Photocatalytic Degradation of N-Nitrosodimethylamine - American

Jul 23, 2005 - The photocatalytic degradation (PCD) reaction of N-ni- trosodimethylamine (NDMA) in water was investigated using pure and surface-modif...
0 downloads 0 Views 268KB Size
Download PDF
Environ. Sci. Technol. 2005, 39, 6800-6807

Photocatalytic Degradation of N-Nitrosodimethylamine: Mechanism, Product Distribution, and TiO2 Surface Modification JAESANG LEE AND WONYONG CHOI* School of Environmental Science and Engineering, Pohang University of Science and Technology, Pohang 790-784, Korea JEYONG YOON School of Chemical Engineering, Seoul National University. Shinlim-dong, Gwanak-gu, Seoul 151-742, Korea

The photocatalytic degradation (PCD) reaction of N-nitrosodimethylamine (NDMA) in water was investigated using pure and surface-modified TiO2. The PCD products of NDMA were methylamine (MA), dimethylamine (DMA), nitrite, nitrate, and ammonium, and their distribution could be changed by modifying the surface of TiO2. The PCD reaction of NDMA seems to be initiated mostly by OH radicals, not valence band holes, because the addition of excess oxalates (hole scavengers) only moderately retarded the PCD rate. The presence of oxalate, however, enabled a new reductive transformation path in which the CO2-• radicals generated from the oxalate converted NDMA into DMA. In acidic suspensions of pure TiO2, the formation of MA was highly favored over DMA and NH3, whereas all degradation products (MA, DMA, and NH3) were generated at comparable concentrations at basic pH. It is suggested that there are three parallel paths depending on the position of the initial attack of OH radical on NDMA and the product distribution is closely related with which path is favored under a specific condition. DMA production is related to the OH radical attack on the nitrosyl nitrogen. Platinum deposition, silica loading, Nafion coating, and surface fluorination were tested to investigate the effects of TiO2 surface modification on the product distribution. The surface platinization of TiO2 had little effect on the PCD reaction of NDMA under air-equilibrated conditions but accelerated the PCD reaction under deaerated conditions. An enhanced PCD reaction of NDMA was achieved with the silicaloaded TiO2 and Nafion-coated TiO2, both of which favored the formation of DMA over MA. The PCD of NDMA on surface-fluorinated TiO2 was also highly enhanced but favored the formation of MA over the formation of DMA.

Introduction N-Nitrosodimethylamine (NDMA) has no commercial use and no industrial production and is generated from the in situ reaction of dimethylamine (DMA) with monochloroamine in the disinfection process or the nitrosation of DMA by nitrite (1-3). Recent advances in the development of analytical techniques that enabled its detection at concen* Corresponding author e-mail: [email protected]; phone: +82-54-279-2283; fax: +82-54-279-8299. 6800

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 17, 2005

trations as low as 1 ng/L (4, 5) brought the issue of NDMA contamination and its regulation in drinking water to the front (6). NDMA is very soluble in water and is rapidly transported and distributed through aquatic environments. NDMA is a probable human carcinogen (1-3, 7, 8). NDMA is known to be very stable against the biological treatment that has been widely used for controlling Ncontaining aquatic pollutants (9) and cannot be removed efficiently through the ozonation alone or ozonation combined with UV (10). Recently, direct UV photolysis (9-11), reduction by zerovalent granular iron (12, 13), and Fenton oxidation (14, 15) were introduced as efficient destruction methods. TiO2-mediated photocatalytic degradation (PCD) is known to mineralize a wide range of organic pollutants with faster rates than conventional treatments (16-19). The remedial power of the TiO2 photocatalyst is mainly ascribed to the nonselective reactivities of hydroxyl radicals generated on the UV-illuminated TiO2 surface. Compared with other UVbased treatment processes that are based on UV-C (primarily 254 nm irradiation) utilization, TiO2 photocatalysis has an obvious merit of using lower-energy photons (UV-A) and even solar light. Unlike most advanced oxidation processes (AOPs) that are based on homogeneous radical chemistry, the heterogeneous photocatalytic reactions strongly depend on the properties of TiO2, such as surface conditions, lattice defects, crystallinity, and particle size. It is now well recognized that even pure TiO2 samples exhibit widely varying photocatalytic reactivities depending on the method of preparation or the commercial provider. In particular, surface properties are critical and the surface modification of TiO2 not only changes the reaction rate (20, 21) but also the mechanism and products (22, 23). Therefore, it is possible to design a modified photocatalyst that is optimized for the degradation of a specific pollutant. For example, ionic pollutants such as tetramethylammonium ((CH3)4N+) and bromate (BrO3-) can be degraded with a much faster rate on surface charge-modifed TiO2 because of the enhanced electrostatic attraction (24, 25). Surface fluorination of TiO2, on the other hand, highly accelerates the OH radical-mediated PCD reactions by enhancing the generation of mobile free OH radicals (26, 27). In this work, we investigated the PCD reaction of NDMA using pure and modified TiO2, which includes surfaceplatinized TiO2 (Pt/TiO2), silica-loaded TiO2 (SiO2/TiO2), Nafion-coated TiO2 (Nf-TiO2), and surface-fluorinated TiO2 (F-TiO2). The effects of the surface modification of TiO2 on the PCD kinetics, mechanism, and product distribution were discussed.

