MnO2(s), in turn, causes brown staining of plumbing fixtures and laundry and “brown water” incidents. The Fort Collins (Colo.) Water Treatment Facility (FCWTF) has experienced Mn2+ levels up to 450 µg/L in Horsetooth Reservoir, one of its two surface water supplies. The source of Mn2+ in the reservoir is the chemical reduction of MnO2(s) in the lower depths of the reservoir, which becomes a reducing environment after stratification and deoxygenation of the reservoir’s hypolimnion. The US Environmental Protection Agency has established a secondary maximum contaminant level of 50 µg/L total Mn for finished water because of its nuisance characteristics. The FCWTF, however, has experienced Mn problems from effluent concentrations as low as 20 µg/L. Sly et al (1990) suggested that the ultimate goal for Mn in finished water should be 10 µg/L, and the FCWTF has adopted this goal. Although Mn2+ oxidation has been practiced for years in water treatment, many studies reported in the literature have not fully explored oxidation of low initial Mn2+ concentrations and dose requirements—issues that may be important to water treatment plants. This study sought to address these gaps in knowledge of Mn2+ oxidation for potable water treatment. D 2003 © American Water Works Association 98 JANUARY 2003 | JOURNAL AWWA • 95:1 | PEER-REVIEWED | GREGORY ET AL Dissolved Mn—µg/L Dissolved Mn—µg/L Dissolved Mn—µg/L The objective of Mn removal processes Bench-scale oxidation of Mn2+ by ClO2 for three initial Mn2+ concentrations FIGURE 1 is to oxidize Mn2+ to MnO2(s), a solid precipitate that can be removed by solid/liquid separation processes such as 300% CIO2 400% CIO2 200% CIO2 sedimentation and filtration. Candidate 60 2+ Initial Mn = 60 µg/L oxidants for the reaction include ozone 50 (O3), chlorine dioxide (ClO2), and potassium permanganate (KMnO4). Free chlo40 rine, an obvious alternative for water treatment, is often impractical because 30 the Mn2+ oxidation reaction kinetics are 20 slow, requiring 2–3 h of contact time at pH 8.0 (Knocke et al, 1990). For this rea10 son, chlorine was excluded as a candidate oxidant for this study. 0 0 50 100 150 200 250 300 The overall objective of the study was Reaction Time—s to evaluate ClO2, KMnO4, and O3 for 100% CIO2 200% CIO2 300% CIO2 the oxidation of Mn2+ to <10 µg/L. Spe220 cific objectives were to understand the 2+ Initial Mn = 200 µg/L 200 effect of initial Mn2+ concentration, par180 ticularly low initial concentrations, on 160 oxidant dose. Because this research was 140 conducted for the FCWTF, it also served 120 100 as an engineering analysis of Mn oxida80 tion for the water treatment plant, which 60 is a fairly typical conventional treatment 40 facility. The basic plant processes include 20 coagulation with alum added in a rapid0 mix basin, flocculation/sedimentation 0 50 100 150 200 250 300 (approximately 60-min detention time Reaction Time—s using inclined plates), rapid-rate filtra100% CIO2 tion with dual-media filters (sand/ 1,100 2+ 1,000 anthracite), and disinfection with chloInitial Mn = 1,000 µg/L 900 rine before distribution. The experimen800 tal conditions (pH, total organic carbon 700 [TOC], temperature, alkalinity) were 600 based on the typical value of each con500 stituent observed at the plant during the 400 5 µg/L 300 late summer/fall season when influent 200 Mn concentrations increase. 100 The other treatment plant parameter 0 important to this analysis was available 0 50 100 150 200 250 300 detention time between the raw water Reaction Time—s intake and the addition of alum in the Relative stoichiometric doses are shown; temperature—9 C, pH—7.0, total organic 2+ rapid-mix basin. This reaction time was an carbon—3.4 mg/L; CIO2—chlorine dioxide, Mn—manganese, Mn —manganous ion important consideration because the FCWTF had observed, while using a temporary KMnO4 system, that depression µg/L. Before this study was conducted, the FCWTF used of the pH from 7.0 to 6.3–6.4 in the rapid mix effectively KMnO4 for Mn oxidation and removal, but results were halted the KMnO4–Mn2+ reaction. During high-flow periods, the detention time is only 120–180 s. Detention time often unsatisfactory in terms of finished water Mn conis approximately 180–300 s during periods of high Mn concentrations, particularly for low influent dissolved Mn centration. Two criteria for selecting the appropriate chemlevels. The FCWTF decided to reassess its use of KMnO4 as well as to consider the alternatives of O3 and ClO2. ical oxidant for the FCWTF were whether the Mn2+–oxidant reaction was completed within approximately 180 s Another Mn removal option considered by FCWTF and whether it produced a final Mn2+ residual of <10 was using MnO2(s)-coated filter media, or the “greeno 2003 © American Water Works Association GREGORY ET AL | PEER-REVIEWED | 95:1 • JOURNAL AWWA | JANUARY 2003 99 TABLE 1 Water quality characteristics of Horsetooth Reservoir Constituent Range pH 7.0–7.1 Temperature—oC 9–10 Alkalinity—mg/L as CaCO3 25–30 Total organic carbon—mg/L 3.4–3.6 SUVA*—L/mg-min 1.7 Dissolved oxygen—mg/L 1.8–2.1 Ambient Mn2+†—µg/L 10–13 Turbidity—ntu 3.5–4.5 *SUVA—specific ultraviolet absorbance †Mn2+—manganese sand” process, a method with demonstrated effectiveness in removing relatively high concentrations of dissolved Mn (Coffey et al, 