Is Combustion the Major Source of Polychlorinated Dibenzo-p-dioxins

Jun 9, 2000 - Atmospheric Chemical Reactions of 2,3,7,8-Tetrachlorinated Dibenzofuran Initiated by an OH Radical: Mechanism and Kinetics Study. Xiaomi...
135 downloads 10 Views 124KB Size
Download PDF
Critical Review

Is Combustion the Major Source of Polychlorinated Dibenzo-p-dioxins and Dibenzofurans to the Environment? A Mass Balance Investigation JOHN I. BAKER AND RONALD A. HITES* School of Public and Environmental Affairs and Department of Chemistry, Indiana University, Bloomington, Indiana 47405

We have summarized some of the significant work characterizing polychlorinated dibenzo-p-dioxin and dibenzofuran (PCDD/F) sources and sinks to and from the atmosphere. Much of this effort has focused on estimating emissions from combustion sources and comparing those estimates to atmospheric deposition measurements. Despite significant growth in the available data for emissions and for deposition, it still appears that total PCDD/F deposition exceeds emissions by well over a factor of 2. We have further investigated this phenomenon by first developing a method to estimate PCDD/F emissions for countries where these data are lacking. Second, we have investigated the global mass balance of PCDD/F on a homologue-specific basis, taking into account degradation by reactions with the OH radical. We have found that most of the mass balance discrepancy is due to the octachlorinated dibenzo-p-dioxin (OCDD) congener. We suggest that the photochemical synthesis of OCDD from pentachlorophenol (PCP) in atmospheric condensed water is the most significant source of OCDD to the environment. Further research directions in this area are suggested.

Introduction Polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/ F) gained widespread notoriety when 2378-TCDD, the most toxic of these compounds, was found as a contaminant in Agent Orange, a defoliant used in Vietnam (1). Acute contamination incidents such as those at Times Beach, MO, and at Seveso, Italy, helped to further familiarize the public with the health hazards of these compounds (1). Prior to the late 1970s, it was thought that these compounds were formed only as byproducts during chemical manufacturing processes and that environmental contamination by PCDD/F was limited to areas where contaminated chemicals had been spilled or applied. In 1977, however, it was discovered that PCDD/Fs were being produced during the incineration of municipal solid waste (2, 3). We now know that PCDD/F are ubiquitous environmental contaminants, which are atmospherically transported from combustion sources to almost all areas of the globe (4). Since that discovery, considerable effort has focused on identifying and quantifying all of the possible combustion sources of these compounds, and at present, more than 20 such sources of PCDD/F have been identified (5). In an ideal situation, the total PCDD/F emissions from these sources into the earth’s atmosphere * Corresonding author e-mail address: [email protected]. 10.1021/es9912325 CCC: $19.00 Published on Web 06/09/2000

 2000 American Chemical Society

should balance the total PCDD/F deposition from the atmosphere to the earth’s surface. Achieving this mass balance has been the subject of several studies.

Mass Balance Studies In 1991, it was noted that, when estimates of PCDD/F emissions from known combustion sources in Sweden were compared with depositional estimates, the values did not agree (6). In fact, it appeared that annual depositional fluxes were on the order of 10-20 times higher than annual emission fluxes. However, these estimates were based on limited data, and it was assumed that the discrepancy was due either to biased sampling or to unidentified sources either in Sweden or outside its borders. Nevertheless, this mass imbalance phenomenon sparked several other PCDD/F mass balance studies. Another mass balance study, conducted in Germany, focused on quantifying all combustion sources of PCDD/F in the city of Hamburg, the second largest city in Germany (7). Emission estimates were compared with measured air concentrations, and it was found that the air above the city contained much higher levels of PCDD/F than could be produced by local sources. Although at first this seemed unusual, the authors concluded that, since long-range transport and meteorology were ignored, one could not assume that there were large missing sources. Still another European mass balance study focused on the United Kingdom (8). Using available source data, these authors found that the annual emission estimates were about an order of magnitude lower than deposition estimates. This discrepancy was attributed to the errors associated with a relatively small data set. In 1997, when a much larger data set was available, the mass balance was updated, and deposition was reported to be about twice as high as emissions (9). In the United States, a number of researchers have addressed this mass imbalance issue. Travis and Hartmeyer reported the first mass balance discrepancy in the United States when they noticed that less than 10% of the 2378TCDD in the environment could be generated by combustion sources (10). Later, Thomas and Spiro wrote an extensive review in which they estimated that about 400 kg of PCDD/F was emitted in the United States in 1989 (11). Upon examining the available data for deposition and emissions, they concluded that there was no evidence for significant missing sources and that the mass balance discrepancy was within the uncertainties of the data (11, 12). Brzuzy and Hites completed a global mass balance study in 1995 (13). These authors calculated a global emission estimate of 3000 kg/yr by multiplying emission factors by VOL. 34, NO. 14, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2879

