Electrochemical Destruction of - ACS Publications - American

May 4, 2010 - in the Accelrys Materials Studio (27) modeling suite. Unre- .... R2. 100 mM NaClO4. 510 ( 28. 0.991. CAP water. 487 ( 19. 0.995 .... 200...
1 downloads 0 Views 602KB Size
Environ. Sci. Technol. 2010, 44, 4264–4269

Electrochemical Destruction of N-Nitrosodimethylamine in Reverse Osmosis Concentrates using Boron-doped Diamond Film Electrodes BRIAN P. CHAPLIN,* GLENN SCHRADER, AND JAMES FARRELL Department of Chemical and Environmental Engineering University of Arizona, Tucson, Arizona 85721

Received December 21, 2009. Revised manuscript received April 8, 2010. Accepted April 20, 2010.

Boron-doped diamond (BDD) film electrodes were used to electrochemically destroy N-nitrosodimethylamine (NDMA) in reverse osmosis (RO) concentrates. Batch experiments were conducted to investigate the effects of dissolved organic carbon (DOC), chloride (Cl-), bicarbonate (HCO3-), and hardness on rates of NDMA destruction via both oxidation and reduction. Experimental results showed that NDMA oxidation rates were not affected by DOC, Cl-, or HCO3- at concentrations present in RO concentrates. However, hydroxyl radical scavenging at 100 mM concentrations of HCO3- and Clshifted the reaction mechanism of NDMA oxidation from hydroxyl radical mediated to direct electron transfer oxidation. In the 100 mM Cl- electrolyte experimental evidence suggests that the in situ production of ClO•3 also contributes to NDMA oxidation. Density functional theory calculations support a reaction mechanism between ClO•3 and NDMA, with an activation barrier of 7.2 kJ/mol. Flow-through experiments with RO concentrate yielded surface area normalized first-order rate constants for NDMA (40.6 ( 3.7 L/m2 h) and DOC (as C) (38.3 ( 2.2 L/m2 h) removal that were mass transfer limited at a 2 mA/cm2 current density. This research shows that electrochemical oxidation using BDD electrodes has an advantage over other advanced oxidation processes, as organics were readily oxidized in the presence of high HCO3- concentrations.

Introduction Reverse osmosis (RO) is increasingly being used to purify impaired drinking water sources and to facilitate the reuse of wastewater effluents (1–4). Although RO produces a high quality effluent, the concentrate is high in dissolved solids and dissolved organic carbon (DOC). The DOC in RO concentrates can contain a mix of organic contaminants (e.g., personal care products, disinfection byproducts, pharmaceuticals, and endocrine disruptors) (2, 3), making them a significant point source for release into the environment. The associated disposal costs can be significant and are a key factor in determining the economic feasibility of RO treatment (5). The disinfection byproduct N-nitrosodimethylamine (NDMA) is a growing health concern (6) and present in * Corresponding author phone: 217-369-5529; fax: 520-621-6048; e-mail: [email protected]. 4264

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 11, 2010

chlorinated wastewater effluents at a concentration of 20-100 ng/L (1, 7), and is further concentrated by RO treatment (1). Oxidation of NDMA has been investigated by a variety of advanced oxidation processes (AOPs) that generate hydroxyl radicals (8–10), but in general AOPs have shown limited effectiveness for target compound oxidation in natural waters because of hydroxyl radical scavenging by bicarbonate and dissolved organic carbon (DOC) (3, 9, 11, 12). Electrochemical oxidation using boron-doped diamond (BDD) film electrodes is an emerging AOP for the removal of organic compounds from water. BDD anodes can oxidize compounds via direct electron transfer at the electrode surface and by hydroxyl radicals produced via water oxidation (13–15). BDD cathodes have shown structurally selective reduction of compounds (15–17) that is thought to occur by charge transfer on the electrode surface (17). The combination of target compound destruction at both the anode and cathode during electrochemical treatment provides an additional mechanism for compound removal over traditional AOPs. BDD electrodes are also resistant to fouling by redox active metals and have a long service-life compared to other electrode materials (18, 19). Although several studies have documented the ability of BDD electrodes to oxidize complex and recalcitrant waste streams (14, 20–22), research is needed to investigate the effects of solution composition on target compound destruction using BDD electrodes. Prior work has shown that the removal of NDMA at BDD electrodes occurs by several mechanisms (e.g., direct electron transfer, oxidation by hydroxyl radicals, and reduction) (15). Therefore, the aim of this study was to investigate the effects of solution composition (DOC, Cl-, HCO3-, and hardness) on the destruction of NDMA using BDD film electrodes. A combination of batch, flow-through, and voltammetry experiments, along with density functional theory modeling were used to determine the effects of dissolved constituents present in RO concentrates on NDMA removal rates and provide mechanistic information on NDMA transformation. In addition, the feasibility of treating NDMA-contaminated RO brines is assessed by estimating the energy requirements for both NDMA and DOC removal.