Experimental Section Chemical and Materials. Chemicals that were used as received in this study include the following: (CH3)2NNO (NDMA, Aldrich), CH3NH3Cl (Sigma), (CH3)2NH2Cl (Sigma), NH4Cl (Aldrich), NaNO2 (Aldrich), NaNO3 (Aldrich), NaF (Samchun), (COOH)2 (Aldrich), Si(OC2H5)4 (Aldrich), and H2PtCl6 (Aldrich). Nafion [-(CF2CF2)n-(CFYCF2)-, Y ) -OCF2CF(CF3)OCF2CF2SO3-M+, 5 wt % solution in a mixture of lower aliphatic alcohol and water] was purchased from Aldrich. Nitrogen gas (> 99% purity, BOC Gas) was used for deaerating the aqueous TiO2 suspensions. The water used was ultrapure (18 MΩcm) and was prepared by a Barnstead purification system. Titanium dioxide (Degussa P25), a mixture of 80% anatase and 20% rutile, was used as a base photocatalyst whose surface was modified in several ways. The B.E.T. surface areas of P25 and other surface modified 10.1021/es0481777 CCC: $30.25

 2005 American Chemical Society Published on Web 07/23/2005

TiO2 were all around 50 m2/g. Other chemicals used were of the highest purity available. Preparation of Modified Photocatalysts. 1. SurfacePlatinzation. Platinization of TiO2 was carried out following a photodeposition method. An aqueous suspension of TiO2 (0.5 g/L) was irradiated with a 200 W mercury lamp for 30 min in the presence of 1 M methanol (electron donor) and 0.1 mM chloroplatinic acid (H2PtCl6). After the suspension was irradiated, the Pt/TiO2 powder was collected by filtration and washed with distilled water. A typical Pt loading on TiO2 was estimated to be ca. 3 wt % (23). A transmission electron micrographic image of Pt/TiO2 showed that the Pt particles were in the size range of 2-5 nm and well dispersed on the TiO2 particles (20-30 nm). 2. Silica Loading. SiO2/TiO2 was prepared using a method proposed by Vohra et al. (24). An aliquot of Si(OC2H5)4 (10100 µL) was added directly to 0.1 g of TiO2 powder (P25) with physical mixing to ensure uniform coating on the TiO2 surface. After the mixture was air dried overnight, the sample was calcined at various temperatures (400-700 °C) for 5 h. The sample calcined at 500 °C exhibited the highest photoactivity for NDMA degradation and hence all SiO2/TiO2 samples used in this study were prepared at 500 °C. The surface atomic compositions of the SiO2/TiO2 catalysts were determined by X-ray photoelectron spectroscopy (XPS, Kratos XSAM 800pci) using Mg KR lines (1253.6 eV) as an excitation source. The spectra were taken for each sample after Ar+ (3 keV) sputter cleaning. The binding energies of all peaks were referenced to the Ti 2p line (458.8 eV) in TiO2. 3. Nafion Coating and Surface Fluorination. The Nafioncoated TiO2 catalyst was prepared as described elsewhere (28). A Nafion solution of 100 µL was added to 0.1 g TiO2 powder; it was then mixed thoroughly and dried overnight at room temperature. The ζ potentials of suspended TiO2 paricles were measured as a function of pH using an electrophoretic light scattering spectrophotometer (ELS 8000, Otsuka). F-TiO2 could be prepared by a simple ligand exchange between fluoride anions and surface hydroxyl groups (reaction 1) (26, 27).

tTi-OH + F- T tTi-F + OH-

(1)

NaF (5 mM) was added to an aqueous suspension of pure TiO2 (0.5 g/L) and then the pH was adjusted to 4 to achieve surface fluorination. A detailed description of the chemistry and photochemistry of F-TiO2 can be found in our recent study (27). Photolysis and Analysis. All pure or modified TiO2 suspensions were prepared at a concentration of 0.5 g/L and were dispersed by simultaneous sonication and shaking for 30 s in an ultrasonic cleaning bath. An aliquot of NDMA stock solution (60 mM) was added to the suspension to make a desired concentration (typically 0.25 mM), and then the initial pH of the suspension was adjusted to a desired value with a 1 M HClO4 or NaOH standard solution. The suspensions were unbuffered. All PCD experiments were carried out using a 300 W Xearc lamp (Oriel) under air-equilibrated or deaerated conditions. For anoxic PCD experiments, the photoirradiation started right after the N2 was purged for 30 min. The dissolved O2 concentration in the suspension was measured using a dissolved-oxygen meter (model 850, Orion). The initial O2 concentration in the N2-purged suspension was typically less than 10 µM and was maintained at this level in a closed reactor throughout the photolysis. No loss of NDMA from volatilization was observed, which is consistent with a previous report (9). Light passed through a 10 cm IR water filter and a UV cutoff filter (λ > 300 nm). The filtered light was focused onto a 60 mL Pyrex reactor with a quartz window. The photocatalytic reactor was filled with minimized head-

space and stirred magnetically. Sample aliquots of 1 mL were withdrawn from the illuminated reactor with a syringe, filtered through a 0.45 µm PTFE filter (Millipore), and injected into a 4 mL glass vial. Multiple photolysis experiments were carried out for a given condition. Quantitative analyses of ionic intermediates and products were performed using an ion chromatograph (IC, Dionex DX-120). The IC system was equipped with a Dionex IonPac AS-14 for anions, a Dionex IonPac CS-12A for cations, and a conductivity detector. The analysis of NDMA was done by using a high performance liquid chromatograph (HPLC Agilent 1100) equipped with a C-18 column (Agilent Zorbax 300SB) and a diode-array detector.

