Environ. Sci. Technol. 2007, 41, 3626-3632
Temperature and Wavelength Dependence of Nitrite Photolysis in Frozen and Aqueous Solutions LIANG CHU AND CORT ANASTASIO* Atmosphere Science Program, Department of Land, Air, and Water Resources, University of California, One Shields Avenue, Davis, California 95616-8627
While the photolysis of nitrite is an important source of hydroxyl radical (•OH) in some natural waters, its wavelength and temperature dependence have not been fully described in solution. In addition, there are no studies of this reaction on ice, although there is evidence of nitrite production in snow. To address these gaps, we have measured the wavelength and temperature dependence of the quantum yields of •OH from the photolysis of frozen and aqueous NO2-. From our solution and ice results, we derive a master equation that describes the •OH quantum yield from NO2- photolysis as a function of both temperature (240-295 K) and illumination wavelength (302-390 • nm): Φ(NO2 f OH )T,λ ) (y0 + a/(1 + exp((λ - c)/b)))exp(((eλ + f)/R) × (1/295 - 1/T)) where y0 ) 0.0204 ( 0.0010, a ) 0.0506 ( 0.0022, b ) 11.2 ( 1.2, c ) 332 ( 1, e ) 20.5 ( 3.2, f ) 7553 ( 1204, uncertainties represent 1 standard error, T is the temperature (K), R is the gas constant (8.314 J mol-1 K-1), and λ is the wavelength (nm). Using these results we predict the pseudo-steady-state concentrations of nitrite on sunlit polar snow grains and compare the relative importance of the photolysis of nitrite, nitrate, and hydrogen peroxide as sources of snow-grain •OH.
Introduction The photolysis of gaseous nitrous acid (HONO) can be a major source of hydroxyl radical (•OH), which in turn is a major sink for many atmospheric trace species (1, 2). In the presence of condensed phasessincluding fog and cloud drops, aerosol particles, and probably the organic surface film on buildingssnitrous acid partitions into solution where it can dissociate into nitrite (NO2-):
HONO(g)h HONO(aq)
(1)
+ HONO(aq)h NO2 + H
(2)
The pKa for nitrous acid is 2.8 and this species can be further protonated in more acidic solutions to form H2ONO+, which has a pKa of 1.7 (3). Although the physical Henry’s law constant for HONO is fairly low (49 M atm-1 at 298 K (2)), the solubility of total N(III) (i.e., NO2-, HONO, and H2ONO+) in atmospheric condensed phases is enhanced with increasing pH because of dissociation of HONO (reaction 2) and accumulation of NO2-. Thus while concentrations of nitrite in acidic cloud drops are on the order of 0.01-1 µM, concentrations in fog * Corresponding author phone: (530) 754-6095; fax: (530) 7521552; e-mail:
[email protected]. 3626
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 10, 2007
and dew drops with near-neutral pH values are typically 10 - 100 µM (4-8). Both aqueous nitrous acid and nitrite undergo photolysis to yield •OH and NO (5, 9, 10). In the case of nitrite this reaction proceeds via
•
NO2- + hν f NO + •O-
(3)
O- + H2O f •OH + OH-
(4)
Note that reaction 3 is the rate-limiting step in this sequence and that •O- will be protonated to •OH in essentially all environmental samples since the pKa for •OH is 11.9 (11). It has been suggested that there is an additional channel for NO2- photolysis, to produce NO2 and a hydrated electron, but this pathway is either very minor relative to reaction 3 or does not occur (10). A few studies have shown that the photolysis of nitrite can be the dominant source of •OH in sunlit fog and dew drops and thus it can play an important role in the transformation of aqueous pollutants (5, 6). Similarly, nitrite photolysis can be important source of •OH in some rivers and in some parts of the ocean (12). While many of the transformations initiated by nitrite photolysis likely lead to less toxic products, these reactions can also form toxic organic compounds such as nitrophenols (13). In addition to this chemistry in solution, photochemical reactions of nitrite and nitrous acid also appear to be important in snow. A number of field studies in the Arctic and Antarctic have shown that