3. SOURCES 3-1

3.1. OVERVIEW OF SOURCES 3-1

3.2. PULP AND PAPER MILLS 3-14

3.2.1. Bleached Chemical Wood Pulp and Paper Mills 3-14

3.2.2. Nonchemical and Nonwood Pulping and Bleaching Mills 3-18

3.2.3. Ongoing Regulatory Investigations 3-19

3.3. PUBLICLY OWNED TREATMENT WORKS (POTWs) 3-20

3.3.1. Sources of CDDs/CDFs 3-20

3.3.2. Releases of CDDs/CDFs 3-22

3.4. CHEMICAL MANUFACTURING AND PROCESSING SOURCES 3-25

3.4.1. Manufacture of Halogenated Organic Chemicals - Overview 3-25

3.4.1.1. Chlorophenols 3-25

3.4.1.2. Chlorobenzenes 3-28

3.4.1.3. Chlorobiphenyls 3-33

3.4.1.4. Aliphatic Chlorine Compounds 3-33

3.4.1.5. Dyes and Pigments 3-35

3.4.2. Manufacture of Halogenated Organic Chemicals - Dioxin/Furan Test Rule Data 3-37

3.4.3. Manufacture of Halogenated Organic Chemicals-Pesticide Data Call-In 3-44

3.4.4. Chlorine Production Using Graphite Electrodes 3-56

3.4.5. Petroleum Refining Catalyst Regeneration 3-59

3.4.6. Additional Chemical Manufacturing and Processing Sources 3-62

3.5. MECHANISMS OF FORMATION OF DIOXIN-LIKE COMPOUNDS DURING COMBUSTION OF ORGANIC MATERIALS 3-63

3.5.1. CDD/CDF Contamination in Fuel as a Source of Combustion Stack Emissions 3-64

3.5.2. Formation of CDDs/CDFs from Precursor Compounds 3-67

3.5.3. The de novo Synthesis of CDDs/CDFs During Combustion of Organic Materials 3-75

3.5.4 Theory on the Emission of Polychlorinated Biphenyls 3-91

3.5.5. Evaluation of Naturally Occurring CDD/CDFs by Examination of Sediment Core Data 3-92

3.5.6. Summary of Theories of CDD/CDF Emissions 3-94

3.6. COMBUSTION AND OTHER HIGH TEMPERATURE SOURCES 3-96

3.6.1. Municipal Solid Waste Incineration 3-97

3.6.2. Hazardous Waste Incineration 3-109

3.6.3. Medical Waste Incineration 3-112

3.6.4. Kraft Black Liquor Recovery Boilers 3-119

3.6.5. Sewage Sludge Incineration 3-120

3.6.6. Primary Nonferrous Metal Smelting/Refining 3-122

3.6.7. Secondary Nonferrous Metal Smelting/Refining 3-123

3.6.7.1 Secondary Aluminum Smelters and Refiners 3-124

3.6.7.2 Secondary Copper Smelters and Refiners 3-124

3.6.7.3 Secondary Lead Smelters and Refiners 3-125

3.6.8. Primary Ferrous Metal Smelting/Refining 3-128

3.6.9. Secondary Ferrous Metal Smelting/Refining 3-129

3.6.10. Scrap Electric Wire Recovery 3-129

3.6.11. Drum and Barrel Reclamation and Incineration 3-131

3.6.12. Tire Combustion 3-132

3.6.13. Motor Vehicle Fuel Combustion 3-134

3.6.14. Wood Burning at Residences 3-143

3.6.15. Industrial Wood-Burning Facilities 3-146

3.6.16. Wood Burned in Forest Fires 3-147

3.6.17. Coal Combustion 3-151

3.6.18. Combustion of Polychlorinated Biphenyls (PCBs) 3-152

3.6.19. Pyrolysis of Brominated Flame Retardants 3-153

3.6.20. Carbon Reactivation Furnaces 3-154

3.6.21. Cement Kilns 3-156

3.6.22. Additional Combustion and High Temperature Sources 3-165

3.7. RESERVOIR SOURCES 3-166

3.8. COMPARING SOURCE EMISSIONS TO DEPOSITION ESTIMATES 3-167

The purpose of this chapter is twofold: (1) to identify sources that release dioxin-like compounds into the environment and (2) to derive national estimates for releases from these sources in the United States. The dioxin-like compounds have been found in all media and all parts of the world. This ubiquitous nature of these compounds suggests that multiple sources exist and that long range transport can occur. An unresolved issue is how the relative impacts from local vs. distant sources compare at a particular location. Presumably, in industrial areas, local sources will dominate, and in rural areas, distant sources will dominate. However, site specific considerations such as stack height, wind patterns, magnitude of local sources, etc. could influence these comparisons.