1993; Knocke et al, 1991a). In this process, Mn2+ adsorbs to the MnO2(s)-coated surface of the media and is subsequently oxidized by free chlorine. This method, however, appears to be most effective for groundwater applications in which the influent dissolved Mn concentration is relatively high (>1 mg/L) and constant throughout the year. Because the dissolved Mn in Horsetooth Reservoir is only an issue for approximately three months of the year and the influent dissolved Mn levels vary during that period, preoxidation appeared to be a more effective alternative. Preoxidation could also provide treatment options for other challenging water quality conditions such as taste and odor compounds. Literature review. It has been shown that the oxidation of Mn2+ by a reactive oxidant involves three mechanisms (Morgan, 1967): solution-phase oxidation, adsorption of the Mn2+ ion to MnO2(s), and surface-catalyzed oxidation of the sorbed Mn2+ ion. These mechanisms are represented by Eqs 1–3. k1 [Mn2+] + [Ox] → [MnO2] [Mn2+] (1) k2 + [MnO2] → [MnO2 ⬅ Mn] k3 [MnO2 ⬅ Mn] + [Ox] → [MnO2] (2) (3) in which MnO2 ⬅ Mn represents MnO2(s) with sorbed Mn2+ ion (Van Benschoten et al, 1991) and Ox is oxidant. Van Benschoten et al (1991) developed rate equations and a kinetic model based on Eqs 1–3 that predict rapid oxidation of Mn2+ by ClO2 and KMnO4—nearly complete oxidation of initial Mn2+ within 30 s. In a study that focused on KMnO4, Carlson and Knocke (1999) showed that the model could be expanded to better characterize Mn2+ oxidation in a natural water that contained significant concentrations of natural organic matter (NOM). KMnO4 oxidation rates were found to be significantly lower in that study. The rate equations were expanded to include a term that accounted for dissolved organic carbon (DOC) concentration, because the DOC concentration in the experiments Van Benschoten et al (1991) used to develop their model was <1 mg/L. Carlson and Knocke (1999) suggested that DOC interfered in the KMnO4–Mn2+ reaction by exerting a KMnO4 demand and possibly complexing a small fraction of Mn2+, rendering it less likely to be oxidized. This approach resulted in a model that more closely characterized Mn 2+ oxidation by KMnO4 in the raw surface water used in the study. The work of Knocke et al (1990) is the most comprehensive investigation of Mn2+ oxidation for water treatment. In addition to their kinetic modeling work, the authors examined the effect of pH, DOC, temperature, and initial Mn2+ concentrations on the oxidation of Mn2+ by ClO2, KMnO4, O3, and other oxidants. They concluded that Mn2+ was oxidized rapidly by ClO2, KMnO4, and O 3 in low-DOC waters over a pH range of pH 5.5–9.0. Two factors, however, complicate the application of these results to water treatment. One is that the lowest concentrations used in the ClO2, KMnO4, and O3 experiments were 500, 200, and 850 µg/L, respectively. Experiments that examined the effect of independent variables such as pH, DOC, and temperature used initial Mn2+ concentrations in the range of 900–1,000 µg/L. Many water treatment facilities must practice Mn2+ oxidation and removal at much lower concentrations, e.g., 50 µg/L, to avoid aesthetic problems. Because the authors did show that Mn2+ oxidation rates clearly decreased with decreasing initial Mn2+ concentrations, additional information on the effect of lower concentrations could benefit the water supply industry. A second factor that could complicate the application of these results to water treatment plants is that although oxidation rates appear to be rapid for all three oxidants, the final Mn2+ residuals are often unacceptable, in terms of full-scale treatment. In the case of KMnO4, Mn2+ residuals were generally in the range of 50–150 µg/L. As discussed earlier, FCWTF desires finished water Mn2+ <10 µg/L. In ClO2 experiments, Mn2+ residuals close to 0 µg/L were only observed when DOC concentrations were <1 mg/L and initial Mn2+ concentrations were approximately 900 µg/L. For other conditions, i.e., lower initial Mn2+ concentrations and the presence of significant DOC, final Mn2+ residuals ranged from 50 to 400 µg/L. Paillard et al (1989) examined, among other things, the effect of TOC on the oxidation of Mn2+ by O3. They performed bench-scale experiments that simulated a conventional full-scale process, including sand filtration, with 2003 © American Water Works Association 100 JANUARY 2003 | JOURNAL AWWA • 95:1 | PEER-REVIEWED | GREGORY ET AL FIGURE 2 300% KMnO4 2+ Initial Mn = 60 µg/L 80 70 60 50 40 30 20 10 0 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 Reaction Time—s MATERIALS AND METHODS 100% KMnO4 220 200 Dissolved Mn—µg/L 200% KMnO4 300% KMnO4 2+ Initial Mn = 200 µg/L 180 160 140 120 100 80 60 40 20 0 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 1,400 1,600 1,800 2,000 Reaction Time—s 100% KMnO4 1,000 900 Dissolved Mn—µg/L Experimental water. The raw water used for all experiments was from Horsetooth Reservoir in Fort Collins. This water originates from high alpine sources near the Continental Divide in the Rocky Mountains. Table 1 provides the water quality characteristics that were used for all bench-scale experiments. Temperature (9 o C), pH (7.0), and TOC (3.4 mg/L) were nearly constant for all experiments. The specific ultraviolet