sunlit snowpacks are a source of both HONO and NOx to the overlying atmosphere (1416). While the emission of NOx appears to be due primarily to photolysis of nitrate on snow grains (17-19), photolysis of nitrite/nitrous acid on snow grains, or in the interstitial air between grains, might also be significant. In part this is because N(III) has a much larger light absorption than nitrate in the UV-A (320 - 400 nm) where actinic fluxes are relatively large compared to the UV-B (280-320 nm) (19, 20). In addition to its possible role in NOx release, nitrite photolysis on snow grains might also be a significant source of •OH (21). Because of the potential importance of nitrite, a number of previous studies have examined its aqueous photochemistry (5, 10, 22, 23). Quantum yields for nitrite photolysis, i.e., the number of molecules of •OH (or NO) formed per photon absorbed by nitrite, are strongly wavelength dependent, decreasing between 300 and 400 nm (10, 22, 23). Unfortunately, the utility of this information is currently limited because past studies have almost all been done at or near room temperature. The one exception is work by Zellner and co-workers (22), who studied nitrite photolysis as a function of temperature between 278 and 353 K, but only at two wavelengths (308 and 351 nm), which is too little information to confidently extrapolate to sunlight conditions. One further limitation to our current understanding of nitrite photochemistry in the environment is that there have been no studies of nitrite photolysis on snow or ice, despite its potential importance in snow chemistry. Our main goal in this work is to determine a master equation that expresses the quantum yield for reaction 3 as a function of illumination wavelength and temperature for both aqueous solution and water ice. To do this, we have measured the quantum yields for •OH formation from nitrite photolysis for temperatures between 240 and 298 K and illumination wavelengths between 302 and 390 nm. From these results we derive a master equation for nitrite photolysis and also examine the potential importance of this photochemistry in snow. 10.1021/es062731q CCC: $37.00
2007 American Chemical Society Published on Web 04/13/2007
Experimental Methods Materials. Sodium nitrite (certified ACS), acetonitrile (Optima), sodium borate (certified ACS), and perchloric acid (Optima) were obtained from Fisher. Sodium benzoate (“BA”; 99%) was from Aldrich, while p-hydroxybenzoic acid (“p-HBA”; 98%) was from TCI America. Purified water (“Milli-Q”) was obtained from a Milli-Q Plus system (g18.2 MΩ cm). Molar Absorptivities of Nitrite as a Function of Temperature. Absorbance spectra of five aqueous NO2- solutions (0-100 mM) were measured in a Shimadzu UV-2501PC spectrophotometer using a stirred, temperature-controlled, 1.0 cm quartz cell and Milli-Q as reference. The molar absorptivity at each wavelength was determined as the slope of the linear regression fit to the data of path lengthnormalized absorbance versus [NO2-]. Ice Sample Preparation. The techniques used for ice pellet preparation and illumination are described fully in Chu and Anastasio (19, 24) and are briefly summarized here. Samples were prepared in a freezing chamber by first pipetting 300 µL of Milli-Q water into a Teflon template placed on a quartz slide backing. After this ice “base” was frozen, 100 µL of sample solution was pipetted onto the ice base and allowed to freeze. Sample solutions contained 40 µM NO2- and 4.0 mM benzoate in Milli-Q water with pH adjusted to 6.0 or 8.0 using sodium borate. All listed concentrations (e.g., benzoate and NO2-) and pH values are for the sample solutions prior to freezing. Because freezing can enhance the O2-mediated oxidation of aqueous nitrite (25) we reduced dissolved O2 levels during preparation by purging the Milli-Q water with helium prior to making each sample solution and by purging the freezing chamber with He during freezing. The frozen samples were then allowed to equilibrate with atmospheric O2 after freezing. Measurements by ion