the total worldwide combustion in each source category. To estimate PCDD/F deposition to land, over 100 soil samples were taken from strategic locations around the world. The PCDD/F concentrations in these samples were converted to soil deposition fluxes, which were grouped together and averaged depending on the climate zone from which the soil was collected. A total PCDD/F depositional flux of 12 000 kg/yr to land was calculated by multiplying these average fluxes by the total land area in each climate zone. Baker and Hites recently studied PCDD/F concentrations over the North Atlantic and found that about 1000 kg/yr PCDD/F is deposited to the oceans (14). Thus, by combining these two estimates, we find about a factor of 4 more deposition (13 000 kg/yr) than emissions (3000 kg/yr), even though care was taken to acquire remote samples that were not biased high by nearby sources. The mass balance estimate of Brzuzy and Hites was later questioned by Eisenberg et al. (15). They suggested that, because the soil data assume a steady state, the deposition estimate is low and does not accurately portray the errors, primarily because these compounds have residence times on the order of 10-100 years and because these compounds have been emitted from sources at variable levels since the 1930s. On the basis of a dynamic modeling framework to further investigate this theory, Eisenberg et al. (15) used distributions reflecting the full range of possible values for environmental parameters such as the degradation half-lives of PCDD/F and other physical/chemical properties. They concluded that the discrepancy between deposition and source emissions should actually be somewhere between 6-fold and 20-fold in favor of deposition. In an attempt to improve the global PCDD/F deposition estimate, Wagrowski and Hites collected 63 soil samples from locations around the world, filling in some of the geographical gaps from the work of Brzuzy and Hites (16). In addition, an alternate method of calculating worldwide PCDD/F depositional fluxes was developed. Rather than extrapolating average PCDD/F fluxes to the corresponding climate zones, Wagrowski and Hites developed a method that correlated PCDD/F deposition with NOx emissions. Since data are available for NOx emissions all over the world, these authors were able to estimate the depositional flux of PCDD/F to areas of the world where no soil had been collected. Using this method of calculating deposition, it was estimated that 3000-10 000 kg/yr PCDD/F was deposited globally, a value slightly less than the estimate of Brzuzy and Hites but still much greater (on average) than global emissions. Despite the large uncertainties in each of these mass balance studies, it is important to note that all of them suggest that there is more PCDD/F deposition from the atmosphere than emissions to it. This is a remarkably consistent conclusion arrived at for different countries and for the earth as a whole and by different authors using different approaches. Taken at face value, this discrepancy suggests either that the emissions have been grossly underestimated, that deposition has been grossly overestimated, or that a major part of the scenario is wrong. We will look at the emission/deposition question first.

Another Estimation of PCDD/F Emissions It is possible that the deposition estimates are biased high because of the extrapolation of PCDD/F deposition data from locations close to sources (thus containing high levels of PCDD/F) to locations far removed from combustion sources (13). (In this case, remote locations would be expected to have much lower PCDD/F concentrations than locations closer to combustion sources because PCDD/F would have more time to diffuse, deposit, or react before depositing to the more remote sinks.) Recent work suggests that biased sampling was part of the problem (13, 14, 16). However, even 2880

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 14, 2000

TABLE 1. PCDD/F and Annual CO2 Emissions and Gross Domestic Product (GDP) for Various European Countries and the United States (5, 17, 18) country

PCDD/F (kg/yr)a

GDP (109 $U.S.)

CO2 emissions (106 Mt)

United States Austria Belgium Switzerland Germany Denmark Spain France Greece Italy Ireland Luxembourg Norway The Netherlands Portugal Sweden Finland United Kingdom

180 7.3 29 11 50 3.0 20 67 7.3 63 2.0 3.0 2.5 7.0 7.6 5.3 4.1 56

5610 164 197 230 1690 130 527 1200 71 1150 44

13460 164 279 114 2650 170 600 1020 199 1100 88 28 160 379 114 146 142 1580

106 291 69 237 125 978

a Emission estimates were reported in kg TEQ/yr; to convert to kg/yr, for comparison with deposition estimates, we multiplied the TEQ by 60 (11, 20).

when remote samples that were clearly far removed from local sources were analyzed and used to estimate deposition to remote locations globally, total PCDD/F deposition estimates were still much higher than the global emission estimate. We believe that biased sampling cannot explain the global mass imbalance. This suggests either that the emission estimates are too low or that there is, in fact, some sort of de novo synthesis of PCDD/F in the atmosphere. We will first look at the global emission estimate again. The United States and many European countries have continued to more accurately estimate PCDD/F emissions (5, 17). Unfortunately there are almost no such data for other countries, particularly those with large populations such as China and India. Therefore, to determine how much PCDD/F these other countries are emitting, it was necessary to find a parameter that correlates well with PCDD/F emissions and that is available for these other countries. Since PCDD/F are emitted from combustion sources, we investigated the correlation of their emission rates with carbon dioxide emission rates, a combustion related rate that is reported for most countries (18). Since environmental data are usually log-normally distributed (19), we plotted the logarithm of the PCDD/F emission estimates versus the logarithm of CO2 emissions using the values in Table 1 for the 18 countries that have characterized their annual PCDD/F emissions (5, 17, 18). We found that PCDD/F emission rates correlate well with carbon dioxide emission rates, giving an r 2 value of 0.796, which is significant with a probability >99% (see Figure 1a). Interestingly, we also found a good correlation, also significant with a probability >99%, between the logarithm of the gross domestic product (GDP) for these countries and the logarithm of their annual PCDD/F emission rates; see Figure 1b. This correlation was fortuitous because a few countries do not report carbon dioxide emissions. Using these simple relationships, we calculated the predicted annual PCDD/F emission rate for each country in the world from the known CO2 emission rate (N ) 156) or from the known GDP (N ) 23) for that country (26 countries did not report either of these data, and we have assumed that their contributions to PCDD/F emissions are minimal). Adding together these predicted PCDD/F emissions gave us a total global emissions estimate of about 1800 kg of PCDD/F per year, which is about half of the estimate of 3000 kg/yr