The chlorinated and brominated dioxins and furans have never been intentionally produced other than on a laboratory-scale basis for use in chemical analyses. Rather, they are generated as byproducts from various combustion and chemical processes. PCBs were produced in relatively large quantities for use in such commercial products as dielectrics, hydraulic fluids, plastics, and paints. They are no longer produced in industrialized countries, but continue to be released to the environment through the use and disposal of these products.

Dioxin-like compounds are released to the environment in a variety of ways and in varying quantities depending upon the source. For example,

• Releases to the air occur primarily from combustors and appear to have the most direct influence on human exposure. As discussed in Chapter 4, atmospheric deposition and subsequent accumulation through the food chain appears to be the major pathway of human exposure to dioxin-like compounds.

• Solid residues such as combustor ash, still bottoms, etc. can contain high levels of CDD/CDFs and can collectively contain more of these compounds than are found in air or water discharges. However, these solid residues are not generally released to the environment in an uncontrolled manner. Rather, they are usually disposed at secure landfills and any leaching to ground water is minimal due to their very low water solubility.

• Water discharges from paper mills, sewage treatment plants, and possibly other industries can contain low levels of CDD/CDFs. These releases can bioaccumulate via the aquatic food chain and ultimately lead to human exposure via fish ingestion.

The major identified sources of environmental release have been grouped into four major types for the purposes of this report:

• Industrial/Municipal Processes: Dioxin-like compounds can be formed through the chlorination of naturally occurring phenolic compounds such as those present in wood pulp. The formation of CDDs and CDFs resulting from the use of chlorine bleaching processes in the manufacture of bleached pulp and paper has resulted in the presence of CDDs and CDFs in paper products as well as in liquid and solid wastes from this industry. Municipal sewage sludge has been found to frequently contain CDDs and CDFs. Influents from industrial facilities, stormwater runoff, microbial metabolism of chlorophenols, and domestic household wastewater have been identified by various researchers as the sources(s) of the CDDs/CDFs.

• Chemical Manufacturing/Processing Sources: Dioxin-like compounds can be formed as by-products from the manufacture of chlorine and such chlorinated compounds as chlorinated phenols, PCBs, phenoxy herbicides, chlorinated benzenes, chlorinated aliphatic compounds, chlorinated catalysts, and halogenated diphenyl ethers. Although the manufacture of many chlorinated phenolic intermediates and products, as well as PCBs, was terminated in the late 1970s in the United States, continued, limited use and disposal of these compounds can result in releases of CDDs, CDFs, and PCBs to the environment. High levels of CDFs have been found in sludge from graphite electrodes used in chloralkali process to manufacture chlorine.

• Combustion and Incineration Sources: Dioxin-like compounds can be generated and released to the environment from various combustion processes when chlorine donor compounds are present. These sources can include incineration of wastes such as municipal solid waste, sewage sludge, hospital, and hazardous wastes; metallurgical processes such as high temperature steel production, smelting operations, and scrap metal recovery furnaces; and the burning of coal, wood, petroleum products, and used tires for power/energy generation.

• Reservoir Sources: The persistent and hydrophobic nature of these compounds causes them to accumulate in soils, sediments, and organic matter and to persist in waste disposal sites. The dioxin-like compounds in these "reservoirs" can be redistributed by various processes such as dust or sediment resuspension resulting in the potential for exposure. Releases from these "reservoirs" are not original sources in a global sense, but can be on a local scale. For example, past air emissions causing deposition onto a watershed with subsequent erosion may have resulted in accumulation in downstream sediments. Future sediment dredging operations could result in short-term significant resuspension of dioxins that had accumulated over a much longer period of time. Similarly, leaf composting operations could lead to releases of the dioxins that had, over the course of a growing season, deposited on or been sorbed to the leaves of deciduous trees in an area. Such leaf reservoirs could also be resuspended during forest fires.

As awareness of these possible sources has grown in recent years, a number of changes have occurred that should reduce the release rates (Rappe, 1992a). For example, releases of dioxin-like compounds have been reduced due to the switch to unleaded automobile fuels (and associated use of catalytic converters and reduction in halogenated scavenger fuel additives), process changes at pulp and paper mills, new emission standards and upgraded emission controls for incinerators, and reductions in the manufacture of chlorinated phenolic intermediates and products.

Some investigators have raised the possibility that major sources exist that have not yet been identified. This suggestion is acknowledged to be quite speculative, but is important to consider. Three studies addressing this issue are summarized below.

Travis and Hattemer-Frey (1991) used the Fugacity Food Chain (FFC) model to predict the contribution of municipal solid waste incinerators, motor vehicles, hospital waste incinerators, residential wood burning, and pulp and paper mill effluents, to the U.S. total environmental input of 2,3,7,8-TCDD. It was estimated that the total input from all five sources combined accounted for only 11 percent of the total 2,3,7,8-TCDD found in the different media in the United States. The authors concluded that this low value indicated: (1) the source term used in the FFC modeling exercise for 2,3,7,8-TCDD may have been too high; (2) some unidentified major source(s) of 2,3,7,8-TCDD exist; or (3) multiple environmental sources of 2,3,7,8-TCDD with no one source dominating the total input.

