Environ. Sci. Technol. 2003, 37, 1933-1940
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 CHARLES M. SHARPLESS AND KARL G. LINDEN* Department of Civil and Environmental Engineering, Duke University, Box 90287, Durham, North Carolina 27708-0287
Both low- and medium-pressure Hg lamps (LP and MP, respectively) were used as ultraviolet light (UV) sources to destroy N-nitrosodimethylamine in a synthetic “natural” water. The lamp performances were directly compared via the UV fluence-based rate constants, which demonstrates that LP and MP have virtually identical photonic efficiencies (fluence-based rate constants of 2.29E-3 and 2.35E-3 cm2/ mJ, respectively). This indicates that the quantum yield for NDMA photolysis is independent of wavelength in the UVC region: a value of 0.30 mol/einstein is found at pH 8.1. Addition of 100 mg/L of H2O2 leads to a 30% increase in the LP fluence-based rate constant but does not alter the MP rate constant, likely due to the tradeoff between light screening by H2O2 and additional radical based degradation. However, in terms of the time-based rate constant, this level of H2O2 slightly enhances the LP performance but hinders the MP performance, suggesting that H2O2 is of little or no economic benefit for NDMA removal by UV. All these effects are explained by modeling the photochemistry according to standard equations. The model predicts that H2O2 may enhance NDMA removal for short optical path lengths but that light-screening by H2O2 may decrease the removal rates for optical path lengths typical of those found in UV reactors.
Introduction N-Nitrosodimethylamine (NDMA) is a contaminant of emerging concern to the North American water treatment industry. It is a known animal carcinogen that is listed as a potential human carcinogen, and levels as high as 10 ppb have been detected in groundwaters. NDMA does not currently fall under U.S. EPA regulations, but it is listed as a priority pollutant on EPA’s National Priorities List. Furthermore, regulations and advisories are enforced in various states across the United States. In Canada, the Ontario Ministry of Environment has established an interim drinking water standard for NDMA of 0.009 µg/L (9 ppt). * Corresponding author phone: (919)660-5196; fax: (919)660-5219; e-mail:
[email protected]. 10.1021/es025814p CCC: $25.00 Published on Web 03/22/2003
2003 American Chemical Society
FIGURE 1. Molar absorption spectrum of NDMA and H2O2 overlain on emission spectra of LP and MP lamps (units are relative einstein cm-2 s-1); LP (...), MP (- - -). LP and MP spectra not to scale relative to one another. Inset shows MP lamp and an expanded wavelength range to illustrate the weak NDMA absorption above 300 nm. Treatment methods for removing NDMA from water are not well established. As a chemical class, N-nitrosamines are very stable in aqueous solution. They are resistant to biodegradation, air stripping, and, due to their high water solubilities, have generally been regarded as resistant to adsorption on activated carbon (although the recent development of an analytical method for NDMA employing an activated carbon fiber felt as a sorbent suggests that this may need reevaluation (1)). They are, however, photochemically labile, and ultraviolet (UV) based processes have shown great promise for treating NDMA contaminated waters (2-4). NDMA strongly absorbs UV radiation in the 200-275 nm wavelength range (Figure 1); its absorption spectrum displays a strong transition centered at 227 nm and a much weaker band centered at 332 nm. Clearly, any UV source with good output below approximately 270 nm should be useful for photolyzing NDMA. Two common UV sources that are used in water treatment are low and medium pressure Hg arc lamps. The former emits essentially monochromatic light at 254 nm, while the latter has various outputs ranging from about 205 nm to above 500 nm. Recent work has shown that irradiation with the polychromatic UV light from medium-pressure Hg lamps (MP) is a very effective treatment process for NDMA (3-5). In contrast, low-pressure Hg lamps (LP) have not been well studied for this purpose, and no direct comparisons of LP and MP degradation of NDMA exist in the scientific or trade literature. The primary issues that need to be addressed are the relative photonic and electrical efficiencies of the two technologies and whether H2O2 can effectively enhance either or both of these processes. This paper presents the results of an experimental and model comparison of NDMA removal rates from drinking water using LP and MP with and without added H2O2, and the work is presented in two parts. In the experimental phase of these studies the principal goals were first, to determine fundamental parameters for NDMA photolysis such as the quantum yield for removal and the related UV energy-based removal rate constant and second, to assess whether H2O2 enhances the process efficiencies as judged both by the UV VOL. 37, NO. 9, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1933
TABLE 1. Experimental Water Compositiona
mg/L a
Ca2+
Mg2+
Na+
CO32-
NO3-
Cl-
SO42-
NOM
alginic acid
29.1
9.99
36.2
90
0.679 as N
40.0
55.0
2.56
5.32
Values are given in mg/L, and for CO32- and SO42- this refers to the total ion mass.