absorbance (SUVA) value for this water is also provided in Table 1. SUVA, the UV absorbance at 254 nm (m–1) divided by the DOC concentration (mg/L), is an indicator of the aromatic character of NOM (Edzwald & Van Benschoten, 1990). The SUVA value increases with an increasing fraction of aromatic or double bond–rich substances that tend to absorb more strongly at 254 nm. The SUVA value of 1.7 L/mg-m for Horsetooth Reservoir is relatively low and indicates that a significant fraction of the NOM is hydrophilic in nature. The low dissolved oxygen values shown in Table 1 are due to stratification of the reservoir in late summer and fall. Generation of oxidant stock solutions/analytical methods. Ozone was produced using a 1-lb/d (0.45 kg/d) pilotscale generator.1 A stainless-steel gas conduit system was constructed to deliver an O3/oxygen gas blend to the Bench-scale oxidation of Mn2+ by KMnO4 for three initial Mn2+ concentrations 100% KMnO4 90 Dissolved Mn—µg/L O3 applied to settled water. They showed that for the same mg O3:TOC concentration ratio, increasing the TOC concentration increases the residual Mn2+ concentration. The initial Mn2+ concentration used was 250 µg/L. Another study that investigated the effect of organics on the oxidation of Mn2+ by O3 offered a contradictory finding. Seby et al (1995) studied the influence of fulvic acid concentrations (0.5–5.0 mg/L) on Mn2+ oxidation efficiency in a synthetic water. They reported that 94% of initial Mn2+ was oxidized at pH 8, regardless of fulvic acid concentration. The authors concluded the oxidation of Mn2+ by O3 has priority over those of fulvic acids. These experiments used a relatively high initial Mn2+ concentration of 660 µg/L. 2+ Initial Mn = 1,000 µg/L 800 700 600 500 400 300 200 100 0 0 200 400 600 800 1,000 1,200 Reaction Time—s o Relative stoichiometric doses are shown; temperature—9 C, pH—7.0, total organic carbon—3.4 mg/L; KMnO4—potassium permanganate, Mn—manganese, 2+ Mn —manganous ion 2003 © American Water Works Association GREGORY ET AL | PEER-REVIEWED | 95:1 • JOURNAL AWWA | JANUARY 2003 101 FIGURE 3 Bench-scale oxidation of Mn2+ by O3 for three initial Mn2+ concentrations O3 = 0.2 mg/L 70 O3 = 0.5 mg/L O3 = 1.5 mg/L O3 = 2.0 mg/L 2+ Initial Mn = 60 µg/L Dissolved Mn—µg/L 60 50 40 30 20 10 0 0 220 200 100 200 O3 = 0.5 mg/L 300 400 Reaction Time—s O3 = 0.75 mg/L 500 O3 = 1.0 mg/L 600 700 O3 = 2.1 mg/L 2+ Initial Mn = 200 µg/L Dissolved Mn—µg/L 180 about 60 mg/L to about 40 mg/L during this time. The O3 stock concentration was measured before each experiment. The aqueous O3 concentration in the O3 stock solutions and the residual O3 in experiments were measured by the indigo trisulfonate spectrophotometric method (Bader & Hoigné, 1981). ClO2 stock solutions were generated by a solid-phase ClO2 generator2 in which a 4% Cl2:96% N2 gas mixture is passed through a packed bed of sodium chlorite. The gas was diffused into cold (2–3oC) DI water for approximately 30 min, producing a ClO2 stock solution concentration of approximately 4,000 mg/L. The purity of each stock solution was measured according to the following formula: 160 % Purity = CClO2/(CClO2 + CChlorite + CFree chlorine) 140 120 100 in which C is the concentration of each species in milligrams per litre. The purity 40 of all stock solutions was >99%. Stock 20 concentrations were measured daily 0 using amperometric titration (for stock 0 100 200 300 400 500 600 700 solutions 0.1 N titrant is used instead of Reaction Time—s the 0.00564 N titrant used for residual analyses) and were found to be stable O3 = 1.0 mg/L O3 = 1.5 mg/L O3 = 2.0 mg/L O3 = 3.0 mg/L when the solution was stored in the dark 1,000 and refrigerated. ClO2 residuals were 2+ 900 Initial Mn = 1,000 µg/L measured using the amperometric titra800 tion method (method 4500-ClO2 E), 700 (Standard Methods, 1998). 600 Stock solutions of KMnO4 (5,000 500 mg/L) were produced by dissolving 400 granular KMnO4(s) in DI water. The 300 concentration of the MnO 4– stock 200 solutions, which were refrigerated at all 100 times, did not decay significantly over 0 a period of several days. Fresh MnO4– 100 200 300 400 500 600 700 0 stock solutions were made weekly. Reaction Time—s Residual Mn analyses were perTemperature—9 C, pH—7.0, total organic carbon—3.4 mg/L; Mn—manganese, 2+ formed by atomic absorption specMn —manganous ion, O3—ozone trometry (method 3500-Mn B) (Standard Methods, 1998). Samples were acidified after filtration, stored in 125bench-scale apparatus located under a fume hood. The gas mL plastic bottles, and refrigerated before analysis. was bubbled into cold (2–3oC) deionized (DI) water in a TOC concentrations were measured by an analyzer 1-L heat-resistant glass flask for 15 min. The O3 stock that uses the persulfate–ultraviolet oxidation method3 (method 5310 C) (Standard Methods, 1998). solutions decayed slowly when sealed and kept in an ice Bench-scale protocol. The general procedure used for bath. Each stock was used for several bench-scale experall experiments was to (1) determine the concentration of iments over 3–4 h. The stock concentration changed from 80 Dissolved Mn—µg/L 60 o 2003 © American Water Works Association 102 JANUARY 2003 | JOURNAL AWWA • 95:1 | PEER-REVIEWED | GREGORY ET AL Dissolved Manganese Residual—µg/L the oxidant stock solution, (2) inject the O3 dosage versus dissolved Mn residual after 7-min reaction time FIGURE 4 appropriate volume of the stock into the sample water at t = 0 to achieve a specific applied dose, and (3) collect and Mn = 0.06 mg/L Mn = 0.2 mg/L Mn = 1.0 mg/L 100 filter Mn samples to characterize the kinetics of Mn2+ oxidation for O3, ClO2, 90 and KMnO4. For ClO2 and KMnO4, 80 the same stock solutions were often used 70 for several days, and the concentrations 60 generally remained constant. Because O3 decays relatively rapidly, new stock 50 solutions were required after three or 40 four experiments. 