Experimental Section Reagents and Water Source. All chemicals were reagent grade and were obtained from Sigma-Aldrich, except N-dimethylnitramine which was obtained from Cambridge Isotope Laboratories, Inc. (Andover, MA). All chemicals were used as received without additional purification. NaCl, NaHCO3, NaClO4, LiClO4, NaClO2, NaClO3, and RO concentrate were used as background electrolyte solutions. NaCl and NaHCO3 were used to investigate the effect of chloride and bicarbonate ions on rates of NDMA oxidation. The perchlorate electrolytes were used as controls, because ClO4- is nonreactive at BDD anodes and cathodes (17, 23). In some experiments, LiClO4 was substituted for NaClO4 to avoid Na+ interference with the detection of cationic amines using the ion chromatograph. CaCl2 and MgCl2 were added to select experiments to investigate the effect of scale formation on the cathode surfaces. The RO concentrate water was obtained from the Central Arizona Project (CAP) aqueduct in Tucson, Arizona. The CAP aqueduct is 541 km long and diverts Colorado River water from Lake Havasu to central and southern Arizona to supplement local drinking water supplies. The CAP water was treated by microfiltration followed by RO. The RO concentrate was subsequently filtered through 0.45 µm filters and stored in glass Pyrex bottles at 4 °C. The CAP water had 10.1021/es903872p

 2010 American Chemical Society

Published on Web 05/04/2010

TABLE 1. Major Dissolved Constituents, pH, and Conductivity of CAP RO Brine concentration 2+

Mg Ca2+ Ba2+ Na+ K+ ClSO42HCO3NO3FDOC (as C) pH conductivity (mS/cm)

constituent (mM) 5.00 7.65 0.003 18.9 0.58 6.20 16.5 4.74 0.05 0.03 1.60 7.5 4.45