Results and Discussion Photocatalytic Degradation of NDMA on Pure and Platinized TiO2. The PCDs of NDMA are summarized in Table 1 when pure TiO2 or Pt/TiO2 was used as a photocatalyst at acidic or basic pH. The two pH conditions represent opposite surface charges for TiO2: positive at pH 4.0 and negative at pH 10.5. The direct photolysis of NDMA in the absence of TiO2 (pH 4, air-equilibrated) converted about 25% of initial NDMA to DMA and NO2- in 4 h (Table 2) because of its weak absorption centered at 332 nm (n f π* transition band,  ) 109 M-1 cm-1) (9, 10). In deaerated solution, the photolysis rate was slightly reduced. The direct photolysis of NDMA at basic pH was negligibly slow. However, the direct photolytic reaction should not contribute to the NDMA degradation observed in the TiO2 suspension because the weak absorption at 332 nm by NDMA should be almost completely blocked by TiO2 particles (through absorption or scattering). From the PCD of NDMA at acidic pH under air equilibration, MA and NO3- evolved as major products. The fact that [MA] was much higher than [DMA] throughout the PCD process implies that MA was not generated from the demethylation of DMA. If the NDMA f DMA conversion were much slower than the DMA f MA conversion, a similar behavior might be observed. However, the rates of the first and second steps are found be comparable (refer to Tables 1 and 2 in ref 23). That is, there should be two parallel paths leading to MA and DMA, respectively. When the PCD reactivity between acidic and basic conditions is compared, NDMA degradation at basic pH was found to proceed at a slower rate and the product distribution was very different than that obtained at acidic pH. The degradation products of NDMA (i.e., DMA, MA, and NH3) were all generated at comparable concentrations at basic pH, whereas MA was the predominant demethylated product and NH3 was absent at acidic pH. The retarded PCD rate of NDMA at basic pH is ascribed to the competition between NDMA and MA/DMA for OH radicals [e.g., compare k(NDMA + •OH) ) 4.3 × 108 M-1 s-1 (29) and k(MA + •OH) ) 1.8 × 1010 M-1 s-1 (30)]. Fewer OH radicals are available to the NDMA + •OH reaction once significant concentrations of DMA and MA are generated. However, the protonated forms of MA/DMA that are dominant at acidic pH have lower reactivity with OH radicals [e.g., k(CH3NH3+ + •OH) ) 3.5 × 107 (31) vs k(CH3NH2 + •OH) ) 1.8 × 1010 M-1 s-1 (30)] and hence interfere less in the NDMA + •OH reaction. The fact that ammonia production was completely absent at acidic pH indicates that the MA demethylation that should be initiated by the MA + •OH reaction is negligible as long as NDMA and DMA are present. The total N balance was satisfactory in most cases, but the Pt-TiO2/N2 cases show a greater than 20% N mass deficit under both acidic and basic conditions, which implies the presence of missing products. There are two kinds of N atoms in NDMA (i.e., amine N bonded to methyl group vs N in nitrosyl group). The breakage of the N-N bond releases an NO group, which is subsequently converted into NO2- and VOL. 39, NO. 17, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6801

TABLE 1. Temporal Course of the Intermediates and Products Concentration (µM) Distribution from the PCD of NDMA (C0 ) 250 µM) in a TiO2 or Pt/TiO2 Suspension pH

catalyst/gas

time (h)

(CH3)2NNO

(CH3)2NH/(CH3)2NH2+

CH3NH2/CH3NH3+

NH3/NH4+

NO2-

NO3-

Ra

total N (%)

acidic pHi 4.0

TiO2/air (pHf 3.7)

0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4

276 211 178 124 92 263 246 231 223 213 272 199 142 91 56 247 173 146 127 111 255 208 173 143 126 247 236 224 212 198 245 196 163 141 118 261 198 170 147 123

0 16 18 28 25 0 4 7 13 20 0 17 23 26 29 0 28 26 40 41 0 7 8 9 10 0 28 34 38 34 0 15 15 24 28 0 19 19 22 20

0 45 89 144 167 0 2 6 9 13 0 59 116 166 202 0 70 94 105 109 0 15 14 19 26 0 8 13 15 16 0 3 8 24 37 0 12 14 17 18

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 2 2 1 0 27 35 43 54 0 4 5 10 12 0 4 9 8 9 0 28 46 65 74

0 4 5 5 6 0 2 3 7 10 0 5 4 8 6 0 0 2 4 5 0 28 48 60 63 0 3 5 5 7 0 31 49 56 53 0 5 12 13 17

0 45 96 165 187 0 1 2 2 3 0 50 111 146 222 0 6 8 14 18 0 24 33 49 68 0 2 4 6 12 0 26 37 53 77 0 8 15 16 19

0 0.77 1.03 1.12 1.05 0 0.18 0.16 0.23 0.26 0 0.74 0.88 0.85 1.05 0 0.08 0.1 0.15 0.17 0 1.12 0.99 0.98 1.02 0 0.45 0.39 0.31 0.39 0 1.16 1.05 1.05 1.02 0 0.13 0.19 0.17 0.19

100 96 102 107 103 100 95 91 91 90 100 97 99 97 105 100 91 86 85 80 100 101 95 91 93 100 104 103 101 97 100 96 91 91 90 100 90 85 82 75

TiO2/N2

Pt-TiO2/air

Pt-TiO2/N2

basic pHi 10.5

TiO2/air (pHf 9.8)

TiO2/N2

Pt-TiO2/air

Pt-TiO2/N2

a

R) ([NO2-] + [NO3-])/∆[NDMA].