a

a

b b

FIGURE 1. Relationship of PCDD/F emissions in the Unites States and in Europe versus annual CO2 emissions (a) and gross domestic product (GDP) (b) for each country; see Table 1. made by Brzuzy and Hites (13). It is important to note that the latter estimate used reported PCDD/F emission rates from the late 1980s and early 1990s, but the calculation reported here uses rate estimates that were derived from 1995 PCDD/F emission rates, from 1995 CO2 emission rates, and from 1995 GDP data. This may indicate that there has been a decrease in the global PCDD/F emission rate in the last 10 years of about 1000 kg/yr. Taken as a whole, however, this analysis suggests that emission estimates of 2000-3000 kg of PCDD/F per year is about right and that the mass imbalance is not due to the emission side of the equation.

Hydroxyl Radical Reactions It is important to note that, in each of these mass balance studies, PCDD/Fs have been treated as one compound by adding together all the tetra- though octa-chlorinated congener concentrations and by comparing the final source and sink values of total PCDD/F. By treating mass balances in this manner, individual homologue information is lost. Figure 2 shows that the homologue profile of PCDD/F sources differs substantially from their sinks. In this case, the homologue profile for the sources is based on data from the U.S. EPA (21), and the homologue profile of the sinks is based on the average of all 170 soil samples taken around the world and analyzed in our laboratory. The latter is consistent with the homologue profiles seen in lake sediment around the world (22-24). The sink profile is dominated by the octachlorinated dioxin, but the source profile is dominated by the tetra- through hexachlorinated furans. It has been hypothesized (25) that this difference is due to relatively fast gas-phase reactions between the PCDD/F congener and the hydroxyl radical (OH). These reactions are more rapid for the less chlorinated PCDD/F homologues, and thus, only the more chlorinated PCDD/F would remain in the atmosphere and be removed by wet and dry deposition processes (25). We will examine this idea further.

FIGURE 2. Average homologue profile for PCDD/F from combustion sources (21) and in environmental sinks (13, 16). The source profile reflects the weighted average of 12 different combustion source categories with error bars depicting the upper limits for each homologue (21). The sink profile reflects the average of 170 soil samples (with standard errors) taken in our laboratory (13, 16). The letters F and D represent furans and dioxins, respectively; the numbers indicate the chlorination levels. Brubaker and Hites have developed a model to examine the effect of PCDD/F + OH reactions on the homologue profiles (25). We will extend the application of this model to quantitatively predict annual deposition of PCDD/F after reactions with OH for each PCDD/F homologue. We can then compare these predicted deposition values to measured deposition values on a homologue-by-homologue basis and see if the results agree. The equation used to calculate deposition, as described by Brubaker and Hites (25), is

(

SD ) SE

kDφ

)

k′OH(1 - φ) + kDφ

where SD and SE are the PCDD/F deposition and emission rates, respectively; kD is the deposition rate constant for PCDD/F in the particle phase; k′OH is the pseudo-first-order rate constant for the reaction of PCDD/F with OH; and φ is the fraction of the PCDD/F associated with the particle phase. This model assumes that no significant photochemistry occurs in the particle phase (25, 26). For the purposes of the following calculation, we will assume that the total annual global PCDD/F emission rate is 2000 kg/yr. Using this value and the source profile from Figure 2, we can apportion the total to the different homologues. This gives us the source emission rates (SE) by homologue shown in the second column of Table 2. The fraction of each PCDD/F homologue in the atmospheric particle phase (φ) depends on the vapor pressure of the homologue, and it varies from about 10% for the tetrachloro homologues to about 100% for the octachloro homologues. The values we used are given in the third column of Table 2. The values k′OH are the product of the average second-order OH reaction rate constant for each homologue group (25) times the 24-h averaged OH concentration of 9.7 × 105 cm-3 (28). Brubaker and Hites experimentally measured the gas-phase OH reaction rate constant for 1,2,3,4-tetrachlorodibenzo-p-dioxin. This measurement verified the rate constants recommended by Atkinson (29) using a structureVOL. 34, NO. 14, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2881

TABLE 2. Parameters Used To Calculated the Loss of PCDD/F through Reaction with the Hydroxyl Radical homologue

sources (kg/yr)

particle fraction (O) (27)

k′OH (×10-6 s-1)