a conductivity of 4.45 mS/cm, which was equivalent to a 75 mM ionic strength electrolyte. The composition of the CAP RO concentrate is shown in Table 1. Rotating Disk Electrode Experiments. Reaction rates for NDMA oxidation and reduction were measured under constant current conditions. Duplicate batch experiments were performed using an initial NDMA concentration of 500 µM (37 mg/L) in 50 mL of CAP RO brine or eight different background electrolyte solutions. The temperature was controlled at 20 °C using a circulating water bath. Currents and electrode potentials were controlled and monitored using a Gamry Series G 750 potentiostat/galvanostat. An 1.1 cmdiameter BDD film on a p-silicon substrate was used as the working electrode (Adamant Technologies, Neutchatel, Switzerland). The working electrode was mounted in a Princeton Applied Research (PAR) model 316 rotating disk electrode (RDE) assembly and rotated at 3000 rpm to eliminate mass transfer limitations on measured reaction rates. The counter electrode was a 12 cm-long by 0.3 mm-diameter platinum wire wrapped in a Nafion membrane (Fuel Cell Scientific, Stoneham, MA) to isolate anodic and cathodic reactions. The reference electrode was a PAR Hg/Hg2SO4 saturated with K2SO4. Potentials are reported versus the standard hydrogen electrode (SHE). To prevent the accumulation of adsorbed organic compounds on the electrode surface, the BDD electrode was preconditioned in a 1 M HClO4 solution at a current density of 10 mA/cm2 for 10 min before each experiment. Cyclic Voltammetry. Cyclic voltammetry (CV) experiments were conducted using the same experimental setup as described in the RDE experiments, except the electrode was stationary. The potential was swept from the open circuit potential to 4.14 V/SHE and back to the open circuit potential at a scan rate of 100 mV/s. Different electrolytes (NaClO4, NaCl, NaHCO3, NaClO3) at a concentration of 100 mM were tested to determine their relative stability to direct oxidation. Flow-through Reactor. The flow-through reactor contained one bipolar and two monopolar BDD films on p-silicon substrates that were 5 cm-long and 2.5 cm-wide (MiniDiaCell, Adamant Technologies). The bipolar electrode was situated between the two monopolar electrodes with an interelectrode gap of 3 mm on both sides. The monopolar electrodes were connected to a Protek (Stayton, OR) model 3005B galvanostatic power supply operated without a reference electrode. The total anode and cathode surface areas were 25 cm2 each, and the solution volume in the reactor was 15 mL, yielding a specific surface area of 1.67 cm-1. The flow-through reactor was operated as a closed-loop at a temperature of 20 °C and a current density of 2 mA/cm2. A 0.25 L solution of 50 µM (3.7 mg/L) NDMA in either CAP water or 75 mM LiClO4 electrolyte was circulated at a rate of 100 mL/min using a peristaltic pump. The reservoir was

connected to a pressure gauge containing a 0.3 mm-diameter platinum wire mesh (Aesar, Ward Hill, MA) to catalyze the reaction between H2 and O2 produced from water electrolysis. Between experiments, the electrode polarity was reversed and a 1 L solution of 100 mM NaClO4 was flushed through the reactor while maintaining a current density of 10 mA/ cm2, to dissolve any precipitate on the cathode. To account for the time that the fluid spent in the reactor, the elapsed electrolysis times (te) were calculated from: te ) t

( ) VR VL

(1)

where t is the elapsed time and VR and VL are the reactor and solution volumes, respectively. Analytical Methods. Concentrations of NO3-, NDMA, and N-dimethylnitramine were determined using high pressure liquid chromatography (HP series II 1090) with ultraviolet absorption detection at a wavelength of 226 nm. Anion and cation concentrations were determined by ion chromatography (Dionex ICS-3000 and Dionex DX 500, respectively). DOC measurements were made using a Shimadzu model VCSH total organic carbon analyzer. Details of the analytical methods are provided in the Supporting Information. Kinetic Modeling. The kinetics of NDMA oxidation by BDD electrodes was represented by the following expression: d[NDMA] 1 ) -kDET - kNDMA,OH•[NDMA][OH•] dt A

[]

(2)

where A is the specific surface area of the electrode (m2/L); kDET is the surface area normalized zero-order rate constant for direct electron transfer of NDMA at the BDD surface; and kNDMA,OH• is the surface area normalized second-order rate constant for reaction between NDMA and hydroxyl radicals. At high overpotentials hydroxyl radicals reach a steady state concentration (24) and kDET , kNDMA,OH•[NDMA] (see Supporting Information for details). Equation 2 can thus be simplified as follows: d[NDMA] 1 ) -kobs[NDMA] dt A

[]

(3)

where kobs is the surface area normalized observed first-order rate constant for NDMA removal. Equation 3 can also be used to represent the kinetics of NDMA reduction. More detail of kinetic model derivation is provided in the Supporting Information. Density Functional Theory Modeling. Density functional theory (DFT) simulations were performed to investigate possible NDMA reaction mechanisms with ClO3•. All DFT calculations were performed using the DMol3 (25, 26) package in the Accelrys Materials Studio (27) modeling suite. Unrestricted spin, all-electron calculations were performed using double-numeric with polarization (DNP) basis sets (28) and the gradient corrected Perdew-Burke-Ernzerhof (PBE) functional for exchange and correlation (29). Implicit solvation was incorporated using the COSMO-ibs (30) polarized continuum model. Transition state searches were performed using a quadratic synchronous transit (QST) method (31) and refined using an eigenvector following method (32). The energy optimized structures and transition states were verified by frequency calculations.