Rappe (1991) found discrepancies between estimated emissions from known sources of CDDs and CDFs into the Swedish environment and calculated aerial deposition rates. The total emissions in Sweden were estimated by Rappe (1991) as 100 to 250 g TEQ/yr. The deposition rates used by Rappe (1991) were derived from a study by Marklund (1990) who made measurements in rural areas of Sweden and found an average deposition rate of 5 ng of TEQ/m² - yr. [Later measurements by Andersson et al. (1992) indicate that deposition in this area has been reduced to about 1 ng of TEQ/m²-yr due to emissions reductions.] Rappe (1991) multiplied the deposition rate of 5 ng TEQ/m²-yr by the total land area of Sweden, yielding a total annual deposition for Sweden of 2,250 g of TEQ/yr, which appears to be 10 to 20 times higher (or 2 to 4 times higher using the deposition rate of Andersson et al. [1992]) than the total emissions from sources originating in Sweden. Possible explanations for this discrepancy are (1) uncertainty in the emission estimates, (2) uncertainty in the deposition estimates, (3) long-range transport of dioxin-like compounds from sources outside of Sweden, or 4) existence of unidentified sources. In an earlier publication, Rappe et al. (1987) compared congener patterns found in human and aquatic life tissue samples with the congener patterns found in various known emission sources and contaminated products. A poor correlation was observed between the congener patterns found in human and environmental samples and the respective potential sources. Rappe et al. (1987) speculated that the observed pattern in human and environmental samples could be the result of a combination of sources, coupled with environmental and biological degradation of the released congeners.

Harrad et al. (1992a; 1992b) have made similar estimates for the United Kingdom. They estimate that the average annual deposition from the atmosphere to the land surface in the United Kingdom is 250 kg of CDD/CDF (on a total mass basis, not TEQ), compared to about 29.1 kg/yr emitted from known sources. As with the other two studies, these discrepancies could be the result of inaccuracies in emission/deposition estimates, long- range transport from outside the country, or unidentified sources. The authors speculated that much of the discrepancy may be accounted for by secondary or "reservoir" sources (i.e. the remobilization and subsequent redeposition of CDD/CDFs already in the environment).

Table 3-1 presents CDD and CDF source-specific air emission estimates reported for West Germany (Fiedler and Hutzinger, 1992), Austria (Riss and Aichinger, 1993), the United Kingdom (ECETOC, 1992), the Netherlands (Koning et al., 1993), Switzerland (Schatowitz et al., 1993), and the United States (based on estimates generated in this document). The emission estimates for West Germany and Switzerland suggest that municipal waste incinerators and metal smelters/refiners are the largest sources of air emissions. In Austria, domestic combustion of wood is believed to be the largest source followed by emissions from the metallurgical industry. In the United Kingdom, municipal waste incinerators and coal combustion are estimated to be the major sources. Municipal waste incinerators are also estimated to be the largest source in the Netherlands. Rappe (1992a) and Lexen et al. (1992) have identified emissions from ferrous and nonferrous metals smelting and refining facilities as potentially the largest current source in Sweden. Rappe (1992a) reported that changes in various industrial practices have led to reductions in dioxin emissions in Sweden from 400 - 600 g of TEQ/yr in 1985 to 100 - 200 g TEQ/yr in 1991.

Similar nationwide emission estimates for the United States have not previously been compiled. This task has been attempted in this document, and the results are presented in Table 3-1 (air emissions only) and in Table 3-2. Table 3-2 lists emission estimates for the major known or suspected sources that could have releases of dioxin-like compounds to the environment. For each source listed in Table 3-2, estimated emissions to air, water, land, and product are listed where appropriate and where data are adequate to enable an estimate to be made. The term "product" in Table 3-2 is defined to include substances or articles (e.g., paper pulp or sewage sludge that is distributed/marketed commercially) that are known to contain dioxin-like compounds and whose subsequent use may result in releases to the environment. Figure 3-1 is a chart that visually displays the range of emission estimates to air that are reported in Table 3-2.

In order to make each source emission estimate, information was required concerning both the "emission factor" term for the source (e.g., grams TEQ per kg of material processed) and the "production" term for the source (e.g., kg of material processed annually in the United States). Because the quantity and quality of the available information for both terms for each emission source varies considerably, a confidence rating scheme was developed. This scheme is based on a consideration of the following factors:

Basis of Estimate - The basis for the emission estimate varied widely from expert judgement to detailed studies. The best studies involved direct emission measurements at multiple facilities. The representativeness of emission samples was evaluated on the basis of the variability in technologies and associated release rates among individual facilities in the source category. The more variability among facilities; the more important it is to test multiple facilities. In other cases, although no direct emission measurements were available, estimates could be derived using indirect techniques. Obviously, these "indirect" estimates are much more uncertain than those based on direct measurements.