F4 F5 F6 F7 F8 D4 D5 D6 D7 D8

430 370 290 180 85 95 130 160 150 140

0.09 0.37 0.73 0.92 0.95 0.13 0.33 0.78 0.97 0.99

0.60 0.29 0.14 0.06 0.02 1.02 0.54 0.26 0.13 0.05

TABLE 3. PCDD/F Emission and Deposition Rates (in kg/yr) Compared by Homologue homologue sources F4 F5 F6 F7 F8 D4 D5 D6 D7 D8 total

a

b

FIGURE 3. PCDD/F homologue emission profiles before and after reactions with the hydroxyl radical (OH). Note the bar symbols are the same as in Figure 2. activity method. Since there are no experimental data for any other PCDD/F, we used Atkinson’s recommended rate constants for these calculations. The resulting pseudo-firstorder rate constants are given in the last column of Table 2. The final parameter kD is the deposition rate constant for PCDD/F isomers in the particle phase. Because the deposition of the particles is not dependent on the chemistry of the associated PCDD/F (it only depends on particle size), this rate constant is the same for all homologues. We used a value based on a reasonable particle phase residence time of 8 days (30), which gave a kD value of 1.45 × 10-6 s-1. The magnitude of PCDD/F remaining after reaction and available for deposition from the atmosphere (SD) was calculated for each homologue group using the parameters given in Table 2, and the results are shown in Figure 3b. The emission profile changes as a result of reactions with the hydroxyl radical, mostly because of a severe depletion of the tetrachloro homologues (and to a lesser extent of the pentachloro homologues). The changes in this profile, however, are not enough to match the experimentally measured deposition profile shown in Figure 2b. This suggests that the OH radical by itself is not responsible for the homologue profile seen in soil and sediment. Thus, a closer look at the individual homologue mass balances is necessary.

Mass Balance by Homologue Since 170 soil samples have been taken from all over the world, primarily from remote locations, we feel that deposi2882

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 14, 2000

a

loss due imbalance to OH after OH rxn deposition factora

430 370 290 180 85 95 130 160 150 140

340 90 10 ∼0 ∼0 80 50 10 ∼0 ∼0

90 280 280 180 85 15 80 150 150 140

280 250 240 230 130 60 100 350 850 5500

2000

∼600

1400

8000

3.1 0.89 0.86 1.3 1.5 4.0 1.3 2.3 5.7 39

Ratio of deposition to sources after OH reactions.

tion estimates have not been underestimated and that a reasonable estimate of deposition is about 8000 kg/yr. By apportioning the 8000 kg/yr total deposition estimate to the fractional profile shown in Figure 2b, we obtain the mass of PCDD/F deposited annually by homologue; these values are shown in Table 3, column 5. When we compare these values to the expected deposition based on the hydroxyl radical model calculations, we see that most homologues balance well, usually within a factor of 2 or less. However, there is a glaring misbalance for OCDD (D8), which seems to have about 40 times more deposition than emissions. There is also a smaller but significant imbalance for HpCDD (D7), which seems to have about 6 times more deposition than emissions. There are two possible explanations for this discrepancy. First, the known combustion sources have been drastically underestimated, and the tetra- through hexachloro dioxins and furans have been almost completely degraded before deposition. Second, there is some other unidentified source of the higher chlorinated dioxins, particularly OCDD. If we assume that once OCDD is formed it is not degraded, then we must have underestimated combustion sources by about 40 times. This seems unreasonable given the exhaustive work directed toward quantifying combustion sources. Therefore, we will take a closer look at the second explanation.

Another Source of OCDD It has been well-documented that technical pentachlorophenol (PCP), a wood preservative, has historically contained large amounts of PCDD/F contamination (31-33). In 1985, worldwide consumption of PCP reached about 108 kg/ yr (34). Because the vapor pressure of OCDD is on the order of 10-8 Pa (35), it is unlikely that volatilization of contaminant OCDD from PCP-treated wood is occurring to any great extent. However, since the vapor pressure of PCP itself is around 0.1 Pa (36), volatilization of PCP from preserved wood is likely. This could lead to ubiquitous contamination of the environment with PCP, and indeed, relatively high levels of PCP have been observed throughout the world (34, 37-39). For example, air samples taken in Canada showed concentrations ranging from 1 to 1200 ng/m3 (37). Soil in the U.K. has PCP concentrations of about 2 µg/kg (34), and rain samples from the U.K., Croatia, and the United States have PCP concentrations ranging from 10 to 100 ng/L (34, 3739). We suggest that some of this PCP can be converted in the atmosphere to OCDD, and to a lesser extent to HpCDD, and that this is the primary source of OCDD and HpCDD in environmental sinks. There are some data in the literature to support this idea. In 1976, Crosby and Wong first reported the photochemical generation of PCDD/F from pentachlorophenol (40). In their experiments, purified PCP was dissolved in a 1% sodium