chromatography showed that there was essentially no oxidation of NO2- in our prepared ice pellets, with the average ((1 σ) concentration after freezing being within (1 ( 2)% of the initial solution value. Ice samples were illuminated at controlled temperatures with monochromatic radiation of wavelength λ using either a 1000 W Xe/Hg lamp (for λ e 366 nm) or a 1000 W Xe lamp (for λ ) 380 and 390 nm) in conjunction with a monochromator (GM250, Spectral Energy Corp.). At the end of illumination the complete ice sample was melted in the dark at room temperature in about 5-10 min, and then the melted mixture was analyzed for p-hydroxybenzoate as described below. •OH Measurements on Ice. Hydroxyl radicals were characterized using a chemical probe technique where photoformed •OH reacts with benzoate (BA) to form phydroxybenzoate (p-HBA), the two other hydroxylated isomers, and other products: (6, 26)
C6H5CO2- + •OH f p-HBA + m-HBA + o-HBA + other (5) Based on room-temperature rate constants for •OH with benzoate and NO2- (27), and our concentration ratio of [benzoate]:[NO2-] )100 in the sample solutions, 98% of photoformed •OH should react with benzoate in our samples. As described below, we make a correction to account for the 2% of •OH not trapped by benzoate. Initial rates of p-hydroxybenzoate formation during illumination with wavelength λ (R*p-HBA,λ, i.e., d[p-HBA]/dt) were determined from plots of [p-HBA] versus illumination time using a linear regression fit. For each illuminated set of ice samples, rates of p-hydroxybenzoate formation were also measured in two controls: (1) in a dark control under identical conditions (sample composition and temperature) except no illumination (RDark) and (2) in an illuminated blank control
with identical conditions as the sample except that no nitrite was added (RBlank,λ). The corrected formation rate of p-HBA in the illuminated sample was then calculated as follows:
Rp-HBA,λ ) R*p-HBA,λ - RDark - RBlank,λ
(6)
There was generally no formation of p-HBA in samples kept in the dark (average rate ((1 σ) of 0.0002 ( 0.0003 nM min-1) and the rate of p-hydroxybenzoate formation in the illuminated blank (0.0037 ( 0.0003 nM min-1) was, on average, 3% of the rate in the corresponding illuminated sample. This small rate of formation in the illuminated blank is likely caused by trace amounts of •OH-generating precursors (e.g., organics) in the Milli-Q water. Note that subtracting both of these controls in eq 6 assumes that their formation mechanisms are independent and that their rates are additive in the samples. While this is a moot point in our current work (since the dark blank is essentially zero), it would need to be verified in possible future cases where both controls have non-negligible rates of •OH formation. As described previously (6), corrected rates of p-hydroxybenzoate formation were converted to rates of •OH formation (ROH,λ) by using the yield of p-HBA formed from the reaction of •OH with benzoate (Yp-HBA) and the fraction of •OH trapped by benzoate (0.98):
ROH,λ ) Rp-HBA,λ/(Yp-HBA × 0.98)
(7)
Values of Yp-HBA were determined as the rate of p-HBA formation divided by the rate of benzoate loss in independent experiments described previously (19). Using the previously measured relationship for ice, Yp-HBA ) -0.0011(pH)2 + 0.0154(pH) + 0.0318 (R2 ) 0.978; (19)), we calculate that Yp-HBA at both pH 6.0 and 8.0 is 0.085 ( 0.002. Calculation of •OH Quantum Yield on Ice. On every day that we performed a nitrite photolysis experiment we also measured the photon flux (Iλ) in the illumination system. To do this we used ice pellets of the same size and composition as the nitrite-containing samples except that 4 µM 2-nitrobenzaldehyde (2NB) was also added as a chemical actinometer. Under our conditions the measured rate constant for 2NB loss (j2NB,λ) is (28)
j2NB,λ ) 2.303 Iλ 2NB,λ Φ2NB,λ l
(8)