Results and Discussion RDE Experiments. The kobs values for NDMA oxidation and reduction by BDD-film electrodes are shown as a function of current density in Figure 1. Individual plots of concentration vs time profiles are provided in the Supporting Information, Figures S-1 and S-2. Under the conditions investigated, both oxidation and reduction rate constants were a VOL. 44, NO. 11, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4265

TABLE 2. Surface Area Normalized First-Order and Zero-Order Rate Constants for NDMA Oxidation at 5 mA/cm2 Measured in Batch Experiments with Different Electrolytesa

FIGURE 1. Surface area normalized observed first-order rate constants for batch experiments measuring NDMA oxidation and reduction as a function of current density. Error bars represent 95% confidence intervals calculated from regression of duplicate experiments. Initial NDMA ) 500 µM; Electrolyte )100 mM NaClO4; T ) 20 °C; 1 cm2 BDD electrode. linear function of current density. NDMA oxidation rate constants were 3 to 16 times faster than NDMA reduction rate constants at similar current densities. The products of NDMA oxidation were consistent with previous work (15). Terminal NDMA oxidation products were CO2, NO3-, and NH4+, with intermediates of dimethylamine (DMA) and methylamine (MA). Terminal NDMA reduction products were DMA and NH4+ (Supporting Information, Figure S-3). The reduction of NDMA occurred by cleavage of the N-N bond, producing stoichiometric concentrations of DMA, but NH4+ production accounted for only 26 ( 2% of the degraded NDMA. Volatile products were not quantified in the batch experiments because the RDE shaft prevented a gastight seal on the reactor, but prior electrochemical studies have shown N2 and N2O as reduction products (33). The pH of the experiments was maintained well below the pKa values of 10.7, 10.6, and 9.2 for DMA, MA, and NH4+, respectively, to prevent volatilization of these compounds. More details on the variation of pH for individual experiments are provided in the Supporting Information, Figures S-4 and S-5. The reduction of NDMA on BDD cathodes can be affected by the precipitation of Ca2+ and Mg2+ minerals (e.g., CaCO3, CaMg(CO3)2, Mg(OH)2) on the cathode surface. The precipitation of these minerals is facilitated by the high HCO3concentration and the production of OH- via water reduction on the cathode. Supporting Information, Figure S-6 compares NDMA concentrations versus electrolysis times in the CAP RO concentrate to those in a NaClO4 control electrolyte at a cathodic current density of 5 mA/cm2. The fraction of NDMA reduced in the RO concentrate after 120 min of reaction (10%) was a factor of 5 less than that in the NaClO4 electrolyte (51%). To confirm that lower removal was caused by mineral precipitation, an additional experiment was performed containing Ca2+, Mg2+, and HCO3- concentrations similar to that of the RO concentrate (Table 1). As shown in Supporting Information, Figure S-6, after 120 min of reaction 10% of the initial NDMA was removed in the Ca2+/Mg2+/HCO3- electrolyte solution, which was similar to that in the RO concentrate. These results suggest that scale formation on the cathode would significantly limit target compound removal by cathodic reactions in RO concentrates. The effect of water composition on oxidation rate constants at a current density of 5 mA/cm2 is shown in Table 2. The value for kobs in the CAP RO concentrate was not significantly different to that in the 100 mM NaClO4 control experiment. This result indicates that dissolved species in the RO concentrate and the bulk pH of the electrolyte did not interfere with NDMA oxidation. The pH in the NaClO4 electrolyte dropped to 2.5 whereas the CAP RO concentrate was stable at 8.5 (Supporting Information, Figure S-4). Additional NDMA oxidation experiments were performed in NaCl and NaHCO3 electrolytes, as Cl- and HCO3- have been 4266

