- Email: [email protected]
Characteristics and applications of controlled–release KMnO4 for groundwater remediation
Chemosphere 66 (2007) 2058–2066 www.elsevier.com/locate/chemosphere
Characteristics and applications of controlled–release KMnO4 for groundwater remediation Eung Seok Lee *, Franklin W. Schwartz Department of Geological Sciences, The Ohio State University, Columbus OH 43210, USA Received 30 May 2006; received in revised form 26 September 2006; accepted 27 September 2006 Available online 30 November 2006
Abstract In situ chemical oxidation (ISCO) using potassium permanganate (KMnO4) has been widely used as a practical approach for remediation of groundwater contaminated by chlorinated solvents like trichloroethylene. The most common applications are active flushing schemes, which target the destruction of some contaminant source by injecting concentrated permanganate (MnO 4 Þ solution into the subsurface over a short period of time. Despite many promising results, KMnO4 flushing is often frustrated by inefficiency associated with pore plugging by MnO2 and bypassing. Opportunities exist for the development of new ISCO systems based on KMnO4. The new scheme described in this paper uses controlled–release KMnO4 (CRP) as an active component in the well-based reactive barrier system. This scheme operates to control spreading of a dissolved contaminant plume. Prototype CRP was manufactured by dispersing fine KMnO4 granules in liquid crystal polymer resin matrix. Scanning electron microscope data verified the formation of micro-scale (ID = 20–200 lm) secondary capillary permeability through which MnO 4 is released by a reaction-diffusion process. Column and numerical simulation data indicated that the CRP could deliver MnO 4 in a controlled manner for several years without replenishment. A proof-of-concept flow-tank experiment and model simulations suggested that the CRP scheme could potentially be developed as a practical approach for in situ remediation of contaminated aquifers. This scheme may be suitable for remediation of sites where accessibility is limited or some low-concentration contaminant plume is extensive. Development of delivery systems that can facilitate lateral spreading and mixing of MnO 4 with the contaminant plume is warranted. 2006 Elsevier Ltd. All rights reserved. Keywords: Controlled–release system; Groundwater; Remediation; Potassium permanganate; Chemical oxidation; TCE; Permeable reactive barrier
1. Introduction Over the past decade, advanced oxidation schemes are being used increasingly to address problems of groundwater contamination by chlorinated solvents like trichloroethylene (TCE). Of the potential oxidants, potassium permanganate (KMnO4) is often used because of its ease of handling and relatively low cost (Schnarr et al., 1998; Siegrist et al., 2001; Seol et al., 2002). Studies have shown that permanganate ðMnO 4 Þ rapidly transforms most chlorinated solvents to short-lived intermediates, and eventu-
*
Corresponding author. Tel.: +1 614 292 0585; fax: +1 614 292 7688. E-mail address: [email protected] (E.S. Lee).
0045-6535/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.chemosphere.2006.09.093
ally innocuous by-products (Yan and Schwartz, 1999, 2000). In most applications, MnO 4 is applied into the subsurface via injection wells in a concentrated aqueous solution as a flood of some source zone. This active flushing scheme treats the problem of contamination through mass removal. When this method works, it is usually because units are permeable and solvents are present at low saturations. When it fails, the main problem is commonly inability to deliver the oxidant due to physical heterogeneity in hydraulic conductivity and/or chemical heterogeneity related to pore plugging by solid reaction products like MnO2 (Seol et al., 2002). Column and flow-tank experiments (Schroth et al., 2001; Lee et al., 2003; Li and Schwartz, 2004) showed how plugging caused the
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
MnO 4 flood to bypass zones of relatively high DNAPL (i.e., a liquid that is denser than water and does not dissolve or mix easily in water) saturation. In field settings, leaving volumes of untreated DNAPL in place contributes to a rebound in aqueous concentrations once flushing is discontinued (e.g., Siegrist et al., 2001). Clearly, then, there are opportunities to use delivery systems in an improved way. This paper examines aspects of using KMnO4 as the active component of a well-based permeable reactive barrier (PRB) system. In this application, KMnO4 is emplaced as a controlled–release solid to create chemically active zones in the subsurface. The solids are resin-KMnO4 composites, cylindrical in form. The controlled–release KMnO4 (CRP) can be applied against a plume in multiple layers of closely spaced wells. Ideally, these forms would dissolve and release low concentrations of MnO 4 at a controlled rate to maintain the predetermined level of the agent within the chemically active zone. Such a barrier system could function by itself over a period of several years and possibly decades without replenishment. When the CRP is exhausted, it could be replaced. The KMnO4 used in this way would provide long-term controlling scheme for aqueous-phase plumes of contaminants. With the typically low dissolved concentrations of contaminants in most plumes, there will be much less tendency for the porous medium to plug up. One of the necessary steps in developing this remedial concept is manufacturing and testing the CRP, which can deliver MnO 4 in a reasonably controlled way over years. This paper describes prototype CRP which has some of the necessary controlled release characteristics and column experiments designed to provide release data. It also presents results of a proof-of-concept experiment with a flow tank, suggesting how the CRP could be used for long-term control of contaminant plume in the subsurface. This paper concludes with presentation of modeling approach capable of quantifying release behaviors of the CRP for the environmental conditions of the contaminated sites.
2059
2. Materials and Methods
all agent level; preservation of agents that can be rapidly destroyed by environmental conditions; reduced need for maintenance; and lower cost (e.g., Langer, 1990). An extruded form of KMnO4 (CAIROX-CR, Carus Chemical Company) is commercially available for removing grease and hydrogen sulfide odor in surface water. This product uses the porous matrix comprising silica and clays as a means of delaying the otherwise instant dissolution of KMnO4 in water. Microcapsules of KMnO4 having slowrelease properties have been also investigated in a column experimental study for possible applications to groundwater remediation (Ross et al., 2005). This material, comprising KMnO4 granule core and waxy polymeric coating, yielded extended duration of permanganate release in the range of 3–80 days. Our implementation of controlled release technology takes advantage of inert organic crystalline matrix to provide for slow dissolution and diffusioncontrolled transport of agent. This form is designed for supply of treatment chemicals to the contaminated aquifers in situ, at a controlled release rate over several years and perhaps decades without replenishment. For this study, a prototype CRP was manufactured by dispersing KMnO4 (Carux Chemical Company, Peru, IL) in organic crystalline matrices (Castin’Craft Clear Polyester Resin, Environmental Technology Inc., Fields Landing, CA). Liquid resin and fine KMnO4 granules were mixed in a specially designed cylindrical mould (ID = 2.5 cm) and allowed to crystallize at room temperature. Ideally, the CRP would have near zero initial permeability, while secondary capillary permeability will form later in the matrix as MnO 4 is released. The crystalline matrix system was designed to yield much smaller permeability for MnO 4 transport as compared to the porous sandy-clay matrix system (CAIROX-CR). The outside diameter of the CRP (2.5 cm) was selected in order to fit inside the delivery wells (2.6 cm) used for the later flow-tank experiments. The CRP used for the column tests was 5 cm-long, had a diameter of 2.5 cm, and contained 35 g of KMnO4. The CRP used for the flow-tank experiment was 10 cm-long with a diameter of 2.5 cm, and contained 100 g of KMnO4.