where Iλ is the photon flux (photon L-1 s-1), (2NB,λ Φ2NB,λ) is the product of the molar absorptivity and quantum efficiency of 2NB, and l is the effective path length of the sample (cm). Values of (2NB,λ Φ2NB,λ) are 640, 300, and 95 L photon-1 cm-1 at 313, 334, and 366 nm, respectively (28), and were calculated to be 857, 45, and 22 L photon-1 cm-1 at 302, 380, and 390 nm, respectively, based on measured molar absorptivities of 2NB and assuming a wavelength independent quantum efficiency of 0.41 (29). Actinometry performed on ice pellets with and without the chemicals used for the •OH measurements (i.e., NaNO2, benzoate and borate) showed that these species had no effect on measured values of j2NB,λ. In addition, we have measured the rate of loss of 2NB in different frozen and liquid portions of the same solution and have found that the phase state does not affect the photolysis rate constant (29). The rate of •OH formation during nitrite illumination with wavelength λ is
ROH,λ ) 2.303 Iλ NO2-,λ Φ(NO2- f •OH)T,λ l [NO2-]
(9)
where NO2-,λ is the base-10 molar absorptivity of nitrite (Table S1), Φ(NO2- f •OH)T,λ is the quantum yield of •OH from nitrite photolysis, and [NO2-] is the molar concentration of nitrite. Note that we express ROH,λ and [NO2-] on the same VOL. 41, NO. 10, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3627
FIGURE 1. Molar absorptivities of aqueous nitrite at 298 K (solid line). The open circles, dashed line, and crosses are data from Zuo and Deng (20), Riordan et al. (3), and Fischer and Warneck (10), respectively. Values of ENO2-,λ as a function of temperature and wavelength determined in this study are tabulated in the Supporting Information (Table S1). volume basis, namely that of the unfrozen sample solution. Combining eqs 8 and 9 produces an expression for the quantum yield of •OH • Φ(NO2 f OH)T,λ )
2NB,λΦ2NB,λROH, λ j2NB,λNO2-, λ[NO2]
(10)
For simplicity, we use the term Φ(NO2- f •OH) in the rest of the text instead of Φ(NO2- f •OH)T,λ. Measurements of Φ(NO2- f •OH) in Aqueous Solution. Aqueous solutions had the same composition as the sample solutions described above and were illuminated in stirred, temperature controlled, 2 cm FUV quartz cells (Spectrocell, Inc.) using the same light described earlier. Aliquots were removed at measured time intervals (generally 5 min) and analyzed for p-hydroxybenzoate. Rates of •OH formation were determined using eq 7, using a newly determined value ((1σ) for Yp-HBA (0.19 ( 0.011) determined as an average of 10 measurements between pH 0.1 and 5.0 and temperatures of 274-298 K. This value is consistent with our previous results obtained at pH 1.9 and 8.3 (6) and indicates that there is no pH or temperature dependence for Yp-HBA in solution between pH 0.1 and 8. The quantum yield of •OH in solution was calculated using eq 7, where j2NB,λ is measured in aqueous 2NB (4 µM) on the same day as the nitrite photolysis experiment.
Results and Discussion Molar Absorptivities of Nitrite as a Function of Temperature. Our absorption spectrum for aqueous NO2- at 298 K shows two weak bands with maxima at 300 and 355 nm, corresponding to nf π* transitions for oxygen and nitrogen as described previously (30). Figure 1 shows that our molar absorptivities generally match very well with the previously reported values from Fischer and Warneck (10), Zuo and Deng (20), and Riordan et al. (3). Average absolute relative percent differences between our results (280-390 nm) and these previous results are 3.1, 3.1, and 4.6%, respectively. At wavelengths from 390 to 400 nm the average absolute relative percent differences between our results and these previous results are significantly larger (37.2, 9.8, and 16.4%, respectively), but the molar absorptivities are small here and do not contribute significantly to the sunlight photolysis of nitrite. As shown in the Supporting Information (Table S1), the molar absorptivities of nitrite are only very weakly dependent upon temperature. For example, between 298 and 274 K there is less than a 1% difference in the maximum 3628
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 10, 2007