30 The oxidant was injected into a 20 closed 1-L Erlenmeyer-type reaction flask at t = 0 and was mixed for several sec10 onds to ensure an even distribution 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 throughout the sample volume. The 0 Ozone Dose— mg/L water temperature was monitored and maintained at approximately 9oC, the Temperature—9 C, pH—7.0, total organic carbon—3.4 mg/L; Mn—manganese, O3—ozone average temperature of Horsetooth Reservoir water during stratification, throughout each experiment using a kinetics of the O3–Mn2+ reaction. Rather, the intent was water bath. All experiments were conducted at pH 7.0. to characterize the latter stages of the interaction and, The 1-L reaction flask was equipped with a bottle-top, more important, to determine the equilibrium concenadjustable-volume dispenser that was flushed before each trations of the dissolved Mn residuals. A majority of the sample was dispensed. The filtration step used an ultraO3–Mn2+ interaction occurs within 60 s, as shown by filtration apparatus with a 30K (30,000 AMU molecular 4 data from other researchers (Reckhow et al, 1991; Knocke weight cut-off) ultrafiltration membrane. The ultrafiltration step was performed under pressure (approximately et al, 1990), and by the Mn2+ oxidation and O3 decay data provided in this article. The time interval between sam30 psi [207 kPa]), which was supplied by connecting a canples was dictated by the minimum time required for the ister of compressed nitrogen gas to the apparatus. Filterfiltration process to be completed and the filter apparaing time was about 150 s per 25-mL sample. This filtratus to be cleaned and prepared for the next sample. In the tion procedure ensured that only dissolved Mn species o The oxidant was injected into a closed 1-L Erlenmeyer-type reaction flask at t = 0 and was mixed for several seconds to ensure an even distribution throughout the sample volume. remained in the final filtrate sample (Carlson et al, 1997; Reckhow et al, 1991). The pH of the filtered samples was checked periodically. It remained virtually constant throughout the experimental procedure. Before use, new 30K membranes were soaked in DI water to remove the glycerin coating from their surfaces. Additionally, 500 mL of DI water was filtered through each new membrane before it was used in experiments. Sample collection. The sample times used were based on oxidant-specific considerations such as the expected kinetics of the oxidant–Mn reaction (whether or not a quenching agent, which eliminates the oxidant residual at the time of sample collection, was used) and the filtration step required to separate dissolved from particulate and colloidal Mn. In the case of O3, for example, sample times were not intended to characterize the initial, rapid O3 experiments, a quenching agent was not used (to avoid reducing residual MnO4– to MnO2 (s)); therefore, samples were filtered immediately. Further Mn oxidation may have occurred during the filtration process for the sample collected at t = 1 min. For the samples collected at t = 3, 7, and 11 min, further oxidation of Mn by O3 was negligible. Quenching agents were used for the ClO2 and KMnO4 experiments, and samples, therefore, did not need to be filtered immediately as in the O 3 experiments. This allowed samples to be collected more frequently. Preliminary experiments indicated that the ClO2–Mn reaction was far more rapid, particularly at lower initial Mn2+ concentrations, than the MnO4––Mn reaction. As a result, the duration of the ClO2 and KMnO4 experiments was 10 and 30 min, respectively. For ClO2, Mn 2003 © American Water Works Association GREGORY ET AL | PEER-REVIEWED | 95:1 • JOURNAL AWWA | JANUARY 2003 103 samples collected at t = 30, 60, 90, 180, 300, and 600 s adequately characterized the reaction kinetics within the first few minutes as well as the equilibrium concentrations. For the KMnO4 experiments, samples were collected at approximately the same frequency for the first 10 min, but two additional Mn samples, at t = 1,200 and 1,800 s, were also obtained. Quenching agents. An important aspect of the benchscale experiments was the selection of the quenching (reducing) agent used to eliminate the oxidant residual and halt the Mn 2+–oxidant reactions. Two features of a quenching agent were desired: immediate and complete reaction with the oxidant residual and the inability to resolubilize previously formed MnO2(s). In the O3 experiments, the primary reason for not using a quenching agent was that one focus of this work the ClO2 experiments. The results in Figure 2 illustrate that oxidation of Mn2+ by KMnO4 is considerably slower than by ClO2 in this water (note that different time scales are used