2.1. Manufacturing controlled–release KMnO4 (CRP)
2.2. Column test and flow-tank experiment
The ability to release a compound at a controlled rate is important in variety of different applications (Gibbs et al., 1999; Wakimoto, 2004). The controlled–release system transfers active agent from a reservoir to a target host, in order to maintain a predetermined concentration or emission level of the agent for a specified period of time or desired fraction (Langer, 1990). This system provides various advantages over conventional delivery systems. For example, risks for toxicity and secondary contamination (overdose) or ineffectiveness (underdose) can be eliminated by maintaining the agent concentrations within the desired range. Other advantages may include in situ construction of stable reaction zones; localized delivery of agents to a particular subsurface compartment, thereby lowering the over-
Column tests were performed using CHROMAFLEX (Kontes) glass column (ID = 4.8 cm, length = 15 cm, v = 270 ml) to determine release rates of CRP in flowing water. Ambient flow rates of 20 ml min1 were maintained using Ismatec peristaltic pumps (Cole Palmer Instrument Company, Chicago, Illinois) to provide perfect sink conditions for CRP release. The CRP used for the column experiment was pre-washed with tap water to remove KMnO4 granules attached to the surface of the pellet. A glass tank was constructed (L · W · D = 100 cm · 20 cm · 50 cm) for flow-tank experiment (Figs. 1 and 2). The tank was filled with 170 kg of silica sand (U.S. Silica, Ottawa, Illinois), which was packed to an approximate bulk density of 1.65 g cm3 and porosity of 37.7%
2060
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
Metal Screen
Multilevel Monitoring Wells (sampling depths: 13, 26, 39 cm below surface)
Input Chamber
Output Chamber
Sands
Center Line
Flow
CRK Delivery Wells
Fig. 1. Schematic diagram of the glass flow-tank setup (L · W · D = 100 cm · 20 cm · 50 cm). Scales are exaggerated to show the compartments. CRP source zone is located at the upstream part of the flow tank. Only the boundaries of the sandy media (L · W · D = 80 cm · 20 cm · 45 cm) are drawn to show well locations in the flow tank. Three peristaltic pumps were used for flow control and sample collection.
(L · W · D = 80 cm · 20 cm · 45 cm). Input and output chambers on the upstream and downstream ends of the tank helped to control inflow and outflow rates. These chambers were separated from the sand by rigid stainless steel screens that were covered with fiber glass to prevent sand from entering the chambers.
Water was pumped into the inflow chamber using Ismatec peristaltic pumps (Cole Palmer Instrument Company, Chicago, Illinois) to create a uniform flow into the upstream end of the tank. The outflow rate was controlled in the same manner. An inflow rate of 19.2 l d1 to the tank provided an ambient flow velocity of 57 cm d1 in the sandy porous medium. At this velocity, it required 34 h (1.4 d) for water to flow through the tank. Effluent from the tank was collected in a waste container and treated to remove TCE and MnO 4 before disposal. The excess MnO 4 was removed from the effluent through reduction by sodium thiosulphate (Na2S2O3). 2.3. Water sampling, flow-tank excavation, SEM, and chemical analyses Four small multi-level wells were emplaced along the centerline of the tank for sample collection (Figs. 1 and 2). They were constructed with 2 lm stainless steel diffusion stones attached to the end of 1.52 mm ID Teflon tubing. Sampling points were set at approximate depths of 13, 26, and 39 cm below the top of the sand medium. Approximately 6 ml of water sample was pumped out from each sampling point using a peristaltic pump. Permanganate concentration was analyzed using a Varian Cary 1 UV– visible spectrophotometer at wavelengths ranging from
Fig. 2. Photos showing the flow-tank setup (a; side view), the delivery wells (top views) before (b) and after (c) installation of CRP, and excavated surface of the sandy media after the experiment (d; top view). Purple permanganate solution is evident in the CRP source zone immediately after removal of the delivery wells (b). In field settings, these PVC pipes can be replaced by screened wells that will remain in the subsurface. Formation of MnO2 reaction products are indicated by the brownish impurities in the sandy media (d). Noticeably, MnO2 precipitates are spread throughout the sandy media.
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
400 to 700 nm. Hydrazine hydrate solution (0.2 ml) was added to samples to quench further reaction between MnO and dissolved TCE. A small volume of the 4 quenched sample (2 ml) was transferred to a vial and mixed with 4 ml of pentane. The vial was shaken for 30 min and stored for 30 min to allow the sample to reach equilibrium. The upper layer, a TCE-pentane phase, of the mixture was extracted and analyzed for TCE. Samples were taken every day for the first 3 weeks and every other day for the next 3 weeks. TCE concentrations in water samples were analyzed using a GC-ECD (Shimadzu GC 17A). A CRP was cut into thin sections, leached in flowing tap water for one month, and dried in vacuum line for high resolution surface imaging with SEM. SEM analyses were performed using JEOL JSM-820 SEM with Oxford eXL energy dispersive X-ray analyzer at the microscopic and chemical analysis research center (MARC) at The Ohio State University. After the flow-tank experiment, the sandy medium was flushed with tap water for two weeks to remove dissolved TCE and MnO 4 before drained. The drained sandy medium was excavated and collected in a large stainless steel container and thoroughly mixed using a spatula (Fig. 2). Sand samples (n = 17) were collected in grid pattern throughout the mixed sandy media for analyses of MnO2 content. The MnO2 analyses were conducted using an ICP-MS (Perkin–Elmer Elan 6000) at the MARC. 2.4. Quantification of CRP release kinetics In the CRP, KMnO4 granules are dispersed throughout the rigid polymeric matrix, which essentially lets the original shape and dimension be maintained as MnO 4 is released. In this case, release occurs via dissolution-diffusion. Release of MnO 4 from a finite-height controlled– release matrix of cylindrical form can be described with an analytical model developed by Roseman and Higuchi (1970). The model was derived for non-porous, dispersed agent (solid) matrix system, which assumes pseudosteady-state (i.e., A Cs); constant diffusion coefficient; a perfect sink condition (i.e., release in flowing water); homogeneous initial agent distribution; no matrix degradation or swelling; and diffusion as the rate-controlling step (i.e., negligible dissolution kinetics). Here, A and Cs denote the amount of available MnO 4 per unit volume of the CRP and solubility of KMnO4, respectively. These conditions match well with experimental conditions of the column experiment for this study. To provide context for our model analyses, the salient points of the analytical model are presented. The quantity (Q 0 ) of permanganate released per unit time is dQ0 dC ¼ 2phDe r dr dt
ð1Þ
satisfying the boundary condition C¼0
t>0
r ¼ r0
h ¼ h0
ð2Þ
2061
and the initial condition C ¼ Cs
t¼0
r¼0
h¼0
ð3Þ
where De is the effective diffusion coefficient, t is time, h and r are the height and the radius of the area under consideration, and r0 and h0 are radius and height of the cylinder, respectively. The solution to Eq. (1) for the given boundary and initial conditions is C s De t r2 r 1 ln þ r20 r2 ¼ A 2 r0 4
ð4Þ
where Q0 ¼ phAðr20 r2 Þ
ð5Þ
Here, Eqs. (4) and (5) define the Q 0 versus t plots, that describe release kinetics of controlled–release cylinders. The validity of this model has been verified against experimental data (Roseman and Higuchi, 1970). 3. Results and discussions 3.1. Release-rate measurements In practice, the effective diffusion coefficient (De) for CRP can be obtained from short-term (several weeks) column release data using an appropriate model. The estimated De value is then used to model the overall release kinetics of the CRP for longer times. Here, modeling of long-term behavior is desired because the CRP is designed to last for years and decades and performing column tests for such a long period of time is not practical. A column experiment was performed to monitor release of MnO 4 from the prototype CRP (OD · h = 2.5 cm · 5 cm; KMnO4 = 32.2 g; resin matrix). Pre-washing of the CRP removed 2.8 g of KMnO4 from the surface based on mass difference between the CRP before and after rinsing. Fig. 3 shows the temporal changes in MnO 4 concentrations in the column discharge over 20 days of testing period. Permanganate concentrations were initially high at 56.6 mg l1 on day 1, but rapidly decreased to 14.3 mg l1 over the next 24 h. Permanganate concentrations then gradually decreased to 3.3 mg l1 on day 20 while fluctuating in the range of 1.3–13.2 mg l1. Mass balance calculations, based on MnO 4 concentration and flow rate, indicated that total of 5.1 g of MnO 4 was released from the CRP. The release pattern of initial, large concentration spike followed by gradual decrease in MnO 4 concentrations is typical for monolithic matrix-type controlled release devices that yield first-order release kinetics due to increasing diffusion length (Higuchi, 1963). This is because secondary permeability forms inwards from outside of the matrix by dissolution–diffusion of KMnO4.The column data suggested that the prototype CRP pellet could deliver MnO 4 into flowing water at controlled rate over an extended period of time.