FIGURE 2. Quantum yields of •OH in aqueous solution (295 K, solid circles) and ice pellets (263 K, gray circles; 240 K, open circles) as a function of illumination wavelength. We measured our solution quantum yields at 298 K and adjusted the values to 295 K (to match the temperature condition of Fischer and Warneck (10)) using the temperature dependence described later in this paper. The squares, triangles, crosses, and diamonds are solution data from Zellner et al. (22), Zafiriou and Bonneau (30), Alif and Boule (23), and Fischer and Warneck (10), respectively. Error bars represent (1 σ, calculated based on propagated uncertainties. molar absorptivities at 355 nm and a 1-2% difference at other wavelengths between 290-410 nm. Quantum Yields of •OH in Solution as a Function of Wavelength. To characterize the wavelength dependence of the •OH quantum yield from NO2- photolysis, we first measured Φ(NO2- f •OH) in solution at 295 K. As shown in Figure 2, our values of Φ(NO2- f •OH) have the same general wavelength dependence that has been reported previously and are within 1-6% of the results of Fischer and Warneck (10), except for the one point at 390 nm, where our value is 20% lower and in better agreement with the empirical functional below. Fitting our results at 295 K along with those of Fischer and Warneck (10) to the empirical function previously suggested by these authors yields:
(
• Φ(NO2 f OH )295K,λ ) y0 +
)
a 1 + exp
(λ -b c)
(11)
where y0 ) 0.0204 ( 0.0010, a ) 0.0506 ( 0.0022, b ) 11.2 ( 1.2, c ) 332 ( 1, uncertainties are one standard error, and wavelength (λ) is in units of nm. The corresponding curve at 295 K, shown as the solid line in Figure 2, illustrates that our data, and most of the past data, are fit very well by this equation. Zafiriou and Bonneau (30) have suggested that this wavelength dependence is not due to excitation into the two different absorption bands (Figure 1), but rather is caused by a wavelength dependence in the relative efficiencies of the excited-state undergoing dissociation (to eventually form •OH) or internal conversion (to go back to NO -). 2 Quantum Yields of •OH in Frozen and Aqueous Solutions as a Function of Temperature and Wavelength. We first characterized the temperature dependence of Φ(NO2- f •OH) by performing experiments in frozen and aqueous samples between 240 and 298 K. As shown in Figure 2, the •OH quantum yields in ice are consistent with the wavelength behavior found in solution, where Φ(NO2- f •OH) decreases with increasing wavelength between a high plateau at λ < 300 nm and a lower plateau at λ > 360 nm. This figure also illustrates that quantum yields at all wavelengths decrease with decreasing temperature. This latter point is shown more clearly in Figure 3, which examines the temperature dependence for both solution and ice at individual illumination wavelengths. Note that for any given wavelength the solution
FIGURE 3. Temperature dependence of Φ(NO2- f •OH) in aqueous and frozen sample solutions (240-298 K). The number next to each series of points indicates the illumination wavelength. The open squares are 366 nm data at pH 8.0, while all other data from this study are for pH 6.0. Errors are (1 σ. Lines are linear regression fits to the solution and ice data at each temperature. The crosses and open diamonds are solution data from Zellner et al. (22).
TABLE 1. Values of Ea and ∆S ((1σ) for the Formation of •OH from Nitrite Photolysis on Ice and in Aqueous Solutiona wavelength (nm) 302 308 313 334 351 366
Ea (kJ mol-1) ∆S (J mol-1 K-1) 14.3 ( 1.2 13.0 ( 3.0 13.7 ( 1.0 13.8 ( 0.9 17.0 ( 3.3 15.4 ( 0.8
26.3 ( 1.9 21.4 ( 4.8 23.8 ( 1.5 20.8 ( 1.6 31.6 ( 6.4 20.3 ( 1.7
reference this study Zellner et al. (22) this study this study Zellner et al. (22) this study
a The activation energy (E ) and change in entropy (∆S) were a calculated based on the linear regression fits to the ln(Φ(NO2- f •OH)) versus 1/T data (Figure 3): Ea ) -slope × R × 10-3 kJ J-1 and ∆S ) y-intercept × R, where R is the gas constant (8.314 J mol-1 K-1). Values of Ea and ∆S from Zellner et al. (22) are calculated from their quantum yield data from 278 to 358 K.