Citation Quality - The quality of the supporting literature varied widely. Whenever possible, only peer reviewed final reports were used. In some cases, however,

draft reports that had undergone some review were used. In a few cases, unpublished references were used such as personal communication with experts.

The confidence rating scheme, presented in Table 3-3, provides criteria for assigning a "high," "medium," or "low" confidence rating for both the emission factor and production terms. As shown in Table 3-2, confidence ratings have been assigned to each emission estimate. The first rating applies to the "production" term, and the second rating applies to the "emission factor" term. In addition to the confidence rating, the uncertainty in these national release estimates is reflected by presenting, where possible, for each source category both a central or "best guess" value and a possible range from a lower to upper estimate. These lower and upper estimates are not intended to be absolute bounds, but reasonable estimates of how much higher or lower the true value might be. Insufficient data were available to statistically derive these ranges. Therefore, a judgement-based approach was developed. This approach uses the average or best guess estimate as the central value of the range (assumed to be a geometric average) and sets the width of the range on the basis of the confidence class as follows:

· Low confidence class: upper end of range is 10 times higher than lower end;

· Medium confidence class: upper end of range is 5 times higher than lower end;

· High confidence class: upper end of range is 2 times higher than lower end.

This approach initially assumes that the range of uncertainty is symmetrical about the central value. However, in some cases it may be more reasonable to shift the uncertainty range upwards or downwards. For example, it may be reasonable to shift the range downwards in cases where there is strong evidence that upgrades have occurred since the emissions testing. Alternatively, it is possible that the range should be shifted upwards if it can be shown that the tested facilities are more representative of the low emitting facilities than the high emitting facilities. It is emphasized that these ranges should be interpreted as judgements which are symbolic of the relative uncertainty among sources, not statistical measures. The remainder of this chapter reviews the available data for estimating CDD/CDF releases from specific source categories and provides the basis for the emission estimates presented in Table 3-2 for the United States.

During 1988, EPA and the U.S. pulp and paper industry jointly conducted a survey of 104 pulp and paper mills in the United States to measure levels of dioxins in effluent, sludge, and pulp (U.S. EPA, 1990a). This study, commonly called the 104-Mill Study, was managed by the National Council of the Paper Industry for Air and Stream Improvement, Inc. (NCASI) with oversight by EPA, and included all U.S. mills where chemically produced wood pulps are bleached with chlorine or chlorine derivatives.

In 1992, the pulp and paper industry conducted its own NCASI-coordinated survey. The collected data were summarized and analyzed in a report entitled "Summary of Data Reflective of the Pulp and Paper Industry Progress in Reducing the TCDD/TCDF Content of Effluents, Pulps, and Wastewater Treatment Sludges" (NCASI, 1993). Although the report is available from NCASI, it has not been peer reviewed nor published in an independent journal. The data used in the report were provided by individual pulp and paper companies and neither NCASI nor EPA can vouch for the accuracy or representativeness of the data. However, NCASI (1993) reports that the pulp and paper industry has taken numerous steps to reduce CDD/CDF releases since 1988, and that NCASI considers the 1992 survey to be more reflective of current conditions than the data generated in the 104-Mill Study (U.S. EPA, 1990a).

As part of its ongoing efforts to develop revised effluent guidelines and standards for the pulp, paper, and paperboard industry, EPA recently published the Development Document for the guidelines and standards being proposed for this industry (U.S. EPA, 1993d). The Development Document presents estimates of the 2,3,7,8-TCDD and 2,3,7,8-TCDF annual discharges in wastewater from the mills in this industry as of January 1, 1993. EPA used the most recent information about each mill from four data bases (104-Mill Study, EPA short-term monitoring studies at 13 mills, EPA long-term monitoring studies at 8 mills, and industry self-monitoring data submitted to EPA) to estimate these discharges. The 104-Mill Study data were used only for those mills that did not report making any process changes subsequent to the 104-Mill Study and did not submit any more recent effluent monitoring data. For the purpose of this report, the release estimates from NCASI (1993) and U.S. EPA (1990a) are presented to show the possible range of releases within recent years, but the U.S. EPA (1993d) estimates are believed to be most reflective of current conditions.