60
4
40
-
-1
6
MnO4- Concentration (mg l )
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
Amount MnO4 Released (g)
2062
Measured Amount Modeled Amount Measured Concentration
2
0 0
100
200
300
400
20
0 500
Time (h) MnO 4
Fig. 3. Temporal variations in the concentrations in the column outflow, measured amount of release MnO 4 , and best-fit model simulation of the release kinetics for the prototype CRP. Outflow samples were taken every 12 h for 20 days.
3.2. Porosity development Fig. 4 presents cross-sectional SEM images of the depleted CRP. Formation of secondary porosity is indicated by dark-colored area. For presentation purposes, pore boundaries were illuminated using Photoshop (25· image). Pore sizes were variable in the range of 10–200 lm. Pore shapes were circular, elongated, or irregular. Formation of secondary permeability is indicated by the smaller darkcolored area within a larger secondary porosity (600· image). Liquid crystal resin completely filled up the pore spaces between the KMnO4 granules, creating a CRP with zero initial porosity. However, KMnO4 granules were not uniform in sizes and shapes and not homogeneously dispersed in the resin matrix. The SEM data verified formation of micro-scale secondary capillary permeability through which MnO 4 can be released by a reaction-diffusion. These optical data, however, also suggested that use of uniform KMnO4 nanoparticles and development of more advanced manufacturing technique is desired for construction of homogeneously dispersed KMnO4-resin matrix systems. 3.3. Proof-of-concept flow-tank experiment A proof-of-concept flow-tank experiment was performed to demonstrate the release and spreading of the CRP by itself in porous medium with an ambient flow of water and the efficacy of the CRP scheme in treating dissolved TCE plume. Three CRP pellets (ID · L = 2.5 cm · 10 cm), each containing 100 g KMnO4 granules were set in each of two PVC pipes (ID · L = 2.6 cm · 40 cm) emplaced in the upstream end of the flow tank (total of six CRP pellets). The PVC pipes were then removed from the porous medium leaving the two vertical lines of CRP (Figs. 1 and 2). Permanganate concentrations at the CRP
source were initially high, ranging from 10 to 70 mg l1 during the first 5 days, and decreased and stayed in the range of 10–30 mg l1 throughout the testing period. When the MnO 4 concentrations became somewhat stabilized on day 6, the inflow water was replaced by TCE solution. TCE solutions with variable TCE concentrations ranging from 42 to 1639 lg l1 were added to the input chamber. Variable TCE provided natural fluctuating conditions of contaminant concentrations for the proof-of-concept experiment. Representative data sets showing the temporal and spatial variations in TCE concentrations across the flow tank are summarized in Table 1. Generally, TCE concentrations high in the inflow water were much lower across the flow tank downstream of the active treatment zone. Clearly, there was active destruction of dissolved TCE by the CRP system. The TCE concentrations, however, remained in many cases above the EPA drinking water standard of 5 lg l1 (Table 1). In addition, the TCE concentrations in the outflow water were higher than in the water samples collected from the monitoring wells that were emplaced along the centerline of the flow tank (Figs. 1 and 2). These data indicated that the TCE plume (<2 mg l1 TCE) was not completely destroyed by CRP in the flow tank. Based on the stoichiometry reaction between TCE and 1 MnO 4 (Eq. (6)), 3.6 mg l þ C2 HCl3 þ 2MnO 4 ! 2MnO2 ðsÞ þ 2CO2 þ 3Cl þ H
ð6Þ (30 lM) of MnO 4 is required to completely destroy up to 2 mg l1 (15 lM) TCE solution. In their detailed studies on the reaction kinetics associated with MnO 4 oxidation of TCE, Yan and Schwartz (1999, 2000) reported that TCE oxidation by MnO 4 is independent of pH and TCE has estimated half-life (T1/2) of 17.8 min. In their batch exper-
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
2063
Fig. 4. Cross-sectional SEM images of the leached prototype CRP. Magnifications are, from the top left clockwise, 25·, 150·, 250·, and 600·, respectively. Dark areas are outlets of secondary capillary permeability, grey areas are resin matrices. Shapes of the leached KMnO4 granules are indicated by the grey circles.
Table 1 Variations in the TCE concentrations in the flow tank Sample ID
Day 3
Day 10
Day 16
Day 40
Input Output 1b 1m 1t 2b 2m 2t 3b 3m 3t 4b 4m 4t
595 345 23 74 104 104 80 69 74 45 30 22 28 9
1639 483 12 14 15 14 8 8 7 27 10 4 13 52
42 18 2 12 10 1 5 8 1 1 5 5 3 2
245 95 10 6 89 7 45 47 1 1 4 39 40 1
Unit: lg/l, input: sample collected from the input chamber; output: sample collected from the output chamber; sampling wells were numbered from 1 to 4 starting from the upstream end; b, m, t indicate sampling points at the bottom, middle, and top of the multi-level sample wells, 13, 26, and 39 cm below the top of the sand medium, respectively.
iment, 0.06 mM of TCE was almost completely destroyed by MnO 4 (1 mM) solution in less than 100 min. The estimated residence time for MnO 4 and TCE plume in the flow
tank was 28 h, which may be long enough for complete destruction of TCE by MnO 4 oxidation. These observations suggested that the incomplete destruction of dissolved TCE in the flow tank was probably due to the lack of lateral spreading of MnO 4 and the resulting incomplete mixing with dissolved TCE. The flow-tank experiment was performed using pure silica sand as porous media. Commercial silica sand is essentially pure, without organic carbons and metals commonly found in natural media. Thus natural oxidant demand in the experiments was negligible. The MnO2 reaction zone in porous media can be used as an indicator for estimating the extent of TCE destruction when the porous medium has a negligible oxidant demand (e.g., Lee et al., 2003). In other words, MnO 4 oxidation of TCE is the only MnO2 producing process, and MnO2 precipitates are immobile. Unlike the dissolved TCE, the MnO2 precipitates will survive the tank excavation process. The average Mn content for sand samples collected in the container was measured as 22.1 mg-Mn kg-sands1 (n = 17, std. = 1.8). As indicated by the small standard deviation of 1.8, the Mn contents showed little variations between samples, indicating excellent mixing of the sandy media in the container. The mean value was therefore considered to be
2064
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
appropriate for calculating the overall mass balance between MnO 4 and TCE in the flow tank. The mean Mn content suggested that 6.3 g (0.072 mol) of MnO2 was produced in the sandy media, indicating destruction of 4.7 g (0.036 mol) of TCE by 8.6 g (0.072 mol) of MnO 4 over 37 days (Eq. (6)). The total amount of MnO 4 removed from the six CRP comprising the two vertical lines of MnO 4 source was measured as 15.5 g (0.13 mol) from the mass difference between the CRP before and after the experiment. To quantify the efficiency of TCE destruction, mass-balance calculations were performed using the reaction stoichiometry of Eq. (6). Mass-balance calculation indicated that only 55.3% of the MnO 4 released from the CRP was used for TCE oxidation, probably due to lack of lateral spreading away from the solid forms. Lack of lateral spreading of soluble agent from an un-pumped well in the subsurface was also reported by other studies (Wilson and Mackey, 1995). Development of delivery system that can facilitate lateral spreading of MnO 4 would be required to provide a practical method for field applications. These data, how-
ever, demonstrated that the CRP scheme would be potentially capable of destroying dissolved TCE plume in a long-term, controlled manner. 3.4. Simulations and potential use of the CRP The analytical model (Eqs. (5) and (6)) was solved numerically using the bisection method (Nakamura, 2002) to simulate release kinetics of the prototype CRP. To fit the simulation and column data, the unit of the column data presented in Fig. 4 (mg l1) was converted to g d1 by multiplying the mean value of two subsequent concentration data by the volume of water discharged over the 12 h period. The volume of water was estimated from the ambient flow rate (28.8 l d1) used for the column experiment. The best-fit modeled data estimated for the total mass of MnO 4 released from the CRP for 20 days (5.18 g) were plotted against the column data (Fig. 3). Many of the results from the column experiments could not be modeled well, probably due to the non-homogeneity of the proto-
30
0.9
20
0.6 Modeled Amount Measured Amount Modeled Release Rate
15
10
0.3
-
-1
MnO4 Release Rate (g d )
Amount MnO4- Released (g)
25
5
a 500
0 1000
750
2500
100
2000
80
1500
60 Amount released Release rate
1000
40
500
20
-1
250
-
Amount MnO4 released (g)
0
MnO4- Release rate (g d )
0
b 0 0
500
1000
1500
2000
0 2500
Time (d) Fig. 5. (a) Model simulations of overall release kinetics for the prototype CRP; (b) simulated release kinetics of a field-scale CRP (r · h = 2.5 cm · 100 cm; mass of KMnO4 = 3000 g, mass of MnO 4 ¼ 2257:9 g).