TABLE 2. Temporal Course of the Intermediates and Products Concentration (µM) Distribution from the Direct Photolytic (λ > 300 nm) Degradation of NDMA (C0 ) 250 µM) condition

time (h)

(CH3)2NNO

(CH3)2NH2+

CH3NH3+

NH4+

NO2-

NO3-

Ra

total N (%)

air-equilibrated pHi 4.0

0 1 2 3 4 0 1 2 3 4

262 239 218 205 192 264 249 236 228 217

0 25 38 48 61 0 16 24 35 47

0 1 2 3 3 0 0 1 1 2

0 0 0 0 0 0 0 0 0 0

0 23 31 52 58 0 3 7 12 19

0 1 1 1 2 0 0 0 1 1

0 1.04 0.72 0.93 0.86 0 0.2 0.25 0.36 0.43

100 101 97 98 97 100 98 95 96 95

N2-saturated pHi 4.0

a

R) ([NO2-] + [NO3-])/∆[NDMA].

NO3-. As a result, the ratio of ([NO2-] + [NO3-])/∆[NDMA] (R in Tables 1 and 2) is close to unity in the air-equilibrated conditions. However, the R is significantly lower than unity in all N2-saturated conditions, which indicates that the NO conversion into NO2-/NO3- needs oxygen as a reagent. In the deaerated condition, the nitrosyl group seems to be released into gaseous NO as reported from the previous studies of direct photolysis of nitrosamines (32, 33). Scheme 1a proposes the reaction pathways for the PCD reaction of NDMA on TiO2. Hydroxyl radicals generated on 6802

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 17, 2005

the UV-illuminated TiO2 surface initiate the PCD reaction of NDMA by attacking one of three positions: methyl group (A), amine nitrogen (B), and nitrosyl nitrogen (C). Path A generates a carbon-centered radical (I) upon H-atom abstraction, which further reacts with O2 generating peroxy and alkoxyl radical intermediates (34-36) to result in a demethylated product, MA. This pathway seems to be responsible for the production of MA as a main PCD product in aerated TiO2 suspensions. However, MA was generated in the deareated suspension as well, though the concentration

SCHEME 1. (a) Proposed Pathways for the PCD of NDMA on TiO2 and (b) Initiation of PCD of NDMA on SiO2/TiO2 or Nf-TiO2 With an Enhanced Proton Concentration in the Interface Region

was much reduced, and this implies the presence of an alternative path that leads to MA. Path B, the reaction of the OH radical with the lone pair electron on the amine N (generating radical II) enables the anoxic demethylation of NDMA to MA. Cationic radical II subsequently decomposes into a NO radical and an iminium ion, CH3NH+dCH2, which hydrolyzes to produce MA and formaldehyde. The production of formate as the oxidized product of formaldehyde was actually detected by IC. Further demethylation of MA produces NH3. Radical II may transform into radical I (37), which makes path A and B interchangeable when oxygen is available. In the presence of O2, path A appears to be favored over path B because the lone pair electron density on the amine nitrogen in NDMA should be much lower than that of other alkylamines. The N-N bond in NDMA, linking the electrondonating amine group to the electron-accepting nitrosyl group, has been reported to have partial double bond character through the resonance of eq 2 (38, 39).

The ground-state electronic configuration of N-nitrosamines (R2NNdO) has been calculated to possess a 48% contribution from polar resonance structure b (38). Polar resonance structure b carries a positive charge on the amine nitrogen (38), which makes the amine N less reactive toward electrophilic OH radicals. Because of this resonance character of the NDMA structure (i.e., the lack of the lone pair electron density on the amine N), the PCD behavior of NDMA is quite different from that of DMA. The most prominent difference between the PCDs of NDMA and DMA is the effect of the

platinization of TiO2. The PCD of neutral DMA in Pt/TiO2/N2 was much faster than that in Pt/TiO2/air, and the PCD product distribution was also very different between the Pt/TiO2/N2 and Pt/TiO2/air systems (23). However the PCD reaction of NDMA in Pt/TiO2/N2 was not significantly different from that in Pt/TiO2/air except for the production of NO2-/NO3-. On the other hand, it is noted that the concentration of demethylated products (MA + NH3) in the Pt/TiO2/N2 suspension was markedly enhanced from that in the TiO2/ N2 suspension. It seems that the catalytic role of Pt stabilizes the transient radical species (II) and consequently accelerates the anoxic demethylation path (B). In a related study (23), we also observed that the anoxic PCD reaction of neutral alkylamines was highly accelerated on Pt/TiO2 because the Pt deposits stabilize transient intermediate radicals such as radical II. However, path A, which needs dioxygen as a reagent, seems to be little affected by the platinization because the PCD reactivity of Pt/TiO2 was similar to pure TiO2 in the aerated condition. This is also consistent with our recent observation that the PCD reactivities for a series of methylated amines were similar between Pt/TiO2 and pure TiO2 as long as dissolved O2 was present (23). It is suggested that DMA is generated in a different manner, path C, in which the OH radical addition to the nitrosyl nitrogen leads to the production of DMA. A γ-radiolytic study suggested that the addition of the OH radical to nitrosyl nitrogen could generate the dimethylaminyl radical ((CH3)2N•) (40). The NDMA-OH adduct (radical III) decomposes into nitrite and the dimethylaminyl radical which subsequently transforms to DMA through a reaction with e-/H+ or an H-atom abstraction from NDMA (10, 40). Since path C does not involve dioxygen as a reagent, the production of DMA in the anoxic condition was comparable to or slightly higher than in the aerated condition. VOL. 39, NO. 17, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6803