and ice data are described by the same Arrhenius relationship. This suggests that nitrite photochemistry on ice is not occurring in bulk ice but rather in quasi-liquid layers, as described previously for nitrate and hydrogen peroxide (19, 24, 31). We combined the solution and ice data in Figure 3 to calculate activation energies (Ea) and entropy changes (∆S) for nitrite photolysis at each wavelength (Table 1). It is interesting to note that Ea for nitrite (∼14 kJ mol-1) is similar to that for nitrate (20 kJ mol-1 (19)) and that these values are much higher than the activation energy for photolysis of hydrogen peroxide (5.7 kJ mol-1 (24)). Fischer and Warneck (10) have suggested that this difference indicates that escape from the solvent cage is more difficult for •O- (which is the initial product from both nitrite and nitrate photolysis) than it is for •OH (which is directly formed from HOOH photolysis). As shown in Figure 4, the activation energy (Ea) for •OH formation from nitrite photolysis increases slightly with wavelength and follows the linear trend
Ea (J mol-1) ) 20.5 × λ (nm) + 7553
(12)
Given the higher relative uncertainties for ∆S values and the scatter in these data points (Figure 4), it appears that values of ∆S are relatively independent of wavelength. Note that there is only one past report of the temperature dependence for nitrite photolysis (22). The data at 308 nm from this previous report are very similar to our Ea and ∆S data (within
FIGURE 4. Activation energies (Ea) (open symbols) and entropy changes (∆S) (filled symbols) as a function of wavelength. Based on all data except for the point at 351 nm, the activation energy (Ea) increases slightly with wavelength and follows the linear fit Ea (kJ mol-1) ) 0.0205 λ + 7.55 (R2 ) 0.559), where λ is the wavelength in nm. The entropy changes (∆S) we measured are essentially independent of wavelength with an average value of 21.5 J mol-1 K-1(dashed line) based on data points at 308, 313, 334, and 366 nm. Circles represent our data while squares are results from Zellner et al. (22). 6%), but the previous data at 351 nm do not match our thermodynamic results (Table 1 and Figure 4) and do not fit in the wavelength or temperature dependence of the nitrite quantum yield (Figures 2 and 3). Based on our results in Figure 3, the temperature dependence of the quantum yield for nitrite photolysis can be expressed as an Arrhenius equation:
( )
Ea • Φ(NO2 f OH)Ti ) A × exp RTi
(13)
where A is an Arrhenius prefactor and Ti is the temperature (K). This equation can be extended to consider the difference in quantum yields at any two temperatures T1 and T2: • Φ(NO2 f OH)T2 ) Φ(NO2 f •
OH)T1 × exp
(( ) (
))
Ea 1 1 × R T1 T2
(14)
Using the quantum yield at 295 K as our reference point (i.e., T1), and substituting eq 12 for Ea into eq 14 yields an expression for Φ(NO2- f •OH) as a function of temperature and wavelength: • Φ(NO2 f OH)T ) Φ(NO2 f •
OH)295K × exp
((eλR+ f) × (2951 - T1)) (15)
where e ) 20.5 ( 3.2, f ) 7553 ( 1204 and errors are one standard error. The final step is to substitute eq 11 for the term Φ(NO2- f •OH)295K in eq 15 to derive the master equation for nitrite photolysis: • Φ(NO2 f OH)T,λ ) (eλ + f) 1 a 1 y0 + exp λ-c R 295 T 1 + exp b
(
(
)
)
(
(
)) (16)
where all variables and parameters are as defined above (i.e., y0 ) 0.0204 ( 0.0010, a ) 0.0506 ( 0.0022, b ) 11.2 ( 1.2, c ) 332 ( 1, e ) 20.5 ( 3.2, f ) 7553 ( 1204, with errors of one standard error). The first term on the right-hand side of eq 16 represents the wavelength dependence of nitrite photolysis, while the VOL. 41, NO. 10, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3629
FIGURE 5. Graphical representation of the master equation for quantum yields of nitrite photolysis (eq 16). Points represent measured quantum yields of •OH (Φ(NO2- f •OH)) in aqueous solution and ice pellets from our experiments (circles) and from those of Fischer and Warneck (10) (squares). The vertical line for each point shows the displacement of the experimental result from the value predicted based on eq 16. The dashed lines represent the wavelength-dependence of the regression equation at 295, 263, and 240 K (see Figure 2), while the dark solid lines highlight the temperature dependence data at 302, 313, 334, and 366 nm (see Figure 3). second term expresses the temperature dependence, which is also wavelength