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
type CRP matrix. Pre-washing of the CRP and standardization of subsequent concentration data may also have contributed to the slight discrepancy. However, the overall pattern of modeled release kinetics and measured data were generally comparable to each other. The De value of 8.61 · 107 cm2 s1 was obtained for the prototype CRP from the best-fit simulation. This value was within the typical range for rubbery polymeric matrices (Narasimhan, 2000). Using this De value, overall release behavior of the CRP could be investigated through model simulations over longer times (Fig. 5a). Simulation data indicated that the duration of the CRP would be 810 days. Over this period, 24.2 g of MnO 4 would be released from the prototype CRP containing 32.2 g of KMnO4. Release rates were initially high at 846 mg d1, then decreased to 110 mg d1 in 7 days, to 50 mg d1 in 30 days, to 20 mg d1 in 150 days, and to 10 mg d1 in 1 yr. At this stage in developing controlled release technology, a scoping calculation is provided to suggest how it might perform under field conditions. We assume a homogeneously dispersed CRP pellet having dimensions r · h = 2.5 cm · 100 cm, Cs = 64 lg mm3, and A (MnO 4Þ ¼ 1150:5 lg mm3 (Total mass of KMnO4 in the pellet = 3000 g). The De value was assumed to be 1.389 · 106 cm2 s1. Eqs. (4) and (5) of the analytical model were solved using these parameters. Simulation results are presented in Fig. 5b. Duration of the CRP was estimated to be 2339 days (6.4 yr). Rates of MnO 4 release were initially high at 65.7 g d1 on day 1, then decreased to 12.7 g d1 in 7 days, to 5.6 g d1 in 30 days, and to 2.4 g d1 in 150 days. Afterwards, release rates gradually decreased to 1.4 g d1 in 1 yr, and to 0.5 g l1 in 1344 days (3.7 yr). Let’s assume a CRP reactive barrier system comprising five discrete barriers installed at 2-m interval downstream normal to an aqueous-phase TCE plume. Each barrier comprises 20 delivery wells spaced in 100-cm intervals, and each well contains 10 CRP (mass of KMnO4 = 3000 g; r · h = 2.5 cm · 100 cm). The simulation data suggest that the well-based CRP reactive barrier (W · L · D = 20 m · 10 m · 10 m) can release more than 500 g of MnO 4 daily over the next 3.7 yrs, creating 2000 m3 of chemically active zone in the subsurface. This type of well-based, long-term, and semi-passive PRB system would be useful for remediation of DNAPL contaminated sites where aqueous-phase plume of pollutants is extensive, but concentrations are low; DNAPL is present in the subsurface as a long-term contaminant source; or accessibility and implementation of other remedial technologies are limited.
4. Conclusions A prototype CRP was constructed using resin as matrix. Column test, SEM analyses, and model simulations verified release of MnO 4 from the CRP in a controlled way through the secondary capillary permeability. Proof-ofconcept flow-tank experiment demonstrated that the CRP
2065
scheme would be capable of destroying dissolved TCE plume in a long-term, controlled manner. In order to take advantage of the merits of controlled–release system in environmental remediation, the release behavior of the CRP needs to be optimized for the hydrologic and environmental conditions of target treatment zone. For example, consideration of natural oxidant demand, lateral spreading of permanganate, and volume and concentration of contaminant plume must be made in field applications. Technically, release rates and life-time of the CRP can be adjusted by changing KMnO4/resin volume ratio or dimensions of the pellets. With these factors well addressed, the CRP scheme could potentially be developed as a practical approach for in situ remediation of contaminated aquifers. Perhaps, the most important research need with this approach is the delivery systems for facilitating lateral spreading and mixing of permanganate with the dissolved contaminant plume.
Acknowledgement This material is based upon work supported by the Department of Energy under Grant FG07-02ER63487.
References Gibbs, G.F., Kermasha, S., Alli, I., Mulligan, C.N., 1999. Encapsulation in the food industry: a review. Int. J. Food Sci. Nutr. 50, 213–225. Higuchi, T., 1963. Mechanism of sustained-action medication: theoretical analysis of rate of release of solid drugs dispersed in solid matrices. J. Pharm. Sci. 52, 1145–1149. Langer, R., 1990. New method of drug delivery. Science 249, 1527–1533. Lee, E.S., Seol, Y., Fang, Y.C., Schwartz, F.W., 2003. Destruction efficiencies and dynamics of reaction fronts associated with the permanganate oxidation of trichloroethylene. Environ. Sci. Technol. 37, 2540–2546. Li, X.D., Schwartz, F.W., 2004. DNAPL remediation with in situ chemical oxidation using potassium permanganate. J. Contam. Hydrol. 68, 39– 53. Nakamura, S., 2002. Numerical analysis and graphic visualization with MATLAB. Prentice Hall, Upper Saddle River, NJ, p. 279. Narasimhan, B., 2000. Accurate models in controlled drug delivery systems. In: Wise, D.L. (Ed.), Handbook of Pharmaceutical Controlled Release Technology. Marcel Dekker, New york, pp. 155–182. Roseman, R.W., Higuchi, W.I., 1970. Release of medroxyprogesterone acetate from a silicone polymer. J. Pharm. Sci. 59, 353–357. Ross, C., Murdoch, L.C., Freedman, D.L., Siegrist, R.L., 2005. Characteristics of potassium permanganate encapsulated in polymer. J. Environ. Eng. 131, 1203–1211. Schnarr, M., Truax, C., Farquhar, G., Hood, E., Gonullu, T., Stickney, B.J., 1998. Laboratory and controlled field experiments using potassium permanganate to remediate trichloroethylene and perchloroethylene DNAPLs in porous media. J. Contam. Hydrol. 29, 205–224. Schroth, M.H., Oostrom, M., Wietsma, T.W., Istok, J.D., 2001. In-situ oxidation of trichloroethene by permanganate: effects on porous medium hydraulic properties. J. Contam. Hydrol. 50, 79–98. Seol, Y., Schwartz, F.W., Zhang, H., 2002. A review of in situ chemical oxidation and heterogeneity. Environ. Eng. Geosci. 9, 37–50. Siegrist, R.L., Urynowicz, M.A., West, O.R., Crimi, M.L., Lowe, K.S., 2001. Principles and practices of in situ chemical oxidation using permanganate. Battelle Press, Columbus, OH, p. 336.
2066
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
Wakimoto, K., 2004. Review: utilization advantages of controlled release nitrogen fertilizer on paddy rice cultivation. JARQ-Jpn. Agr. Res. Q. 38, 15–20. Wilson, R.D., Mackey, D.M., 1995. A method for passive release of solutes from an unpumped well. Ground Water 33, 936–945.
Yan, Y.E., Schwartz, F.W., 1999. Oxidative degradation and kinetics of chlorinated ethylenes by potassium permanganate. J. Contam. Hydrol. 37, 343–365. Yan, Y.E., Schwartz, F.W., 2000. Kinetics and mechanisms for TCE oxidation by permanganate. Environ. Sci. Technol. 34, 2535–2541.