TABLE 3. Temporal Course of the Intermediates and Products Concentration (µM) Distribution from the PCD of NDMA in a TiO2 Suspension in the Presence of Excess Oxalate or tert-butyl Alcohola catalyst/additive TiO2

TiO2/oxalate

TiO2/tert-butyl alcohol

a

time (h)

(CH3)2NNO

(CH3)2NH2+

CH3NH3+

NH4+

NO2-

NO3-

total N (%)

0 1 2 3 4 0 1 2 3 4 0 1 2 3 4

276 211 178 124 92 262 222 187 156 121 259 240 231 219 207

0 16 18 28 25 0 32 45 61 84 0 4 6 8 10

0 45 89 144 167 0 31 65 69 76 0 8 19 27 33

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 4 5 5 6 0 28 30 32 26 0 0 0 0 0

0 45 96 165 187 0 10 30 56 103 0 17 18 30 38

100 96 102 107 103 100 104 104 101 103 100 98 97 97 96

C0 ) 250 µM, [oxalate]0 ) [tert-butyl alcohol]0 ) 2.5 mM, pHi ) 4.0, air-equilibrated.

Roles of OH Radicals, VB Holes, and CO2 Radicals in the PCD of NDMA. Path B might also be initiated by valence band (VB) holes. To assess the possible role of VB holes in NDMA degradation, the PCD experiment was carried out in the presence of excess oxalates as an efficient hole scavenger (41). Table 3 compares the PCD of NDMA and the concurrent product generation in a pure TiO2 suspension with or without oxalate. Although most VB holes should be scavenged by oxalates, the PCD rate of NDMA was only moderately retarded. However, the addition of excess tert-butyl alcohol as a free OH radical scavenger (19) significantly inhibited the PCD of NDMA, which also indicates that the major oxidants are OH radicals, not VB holes. On the other hand, it should be noted that the production of DMA was enhanced in the presence of oxalate (Table 3). This implies that another new path that leads to DMA formation exists in this case. The most probable scenario is that the CO2-• radical generated from the reaction of VB hole with oxalate (41) reacts with NDMA to produce DMA (reactions 3 and 4).

(COO)22- + hvb+ f -OOC-COO• f CO2-• + CO2 (3) (CH3)2NNO + CO2-• + H2O f

(CH3)2NH + CO2 + OH- + •NO (4)

CO2-• is a very strong reductant having the standard reduction potentials, E°(CO2/CO2-•) ) -1.8 V (vs NHE) (42), and it has been reported to reduce Ni2+ and NO2- (43). Since the electroreduction of NDMA is allowed at potentials more negative than -1.3 V (12, 13), the CB electrons of TiO2 cannot directly reduce NDMA. To confirm that DMA production in the presence of oxalate is mediated via a reductive path, the PCD of NDMA with oxalate was carried out in a deaerated TiO2 suspension where the quenching of CO2-• by O2 should be inhibited (44) (Figure 1).

CO2-• + O2 f CO2 + O2-• (k ) 2.4 × 109 M-1s-1) (5) The fact that both the anoxic degradation of NDMA and the production of DMA were higher in the presence of oxalate supports the CO2 radical-mediated pathway (reactions 3 and 4). Effects of Silica Loading or Nafion Coating on TiO2. Figure 2a shows that the distribution of products generated from the PCD of NDMA on SiO2/TiO2 is significantly influenced by the silica content. The PCD rate of NDMA on SiO2/TiO2 (Si atomic 4.6%) was faster by about 30% compared 6804

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 17, 2005

FIGURE 1. PCD of NDMA in a pure TiO2 suspension with or without excess oxalates (hole scavengers) under N2-saturated conditions ([NDMA]0 ) 250 µM, [oxalate]0 ) 2.5 mM, pHi ) 4). with that on pure TiO2. The XPS analysis of the SiO2/TiO2 samples (Figure 2b) shows that increasing the silica content gradually shifted the Si 2p binding energy to 103.35 eV (corresponding to bulk SiO2), which indicates the development of the Si-O-Si networks. The most outstanding trend found in Figure 2a is that the fraction of DMA steadily increases, whereas that of MA decreases with increasing Si content. In a similar manner, [NO3-] decreases but [NO2-] increases with increasing silica load. This indicates that the PCD mechanism is controlled by the surface property of the photocatalyst. In terms of the mechanisms suggested in Scheme 1a, this implies that path C is highly favored at higher surface Si concentrations. The most direct effect of the silica loading on TiO2 is the surface charge and acidity modification (45). Figure 3a compares the change of the ζ potentials of pure TiO2 and the SiO2/TiO2 particles suspended in water as a function of pH. The point of zero ζ potential (PZZP) of TiO2 is about 6.5 while that of SiO2/TiO2 is significantly shifted to lower pH (∼2). Therefore, the surface charge of SiO2/TiO2 is dominantly negative in the region of pH >2. To counterbalance the negative surface charge, protons should be more concentrated at the catalyst/water interface. Scheme 1b illustrates that the proton-NDMA interaction in the interfacial region of SiO2/TiO2 favors the polar structure (eq 2) carrying a partial negative charge on the nitrosyl oxygen. It is known that NDMA forms a polar complex with Lewis acid (or protons) (38). An alternative description of this situation