in Figures 1 and 2). For initial Mn2+ = 60 µg/L, about 1,800 s were required for complete oxidation with a 300% stoichiometric dose (the stoichiometry of the reaction is 1.9 mg KMnO4/mg Mn2+). Even after this reaction time, the final Mn2+ residual was >10 µg/L. For initial Mn2+ = 200 µg/L, 1,200 s were required to produce Mn2+ residuals near 10 µg/L. At initial Mn2+ = 1,000 µg/L, the oxidation rate and final Mn2+ residual were comparable to the ClO2 results. Figure 2 shows that the impact of initial Mn2+ concentration is more significant for KMnO4 than ClO2, when considering the finished water Mn goal of 10 µg/L. Final Mn2+ residual <10 µg/L was only observed when the An important aspect of the bench-scale experiments was the selection of the quenching (reducing) agent used to eliminate the oxidant residual and halt the Mn 2+–oxidant reactions. was the formation of MnO4–, which appeared to occur in a majority of the experiments. The use of a quenching agent would have reduced some of the residual MnO4– to MnO2(s), which would have been removed during the filtration process and made it impossible to determine the true concentration of total dissolved Mn, not just Mn2+, remaining in solution following ozonation. In ClO2 experiments, a 0.1 N potassium iodide solution was a satisfactory quenching agent. For KMnO4 experiments, 0.00564 N phenylarsine oxide was used. Experiments were conducted to determine the minimum effective dose of each quenching agent to avoid overdosing the quenching agents and the potential to reduce MnO2(s) to soluble forms of Mn. RESULTS AND DISCUSSION ClO2. Figure 1 shows the results of bench-scale oxidation experiments using ClO2 with three initial Mn2+ concentrations: 60, 200, and 1,000 µg/L. The relative stoichiometric doses of ClO2 used ranged from 100 to 400% (the stoichiometry of the reaction is 2.5 mg ClO2/mg Mn2+). Mn2+ was oxidized to <10 µg/L within 120 s for initial Mn2+ = 200 and 1,000 µg/L. At the lowest Mn2+ concentration of 60 µg/L, the dissolved Mn concentration was <10 µg/L after 300 s. The only experimental conditions using ClO2 that resulted in a final Mn2+ residual >10 µg/L were at the lowest initial Mn2+ level and a dose of 200% stoichiometry (0.3 mg/L). Although oxidation was rapid for all experimental conditions, it was clear that increasing initial Mn2+ concentrations increased oxidation rates. KMnO4. Figure 2 shows the results of bench-scale oxidation experiments using KMnO4. The initial Mn2+ concentrations were approximately the same as those used for initial Mn2+ = 1,000 µg/L. The oxidation rates at this concentration are far more rapid than for initial Mn2+ = 60 and 200 µg/L. The rapid oxidation rate observed in experiments using an initial Mn2+ concentration of 1,000 µg/L indicates that solution-phase oxidation proceeds as rapidly as surface-catalyzed oxidation in these conditions. These results suggest that for KMnO4, sorption of Mn2+ onto MnO2(s) and subsequent surface oxidation are the primary mechanism at low initial Mn2+ concentrations (when the solution-phase reaction proceeds slowly). This was not the case for ClO2. O3. Figure 3 shows the results of Mn2+ oxidation versus time by O3. The trendlines in Figure 3 are generally flat after the 60-s point, indicating that the O3–Mn2+ reaction reached completion in less than 60 s. Figure 3 and experimental observations indicate that the reaction is rapid, probably reaching completion within 30 s. Final Mn2+ residuals <10 µg/L were only observed in experiments using initial Mn2+ = 1,000 µg/L. Figure 4 shows the effect of O3 dose on Mn2+ residuals after 7 min of reaction time for the three initial Mn2+ concentrations used in Figure 3. The 7-min samples were used to represent equilibrium conditions. Figure 4 shows that increasing O3 dosages above the optimum increases Mn2+ residuals, and Mn2+ residuals from initial Mn2+ = 60 and 200 µg/L are always >20 µg/L. The increasing Mn2+ residuals observed at higher O3 dosages appear to be due to the formation of MnO4–, which is a dissolved species and results from the oxidation of a small fraction of Mn(II) to Mn(VII). Mn2+ is essentially oxidized beyond the desired Mn(IV) state to the undesirable dissolved state, Mn (VII). Optimizing O3 dosages for minimum Mn2+ residuals required a balance between overcoming O3 demand 2003 © American Water Works Association 104 JANUARY 2003 | JOURNAL AWWA • 95:1 | PEER-REVIEWED | GREGORY ET AL experimental pH, TOC, and temperature conditions showed that dosages below this range increase Mn2+ residuals because of competing O3 demand 300% KMnO4 O3 = 1.5 mg/L 400% CIO2 from NOM (Gregory & Carlson, 90 2001). Increasing dissolved Mn resid80 uals observed for higher O3 dosages 70 were due primarily to the formation of 60 MnO4–. 