Characteristics and applications of controlled–release KMnO4 for groundwater remediation Eung Seok Lee *, Franklin W. Schwartz Department of Geological Sciences, The Ohio State University, Columbus OH 43210, USA Received 30 May 2006; received in revised form 26 September 2006; accepted 27 September 2006 Available online 30 November 2006
Abstract In situ chemical oxidation (ISCO) using potassium permanganate (KMnO4) has been widely used as a practical approach for remediation of groundwater contaminated by chlorinated solvents like trichloroethylene. The most common applications are active flushing schemes, which target the destruction of some contaminant source by injecting concentrated permanganate (MnO 4 Þ solution into the subsurface over a short period of time. Despite many promising results, KMnO4 flushing is often frustrated by inefficiency associated with pore plugging by MnO2 and bypassing. Opportunities exist for the development of new ISCO systems based on KMnO4. The new scheme described in this paper uses controlled–release KMnO4 (CRP) as an active component in the well-based reactive barrier system. This scheme operates to control spreading of a dissolved contaminant plume. Prototype CRP was manufactured by dispersing fine KMnO4 granules in liquid crystal polymer resin matrix. Scanning electron microscope data verified the formation of micro-scale (ID = 20–200 lm) secondary capillary permeability through which MnO 4 is released by a reaction-diffusion process. Column and numerical simulation data indicated that the CRP could deliver MnO 4 in a controlled manner for several years without replenishment. A proof-of-concept flow-tank experiment and model simulations suggested that the CRP scheme could potentially be developed as a practical approach for in situ remediation of contaminated aquifers. This scheme may be suitable for remediation of sites where accessibility is limited or some low-concentration contaminant plume is extensive. Development of delivery systems that can facilitate lateral spreading and mixing of MnO 4 with the contaminant plume is warranted. 2006 Elsevier Ltd. All rights reserved. Keywords: Controlled–release system; Groundwater; Remediation; Potassium permanganate; Chemical oxidation; TCE; Permeable reactive barrier
1. Introduction Over the past decade, advanced oxidation schemes are being used increasingly to address problems of groundwater contamination by chlorinated solvents like trichloroethylene (TCE). Of the potential oxidants, potassium permanganate (KMnO4) is often used because of its ease of handling and relatively low cost (Schnarr et al., 1998; Siegrist et al., 2001; Seol et al., 2002). Studies have shown that permanganate ðMnO 4 Þ rapidly transforms most chlorinated solvents to short-lived intermediates, and eventu-
*
Corresponding author. Tel.: +1 614 292 0585; fax: +1 614 292 7688. E-mail address: [email protected] (E.S. Lee).
0045-6535/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.chemosphere.2006.09.093
ally innocuous by-products (Yan and Schwartz, 1999, 2000). In most applications, MnO 4 is applied into the subsurface via injection wells in a concentrated aqueous solution as a flood of some source zone. This active flushing scheme treats the problem of contamination through mass removal. When this method works, it is usually because units are permeable and solvents are present at low saturations. When it fails, the main problem is commonly inability to deliver the oxidant due to physical heterogeneity in hydraulic conductivity and/or chemical heterogeneity related to pore plugging by solid reaction products like MnO2 (Seol et al., 2002). Column and flow-tank experiments (Schroth et al., 2001; Lee et al., 2003; Li and Schwartz, 2004) showed how plugging caused the
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
MnO 4 flood to bypass zones of relatively high DNAPL (i.e., a liquid that is denser than water and does not dissolve or mix easily in water) saturation. In field settings, leaving volumes of untreated DNAPL in place contributes to a rebound in aqueous concentrations once flushing is discontinued (e.g., Siegrist et al., 2001). Clearly, then, there are opportunities to use delivery systems in an improved way. This paper examines aspects of using KMnO4 as the active component of a well-based permeable reactive barrier (PRB) system. In this application, KMnO4 is emplaced as a controlled–release solid to create chemically active zones in the subsurface. The solids are resin-KMnO4 composites, cylindrical in form. The controlled–release KMnO4 (CRP) can be applied against a plume in multiple layers of closely spaced wells. Ideally, these forms would dissolve and release low concentrations of MnO 4 at a controlled rate to maintain the predetermined level of the agent within the chemically active zone. Such a barrier system could function by itself over a period of several years and possibly decades without replenishment. When the CRP is exhausted, it could be replaced. The KMnO4 used in this way would provide long-term controlling scheme for aqueous-phase plumes of contaminants. With the typically low dissolved concentrations of contaminants in most plumes, there will be much less tendency for the porous medium to plug up. One of the necessary steps in developing this remedial concept is manufacturing and testing the CRP, which can deliver MnO 4 in a reasonably controlled way over years. This paper describes prototype CRP which has some of the necessary controlled release characteristics and column experiments designed to provide release data. It also presents results of a proof-of-concept experiment with a flow tank, suggesting how the CRP could be used for long-term control of contaminant plume in the subsurface. This paper concludes with presentation of modeling approach capable of quantifying release behaviors of the CRP for the environmental conditions of the contaminated sites.
2059
2. Materials and Methods
all agent level; preservation of agents that can be rapidly destroyed by environmental conditions; reduced need for maintenance; and lower cost (e.g., Langer, 1990). An extruded form of KMnO4 (CAIROX-CR, Carus Chemical Company) is commercially available for removing grease and hydrogen sulfide odor in surface water. This product uses the porous matrix comprising silica and clays as a means of delaying the otherwise instant dissolution of KMnO4 in water. Microcapsules of KMnO4 having slowrelease properties have been also investigated in a column experimental study for possible applications to groundwater remediation (Ross et al., 2005). This material, comprising KMnO4 granule core and waxy polymeric coating, yielded extended duration of permanganate release in the range of 3–80 days. Our implementation of controlled release technology takes advantage of inert organic crystalline matrix to provide for slow dissolution and diffusioncontrolled transport of agent. This form is designed for supply of treatment chemicals to the contaminated aquifers in situ, at a controlled release rate over several years and perhaps decades without replenishment. For this study, a prototype CRP was manufactured by dispersing KMnO4 (Carux Chemical Company, Peru, IL) in organic crystalline matrices (Castin’Craft Clear Polyester Resin, Environmental Technology Inc., Fields Landing, CA). Liquid resin and fine KMnO4 granules were mixed in a specially designed cylindrical mould (ID = 2.5 cm) and allowed to crystallize at room temperature. Ideally, the CRP would have near zero initial permeability, while secondary capillary permeability will form later in the matrix as MnO 4 is released. The crystalline matrix system was designed to yield much smaller permeability for MnO 4 transport as compared to the porous sandy-clay matrix system (CAIROX-CR). The outside diameter of the CRP (2.5 cm) was selected in order to fit inside the delivery wells (2.6 cm) used for the later flow-tank experiments. The CRP used for the column tests was 5 cm-long, had a diameter of 2.5 cm, and contained 35 g of KMnO4. The CRP used for the flow-tank experiment was 10 cm-long with a diameter of 2.5 cm, and contained 100 g of KMnO4.