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 11, 2010

electrolyteb

kobs (L/m2 hr)

R2

100 mM NaClO4 CAP water 5 mM NaHCO3 5 mM NaCl 100 mM NaClO3 100 mM NaHCO3 100 mM NaCl

510 ( 28 487 ( 19 524 ( 20 498 ( 22 600 ( 51

0.991 0.995 0.996 0.994 0.983

kDET (mmol/h m2)

R2

76.2 ( 4.5 29.2 ( 4.7

0.989 0.925

a Confidence intervals (95%) were calculated from regression of duplicate experiments, and R2 values represent the coefficient of correlation. Initial NDMA ) 500 µM; T ) 20 °C; 1 cm2 BDD electrode. b Electrolyte ionic strength kept constant by addition of NaClO4

FIGURE 2. Cyclic voltammetry scans in the presence of different electrolytes, each at a concentration of 100 mM. Scan rate )100 mV/s; T ) 20 °C; 1 cm2 BDD electrode. shown to affect the performance of other AOPs (3, 9, 11, 12, 34). As shown in Table 2, 5 mM concentrations of Cl- and HCO3had no effect on kobs relative to the control. However, 100 mM HCO3- and Cl- concentrations significantly slowed the NDMA oxidation rate and changed the reaction kinetics compared to the control, which is discussed in detail later. The lower rate of NDMA oxidation in the presence of 100 mM HCO3- can be attributed to hydroxyl radical scavenging. A second-order rate constant of kHCO3-,OH• ) 8.5 × 106 M-1 s-1 has been reported (35). CV scans in the 100 mM NaHCO3 electrolyte produced lower currents than the 100 mM NaClO4 control (Figure 2), indicating that HCO3- does not undergo significant direct electron transfer at the anode surface and that it also blocks the oxidation of water relative to the control. Thus, in addition to hydroxyl radical scavenging, HCO3- may decrease rates of NDMA oxidation by decreasing the rate of hydroxyl radical production via water oxidation. The fact that 5 mM HCO3- had no effect on kobs is contrary to studies using other AOPs. For example, in a study using an ozone/H2O2 AOP to generate hydroxyl radicals, NDMA oxidation rates declined in the presence of HCO3- at concentrations greater than 1 mM (9). NDMA oxidation by hydroxyl radicals within the diffuse layer adjacent to BDD anodes may be less affected by HCO3- as a hydroxyl radical scavenger because the diffuse layer is highly acidic because of H+ generation via water oxidation. The low pH in this region would convert HCO3- to H2CO3, which does not readily react with hydroxyl radicals (35). However, at high enough bulk HCO3- concentrations the flux of HCO3- molecules to the anode surface is greater than the production of H+, and thus hydroxyl radical scavenging by HCO3- could occur. Using the Levich equation (36) the estimated flux of HCO3- to the electrode surface becomes larger than the H+ production rate at HCO3- concentrations >16 mM; using a 3000 rpm

FIGURE 3. Concentration vs time profiles for duplicate batch experiments measuring NDMA oxidation at 5 mA/cm2 in the presence of (a) 100 mM NaHCO3 and (b) 100 mM NaCl. T ) 20 °C; 1 cm2 BDD electrode. Concentration was normalized by initial NDMA concentration (Co) on an N-atom basis. Solid line in (b) represents second order model fit. electrode rotation speed, a current density of 5 mA/cm2, and assuming all current was expended on water oxidation. In the 100 mM NaHCO3 electrolyte solution the flux of HCO3to the electrode surface (3.23 × 10-7 mol/cm2 s) is an order of magnitude greater than the H+ production rate (5.18 × 10-8 mol/cm2 s). Further details are provided in the Supporting Information. Not only was the rate of NDMA removal affected, but the product distribution also changed in the presence of 100 mM HCO3-, as compared to the 100 mM NaClO4 control. As shown in Figure 3a, oxidation of NDMA in 100 mM HCO3yielded near stoichiometric production of N-dimethylnitramine (DMNA), with trace levels of NO3-, NH4+, and MA (