dependent. The formula is applicable for wavelengths between approximately 280-420 nm and a temperature range of at least 240-298 K. Equation 16 is shown graphically as the grid in Figure 5, along with our solution and ice data and the solution data of Fischer and Warneck (10). It is interesting to note that the wavelength dependence of nitrite photolysis on ice is consistent with the solution behavior, but with quantum yields that are all decreased by roughly the same factor. This is shown more clearly in Figure 2, where sections of Figure 5 at 240 and 263 K are shown in comparison with the 295 K solution data. Quantum Yield of •OH on Ice as a Function of Ionic Strength and pH. Experiments with added Na2SO4 to adjust the ionic strength (I) show that there is no effect on Φ(NO2f •OH) within the range tested (I ) 4.0 - 7.0 mM). This is consistent with previous work of Zafiriou and Bonneau (30), who found no dependence within the range of I ) 35 - 900 mM. We have also examined the effect of pH on the •OH quantum yield. At pH 6.0, the value in most of our solutions, 99.94% of N(III) is present as NO2- while 0.06% exists as HNO2. Thus the photochemistry in our experiments should be that of nitrite rather than HNO2 or H2ONO+. To test this, we performed a series of experiments at pH 8.0, where 99.99% of the N(III) is present as NO2-. As shown on the 366 nm line of Figure 3, values of Φ(NO2- f •OH) at pH 8 (0; 278-298 K) match the results from pH 6 (9; 240-298 K), confirming that our experimental results for NO2- are not affected by trace amounts of HNO2. This is consistent with past results from Zellner et al. (22), where quantum yields of •OH in solution (308 nm, 278 K) are constant between pH 5 and 9. Environmental Implications: Nitrite Photolysis on Arctic and Antarctic Snow Grains. As described in the introduction, nitrite appears to play a significant role in the chemistry of snowpacks, but little is known about its budget or its contribution to the formation of •OH on snow grains. Our goal here is to use our measured quantum yields and molar absorptivities for nitrite to estimate its snow-grain 3630
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 10, 2007
FIGURE 6. Action spectra for •OH formation from the photolysis of HOOH, NO3-, and NO2- on surface snow grains at Neumayer, Antarctica on midday of Dec. 21. The top panel shows molar absorptivities of hydrogen peroxide (solid line; (24)), nitrate (dashed line; (19)), and nitrite (dot-dashed line; this work) at 274 K. The middle panel shows the rate constants for •OH formation from photolysis of HOOH (4.8 × 10-6 s-1, solid line), nitrate (2.8 × 10-7 s-1, dashed line), and nitrite (3.5 × 10-5 s-1, dot-dashed line) for surface snow at 263 K. The bottom panel shows the rates of •OH formation from photolysis of HOOH (solid line; total rate ) 2.3 × 10-11 M s-1), nitrate (dashed line; 3.9 × 10-13 M s-1), and nitrite (dot-dashed line; 1.8 × 10-13 M s-1), calculated by multiplying the photolysis rate constants by estimated snow-grain concentrations for HOOH, NO3-, and NO2- (4.8, 1.4, and 0.0052 µM, respectively). Note that rates of photolysis for nitrate and nitrite in the bottom panel are multiplied by 10 to be seen at this scale. concentrations and its role in •OH formation. We estimate concentrations at four polar locations by assuming that NO2on the snow grains is at a photostationary state determined by the relative sizes of its sources and sinks. We consider two sources of nitrite: direct formation from the minor channel of nitrate photolysis (32)
NO3- + hν f NO2- + O(3P)
(17)
and via hydrolysis of NO2 (33) formed from the major channel of nitrate photolysis (19)
NO3- + hν f NO2 + •O-
(18)
2 NO2 + H2O f HONO + HNO3
(19)
Since the kinetics of this latter reaction on ice are not known (33), we assume that the NO2 formed in reaction 18 quickly forms HONO via reaction 19 and then dissociates to nitrite on the snow grains. Based on calculated nitrate photolysis
TABLE 2. Calculated Lifetimes and Steady-state Concentrations of Nitrite on Polar Snow Grains in the Surface Snowpack lifetime of nitrite (hr)g -]c
-
-)d
j(NO3 f NO2 (s-1)
-
j(NO3 f NO2 (s-1)
locationa
dateb
[NO3 (µM)
Alert, Nunavut Summit, Greenland South Pole Neumayer, Antarctic Neumayer, Antarctic (midnight)
Mar. 21
4.2
2.2 × 10-9
6.0 × 10-9
Jun. 21
4.0
8.2 × 10-8
Dec. 21 Dec. 21
1.6 1.4
Dec. 21
1.4