2.1. Manufacturing controlled–release KMnO4 (CRP)
2.2. Column test and flow-tank experiment
The ability to release a compound at a controlled rate is important in variety of different applications (Gibbs et al., 1999; Wakimoto, 2004). The controlled–release system transfers active agent from a reservoir to a target host, in order to maintain a predetermined concentration or emission level of the agent for a specified period of time or desired fraction (Langer, 1990). This system provides various advantages over conventional delivery systems. For example, risks for toxicity and secondary contamination (overdose) or ineffectiveness (underdose) can be eliminated by maintaining the agent concentrations within the desired range. Other advantages may include in situ construction of stable reaction zones; localized delivery of agents to a particular subsurface compartment, thereby lowering the over-
Column tests were performed using CHROMAFLEX (Kontes) glass column (ID = 4.8 cm, length = 15 cm, v = 270 ml) to determine release rates of CRP in flowing water. Ambient flow rates of 20 ml min1 were maintained using Ismatec peristaltic pumps (Cole Palmer Instrument Company, Chicago, Illinois) to provide perfect sink conditions for CRP release. The CRP used for the column experiment was pre-washed with tap water to remove KMnO4 granules attached to the surface of the pellet. A glass tank was constructed (L · W · D = 100 cm · 20 cm · 50 cm) for flow-tank experiment (Figs. 1 and 2). The tank was filled with 170 kg of silica sand (U.S. Silica, Ottawa, Illinois), which was packed to an approximate bulk density of 1.65 g cm3 and porosity of 37.7%
2060
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
Metal Screen
Multilevel Monitoring Wells (sampling depths: 13, 26, 39 cm below surface)
Input Chamber
Output Chamber
Sands
Center Line
Flow
CRK Delivery Wells
Fig. 1. Schematic diagram of the glass flow-tank setup (L · W · D = 100 cm · 20 cm · 50 cm). Scales are exaggerated to show the compartments. CRP source zone is located at the upstream part of the flow tank. Only the boundaries of the sandy media (L · W · D = 80 cm · 20 cm · 45 cm) are drawn to show well locations in the flow tank. Three peristaltic pumps were used for flow control and sample collection.
(L · W · D = 80 cm · 20 cm · 45 cm). Input and output chambers on the upstream and downstream ends of the tank helped to control inflow and outflow rates. These chambers were separated from the sand by rigid stainless steel screens that were covered with fiber glass to prevent sand from entering the chambers.
Water was pumped into the inflow chamber using Ismatec peristaltic pumps (Cole Palmer Instrument Company, Chicago, Illinois) to create a uniform flow into the upstream end of the tank. The outflow rate was controlled in the same manner. An inflow rate of 19.2 l d1 to the tank provided an ambient flow velocity of 57 cm d1 in the sandy porous medium. At this velocity, it required 34 h (1.4 d) for water to flow through the tank. Effluent from the tank was collected in a waste container and treated to remove TCE and MnO 4 before disposal. The excess MnO 4 was removed from the effluent through reduction by sodium thiosulphate (Na2S2O3). 2.3. Water sampling, flow-tank excavation, SEM, and chemical analyses Four small multi-level wells were emplaced along the centerline of the tank for sample collection (Figs. 1 and 2). They were constructed with 2 lm stainless steel diffusion stones attached to the end of 1.52 mm ID Teflon tubing. Sampling points were set at approximate depths of 13, 26, and 39 cm below the top of the sand medium. Approximately 6 ml of water sample was pumped out from each sampling point using a peristaltic pump. Permanganate concentration was analyzed using a Varian Cary 1 UV– visible spectrophotometer at wavelengths ranging from
Fig. 2. Photos showing the flow-tank setup (a; side view), the delivery wells (top views) before (b) and after (c) installation of CRP, and excavated surface of the sandy media after the experiment (d; top view). Purple permanganate solution is evident in the CRP source zone immediately after removal of the delivery wells (b). In field settings, these PVC pipes can be replaced by screened wells that will remain in the subsurface. Formation of MnO2 reaction products are indicated by the brownish impurities in the sandy media (d). Noticeably, MnO2 precipitates are spread throughout the sandy media.
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
400 to 700 nm. Hydrazine hydrate solution (0.2 ml) was added to samples to quench further reaction between MnO and dissolved TCE. A small volume of the 4 quenched sample (2 ml) was transferred to a vial and mixed with 4 ml of pentane. The vial was shaken for 30 min and stored for 30 min to allow the sample to reach equilibrium. The upper layer, a TCE-pentane phase, of the mixture was extracted and analyzed for TCE. Samples were taken every day for the first 3 weeks and every other day for the next 3 weeks. TCE concentrations in water samples were analyzed using a GC-ECD (Shimadzu GC 17A). A CRP was cut into thin sections, leached in flowing tap water for one month, and dried in vacuum line for high resolution surface imaging with SEM. SEM analyses were performed using JEOL JSM-820 SEM with Oxford eXL energy dispersive X-ray analyzer at the microscopic and chemical analysis research center (MARC) at The Ohio State University. After the flow-tank experiment, the sandy medium was flushed with tap water for two weeks to remove dissolved TCE and MnO 4 before drained. The drained sandy medium was excavated and collected in a large stainless steel container and thoroughly mixed using a spatula (Fig. 2). Sand samples (n = 17) were collected in grid pattern throughout the mixed sandy media for analyses of MnO2 content. The MnO2 analyses were conducted using an ICP-MS (Perkin–Elmer Elan 6000) at the MARC. 2.4. Quantification of CRP release kinetics In the CRP, KMnO4 granules are dispersed throughout the rigid polymeric matrix, which essentially lets the original shape and dimension be maintained as MnO 4 is released. In this case, release occurs via dissolution-diffusion. Release of MnO 4 from a finite-height controlled– release matrix of cylindrical form can be described with an analytical model developed by Roseman and Higuchi (1970). The model was derived for non-porous, dispersed agent (solid) matrix system, which assumes pseudosteady-state (i.e., A Cs); constant diffusion coefficient; a perfect sink condition (i.e., release in flowing water); homogeneous initial agent distribution; no matrix degradation or swelling; and diffusion as the rate-controlling step (i.e., negligible dissolution kinetics). Here, A and Cs denote the amount of available MnO 4 per unit volume of the CRP and solubility of KMnO4, respectively. These conditions match well with experimental conditions of the column experiment for this study. To provide context for our model analyses, the salient points of the analytical model are presented. The quantity (Q 0 ) of permanganate released per unit time is dQ0 dC ¼ 2phDe r dr dt
ð1Þ
satisfying the boundary condition C¼0
t>0
r ¼ r0
h ¼ h0
ð2Þ
2061
and the initial condition C ¼ Cs
t¼0
r¼0
h¼0
ð3Þ
where De is the effective diffusion coefficient, t is time, h and r are the height and the radius of the area under consideration, and r0 and h0 are radius and height of the cylinder, respectively. The solution to Eq. (1) for the given boundary and initial conditions is C s De t r2 r 1 ln þ r20 r2 ¼ A 2 r0 4
ð4Þ
where Q0 ¼ phAðr20 r2 Þ
ð5Þ
Here, Eqs. (4) and (5) define the Q 0 versus t plots, that describe release kinetics of controlled–release cylinders. The validity of this model has been verified against experimental data (Roseman and Higuchi, 1970). 3. Results and discussions 3.1. Release-rate measurements In practice, the effective diffusion coefficient (De) for CRP can be obtained from short-term (several weeks) column release data using an appropriate model. The estimated De value is then used to model the overall release kinetics of the CRP for longer times. Here, modeling of long-term behavior is desired because the CRP is designed to last for years and decades and performing column tests for such a long period of time is not practical. A column experiment was performed to monitor release of MnO 4 from the prototype CRP (OD · h = 2.5 cm · 5 cm; KMnO4 = 32.2 g; resin matrix). Pre-washing of the CRP removed 2.8 g of KMnO4 from the surface based on mass difference between the CRP before and after rinsing. Fig. 3 shows the temporal changes in MnO 4 concentrations in the column discharge over 20 days of testing period. Permanganate concentrations were initially high at 56.6 mg l1 on day 1, but rapidly decreased to 14.3 mg l1 over the next 24 h. Permanganate concentrations then gradually decreased to 3.3 mg l1 on day 20 while fluctuating in the range of 1.3–13.2 mg l1. Mass balance calculations, based on MnO 4 concentration and flow rate, indicated that total of 5.1 g of MnO 4 was released from the CRP. The release pattern of initial, large concentration spike followed by gradual decrease in MnO 4 concentrations is typical for monolithic matrix-type controlled release devices that yield first-order release kinetics due to increasing diffusion length (Higuchi, 1963). This is because secondary permeability forms inwards from outside of the matrix by dissolution–diffusion of KMnO4.The column data suggested that the prototype CRP pellet could deliver MnO 4 into flowing water at controlled rate over an extended period of time.