50 Oxidation of Mn2+ by O3 proceeds 40 rapidly. The issue is that the potential for MnO 4– formation exists for all 30 doses in which the stoichiometry for 20 2+ Initial Mn = 60 µg/L the MnO4–-forming reaction (2.2 mg 10 O3/mg Mn2+) is satisfied. The distinc0 tive pink color of MnO4– was observed 0 100 200 300 400 500 600 in many experiments. In this study, to Reaction Time—s achieve effective Mn2+ removal for all 300% KMnO4 O3 = 1.0 mg/L 200% CIO2 but the highest initial Mn2+ concentra200 tion, the background demand exerted 180 by NOM required the use of O 3 160 dosages that were sufficient for the pro140 duction of MnO4–. As a result, dis120 solved Mn residuals <10 µg/L were not 100 attainable for initial Mn2+ concentra80 tions of 60 and 200 µg/L. 2+ Initial Mn = 200 µg/L 60 The formation of MnO4– in the O3 40 experiments implies that the dissolved 20 Mn residuals measured were due to 0 MnO4– or a combination of MnO4– 0 100 200 300 400 500 600 and Mn2+. Because the samples from Reaction Time—s these experiments were not quenched, no distinction was made between the O3 = 1.5 mg/L 100% CIO2 100% KMnO4 two species. Therefore, the results that 1,200 show “dissolved Mn” residuals for these 1,000 experiments include Mn2+ and MnO4–. Comparison of oxidants. Figure 5 800 compares Mn2+ removal for the most 600 effective doses of each oxidant at each initial Mn2+ concentration. When com2+ 400 Initial Mn = 1,000 µg/L pared with the results for KMnO4 and O3, the data in Figure 5 indicate that 200 ClO2 was the most effective oxidant 0 for reducing dissolved Mn residuals to 0 100 200 300 400 500 600 <10 µg/L. Previous work (Knocke et Reaction Time—s al, 1990) suggested that Mn2+ oxidation 2+ rates by KMnO4 and O3 were similar to Oxidant doses shown are the optimum at each initial Mn ; temperature—9 C, pH—7.0, total organic carbon—3.4 mg/L; CIO2—chlorine dioxide, KMnO4—potassium ClO2 and that all three were rapid, 2+ permanganate, Mn—manganese, Mn —manganous ion, O3—ozone effective oxidants for Mn2+. These studies, however, generally used high iniexerted by NOM and overdosing, which causes the fortial Mn2+ concentrations (close to 1 mg/L), and the focus mation of MnO4–. As a result, there is a relatively narrow was not on attaining final Mn residuals of <10 µg/L. The optimum dose—particularly for higher initial Mn2+ concurrent study emphasized attaining low dissolved Mn centrations. Dosages above and below this range result in residuals within a relatively short period of time (approxsignificantly higher dissolved Mn residuals. A previous imately 180 s). Figures 1–3 and Figure 5 indicate that study using Horsetooth Reservoir water and the same there can be significant differences between the reaction Bench-scale comparison of oxidation of Mn2+ by KMnO4, ClO2, and O3 at three initial Mn2+ concentrations Dissolved Mn—µg/L Dissolved Mn—µg/L Dissolved Mn—µg/L FIGURE 5 o 2003 © American Water Works Association GREGORY ET AL | PEER-REVIEWED | 95:1 • JOURNAL AWWA | JANUARY 2003 105 MnO2 ≡ Mn—mg/L MnO2 (s)—mg/L +2 Mn /Mn +2 0 rates and final Mn residuals for each Results of modeling analysis showing effect of initial Mn2+ concentration on FIGURE 6 oxidant. concentrations of Mn2+, MnO2(s), and MnO2 ⬅ Mn2+ concentrations versus From these results, ClO2 appeared time for oxidation by KMnO4 to be the optimum alternative for two reasons. First, when compared with the KMnO4 results, the ClO2–Mn2+ reacInitial Mn = 0.06 mg/L Initial Mn = 0.2 mg/L tion kinetics were significantly more Initial Mn = 1.0 mg/L rapid (Figures 1 and 2), particularly for 1.2 initial Mn2+ = 60 and 200 µg/L. Sec1.0 ond, ClO 2 did not appear to create – MnO 4 , an undesirable dissolved 0.8 species, as did O3. Although MnO4– was not directly measured in the ClO2 0.6 experiments, its distinctive pink color 0.4 was not observed in any of the ClO2 experiments, and the formation of 0.2 MnO4– from the reaction between ClO2 and Mn2+ has not previously been re0.0 0 300 600 900 1,200 1,500 1,800 ported in the literature. In the O3 expers Time— iments, the water turned pink when the 0.25 reaction stoichiometry of 2.2 mg O3/mg 2+ Mn was met. The formation of MnO4– upon ozonation of water con0.20 taining Mn 2+ has previously been reported (Gregory & Carlson, 2001; 0.15 Paillard et al, 1989). The ClO2 results also indicated that 0.10 Mn2+ ions may be oxidized before the NOM in Horsetooth Reservoir. This 0.05 hypothesis is supported by Figure 1 for the cases of initial Mn 2+ = 200 and 0.00 0 300 600 900 1,200 1,500 1,800 1,000 µg/L. In both cases, ClO2 doses Time— s of 100% stoichiometry (based on 2.45 mg ClO2/mg Mn2+) oxidized Mn2+ to 0.8 <5 µg/L. Significant ClO2 demand from 0.7 the NOM would have prevented the 2+ nearly 100% oxidation of Mn that 0.6 was observed. 0.5 Another hypothesis, which may 0.4 account for the low Mn 2+ residuals observed in ClO2 experiments and the 0.3 higher residuals observed in the 0.2 KMnO4 experiments, is that ClO2 is 0.1 able to oxidize NOM-complexed Mn2+ ions and that KMnO 4, which has a 0 0 300 600 900 1,200 1,500 1,800 lower oxidation–reduction potential, Time—s may not. Although Mn2+ is not readAssumed oxidant dose—125% theoretical stoichiometry; KMnO4—potassium ily complexed by NOM (unlike iron, 2+ permanganate, Mn—manganese, Mn —manganous ion, MnO2—manganese oxide for instance), it has been shown that a small fraction, probably <10%, will complex with NOM (Salbu & complexed fraction of the dissolved Mn and reduce Steinnes, 1995; Knocke et al, 1990). If, for example, Mn2+ concentrations to <5 µg/L. 5% of the Mn2+ in a 200-µg/L solution is complexed and 2+ unavailable for oxidation by KMnO4, the final Mn Modeling results. To characterize the effect of initial residual will be at least 10 µg/L. This concentration can Mn2+ concentration on oxidation kinetics and examine the be significant. ClO2, however, appears to