NCASI (1993) found that less than 10 percent of mills had 2,3,7,8-TCDD and 2,3,7,8-TCDF concentrations in effluent above the nominal detection limits of 10 ppq and 100 ppq, respectively. Similar results were obtained in the short- and long-term sampling reported for 18 mills in U.S. EPA (1993d). 2,3,7,8-TCDD was detected at four mills, and 2,3,7,8-TCDF was detected at nine mills. Wastewater sludges at most mills (75 to 90 percent) were reported by NCASI (1993) to contain less than 10 ppt of 2,3,7,8-TCDD and less than 100 ppt of 2,3,7,8-TCDF. U.S. EPA (1993d) reported similar results but did find detectable levels of 2,3,7,8-TCDD and 2,3,7,8-TCDF in sludges from 64 percent and 85 percent of the facilities sampled, respectively. NCASI (1993) reported that nearly 90 percent of the bleached pulps contained less than 2 ppt of 2,3,7,8-TCDD and less than 20 ppt of 2,3,7,8-TCDF. The final levels in white paper products would correspond to levels in bleached pulp, so bleached paper products would also be expected to contain less than 2 ppt of 2,3,7,8-TCDD. Overall, NCASI (1993) reports a 90 percent reduction in TEQ generation from 1988 to 1992.

The 104-Mill Study and the NCASI study measured only 2,3,7,8-TCDD and 2,3,7,8-TCDF because these two congeners are the primary contributors (90 percent or more) to the TEQ total found in pulp, sludge, and effluent (U.S. EPA, 1990b). Ninety-four mills participated in the NCASI study, and the remaining 10 (of 104) were assumed by NCASI to be operating at the same levels as measured in the 1988 104 Mill Study. All not detected values were counted as half the detection limit. If detection limits were not reported, they were assumed to be 10 ppq for effluent and 1 ppt for sludge or bleached pulp.

The U.S. annual discharge rates of 2,3,7,8-TCDD, 2,3,7,8-TCDF, and TEQs due to these two compounds are summarized in Table 3-4 for each study. As stated previously, the 1993 discharge estimate for effluent (U.S. EPA, 1993d) is believed to be the best estimate of current emissions. During the period between the conduct of the 104 Mill Study and the issuance of the U.S. EPA Development Document (U.S. EPA, 1993d), the U.S. pulp and paper industry has reduced releases of CDD/CDFs primarily by instituting numerous process changes to reduce the formation of CDD/CDFs during the production of chemically bleached wood pulp. U.S. EPA (1993d) did not provide extensive sampling of sludge and pulp samples from bleached chemical wood pulp and paper mills comparable to that provided for effluents. However, because most of the reduction between 1988 and 1993 can be attributed to process changes of a pollution prevention nature, the percentage reduction observed in effluent emissions (from 356 g TEQ/yr to 105 g TEQ/yr or 70 percent reduction) is likely representative of the reduction that has been achieved in sludge and pulp emissions over this same time period. Table 3-4 presents best estimates of emissions in sludge and pulp of 100 g TEQ/yr and 150 g TEQ/yr, respectively, using this assumption. The confidence ratings for these release estimates were judged to be H/H based on the fact that direct measurements have been made at virtually all facilities, indicating a high level of confidence in both the production and emission factor estimates. Based on these high confidence ratings, the estimated ranges of potential annual emissions for effluent, sludge, and pulp are assumed to vary by a factor of 2 between the low and high ends of the ranges. Assuming that the best estimates of annual emissions (i.e., the 1993 discharge-based estimates presented in Table 3-4) are the geometric means of these ranges, then the ranges are calculated to be 74 to 150 g TEQ/yr for effluent, 71 to 140 g TEQ/yr for sludge, and 105 to 210 g TEQ/yr for pulp.

In 1990, approximately 20.5 percent or 500 million dry kg of the pulp and paper mill wastewater sludge generated by facilities employing chlorine bleaching of pulp were incinerated (U.S. EPA, 1993e). The majority of the wastewater sludge generated by these facilities is landfilled or placed in surface impoundments (79.5 percent) with the remainder incinerated (20.5 percent), applied to land directly or as compost (4 percent), or distributed as a commercial product (less than 1 percent) (U.S. EPA, 1993e). Black liquor recovery boilers used in the Kraft process for the production of paper pulp are potential sources of

CDDs/CDFs. Estimates of potential CDD/CDF emissions to air from these sources are discussed in Section 3.6.4.

Although the EPA Office of Water does not believe that secondary fiber mills (i.e., mills using recycled paper as a source of pulp) are significant sources of CDDs and CDFs, EPA is considering whether to establish effluent limitations guidelines and standards for CDD/CDFs for these mills based primarily upon data generated for the Development Document (U.S. EPA, 1993d). These data, collected by EPA or provided to EPA by industry, indicate detectable levels of 2,3,7,8-TCDD in the effluents of 2 of the 12 mills with reported monitoring data and detectable levels of 2,3,7,8-TCDF in the effluents of 4 of the 7 mills with data (U.S. EPA, 1993d).