60
4
40
-
-1
6
MnO4- Concentration (mg l )
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
Amount MnO4 Released (g)
2062
Measured Amount Modeled Amount Measured Concentration
2
0 0
100
200
300
400
20
0 500
Time (h) MnO 4
Fig. 3. Temporal variations in the concentrations in the column outflow, measured amount of release MnO 4 , and best-fit model simulation of the release kinetics for the prototype CRP. Outflow samples were taken every 12 h for 20 days.
3.2. Porosity development Fig. 4 presents cross-sectional SEM images of the depleted CRP. Formation of secondary porosity is indicated by dark-colored area. For presentation purposes, pore boundaries were illuminated using Photoshop (25· image). Pore sizes were variable in the range of 10–200 lm. Pore shapes were circular, elongated, or irregular. Formation of secondary permeability is indicated by the smaller darkcolored area within a larger secondary porosity (600· image). Liquid crystal resin completely filled up the pore spaces between the KMnO4 granules, creating a CRP with zero initial porosity. However, KMnO4 granules were not uniform in sizes and shapes and not homogeneously dispersed in the resin matrix. The SEM data verified formation of micro-scale secondary capillary permeability through which MnO 4 can be released by a reaction-diffusion. These optical data, however, also suggested that use of uniform KMnO4 nanoparticles and development of more advanced manufacturing technique is desired for construction of homogeneously dispersed KMnO4-resin matrix systems. 3.3. Proof-of-concept flow-tank experiment A proof-of-concept flow-tank experiment was performed to demonstrate the release and spreading of the CRP by itself in porous medium with an ambient flow of water and the efficacy of the CRP scheme in treating dissolved TCE plume. Three CRP pellets (ID · L = 2.5 cm · 10 cm), each containing 100 g KMnO4 granules were set in each of two PVC pipes (ID · L = 2.6 cm · 40 cm) emplaced in the upstream end of the flow tank (total of six CRP pellets). The PVC pipes were then removed from the porous medium leaving the two vertical lines of CRP (Figs. 1 and 2). Permanganate concentrations at the CRP
source were initially high, ranging from 10 to 70 mg l1 during the first 5 days, and decreased and stayed in the range of 10–30 mg l1 throughout the testing period. When the MnO 4 concentrations became somewhat stabilized on day 6, the inflow water was replaced by TCE solution. TCE solutions with variable TCE concentrations ranging from 42 to 1639 lg l1 were added to the input chamber. Variable TCE provided natural fluctuating conditions of contaminant concentrations for the proof-of-concept experiment. Representative data sets showing the temporal and spatial variations in TCE concentrations across the flow tank are summarized in Table 1. Generally, TCE concentrations high in the inflow water were much lower across the flow tank downstream of the active treatment zone. Clearly, there was active destruction of dissolved TCE by the CRP system. The TCE concentrations, however, remained in many cases above the EPA drinking water standard of 5 lg l1 (Table 1). In addition, the TCE concentrations in the outflow water were higher than in the water samples collected from the monitoring wells that were emplaced along the centerline of the flow tank (Figs. 1 and 2). These data indicated that the TCE plume (<2 mg l1 TCE) was not completely destroyed by CRP in the flow tank. Based on the stoichiometry reaction between TCE and 1 MnO 4 (Eq. (6)), 3.6 mg l þ C2 HCl3 þ 2MnO 4 ! 2MnO2 ðsÞ þ 2CO2 þ 3Cl þ H
ð6Þ (30 lM) of MnO 4 is required to completely destroy up to 2 mg l1 (15 lM) TCE solution. In their detailed studies on the reaction kinetics associated with MnO 4 oxidation of TCE, Yan and Schwartz (1999, 2000) reported that TCE oxidation by MnO 4 is independent of pH and TCE has estimated half-life (T1/2) of 17.8 min. In their batch exper-
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
2063
Fig. 4. Cross-sectional SEM images of the leached prototype CRP. Magnifications are, from the top left clockwise, 25·, 150·, 250·, and 600·, respectively. Dark areas are outlets of secondary capillary permeability, grey areas are resin matrices. Shapes of the leached KMnO4 granules are indicated by the grey circles.
Table 1 Variations in the TCE concentrations in the flow tank Sample ID
Day 3
Day 10
Day 16
Day 40
Input Output 1b 1m 1t 2b 2m 2t 3b 3m 3t 4b 4m 4t
595 345 23 74 104 104 80 69 74 45 30 22 28 9
1639 483 12 14 15 14 8 8 7 27 10 4 13 52
42 18 2 12 10 1 5 8 1 1 5 5 3 2
245 95 10 6 89 7 45 47 1 1 4 39 40 1
Unit: lg/l, input: sample collected from the input chamber; output: sample collected from the output chamber; sampling wells were numbered from 1 to 4 starting from the upstream end; b, m, t indicate sampling points at the bottom, middle, and top of the multi-level sample wells, 13, 26, and 39 cm below the top of the sand medium, respectively.
iment, 0.06 mM of TCE was almost completely destroyed by MnO 4 (1 mM) solution in less than 100 min. The estimated residence time for MnO 4 and TCE plume in the flow
tank was 28 h, which may be long enough for complete destruction of TCE by MnO 4 oxidation. These observations suggested that the incomplete destruction of dissolved TCE in the flow tank was probably due to the lack of lateral spreading of MnO 4 and the resulting incomplete mixing with dissolved TCE. The flow-tank experiment was performed using pure silica sand as porous media. Commercial silica sand is essentially pure, without organic carbons and metals commonly found in natural media. Thus natural oxidant demand in the experiments was negligible. The MnO2 reaction zone in porous media can be used as an indicator for estimating the extent of TCE destruction when the porous medium has a negligible oxidant demand (e.g., Lee et al., 2003). In other words, MnO 4 oxidation of TCE is the only MnO2 producing process, and MnO2 precipitates are immobile. Unlike the dissolved TCE, the MnO2 precipitates will survive the tank excavation process. The average Mn content for sand samples collected in the container was measured as 22.1 mg-Mn kg-sands1 (n = 17, std. = 1.8). As indicated by the small standard deviation of 1.8, the Mn contents showed little variations between samples, indicating excellent mixing of the sandy media in the container. The mean value was therefore considered to be
2064
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
appropriate for calculating the overall mass balance between MnO 4 and TCE in the flow tank. The mean Mn content suggested that 6.3 g (0.072 mol) of MnO2 was produced in the sandy media, indicating destruction of 4.7 g (0.036 mol) of TCE by 8.6 g (0.072 mol) of MnO 4 over 37 days (Eq. (6)). The total amount of MnO 4 removed from the six CRP comprising the two vertical lines of MnO 4 source was measured as 15.5 g (0.13 mol) from the mass difference between the CRP before and after the experiment. To quantify the efficiency of TCE destruction, mass-balance calculations were performed using the reaction stoichiometry of Eq. (6). Mass-balance calculation indicated that only 55.3% of the MnO 4 released from the CRP was used for TCE oxidation, probably due to lack of lateral spreading away from the solid forms. Lack of lateral spreading of soluble agent from an un-pumped well in the subsurface was also reported by other studies (Wilson and Mackey, 1995). Development of delivery system that can facilitate lateral spreading of MnO 4 would be required to provide a practical method for field applications. These data, how-
ever, demonstrated that the CRP scheme would be potentially capable of destroying dissolved TCE plume in a long-term, controlled manner. 3.4. Simulations and potential use of the CRP The analytical model (Eqs. (5) and (6)) was solved numerically using the bisection method (Nakamura, 2002) to simulate release kinetics of the prototype CRP. To fit the simulation and column data, the unit of the column data presented in Fig. 4 (mg l1) was converted to g d1 by multiplying the mean value of two subsequent concentration data by the volume of water discharged over the 12 h period. The volume of water was estimated from the ambient flow rate (28.8 l d1) used for the column experiment. The best-fit modeled data estimated for the total mass of MnO 4 released from the CRP for 20 days (5.18 g) were plotted against the column data (Fig. 3). Many of the results from the column experiments could not be modeled well, probably due to the non-homogeneity of the proto-
30
0.9
20
0.6 Modeled Amount Measured Amount Modeled Release Rate
15
10
0.3
-
-1
MnO4 Release Rate (g d )
Amount MnO4- Released (g)
25
5
a 500
0 1000
750
2500
100
2000
80
1500
60 Amount released Release rate
1000
40
500
20
-1
250
-
Amount MnO4 released (g)
0
MnO4- Release rate (g d )
0
b 0 0
500
1000
1500
2000
0 2500
Time (d) Fig. 5. (a) Model simulations of overall release kinetics for the prototype CRP; (b) simulated release kinetics of a field-scale CRP (r · h = 2.5 cm · 100 cm; mass of KMnO4 = 3000 g, mass of MnO 4 ¼ 2257:9 g).