hydroxide solution at a concentration of 1 g/L, and this solution was irradiated in a sunlight-simulating photoreactor for 16 h. The reaction generated 26 mg/L OCDD and 5 mg/L hexachlorodibenzo-p-dioxin (HxCDD), though it is not clear how many congeners of HxCDD were formed. These results suggested the possibility of natural formation of PCDD/F from PCP in the environment; however, this high reactant concentration would only be found in areas where PCP had been used, and it was unlikely that such an alkaline environment would exist in such cases. In 1994, Vollmuth et al. investigated the generation of PCDD/Fs by the photochemical treatment of seepage water or wastewater contaminated with PCP (41). For their experiments, they used a photochemical reaction system that was optimized for emission at 254 nm (the previous work had used wavelengths above 300 nm). They used purified PCP dissolved in water at a concentration of 1 mg/L, a level representative of landfill seepage water, and they irradiated this solution for 5 h. OCDD was generated at a concentration of about 100 ng/L along with lesser amounts of OCDF, HpCDD and HpCDF. A similar study using wavelengths between 200 and 300 nm also found that these compounds could be generated from a solution of PCP (42). Recently other researchers have investigated the photodegradation of solid PCP using wavelengths above 290 nm (43). For these experiments, PCP was dissolved either in methanol or in a 50% mixture of methanol and water. The solution was then placed in a reactor and evaporated so that a homogeneous layer of PCP was formed on the bottom of this reactor. In both cases, OCDD and lesser amounts of HpCDD were formed. Because relatively high concentrations of PCP have been detected in rain (34, 38-39), it is possible that PCP at environmental levels may react photochemically in condensed water to form OCDD and HpCDD. In fact, this reaction could be the primary source of OCDD in the environment. To mimic more environmentally relevant conditions, we repeated the experiment described by Vollmuth et al. (41); these results are described in the Appendix. We eliminated wavelengths below 290 nm by enclosing a medium pressure mercury lamp in a water-cooled glass sleeve. The initial pH levels in our irradiated solutions were about 5.5, similar to the pH of rain. In this experiment, we found conversion to OCDD with yields as high as 0.1% by mass. Lower levels of both HpCDD congeners were also detected with yields as high as 0.01%. Since conversion of PCP to OCDD clearly occurs in water at wavelengths above 290 nm, it follows that the photochemical synthesis of OCDD from PCP may be occurring in condensed water in the environment. The next question is whether the environmental levels of PCP and its OCDD yield are sufficiently high to account for the mass balance differences. It is clear that OCDD is the major contributor to the mass imbalance; therefore, we will examine the necessary parameters that are required to bring it into balance. Using the mass balance values described earlier (see Table 3), about 5000 kg/yr more OCDD appears to be depositing to the surface of the earth annually than is being emitted from combustion sources. We know that about 5.2 × 1014 m3 of rain reaches the earth’s surface each year (44). PCP concentrations in rain are about 10-100 ng/L (34, 38, 39). On the basis of these data, we will assume that the average concentration of PCP in rain is about 20 ng/L. Multiplying this concentration by the rainfall rate, we get a total PCP wet deposition to the surface of the earth of about 1 × 107 kg/yr. If only 0.05% (500 ppm) of this PCP is converted into OCDD, this would give the 5000 kg/yr OCDD necessary to close the mass balance for this compound. To account for the excess HpCDD seen in deposition, only about 700 kg/yr must be

a

b

FIGURE 4. Average PCDD/F concentrations in rain (with standard errors) taken from Indianapolis, IN (a, N ) 15), and from Bloomington, IN (b, N ) 13) (45, 46). The bar symbols are the same as in Figure 2. produced from this reaction, which is a conversion rate of about 0.007%. From the photochemical conversion studies cited here, it is clear that these yields are well within the range of experimental values. Of course, we have neglected kinetics in this calculation. It is important to know if the residence time of the condensed water in the atmosphere is long enough to produce OCDD at the conversion rate that we see. A study of the PCP to OCDD reaction kinetics has not been done; however, we note that water particle residence times are on the order of a few hours (45), which is essentially the same as the time used for the photochemical experiments reported here and in the literature. Because we are suggesting that PCP is converted to OCDD and HpCDD in rain, we expect rain to contain higher levels of these two compounds as compared to the other PCDD/F. Unfortunately, measurements of PCDD/F in rain are sparse, but when these measurements have been made (46, 47), the homologue profiles closely resemble the average homologue profile of soil and sediment. For example, Figure 4 shows the average homologue profile for rain taken in Bloomington, IN, and in Indianapolis, IN (46, 47). Note the agreement with the sink profile shown in Figure 2b. Because most atmospheric condensed water particles never become rain (they evaporate) (45), much of the PCDD/F formed from PCP in these water particles will be subject to dry deposition, and we expect the homologue profile for dry deposition to be similar to the homologue profile for wet deposition. Indeed this is what has been observed (46, 47). Incidentally, the formation of octachlorodibenzo-p-dioxin in condensed water in the atmosphere is probably not of much importance in terms of additional risk to man or the environment because this particular congener has a very low toxic equivalent factor (11, 20). Thus, the additional toxicity created by this pathway is probably low.