Data on the presence of more chlorinated (i.e., penta-through octachlorinated) CDDs and CDFs in the effluents of these facilities were not generated for the Development Document (U.S. EPA, 1993d). However, Berry et al. (1993) reports that trace levels of these higher chlorinated homologs were commonly observed in the effluents from Canadian pulp mills that use recycled paper for fiber furnish (i.e., the raw materials used to manufacture pulp) and/or that do not practice chlorine bleaching. Similar results were reported by Rappe et al. (1990). The congener profile observed is not dominated by the tetra-CDDs/CDFs, as is the case with bleach plant wastewater, but rather by the higher chlorinated congeners more consistent with the congener profile found in ambient air, soil, and adipose tissue. These results lead to the hypothesis that paper and paperboard products, during their useful life, can accumulate trace amounts of CDD/CDFs from the ambient environment.

As a step in evaluating this hypothesis, Berry et al. (1993) analyzed the CDD/CDF content of pulp and paper samples from Canadian mills that use neither chlorine-containing bleaching compounds nor fibers that have been bleached with chlorine-containing compounds as well as papers from mills that use recycled paper as a furnish. All samples analyzed had detectable levels of one or more CDD/CDFs. The congener profiles of the samples were similar with the higher chlorinated congeners dominating in terms of concentration. The order of degree of contamination on a TEQ basis, from high to low, is recycled linerboard (1 sample--2.5 ng TEQ/kg) > "totally chlorine-free" bleached kraft paper (1 sample--0.35 ng TEQ/kg) > pulp from de-inked recycled paper (1 sample--0.19 ng TEQ/kg) > newsprint (17 samples--mean = 0.07 ng TEQ/kg) > unbleached kraft paper (2 samples--mean = 0.02 ng TEQ/kg). Rappe et al. (1990) also reported finding higher levels of CDD/CDFs, particularly the hepta- and octa-chlorinated congeners, in recycled paper pulps than in virgin bleached and unbleached pulps. Based on the results of their study, Berry et al. (1993) concluded that, although it may be possible to produce a dioxin-free pulp, it is likely that all papers will become contaminated during their first life cycle by contact with dioxin-laden dust, and contamination is inevitable if they are recycled multiple times.

The U.S. EPA is currently under court order to develop revised effluent guidelines (i.e., Best Available Technology and Pretreatment Standards for Existing Sources) for the chemical pulping and bleaching subcategories of the pulp and paper industry. These revised effluent guidelines and standards which address control of CDDs and CDFs from bleached chemical wood pulp and paper mills were proposed by EPA on December 17, 1993 (Federal Register, 1993a). In addition, the Clean Air Act Amendments of 1990 require EPA to promulgate Most Achievable Control Technology (MACT) standards for hazardous air pollutants from this industry by 1997. To that end, the Office of Air and Radiation, in coordination with the Office of Water, proposed control technology standards for non-combustion sources on December 17, 1993, (Federal Register, 1993a) and will propose control technology standards for combustion sources by October 1994 with promulgation of both by September 1995 (U.S. EPA, 1992d).

Based on the results of an in-depth risk assessment, EPA's Office of Solid Waste concluded that dioxin contained in pulp and paper mill sludges does not pose an unreasonable probability of adverse effects on human health and the environment when disposed in landfills and surface impoundments and that further regulation of these facilities under Subtitle D of the Resource Conservation and Recovery Act (RCRA) to reduce potential dioxin-related risks was not warranted (U.S. EPA, 1991a).

However, EPA did find that improper land application of pulp and paper mill sludge for soil conditioning purposes can pose a significant risk to wildlife. In 1991, EPA proposed a regulation under the Toxic Substances Control Act (TSCA) to limit the concentration of CDDs/CDFs in soil conditioned with sludge and also to establish site management practices for land application of the sludge. EPA deferred finalizing the rule until issuance of the final integrated regulations for effluent guidelines and MACT standards. These regulations could make TSCA rulemaking unnecessary. In the interim, EPA is negotiating a voluntary agreement with the American Forest and Paper Association to establish CDD/CDF standards and management practices for the use of sludge as a conditioner (U.S. EPA, 1993b).

CDD/CDFs have been measured in sewage sludge, though the origins have not been well established. In fact, Oberg et al. (1992) reported that low levels of HpCDDs and OCDD are formed, probably as a result of microbial action, in aerated sewage sludge spiked with pentachlorophenol. Potential sources of the CDD/CDFs include industrial inputs, runoff to sewers from lands or urban surfaces contaminated by product uses or deposition of emissions from combustion sources, household wastewater, chlorination operations within the wastewater treatment facility, or a combination of all the above (Rappe, 1992a; Rappe et al., 1989; Horstmann et al., 1992). The major source(s) for a given treatment plant is likely to be site-specific. For example, Rieger and Ballschmiter (1992) traced the origin of CDDs and CDFs found in municipal sewage sludge in Ulm, Germany, to metal manufacturing and urban sources. The characteristics of both sources were similar and suggested generation via thermal processing. The presence of CDD/CDFs in sewage sludge suggests that CDD/CDFs may also be present in the wastewater effluent discharges of POTWs; however, no published studies reporting the results of effluent analyses for CDD/CDFs could be found.