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
type CRP matrix. Pre-washing of the CRP and standardization of subsequent concentration data may also have contributed to the slight discrepancy. However, the overall pattern of modeled release kinetics and measured data were generally comparable to each other. The De value of 8.61 · 107 cm2 s1 was obtained for the prototype CRP from the best-fit simulation. This value was within the typical range for rubbery polymeric matrices (Narasimhan, 2000). Using this De value, overall release behavior of the CRP could be investigated through model simulations over longer times (Fig. 5a). Simulation data indicated that the duration of the CRP would be 810 days. Over this period, 24.2 g of MnO 4 would be released from the prototype CRP containing 32.2 g of KMnO4. Release rates were initially high at 846 mg d1, then decreased to 110 mg d1 in 7 days, to 50 mg d1 in 30 days, to 20 mg d1 in 150 days, and to 10 mg d1 in 1 yr. At this stage in developing controlled release technology, a scoping calculation is provided to suggest how it might perform under field conditions. We assume a homogeneously dispersed CRP pellet having dimensions r · h = 2.5 cm · 100 cm, Cs = 64 lg mm3, and A (MnO 4Þ ¼ 1150:5 lg mm3 (Total mass of KMnO4 in the pellet = 3000 g). The De value was assumed to be 1.389 · 106 cm2 s1. Eqs. (4) and (5) of the analytical model were solved using these parameters. Simulation results are presented in Fig. 5b. Duration of the CRP was estimated to be 2339 days (6.4 yr). Rates of MnO 4 release were initially high at 65.7 g d1 on day 1, then decreased to 12.7 g d1 in 7 days, to 5.6 g d1 in 30 days, and to 2.4 g d1 in 150 days. Afterwards, release rates gradually decreased to 1.4 g d1 in 1 yr, and to 0.5 g l1 in 1344 days (3.7 yr). Let’s assume a CRP reactive barrier system comprising five discrete barriers installed at 2-m interval downstream normal to an aqueous-phase TCE plume. Each barrier comprises 20 delivery wells spaced in 100-cm intervals, and each well contains 10 CRP (mass of KMnO4 = 3000 g; r · h = 2.5 cm · 100 cm). The simulation data suggest that the well-based CRP reactive barrier (W · L · D = 20 m · 10 m · 10 m) can release more than 500 g of MnO 4 daily over the next 3.7 yrs, creating 2000 m3 of chemically active zone in the subsurface. This type of well-based, long-term, and semi-passive PRB system would be useful for remediation of DNAPL contaminated sites where aqueous-phase plume of pollutants is extensive, but concentrations are low; DNAPL is present in the subsurface as a long-term contaminant source; or accessibility and implementation of other remedial technologies are limited.
4. Conclusions A prototype CRP was constructed using resin as matrix. Column test, SEM analyses, and model simulations verified release of MnO 4 from the CRP in a controlled way through the secondary capillary permeability. Proof-ofconcept flow-tank experiment demonstrated that the CRP
2065
scheme would be capable of destroying dissolved TCE plume in a long-term, controlled manner. In order to take advantage of the merits of controlled–release system in environmental remediation, the release behavior of the CRP needs to be optimized for the hydrologic and environmental conditions of target treatment zone. For example, consideration of natural oxidant demand, lateral spreading of permanganate, and volume and concentration of contaminant plume must be made in field applications. Technically, release rates and life-time of the CRP can be adjusted by changing KMnO4/resin volume ratio or dimensions of the pellets. With these factors well addressed, the CRP scheme could potentially be developed as a practical approach for in situ remediation of contaminated aquifers. Perhaps, the most important research need with this approach is the delivery systems for facilitating lateral spreading and mixing of permanganate with the dissolved contaminant plume.
Acknowledgement This material is based upon work supported by the Department of Energy under Grant FG07-02ER63487.
References Gibbs, G.F., Kermasha, S., Alli, I., Mulligan, C.N., 1999. Encapsulation in the food industry: a review. Int. J. Food Sci. Nutr. 50, 213–225. Higuchi, T., 1963. Mechanism of sustained-action medication: theoretical analysis of rate of release of solid drugs dispersed in solid matrices. J. Pharm. Sci. 52, 1145–1149. Langer, R., 1990. New method of drug delivery. Science 249, 1527–1533. Lee, E.S., Seol, Y., Fang, Y.C., Schwartz, F.W., 2003. Destruction efficiencies and dynamics of reaction fronts associated with the permanganate oxidation of trichloroethylene. Environ. Sci. Technol. 37, 2540–2546. Li, X.D., Schwartz, F.W., 2004. DNAPL remediation with in situ chemical oxidation using potassium permanganate. J. Contam. Hydrol. 68, 39– 53. Nakamura, S., 2002. Numerical analysis and graphic visualization with MATLAB. Prentice Hall, Upper Saddle River, NJ, p. 279. Narasimhan, B., 2000. Accurate models in controlled drug delivery systems. In: Wise, D.L. (Ed.), Handbook of Pharmaceutical Controlled Release Technology. Marcel Dekker, New york, pp. 155–182. Roseman, R.W., Higuchi, W.I., 1970. Release of medroxyprogesterone acetate from a silicone polymer. J. Pharm. Sci. 59, 353–357. Ross, C., Murdoch, L.C., Freedman, D.L., Siegrist, R.L., 2005. Characteristics of potassium permanganate encapsulated in polymer. J. Environ. Eng. 131, 1203–1211. Schnarr, M., Truax, C., Farquhar, G., Hood, E., Gonullu, T., Stickney, B.J., 1998. Laboratory and controlled field experiments using potassium permanganate to remediate trichloroethylene and perchloroethylene DNAPLs in porous media. J. Contam. Hydrol. 29, 205–224. Schroth, M.H., Oostrom, M., Wietsma, T.W., Istok, J.D., 2001. In-situ oxidation of trichloroethene by permanganate: effects on porous medium hydraulic properties. J. Contam. Hydrol. 50, 79–98. Seol, Y., Schwartz, F.W., Zhang, H., 2002. A review of in situ chemical oxidation and heterogeneity. Environ. Eng. Geosci. 9, 37–50. Siegrist, R.L., Urynowicz, M.A., West, O.R., Crimi, M.L., Lowe, K.S., 2001. Principles and practices of in situ chemical oxidation using permanganate. Battelle Press, Columbus, OH, p. 336.
2066
E.S. Lee, F.W. Schwartz / Chemosphere 66 (2007) 2058–2066
Wakimoto, K., 2004. Review: utilization advantages of controlled release nitrogen fertilizer on paddy rice cultivation. JARQ-Jpn. Agr. Res. Q. 38, 15–20. Wilson, R.D., Mackey, D.M., 1995. A method for passive release of solutes from an unpumped well. Ground Water 33, 936–945.
Yan, Y.E., Schwartz, F.W., 1999. Oxidative degradation and kinetics of chlorinated ethylenes by potassium permanganate. J. Contam. Hydrol. 37, 343–365. Yan, Y.E., Schwartz, F.W., 2000. Kinetics and mechanisms for TCE oxidation by permanganate. Environ. Sci. Technol. 34, 2535–2541.