Future Directions We are suggesting here that the photochemical conversion of PCP in the aqueous phase in the atmosphere is the most important source of OCDD (and to a lesser extent, of HpCDD) to the surface of the earth. We base this suggestion on (a) the VOL. 34, NO. 14, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2883

observation that more OCDD is depositing from the atmosphere than is entering it (the mass balance problem) and (b) the observation that OCDD and HpCDD can be produced in water from PCP. Interestingly, the PCDD/F homologue profile produced by these PCP reactions closely resembles the profile observed in environmental sinks. The remaining PCDD/F can easily be accounted for by combustion source emissions. Clearly, the de novo synthesis of OCDD from PCP in the atmospheric condensed water phase is a reaction that must be studied further in order to more accurately assess the magnitude of PCDD/F produced by this source. A photochemical reaction system that uses a mercury lamp and purified reactants does not mimic the conditions that occur in condensed water in the atmosphere. Such an experiment serves only to demonstrate that wavelengths above 290 nm will convert PCP to OCDD and that the yield under these conditions is sufficient to account for the discrepancy in the mass balance. There are many conditions that may cause this yield to increase or decrease. To more accurately assess the importance of this reaction, photochemical studies that better mimic the conditions in the environment should be conducted. For example, it is important to determine how the matrix of compounds that exist in a rain drop affects the rate and yield of this reaction. A simple experiment using rain and labeled PCP would be a first step. It will also be important to monitor the effect of the hydroxyl radical on this reaction. Previous work has suggested that OH may promote the breakdown of PCP (42), and our studies demonstrated a better yield of OCDD when low levels of OH (generated from H2O2) were present in the solution (see Appendix). Condensed water in the atmosphere exists as fine droplets in clouds or fog. These conditions are different from the large volume solutions used in the photochemical experiments described here. Fine droplets have a much higher surface to volume ratio, and if this PCP to OCDD reaction occurs near the droplet surface, it would be much more efficient in fine droplets than in bulk solution. An experimental design that could examine this reaction in small droplets is needed. Research to determine the wavelengths that initiate this reaction should also be conducted. Knowing this wavelength and the intensity needed to produce OCDD combined with a working knowledge of the effects of a true environmental matrix will allow a much more accurate estimate of the amount of OCDD (and HpCDD) produced globally in condensed water particles in the environment. Finally, little data exist regarding the concentration of PCP in precipitation. Since we believe that the conversion of PCP to PCDD/F is the major source of these compounds to the environment, it will be important to obtain precipitation PCP concentrations from all over the world. These data can then be used in loadings calculations to more accurately quantify the contribution of PCDD/F from this source to the global environment.

Acknowledgments We would like to thank the National Science Foundation (Grant BES93-21274) for funding this work.

Appendix Pentachlorophenol (PCP), a fungicide, has been widely used for the preservation of lumber, and in 1985, its production peaked at about 108 kg/yr (34). This widespread use has led to high levels of PCP contamination in a variety of matrixes including rain (34, 37-39). In 1976, it was found that high concentrations of aqueous PCP under basic conditions can be converted to octachlorodibenzo-p-dioxin (OCDD) (40). Further research showed that when aqueous PCP at a concentration as low as 1 mg/L is irradiated at wavelengths of less than 290 nm, OCDD and lesser amounts of heptachlorodibenzo-p-dioxin (HpCDD) are generated (41, 42). We have investigated this reaction further, and we have confirmed that this reaction occurs when environmentally significant wavelengths greater than 290 nm are used even at PCP levels as low as 100 ppb and at a realistic rain pH of 5.5. Experimental. The 99% pure 13C6-PCP was purchased from Cambridge Isotope Laboratories, Inc. (Andover, MA) and diluted with acetone to a concentration of 1.2 µg/mL. The PCP solution was spiked into a 1-L quartz Erlenmeyer flask (Ace Glass, Louisville, KY) and diluted with HPLC grade water to give a PCP concentration between 100 and 1000 µg/L. This allowed us to investigate this reaction at concentrations as much as an order of magnitude below the lowest concentrations previously studied. The flasks were placed 30 cm from a medium-pressure mercury lamp (Ace Glass, Louisville, KY) and stirred. The lamp was enclosed by a water-cooled glass sleeve that eliminated wavelengths below 290 nm. These solutions were not buffered; however, the initial pH for each flask was about 5.5, similar to the pH of rain. The solutions were then irradiated for a period between 45 min and 4 h. The light power was monitored using a PMA 2100 pyranometer (Solar Light Co., Philadelphia, PA) and integrated over the reaction time. When the reaction time was completed, a solution of unlabeled PCDD/F with congeners representing each homologue group was spiked onto the solution as the internal standard. Dark experiments were also performed alongside the above reaction vessels; for these experiments, the flask was wrapped in aluminum foil to prevent light from entering the solution. All solvents were purchased from EM Science, Gibbstown, NJ. Upon completion of the reaction, 50 mL of hexane was added to the flask. About 50 g of NaCl was also added to the flask to facilitate the extraction of PCDD/F from the water layer. This solution was transferred to a separatory funnel and shaken. The flask was then rinsed with 50 mL of hexane that was then transferred to the separatory funnel. After two layers had formed, the hexane fraction was placed in a roundbottom flask, and the water was extracted two more times with 50 mL of hexane. The extracts were combined, and the residual water in the hexane fraction was removed using anhydrous sodium sulfate. The hexane fraction was then reduced to a volume of about 1 mL by rotary evaporation and placed in an 8-mL amber vial, where its volume was further reduced by nitrogen blowdown to 100 µL.

TABLE 4. Conversion of 13C6-PCP to 13C12-PCDD under Varying Irradiation Conditions

a

mass of PCPa (µg)

vol of H2O2 (µL)

time irradiated (total energy)

100 1000 100 1000 1000

100 100 0 0 0

4 h (70 J/cm2) 4 h (70 J/cm2) 45 min (10 J/cm2) 45 min (20 J/cm2) 45 min (20 J/cm2)

In 1 L of water.