FIGURE 2. Diagram of the collimated beam setup for experiments. The diagram shows the arrangement for the MP lamp. For the LP experiments, the setup is very similar, except that the housing holds four lamps. fluence and time required to achieve a given amount of NDMA removal. Unexpected differences between how H2O2 affected the experimental performances of LP and MP led to efforts to model how these effects are related to the lamp spectral distributions as well as their dependence upon crucial factors such as H2O2 concentration and optical path length in solution.
Materials and Methods Reagents. All reagents other than NDMA and the natural organic matter used to prepare the experimental water (described below) were analytical grade and were used as received. NDMA was purchased as the neat solvent (SigmaAldrich) and was used as received. Two types of natural organic matter (NOM) were used to prepare the experimental water: Suwannee River natural organic matter (reverse osmosis isolate) was obtained from the International Humic Substances Society; alginic acid was obtained from Dr. Susan Andrews at the University of Waterloo, Ontario, Canada. The experimental water was prepared according to a specific recipe and contained most of the major cations and anions as well as approximately 3 mg/L of dissolved organic carbon (DOC). The water composition is given in Table 1, and its pH was approximately 8.0. Photolysis Experiments. Photolysis experiments were carried out in bench scale UV “collimated beam” instruments. This is a simple experimental geometry in which the lamp(s) are housed in a shuttered box, and the sample is placed in an open dish at some distance below the shutter (Figure 2). For LP, the instrument was made in our laboratory and houses four 15 W germicidal lamps (ozone-free, General Electric #G15T8). For MP, a Calgon Carbon (Pittsburgh, PA) benchtop instrument was used in conjunction with a single 1 kW MP lamp (ozone-free, Hanovia #6806A441). Incident irradiances were determined by chemical actinometry (see below for methods) using identical geometry and solution volumes as in photolysis experiments. The delivered UV fluence (mJ/ cm2) was calculated as the average irradiance (incident 1934
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 9, 2003
irradiance corrected for the water absorbance and path length (6)) multiplied by the exposure time. During the course of the experiments, the solution absorbance decreased; this was continuously monitored and the data were used to determine the average UV fluence delivered to the solution at sampling time points by assuming the average irradiance between two sampling points to be equal to the mean of the irradiances at those points. For MP, the UV fluence was defined as the energy over the 200-300 nm range measured as described below. In a typical photolysis experiment, 120 mL of water was placed in a 70 × 50 mm crystallization dish (area 34.2 cm2, solution depth approximately 3.5 cm, open to the atmosphere) and spiked with NDMA to achieve a starting concentration of 1 µM (74 ppb). This concentration is low enough that NDMA contributes negligibly to the solution absorbance (Figure 6); the fraction of the total absorbance due to NDMA was 0.015 (1.5%) without H2O2 and 0.012 (1.2%) with 100 mg/L of H2O2. In some cases, 100 mg/L of H2O2 (2.94 mM) was also added by spiking an appropriate amount of 30% H2O2 (Ultrex brand, J. T. Baker). Under these conditions, typical average irradiance values (200 to 300 nm) were approximately 2.0 and 2.4 mW cm-2 for MP with and without H2O2, respectively, with the corresponding values for LP being 1.2 and 1.4 mW cm-2. At appropriate times, 0.3 mL samples were removed from the dish, filtered, and stored for an hour or two prior to analysis by HPLC. The sample removed at 0 mJ/cm2 serves as a dark control to ensure that no loss of NDMA is occurring via reactions other than photolysis: the concentration at this fluence was always equal to the spiked concentration. No loss of NDMA to volatilization was observed in unexposed stirred samples, which is consistent with its high water solubility (approximately 5-10 M based on its vapor pressure and Henry’s Law constant (7)) and the results of fugacity modeling (8). Actinometry. Incident irradiance values (mW cm-2) over the 200-300 nm range were measured in separate experiments by chemical actinometry in conjunction with spectral measurements. Chemical actinometry was used to quantify the incident photon irradiance (einstein cm-2 s-1). Two actinometers were used: iodide/iodate actinometry (9) for LP and ferrioxalate actinometry (0.006 M K3Fe(Ox)3) (10) for MP. Actinometric photon irradiances were correlated to radiometer readings (Model IL1700, SED240 detector, International Light, Co., Newburyport, MA), and the radiometer was used to adjust the actinometry results for daily fluctuations in the lamp intensity. With LP, the irradiance (mW cm-2) is simply calculated from the photon irradiance (determined from the rate of I3- production, (9)) and the photon energy at 254 nm. For MP, the irradiance was defined as the energy delivered below 300 nm. The photon irradiance in this range was calculated from the difference between the Fe2+ production rates with and without a 3 mm thick 295 nm long-pass filter (Schott, WG295) placed in the light path. The results of a typical actinometry experiment are given in Table 2; the actinometry data are converted into the photon irradiance below 300 nm using the formula (see Appendix for derivation)
E0p,