FIGURE 2. (a) Distribution of products from PCD of NDMA on SiO2/ TiO2 photocatalysts having different Si contents, after 4 h of UV irradiation (C0 ) 250 µM, [SiO2/TiO2] ) 0.5 g/L, pHi ) 4, airequilibrated condition). The pure TiO2 (P25), referred as “0.0%” in the figure, was also calcined at 500 °C like the other SiO2/TiO2 samples to correct any change in the surface condition upon calcining. The initial removal rates of NDMA as a function of the Si content are also shown along with the product distribution. (b) XPS spectra of Si (2p band) on SiO2/TiO2 that was loaded with different Si contents. is that Lewis acid sites on the SiO2 domain interact with NDMA to form a polar surface complex. With this NDMAacid complex (Scheme 1b), the amine nitrogen carries a partial positive charge and should be less reactive toward OH radicals. As a result, path B is less favored, and the contribution from path C (leading to DMA formation) becomes more important in the overall PCD reaction with increasing the Si content in SiO2/TiO2 (Figure 2a and Scheme 1b). The concentration of NDMA in the surface region of SiO2/TiO2 even facilitates the evolution of DMA through enhancing the reaction between the dimethylaminyl radical ((CH3)2N•) and NDMA. From the photolysis of NDMA (10, 11, 33), it was reported that the dialkylaminyl radical (R1R2N•) produced from the direct photolysis of the N-N bond in nitrosoamines can abstract an H atom from the parent nitrosoamines to generate amines (R1R2NH) while it has very low reactivity toward dissolved O2 (10, 33). The surface acidity of TiO2 could be even more enhanced by coating Nafion polymer on TiO2 surface (Nf-TiO2). The sulfonate groups in the Nafion coating induce highly negative surface charges on TiO2 as shown in Figure 3a and enable the accumulation of protons within the Nafion matrix (Scheme 1b). It is well-known that the pH in the Nafion internal channel is much lower than the pH in the aqueous bulk phase (46). Figure 3b compares the PCD of NDMA among

FIGURE 3. (a) ζ potentials of pure TiO2, SiO2/TiO2, and Nf-TiO2 particles in aqueous suspensions ([catalyst]0 ) 2 mg/L) as a function of pH. (b) Comparison of the PCD rates of NDMA on pure TiO2 (P25 as received), SiO2/TiO2 (calcined at 500 °C), and Nf-TiO2 (pHi ) 4). pure TiO2, SiO2/TiO2, and Nf-TiO2 systems and Figure 4a shows the time-dependent generation of products from NDMA degradation in a Nf-TiO2 suspension. Nf-TiO2 induced faster PCD of NDMA than did SiO2/TiO2. Over 90% of the NDMA was removed within 2 h of irradiation, and DMA evolved as a main product, as in the case of SiO2/TiO2. The cases of both SiO2/TiO2 and Nf-TiO2 show that the proton (or Lewis acid site) accumulation in the surface region favors the formation of DMA. Effects of Surface Fluorination of TiO2. When the surface of TiO2 is fluorinated through reaction 1, the PCD reactions of several organic compounds such as phenol (26), 4-chlorophenol (47), tetramethylammonium (48), and acid orange 7 (27) were accelerated. This has been attributed to the fact that the OH radicals generated on F-TiO2 surface are more mobile (reaction 6) than those generated on pure TiO2 (reaction 7).

tTi-F + H2O + hvb+ f tTi-F + •OHfree + H+ (6) tTi-OH + hvb+ f tTi-OH•+

(7)

As a result, substrates that react mainly through an OH radical-mediated pathway are more rapidly degraded in the F-TiO2 suspension. On the other hand, substrates whose degradation is initiated by direct hole transfer are less degraded on F-TiO2 because of the hindered adsorption of substrates on F-TiO2 (27). VOL. 39, NO. 17, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6805

Acknowledgments This work was supported by KOSEF (Grant No. R01-2003000-10053-0) and the Center for Integrated Molecular Systems (CIMS).

Literature Cited

FIGURE 4. PCD of NDMA on (a) Nf-TiO2 and (b) F-TiO2 ([NaF]0 ) 5 mM, pHi ) 4). Figure 4b shows that the PCD reaction of NDMA on F-TiO2 is twice as fast as that on pure TiO2, which implies that the NDMA degradation is initiated by OH radical attack, not VB holes. The product distribution from PCD on F-TiO2 was very different from that obtained with SiO2/TiO2 or Nf-TiO2 (compare Figure 4a versus 4b). MA was the main product in the F-TiO2 suspension, whereas DMA was dominantly produced with SiO2/TiO2 or Nf-TiO2. This implies that path A (H-atom abstraction) is favored with F-TiO2, as in the case of tetramethylammonium (48). It seems that mobile OH radicals are more likely to react with position A rather than with position C. As the production of DMA, a precursor of NDMA formation, is not desired, F-TiO2 should be more suitable as a practical photocatalyst for NDMA degradation than SiO2/TiO2 or Nf-TiO2. The surface modification of TiO2 should affect the PCD of intermediates as well as the parent substrate, NDMA. Because NDMA and its degradation intermediates (MA and DMA) are structurally similar, the effects of surface modification on the PCD of intermediates should not be very different from that of NDMA. Finally, it should be noted that the NDMA concentration used in this study is several orders of magnitude higher than that found in polluted waters (∼100 ng/L). The detailed kinetic and mechanistic behaviors observed in the high concentration region may not be directly extrapolated to low concentration conditions. Further studies employing the environmentally more relevant condition are required to confirm the findings from this study. 6806