oxidize the relative concentrations of Mn2+, MnO2(s), and MnO2 ⬅ 2003 © American Water Works Association 106 JANUARY 2003 | JOURNAL AWWA • 95:1 | PEER-REVIEWED | GREGORY ET AL Mn (MnO2 with sorbed Mn2+ ion) during the reaction period, the kinetic model and equations developed by Carlson and Knocke (1999) for oxidation by KMnO4 were used. This model was based on the autocatalytic mathematical model originally developed by Van Benschoten et al (1991). Carlson and Knocke’s (1999) work expanded the original rate expressions to include a term to account for DOC concentration and were able to more accurately characterize the KMnO4–Mn2+ reaction for water treatment applications. The model in this study used Carlson and Knocke’s second-order rate constants: log k1 = 2.32, log k2 = 4.01, log k3 = 2.61, and log k4 = –0.37 s–1 M–1. The rate constant k4 accounts for a firstorder reaction between organic matter and MnO4–. The aim of this analysis was not to compare this study’s observed Mn2+ removal results with model-predicted results. Rather, the aim was to assist in the understanding of the observed data, which indicated that initial Mn2+ centrations, however, the reaction proceeds relatively slowly, and an excess of free MnO2(s) is not available— contributing to the higher final Mn2+ residuals that were observed for these conditions. Effect of pH and temperature on Mn 2+ oxidation. Although the effects of pH and temperature on Mn2+ oxidation were not directly investigated in this study, a discussion of the effects of these variables will assist in applying these results to other water quality conditions. Mn2+ oxidation rates increase and final Mn2+ residuals decrease with increases in pH and temperature (Reckhow et al, 1991; Knocke, 1991b; Knocke et al, 1990). Knocke (1991b) showed that for ClO2 and an initial Mn2+ concentration of approximately 0.95 mg/L, oxidation was completed within 5 s at pH 7.0, whereas at pH 5.5, complete oxidation required about 20 s. For drinking water treatment, these differences may not be significant if sufficient contact time is available. Tem- The objectives of this study were to evaluate ClO 2, KMnO 4, and O 3 for the oxidation of Mn 2+ to <10 µg/L and to understand the effect of initial Mn 2+ concentration and oxidant dose. concentrations have a large effect on oxidation kinetics and final dissolved Mn residuals and examine the relative concentrations of Mn2+, MnO2(s), and MnO2 ⬅ Mn2+ throughout the reaction. The model outputs (Figure 6) show the effect of initial Mn2+ concentration on the relative concentrations of Mn2+, MnO2(s), and MnO2 ⬅ Mn. The assumed KMnO4 dose was 125% of the theoretical stoichiometric requirement. Although the oxidation rates are somewhat slower than those shown by the experimental data, the impact of initial Mn2+ on oxidation kinetics is clearly illustrated and is similar to the observed data presented in Figure 2. Figure 6 shows that the MnO2 ⬅ Mn2+ complex is the dominant species as the reaction proceeds. In the early stages of the reaction, the concentration of free MnO2(s) is low because Mn2+ quickly sorbs to newly formed MnO2(s). This result was expected, given that the sorption rate constant k2 (4.01 s–1 M–1) is approximately 1.7 orders of magnitude greater than the solution-phase oxidation constant k1 (2.32 s–1 M–1). As illustrated in Figure 6 at initial Mn2+ = 1.0 mg/L, only when the Mn2+ concentration becomes low does the MnO2(s) concentration increase significantly. These results suggest that a mechanism that accounts for the low final Mn2+ residuals observed in all experiments using an initial Mn2+ = 1.0 mg/L would be that an excess of free MnO2(s) serves to scavenge Mn2+ from solution. At high initial Mn2+ concentrations, the reaction kinetics will be rapid because the concentration of both reactants is high. As the reaction progresses and the concentration of the reactants decreases, Mn2+ continues to be efficiently removed from solution because of the sorption mechanism k2. For low initial Mn2+ con- perature effects were similar. Complete oxidation with ClO 2 was achieved within 60 s at 10 oC and within approximately 10 s at 25oC. For KMnO4, similar effects of pH and temperature were observed. For pH 5.5 and 7.0, for example, there were significant differences in the rate of oxidation, but the final residuals were almost equal after 15 s. Knocke et al (1990) observed that for ClO2 and KMnO4, second-order rate constants increased by approximately one order of magnitude when the pH was increased from 6.0 to 7.0. For O3, Reckhow et al (1991) also reported an increase in oxidation rate with increasing pH. At pH 6, a secondorder rate constant of 5 × 103 M–1s–1 was observed, whereas at pH 7 the observed rate was 2 × 104 M–1s–1. CONCLUSIONS The objectives of this study were to evaluate ClO2, KMnO4, and O3 for the oxidation of Mn2+ to <10 µg/L and to understand the effect of initial Mn2+ concentration and oxidant dose. The following conclusions are based on the specific raw water quality conditions of this study: pH 7.0, TOC = 3.4 mg/L, and temperature = 9oC. ClO2 was the most effective oxidant for removing lower initial Mn2+ concentrations (60 and 200 µg/L) and for consistently meeting the treatment goal of <10 µg/L final Mn2+. Reaction kinetics between ClO2 and Mn2+ were significantly