)e
-
j(NO2 f (s-1)
•OH)f
photolysis
overall
steady state [NO2-] (nM)
1.9 × 10-6
150
8.8
0.7
2.2 × 10-7
2.9 × 10-5
10
4.8
13
2.5 × 10-8 1.0 × 10-7
6.8 × 10-8 2.8 × 10-7
1.3 × 10-5 3.5 × 10-5
21 8
6.5 4.3
2 5
2.3 × 10-9
6.2 × 10-9
1.9 × 10-6
150
8.8
0.2
a Locations of sites: 82.5 °N, 62.3 °W (Alert), 72.6 °N, 38.5 °W (Summit), 90 °S (South Pole), and 70.7 °S, 8.3 °W (Neumayer). The actinic fluxes used in our calculations are snow surface values from the NCAR TUV model (37) using altitudes, snow albedo, and ozone columns described previously (24). b Temperatures used for calculations at Alert, Summit, South Pole, and Neumayer were 243, 263, 253, and 268 K, respectively. Calculations are for midday (solar noon) on the summer solstice except for Alert (midday on spring equinox) and the second Neumayer entry. c Values are from Toom-Sauntry and Barrie (38) for Alert, Dibb et al. (39) for Summit, Wolff et al. (40) for South Pole, and Mulvaney et al. (41) and Wolff et al. (40) for Neumayer. d Rate constant for nitrite formation from the photolysis of nitrate, calculated using molar absorptivities from Chu and Anastasio (19), the quantum yield from Dubowski et al. (32) (Φ(NO3-f NO2-) ) 1.5 × 10-3 at 263 K) and assuming the activation energy for this channel is the same as for the nitrate photolysis pathway that makes •OH (19). e Rate constant for NO2 formation from nitrate photolysis, calculated based on data from Chu and Anastasio (19). f Rate constant for •OH formation from nitrite photolysis, calculated using eq 16 and the 274 K molar absorptivities for nitrite from Table S1. g The photolytic lifetime of NO2- was calculated as the inverse of the rate constant for photolysis (j(NO2- f •OH)). The overall lifetime was calculated considering photolysis and snow-grain reactions with •OH and O3. The lifetime of snow-grain nitrite with respect to reaction with •OH is calculated to be 59 h using an estimated •OH concentration of 1.0 × 10-15 M in snow quasi-liquid layers (36) and a rate constant of 4.7 × 109 M-1 s-1 (27). The lifetime of nitrite with respect to reaction with O3 in the quasi-liquid layer is calculated to be 11 h based on Henry’s law partitioning of 30 ppbv of firn air ozone and a rate constant of 2.6 × 104 M-1 s-1 (calculated based on the roomtemperature rate constant (27) with an estimated activation energy of 45 kJ mol-1).
rate constants (Table 2), these two sources of snow-grain nitrite are roughly comparable and have a combined rate of formation on the order of 10-14 to 10-12 mol L-1-snow s-1 under the conditions examined. In terms of the loss of NO2-, based on our calculations in Table 2, the major sinks on polar snow grains appear to be direct photolysis (reaction 3) and oxidation by ozone:
O3 + NO2- f •O3- + NO2
(20)
Interestingly, both of these sinks for nitrite make •OH: photolysis produces •OH as described in reactions 3 and 4, while the ozone reaction (20) produces •OH via the subsequent rapid protonation and decomposition of •O3- into •OH and O . The lifetime of nitrite with respect to reaction 2 (20) is 11 h assuming the snow grains are in equilibrium with 30 ppbv of ozone in the near-surface firn air. This reaction with ozone, in combination with direct photolysis (and a minor contribution from reaction with •OH), lead to an overall lifetime for nitrite on near-surface snow grains of 4-7 h during summer at midday (Table 2). Note that this NO2- lifetime should increase with depth in the snowpack since mixing ratios of O3, and the rate constant for nitrite photolysis, both decrease with depth. Combining the sources and sinks for nitrite yields steadystate concentrations on the near-surface snow grains of approximately 0.2-13 nmol L-1-snow, with the highest amount at Summit (Table 2). These concentrations are somewhat lower than reported values in springtime surface snow at Barrow, Alaska, which were typically near the detection limit of 40 nmol L-1 (34). This discrepancy suggests that our simple treatment is missing one or more sources of nitrite, e.g., partitioning from the firn air, but other factors will also affect nitrite levels, such as actinic flux, temperature, concentrations of nitrate and ozone, and perhaps snow type. In comparison, in a recent paper that modeled nitrate and nitrite photochemistry in the surface snow at Summit during the summer, the estimated snow-grain concentration of NO2is