2884

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 14, 2000

13C

12-HxCDD formed (ng)

0 30 0 0 0

13C

12-HpCDD formed (ng)

13C -OCDD 12 formed (ng)

0 140 0.8 0 0

0.5 1,100 6 7 2

PCDD/F were quantitated using the internal standard approach, and analysis was performed on a Hewlett-Packard 5973 gas chromatographic mass spectrometer operating in the electron capture, negative ionization mode. Chromatographic separation was achieved using a 30-m, DB-5MS, capillary column (250 µm i.d.; 0.25 µm film thickness; J&W Scientific, Folsom, CA). Helium was used as the carrier gas. The sample (2 µL) was injected with a 25 psi pulsed injection in the splitless mode. Temperature programming was as follows: isothermal for 2 min at 110 °C, 30 °C/min to 210 °C, 2 °C/min to 280 °C, isothermal for 10 min. The flow of the reagent gas, methane, in the mass spectrometer’s ion source was maintained at 40% of the total flow or at 2 cm3/min, which gave a manifold pressure of 2 × 10-4 Torr. The ion source temperature was held at 150 °C. Two ions from each homologue group were monitored using selected ion monitoring to enhance sensitivity, and a peak was only classified as a PCDD/F if its mass spectral intensities were in the correct, predicted isotopic ratio (4). Results and Discussion. We first analyzed the 13C6-PCP spiking solution by GC/MS to make sure that any PCDD/F detected was not a contaminant in the original standard. The mass chromatograms showed that the standard was clear of any PCDD/F. As a second quality control measure, dark experiments were conducted so that we could rule out any pathway of PCDD/F production other than a photochemical reaction. No PCDD/F were detected in any dark experiments. Table 4 gives the mass of PCDD/F produced in the five different experiments. To investigate the effect of the hydroxyl radical (OH) on this reaction, two of these experiments included hydrogen peroxide at a concentration of 100 ppb. The best conversion rate is seen in the experiment containing hydrogen peroxide and 1 mg/L PCP. Irradiation of this experiment lasted for 4 h, giving a yield of 0.1% by mass for the conversion of PCP to OCDD. This experiment also showed the formation of HpCDD with a yield of about 0.01% and the formation of HxCDD with a yield of about 0.003%. The two experiments that had the lowest concentration of PCP still showed conversion to OCDD, but at a lower yield. Since the experiment that had 100 µg of PCP and 100 ppb H2O2 had a lower yield as compared to the experiment with 1 mg of PCP and 100 ppb H2O2, it seems likely that the proportion of H2O2 to the reactant is important. This ratio is high in the first experiment, so the PCP may be degrading by a different pathway. It is also important to note that the experiments in which no H2O2 was added still showed a conversion of PCP to OCDD. These experiments were irradiated for 45 min and showed yields between 0.006% and 0.0007%. The purpose of this experiment was to confirm that PCDD/F is generated from PCP photochemically at environmentally relevant pH levels and to show that wavelengths less than 290 nm are not necessary to initiate this reaction. This is important because sunlight with wavelengths less than 290 nm does not reach the troposphere and because previous work had used pH levels of 7.0 or higher (40, 42). Our data suggest that available sunlight in the troposphere may convert environmental levels of PCP in the atmosphere, particularly in the condensed aqueous phase, to PCDD/F.

Literature Cited (1) Gough, M. In Dioxin, Agent Orange: The Facts. Plenum Press: New York, 1986. (2) Olie, K.; Vermeulen, P. L.; Hutzinger, O. Chemosphere 1977, 6, 455-459. (3) Buser, H. R.; Busshardt, H.-P.; Rappe, C. Chemosphere 1978, 7, 165-172. (4) Czuczwa, J. M.; Hites, R. A. Environ. Sci. Technol. 1984, 18, 444450. (5) U.S. Environmental Protection Agency. The Inventory of Sources of Dioxin in the United States; The Office of Research and

(6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20)

(21)

(22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32) (33) (34) (35) (36) (37) (38) (39) (40) (41)