faster than those observed for the KMnO4–Mn2+ reaction. For ClO2, Mn2+ residuals were reduced to <10 µg/L within 90 s at initial Mn2+ = 200 and 1,000 µg/L and within 180 s when the initial Mn2+ = 60 µg/L. For KMnO4, the reaction time required 2003 © American Water Works Association GREGORY ET AL | PEER-REVIEWED | 95:1 • JOURNAL AWWA | JANUARY 2003 107 to produce similar results was approximately 30 min for initial Mn2+ = 60 µg/L and 10–20 min for initial Mn2+ = 200 µg/L. For Mn2+ oxidation by O3, there is a relatively narrow dosage range in which minimum dissolved Mn residuals can be achieved. Dosages below this range significantly increase Mn2+ residuals because of competing O 3 demand from NOM. Dosages above this range increase the potential for MnO4– formation and increase dissolved Mn residuals. Optimum O3 dosages for the lower initial Mn2+ levels resulted in final dissolved Mn residuals >20 µg/L. For the conditions of this research, Mn2+ oxidation by KMnO4 did not provide acceptable results for the two lowest initial Mn2+ concentrations. Oxidation of Mn2+ to <10 µg/L by KMnO4 was possible only at initial Mn2+ = 1,000 µg/L. For an initial Mn2+ = 60 µg/L, a reaction time of 30 min was required to achieve Mn2+ <20 µg/L. Thus, KMnO4 requires considerably longer reaction time than O3 and ClO2 to achieve acceptable Mn2+ oxidation. For the highest initial Mn2+ concentration, 1,000 µg/L, all oxidants met the treatment goal of final Mn2+ residuals <10 µg/L within approximately 180 s. The higher concentrations of the reacting species, Mn2+, and oxidant cause an increase in the reaction rate for all three Mn2+ removal mechanisms: solution-phase oxidation, sorption of Mn2+ to MnO2(s), and surface oxidation. ACKNOWLEDGMENT The authors express gratitude to Kevin Gertig and Ben Alexander of Fort Collins Utilities. Their enthusiastic REFERENCES Bader, H. & Hoigné, J., 1981. Determination of Ozone in Water by the Indigo Method. Water Res., 15:4:449. Carlson, K.H. & Knocke, W.R., 1999. Modeling Mn Oxidation With Potassium Permanganate. Jour. Envir. Engrg., 125:10:892. Carlson, K.H. et al, 1997. The Relationship of Speciation to Iron and Mn Removal Strategies. Jour. AWWA, 89:4:162. Coffey, B. et al, 1993. Modeling Soluble Mn Removal by Oxide-coated Filter Media. Jour. Envir. Engrg., 119:4:679. support allowed this research to be successful. This work was funded by the city of Fort Collins. ABOUT THE AUTHORS: Dean Gregory5 is a PhD candidate in the Department of Civil Engineering at Colorado State University, Fort Collins, CO 80523; e-mail [email protected]. He has a BA degree from Colgate University in Hamilton, N.Y., and an MS degree in environmental engineering from Colorado State University. Gregory worked as a process engineer for the Fort Collins Utilities investigating Mn issues and conducting pilot-scale research on other treatment problems before beginning this doctoral program. His work has been published previously in Ozone Science & Engineering and Journal of Environmental Engineering. Kenneth Carlson is an assistant professor in the Department of Civil Engineering at Colorado State University. FOOTNOTES 1CDG Technology, Bethlehem, Pa. Technology, Bethlehem, Pa. Instruments, model 800 TOC analyzer, Boulder, Colo. 4Amicon Inc., Beverly, Mass. 5To whom correspondence should be addressed 2CDG 3Sievers If you have a comment about this article, please contact us at [email protected]. Knocke, W.R. et al, 1991a. Removal of Soluble Mn by Oxide-coated Filter Media: Sorption Rate and Removal Mechanism Issues. Jour. AWWA, 83:8.53. Knocke, W.R. et al, 1991b. Kinetics of Mn and Iron Oxidation by Potassium Permanganate and Chlorine Dioxide. Jour. AWWA, 83:6:80. Knocke, W.R. et al, 1990. Alternative Oxidants for the Removal of Soluble Iron and Mn. AWWA Res. Fdn., Denver. Edzwald, J.K., & Van Benschoten, J.E., 1990. Aluminum Coagulation of Natural Organic Matter. Chemical Water and Wastewater Treatment, Springer-Verlag, Berlin. Morgan, J.J., 1967. Chemical Equilibria and Kinetic Properties of Mn in Natural Waters. Principles and Applications of Water Chemistry (S.D. Faust and J.V. Hunter, editors). John Wiley & Sons, New York. Gregory, D. & Carlson, K.H., 2001. Ozonation of Dissolved Mn in the Presence of Natural Organic Matter. Ozone Sci. & Engrg., 23:2:149. Paillard, H. et al, 1989. Iron and Mn Removal With Ozonation in the Presence of Humic Substances. Ozone Sci. & Engrg., 11:93. 2003 © American Water Works Association 108 JANUARY 2003 | JOURNAL AWWA • 95:1 | PEER-REVIEWED | GREGORY ET AL Reckhow, D.A. et al, 1991. Oxidation of Iron and Mn by Ozone. Ozone Sci. & Engrg., 13:675. Salbu, B. & Steinnes, E. (editors), 1995. Trace Elements in Natural Waters. CRC Press, Boca Raton, Fla. Seby, F. et al, 1995. Study of the Ozone-Mn Reaction and the Interactions of Disulfonate Indigo Carmin/Oxidized Mn Forms as a Function of pH. Ozone Sci. & Engrg., 17:135. Sly, L.I. et al, 1990. Deposition of Mn in a Drinking Water Distribution System. Applied & Envir. Microbiol., 56:3:628. Standard Methods for the Examination of Water and Wastewater, 1998 (20th ed.). APHA, AWWA, and WEF, Washington. Van Benschoten, J.E. et al, 1991. Kinetic Modeling of Mn (II) Oxidation by Chlorine Dioxide and Potassium Permanganate. Envir. Sci. & Technol., 26:7:1327.
© Copyright 2026 Paperzz