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 17, 2005

(1) Mitch, W. A.; Sedlak, D. Formation of N-nitrosodimethylamine (NDMA) from dimethylamine during chlorination. Environ. Sci. Technol. 2002, 36, 588-595. (2) Gerecke, A. C.; Sedlak, D. Precursors of N-nitrosodimethylamine in natural water. Environ. Sci. Technol. 2003, 37, 1331-1336. (3) Choi, J.; Valentine, R. L. Formation of N-nitrosodimethylamine (NDMA) from reaction of monochloramine: a new disinfection byproduct.Wat. Res. 2002, 36, 817-824. (4) Tomkins, B. A.; Griest, W. H.; Higgins, C. E. Determination of N-nitrosodimethylamine at part-per-trillion levels in drinking waters and contaminated groundwaters. Anal. Chem. 1995, 67, 4387-4395. (5) Tomkins, B. A.; Griest, W. H. Determination of N-nitrosodimethylamine at part-per-trillion levels in contaminated groundwaters and drinking waters featuring carbon-based membrane extraction disks. Anal. Chem. 1996, 68, 2533-2540. (6) Fleming, E. C.; Pennington, J. C.; Wachob, B. G.; Howe, R. A.; Hill, D. O. Removal of N-nitrosodimethylamine from waters using physical-chemical techniques. J. Hazard. Mater. 1996, 51, 151-164. (7) Lin, H.-L.; Hollenberg, P. F. N-nitrosodimethylamine-mediated formation of oxidized and methylated DNA bases in cytochrome P450 2E1 expressing cell line. Chem. Res. Toxicol. 2001, 14, 562566. (8) Souliotis, V.; Henneman, J. R.; Reed, C. D.; Chhabra, S. K.; Diwan, B. A.; Anderson, L. M. Kryptopoulos, S. A. DNA adduct and liver DNA replication in rats during chronic exposure to N-nitrosodimethylamine (NDMA) and their relationships to the dosedependence of NDMA hepatocarcinogenesis. Mutat. Res. 2002, 500, 75-87. (9) Sharpless, C. M.; Linden, K. G. Experimental and model comparisons of low-and medium-pressure Hg lamps for the direct and H2O2 assisted UV photodegradation of N-nitrosodimethylamine in simulated drinking water. Environ. Sci. Technol. 2003, 37, 1933-1940. (10) Stefan, M. I.; Bolton, J. R. UV direct photolysis of N-nitrosodimethylamine (NDMA): kinetics and product study. Helv. Chim. Acta 2002, 85, 1416-1426. (11) Fiz, M.; Usero J. L.; Casado, J. Photolysis of N-nitrosoamines in neutral media. Int. J. Chem. Kinet. 1993, 25, 341-351. (12) Gui, L.; Gillham, R. W.; Odziemkowski, M. S. Reduction of N-nitrosodimethylamine with granular iron and nickel-enhanced iron. 1. Pathways and kinetics. Environ. Sci. Technol. 2000, 34, 3489-3494. (13) Odiziemkowski, M. S.; Gui, L.; Gillham R. W. Reduction of N-nitrosodimethylamine with granular iron and nickel-enhanced iron. 2. Mechanistic studies. Environ. Sci. Technol. 2000, 34, 3495-3500. (14) Heur, Y.-H.; Streeter, A. J.; Nims, R. W.; Keefer, L. K. The fenton degradation as nonenzymic model for microsomal denitrosation of N-nitrosodimethylamine. Chem. Res. Toxicol. 1989, 2, 247253. (15) Hiramoto, K.; Ryuno, Y.; Kikugawa, K. Decomposition of N-nitrosoamines, and concomitant release of nitric oxide by Fenton reagent under physiological conditions. Mutat. Res. 2002, 520, 103-111. (16) Hoffmann, M. R.; Martin, S. T.; Choi, W.; Bahnemann, D. W. Environmental applications of semiconductor photocatalysis. Chem. Rev. 1995, 95, 69-96. (17) Photocatalysis-Fundamentals and Applications; Serpone, N., Pelizzetti, E., Eds.; Wiley-Interscience: New York, 1989. (18) Choi, W.; Hong, S. J.; Chang, Y. S.; Cho, Y. Photocatalytic degradation of polychlorinated dibenzo-p-dioxins on TiO2 film under UV or solar light irradiation. Environ. Sci. Technol. 2000, 34, 4810-4815. (19) Kim, S.; Choi, W. Kinetics and mechanisms of photocatalytic degradation of (CH3)nNH4-n+ (0e ne 4) in TiO2 suspensions: The role of OH radicals. Environ. Sci. Technol. 2002, 36, 20192025. (20) Hwang, S.; Lee, M. C.; Choi, W. Highly enhanced photocatalytic oxidation of CO on titania deposited with Pt nanoparticles:

(21) (22) (23)

(24) (25)

(26)

(27) (28) (29) (30) (31) (32) (33) (34) (35)