In a series of recent studies, Horstmann et al. (1992; 1993a; 1993b) and Horstmann and McLachlan (1994) demonstrated that wastewater from household washing machines could be the major source at many, if not all, POTWs that serve primarily residential populations. Horstmann et al. (1992) provided initial evidence that household wastewater could be a significant source. Horstmann et al. (1993a) measured CDD/CDF levels in the effluent from four different loads of laundry from two different domestic washing machines. The concentrations of total CDD/CDF in the four samples ranged from 3,900 to 7,100 pg/L and were very similar in congener profile with OCDD being the dominant congener followed by the hepta- and hexa-CDDs. Based on the similar concentrations and congener profiles found, Horstmann et al. (1993a) concluded that the presence of CDD/CDF in washing machine wastewater is widespread. A simple mass balance performed using the results showed that the CDD/CDFs found in the four washing machine wastewater samples could account for 27 to 94 percent of the total CDD/CDF measured in the sludge of the local wastewater treatment plant (Horstmann and McLachlan, 1994).

Horstmann et al. (1993a) also performed additional experiments that showed that detergents, commonly used bleaching agents, and the washing cycle process itself were not responsible for the observed CDD/CDFs. Rappe and Andersson (1992) had previously reported that wastewater from clothing and dish washing machines in which sodium hypochlorite-containing detergents were used contained low levels of CDD/CDFs.

To determine if the textile fabric or fabric finishing processes could account for the observed CDD/CDFs, Horstmann et al. (1993b) analyzed the CDD/CDF content of eight different raw (unfinished) cotton cloths containing fiber from different countries and five different white synthetic materials (acetate, viscose, bleached polyester, polyamide, and polyacrylic). The maximum concentrations found in the textile fabrics were 30 ng/kg in the cotton products and 45 ng/kg in the synthetic materials. Also, a cotton finishing scheme was developed in which one of the cotton materials was subjected to a series of 16 typical cotton finishing processes; one sample was analyzed following each step. The fabric finishing processes showing the greatest effect on CDD/CDF concentration were the application of an indanthrene dye and the "wash and wear" finishing process which together resulted in a CDD/CDF concentration of about 100 ng/kg. Based on the concentrations found, the authors concluded that neither unfinished new fabrics nor common cotton finishing processes can explain the CDD/CDF levels found in wastewater.

Horstmann and McLachlan (1994) analyzed 35 new textile samples, primarily cotton products, for CDD/CDFs. Low levels were found in many cases (total CDD/CDF less than 50 ng/kg). However, several colored T-shirts from a number of clothing producers had extremely high levels, with concentrations up to 290,000 ng/kg. Because the concentrations in identical T-shirts purchased at the same store varied by up to a factor of 20, the authors concluded that the source of CDD/CDFs is not a textile finishing process because a process source would have resulted in a more consistent level of contamination.

Horstmann and McLachlan (1994) conducted additional experiments that demonstrated that the small percentage of clothing items with high CDD/CDF levels could be responsible for the quantity of CDD/CDFs observed in household wastewater and sewage sludge. They were able to demonstrate that the CDD/CDFs can be gradually removed from the fabric during washing, can be transferred to the skin and subsequently transferred back to other textiles and then washed out, or can be transferred to other textiles during washing and then removed during subsequent washings.

EPA conducted the National Sewage Sludge Survey in 1988 to obtain national data on sewage sludge quality and management. As part of this survey, EPA analyzed sludges from 175 POTWs for CDD/CDF content; sludges from 15 of the POTWs had detectable levels of 2,3,7,8-TCDD. All sludges had detectable levels of at least one CDD/CDF congener (Rubin and White, 1992). TEQ concentrations ranged from 0.7 to 1,816 ng TEQ/kg dry weight. If all not detected values are assumed to be zero, then the mean and median concentrations are 50 and 9 ng TEQ/kg, respectively. If the not detected values are set equal to the detection limit, then the mean and median concentrations are 86 and 50 ng TEQ/kg, respectively (Rubin and White, 1992).

Approximately 5.4 million dry metric tons of sewage sludge are estimated by EPA to be generated annually in the United States (Federal Register, 1993b). Table 3-5 lists the volume of sludge disposed annually by use and disposal practices. Table 3-5 also lists the estimated amount of TEQs that may be present in sewage sludge and potentially be released to the environment. These values were estimated using the mean TEQ concentration value (not detected values assumed to be zero) reported by Rubin and White (1992) (i.e., 50 ng TEQ/kg). Multiplying this mean concentration by the sludge volumes generated, yields an annual potential total release of 208 grams of TEQ for nonincinerated sludges. Of this 208 grams of TEQ, 3.6 grams enter commerce as a product for distribution and marketing. The remainder is applied to land or is landfilled.