Development, National Center for Environmental Assessment: Washington, DC, April 1998; External Review Draft, EPA/600/ P-98/002Aa. Rappe, C. Banbury Report 35: Biological Basis for Risk Assessment of Dioxins and Related Compounds; Cold Spring Harbor Press: Cold Springs Harbor, NY, 1991. Lau, C.; Fiedler, H.; Hutzinger, O.; Rippen, G.; Wesp, H. F.; Sievers, S.; Friesel, P.; Schacht, U.; Gras, B.; Reich, T.; Vahrenholt, F. Organohalogen Compd. 1996, 28, 83-88. Harrad, S. J.; Jones, K. C. Sci. Total Environ. 1992, 126, 89-107. Duarte-Davidson, R.; Sewart, A.; Alcock, R. E.; Cousins, I. T.; Jones, K. C. Environ. Sci. Technol. 1997, 31, 1-11. Travis, C. C.; Hattemeyer-Frey, H. A.; Sci. Total Environ. 1991, 104, 97-127. Thomas, V. M.; Spiro, T. G. Toxicol. Environ. Chem. 1995, 50, 1-37. Thomas, V. M.; Spiro, T. G. Environ. Sci. Technol. 1996, 30, 82A-85A. Brzuzy, L. P.; Hites, R. A. Environ. Sci. Technol. 1996, 30, 17971804. Baker, J. I.; Hites, R. A. Environ. Sci. Technol. 1999, 33, 14-20. Eisenberg, J. N. S.; Bennett, D. H.; Mckone, T. E. Environ. Sci. Technol. 1998, 32, 115-123. Wagrowski, D. M.; Hites, R. A. Environ. Sci. Technol. 2000, 34, 2952-2958. Quass, U.; Fermann, M. W.; Bro¨ker, G. Organohalogen Compd. 1998, 36, 7-10. United Nations Statistical Division. World Statistics Pocketbook; Department for Economic and Social Information and Policy Analysis, United Nations: New York, 1995. Parkhurst, D. F. Environ. Sci. Technol. 1998, 32, 92A-98A. Thomas, V. M.; Spiro, T. G. An Estimation of Dioxin Emissions in the United States PU/CEES Report 285; Center for Energy and Environmental Studies, Princeton University: Princeton, NJ, 1994. U.S. Environmental Protection Agency. Database of Sources of Environmental Releases of Dioxin-like Compounds in the United States; The Office of Research and Development, National Center for Environmental Assessment: Washington, DC, April 1998; External Review Draft, EPA/600/P-98/002Ab. Baker, J. I.; Hites, R. A. Environ. Sci. Technol. 2000, 34, 28872891. Pearson, R. F.; Swackhamer, D. L.; Eisenreich, S. J.; Long, D. T. J. Great Lakes Res. 1998, 24, 65-82. Czuczwa, J. M.; Niessen, F.; Hites, R. A. Chemosphere 1985, 14, 1175-1179. Brubaker, W. W.; Hites, R. A. Environ. Sci. Technol. 1997, 31, 1805-1810. Koester, C. J.; Hites, R. A. Environ. Sci. Technol. 1992, 26, 502507. Eitzer, B. D.; Hites, R. A. Environ. Sci. Technol. 1989, 23, 13891395. Prinn, R. G.; Weiss, R. F.; Miller, B. R.; Huang, J.; Alyea, F. N.; Cunnold, D. M.; Fraser, P. J.; Hartley, H.; Simmonds, P. G. Science 1995, 259, 187-192. Atkinson, R. Issues in Environmental Science and Technology; The Royal Society of Chemistry: Cambridge, U.K., 1996. Muller, J. Sci. Total Environ. 1984, 36, 339-346. Gurprasad, N.; Constable, M.; Haidar, N.; Cabalo, E. Organohalogen Compd. 1995, 24, 501-504. Hagenmaier, H.; Berchtold, A. Chemosphere 1986, 15, 19911994. Hagenmaier, H.; Brunner, H. Chemosphere 1987, 16, 17591764. Wild, S. R.; Harrad, S. J.; Jones, K. C. Chemosphere 1992, 24, 833-845. Mackay, D.; Shiu, W. Y.; Ma, K. C. Illustrated Handbook of Physical-Chemical Properties and Environmental Fate for Organic Chemicals: Vol. II: Lewis Publishers: Chelsea, MI, 1992. Mackay, D.; Shiu, W. Y.; Ma, K. C. Illustrated Handbook of Physical-Chemical Properties and Environmental Fate for Organic Chemicals: Vol. V; Lewis Publishers: Chelsea, MI, 1992. Waite, D. T.; Gurprasad, N. P.; Cessna, A. J.; Quiring, D. V. Chemosphere 1998, 27, 2251-2260. Leuenberger, C.; Ligocki, M. P.; Pankow, J. F. Environ. Sci. Technol. 1985, 19, 1053-1058. Fingler, S.; Tkalcevic, B.; Frobe, Z.; Drevenkar, V. Analyst 1994, 119, 1135-1140. Crosby, D. G.; Wong, A. S. Chemosphere 1976, 5, 327-332. Vollmuth, S.; Zaic, A.; Niessner, R. Environ. Sci. Technol. 1994, 28, 1145-1149.

VOL. 34, NO. 14, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2885

(42) Waddell, D. S.; Bonek-Oclesa, Gobran, T. Organohalogen Compd. 1995, 23, 407-412. (43) Piccinini, P.; Pichat, P.; Guillard, C. J. Photochem. Photobiol. A 1998, 119, 137-142. (44) Harte, J. Consider a Spherical Cow: A Course in Environmental Problem Solving; University Science Books: Mill Valley, CA, 1988. (45) Seinfeld, J. H.; Pandis, S. N. In Atmospheric Chemistry and Physics: From Air Pollution to Climate Change; John Wiley & Sons: New York, 1998.

2886

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 14, 2000

(46) Koester, C. J.; Hites, R. A. Environ. Sci. Technol. 1992, 26, 13751382. (47) Eitzer, B. D.; Hites, R. A. Environ. Sci. Technol. 1989, 23, 13961402.

Received for review November 3, 1999. Revised manuscript received February 24, 2000. Accepted March 2, 2000. ES9912325