Kinetics and mechanisms. Appl. Catal. B: Environ. 2003, 46, 49-63. Ryu, J.; Choi, W. Effects of TiO2 surface modifications on photocatalytic oxidation of arsenite: The role of superoxides. Environ. Sci. Technol. 2004, 38, 2928-2933. Lee, J.; Park, H.; Choi, W. Selective photocatalytic oxidation of NH3 to N2 on platinized TiO2. Environ. Sci. Technol. 2002, 36, 3872-3878. Lee, J.; Choi, W. Effect of Pt deposits on TiO2 on the anoxic photocatalytic degradation pathways of alkylamines in water: Dealkylation and N-alkylation. Environ. Sci. Technol. 2004, 38, 4026-4033. Vohra, M. S.; Lee, J.; Choi, W. Enhanced photocatalytic degradation of tetramethylammonium on silica-loaded TiO2. J. Appl. Electrochem. 2005, 35, 757. Noguchi, H.; Nakajima, A.; Watanabe, T.; Hashimoto, K. Design of a photocatalyst for bromate decomposition: Surface modification of TiO2 by pseudo-boehmite. Environ. Sci. Technol. 2003, 37, 153-157. Minero, C.; Mariella, G.; Maurino, V.; Pellizzetti, E. Photocatalytic Transformation of Organic Compounds in the Presence of Inorganic Anions. 1. Hydroxyl-Mediated and Direct ElectronTransfer Reactions of Phenol on a Titanium Dioxide-Fluoride System. Langmuir 2000, 16, 2632-2641. Park, H.; Choi, W. Effects of TiO2 surface fluorination on photocatalytic reactions and photoelectrochemical behaviors. J. Phys. Chem. B 2004, 108, 4086-4093. Vohra, M. S.; Tanaka, K. Enhanced photocatalytic activity of Nafion-coated TiO2. Environ. Sci. Technol. 2001, 35, 411-415. Mezyk, S. P.; Cooper, W. J.; Madden, K. P.; Bartels, D. M. Free radical destruction of N-nitrosodimethylamine in water. Environ. Sci. Technol. 2004, 38, 3161-3167. Getoff, N.; Schwo¨rer, F. Pulsradiolytishe bestimmung von geschwindigheitskonstanten der reacktionen einiger amine mit OH and eaq-. Int. Radiat. Phys. Chem. 1970, 2, 81-89. Getoff, N.; Schwo¨rer, F. Pulsradiolyse von methylamin in wa¨sseriger lo¨sung. Int. Radiat. Phys. Chem. 1971, 3, 429-439. Hiramoto, K.; Ohkawa, T.; Kikugawa, K. Release of nitric oxide together with carbon-centered radicals from N-nitrosamines by ultraviolet light irradiation Free Rad. Res. 2001, 35, 803-813. Geiger, G.; Huber, J. R. Photolysis of dimethylnitrosamine in the gas phase. Helv. Chim. Acta 1981, 64, 989-995. Maillard, B.; Ingold, K. U.; Scaiano, J. C. Rate constants for the reactions of free radicals with oxygen in solution. J. Am. Chem. Soc. 1983, 105, 5095-5099. Alfassi, Z. B.; Mosseri, S.; Neta, P. Halogenated alkylperoxyl radicals as oxidants: effects of solvents and of substituents on rates of electron transfer. J. Phys. Chem. 1987, 91, 3383-3385.

(36) Russell, G. A. Deuterium-isotope effects in the autoxidation of aralkyl hydrocarbons. Mechanism of interaction of peroxyl radicals. J. Am. Chem. Soc. 1957, 79, 3871-3877. (37) Das, S.; Sonntag, C. Z. The oxidation of trimethylamine by OH radicals in aqueous solution, as studied by pulse radiolysis, ESR, and product analysis. The reactions of the alkylamine radical cation, the aminoalkyl radical, and the protonated aminoalkyl radical. Z. Naturforsch. 1986, 41b, 505-513. (38) Chow, Y. L. Nitrosamine photochemistry: Reactions of aminium radicals. Acc. Chem. Res. 1973, 6, 354-360. (39) Looney, C. E.; Philip, W. D.; Reiley, E. L. Nuclear magnetic resonance and infrared study of hindered rotation in nitrosamines. J. Am. Chem. Soc. 1957, 79, 6136-6142. (40) Wink, D. A.; Desrosiers, M. F. Unusual spin-trap chemistry for the reaction of hydroxyl radical with the carcinogen Nnitrosodimethylamine. Radiat. Phys. Chem. 1991, 38, 467-472. (41) Mao, Y.; Scho¨neich, C.; Asmus, K.-D. Identification of organic acids and other intermediates in oxidative degradation of chlorinated ethanes on titania surfaces en route to mineralization: A combined photocatalytic and radiation chemistry study. J. Phys. Chem. 1991, 95, 10080-10089. (42) Perissinotti, L. L.; Brusa, M. A.; Grela, M. A. Yield of carboxyl anion radicals in the photocatalytic degradation of formate over TiO2 particles. Langmuir 2001, 17, 8422-8427. (43) Forouzan, F.; Richards, T. C.; Bard, A. J. Photoinduced reaction at TiO2 particles. Photodeposition from NiII solutions with oxalate. J. Phys. Chem. 1996, 100, 18123-18127. (44) Jeong, J.; Yoon, J. Dual roles of CO2-• for degrading synthetic organic chemicals in the photo/ferrioxalate system. Wat. Res. 2004, 38, 3531-3540. (45) Anderson, C.; Bard, A. J. Improved photocatalytic activity and characterization of mixed TiO2/SiO2 and TiO2/Al2O3 materials. J. Phys. Chem. B 1997, 101, 2611-2616. (46) Sondheimer, S. J.; Bunce, N. J.; Fyfe, C. A. Structure and Chemistry of Nafion-H: A Fluorinated Sulfonic Acid Polymer. J. Macromol. Sci., Polym. Rev. 1986, C26, 353. (47) Kim, S.; Park, H.; Choi, W. Comparative Study of Homogeneous and Heterogeneous Photocatalytic Redox Reactions: PW12O403vs TiO2. J. Phys. Chem. B 2004, 108, 6402-6411. (48) Vohra, M. S.; Kim, S.; Choi, W. Effects of surface fluorination of TiO2 on the photocatalytic degradation of tetramethylammonium. J. Photochem. Photobiol. A 2003, 160, 55-60.

Received for review November 19, 2004. Revised manuscript received June 22, 2005. Accepted June 23, 2005. ES0481777

VOL. 39, NO. 17, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6807