This release estimate is assigned a H/H confidence rating indicating high confidence in both the production and emission factor estimates. The high rating was based on the judgement that the 175 tested facilities were reasonably representative of the variability in the POTW technologies and sewage characteristics. Based on this high confidence rating, the estimated range of potential annual emissions is assumed to vary by a factor of 2 between the low and high ends of the range. Assuming that the best estimate of annual emission to land (105 g TEQ/yr) is the geometric mean of this range, then the range is calculated to be 145 to 290 g TEQ/yr. Assuming that the best estimate of 3.6 g TEQ annual emissions in product (i.e., the fraction of sludge that is distributed and marketed as a product) is the geometric mean of the range, then the range is calculated to be 2.5 to 5.0 g TEQ/yr.

An additional 10 to 52 grams of TEQ (central estimate of 23 g TEQ/yr) are estimated to be released to the atmosphere annually by the incineration of sewage sludge. The basis of these incineration release estimates is presented in Section 3.6.5. It is interesting to note that CDDs and CDFs detected in ambient air in Ohio have been linked to sewage sludge combustion (Edgerton et al., 1989). In this study, total CDD/CDF in ambient air ranged from 1,900 to 9,900 fg/m³; no 2,3,7,8-TCDD was detected in any of the samples with a detection limit of less than 240 fg/m³.

Several chemical production processes have been shown to generate CDDs and CDFs (Versar, 1985; Hutzinger and Fiedler, 1991a). CDDs and CDFs can be formed during the manufacture of chlorophenols, chlorobenzenes, and chlorobiphenyls (Versar, 1985; Ree et al., 1988). Consequently, disposal of industrial wastes from manufacturing facilities producing these compounds may result in the release of CDDs and CDFs to the environment. Also, the products themselves may contain these compounds, and when used/consumed, may result in additional releases to the environment. CDD and CDF congener distribution patterns indicative of noncombustion sources have been observed in sediments in southwest Germany and the Netherlands. The congener patterns found suggest that wastes from the production of chlorinated organic compounds may be important sources of CDD and CDF contamination in these regions (Ree et al., 1988). The production and use of many of the chlorophenols, chlorophenoxy herbicides, and PCB products have been banned or strictly regulated in most countries. However, these products may have been a source of the environmental contamination that occurred prior to the 1970s and may continue to be a source of environmental releases based on limited use and disposal conditions (Rappe, 1992a).

The two major manufacturing processes used to produce chlorophenols include: (1) electrophilic chlorination of phenol by chlorine gas in the presence of catalytic amounts of aluminum chloride and organic chlorination promoters and stabilizers; and (2) alkaline hydrolysis of chlorobenzenes using aqueous methanolic sodium hydroxide and heat (Ree et al., 1988). CDD and CDF formation is promoted by the high temperatures and/or alkaline conditions used in these processes. CDDs and CDFs may be formed by nucleophilic substitution, radical reactions, and pyrolysis mechanisms (Versar, 1985; Ree et al., 1988). The major CDD/CDF congeners generated by chlorophenol manufacture are the hexa- through octa-chlorinated congeners (Versar, 1985).

The concentrations of CDD/CDFs in chlorophenols analyzed in the 1970s and early 1980s were assembled and summarized by Versar (1985) and Hutzinger and Fiedler (1991a). Hagenmaier and Brunner (1987) reported the results of analyses of four pentachlorophenol products commercially available during the late 1980s; the total TEQ concentrations in these four products ranged from 0.08 to 2.32 mg/kg. Table 3-6 presents a summary of the data from these three studies. No more recent data on concentrations of CDDs and CDFs in chlorophenols could be found in the literature. However, the mono- through tetra- substituted chlorophenols and bromophenols are subject to reporting under the Dioxin/Furan Test Rule (discussed in Section 3.4.2) and/or the Dioxin/Furan Pesticide Data Call-In. (See Section 3.4.3.) CDDs and CDFs have also been found in numerous chlorophenol-based biocides according to Versar (1985) and Hutzinger and Fiedler (1991a). (See Section 3.4.3 for information on current EPA efforts to obtain data on contamination levels in pesticides.)

Several studies have provided evidence of localized environmental contamination resulting from the production or use of chlorophenols. For example, Tong et al. (1990) observed that sediment samples collected from a site near a chemical manufacturing facility where 2,4,5-T had been synthesized were highly contaminated with CDDs and CDFs. In addition, the CDD and CDF congener distribution pattern in the sediment was similar to that of 2,4,5-T, suggesting the manufacture found in 2,4,5-T as a primary source of contamination.

As indicated in Table 3-6, pentachlorophenol (PCP) products have been reported to be the most contaminated chlorophenol products. The major congener found in PCP is OCDD, but lower chlorinated congeners are also found (Rappe et al., 1987; Hutzinger and Fiedler, 1991a). High levels of CDD/CDFs have also been found in sludges from the production of PCP (Versar, 1985; Hutzinger and Fiedler, 1991a). McKee et al. (1990) surveyed harbor sediments adjacent to a wood preserving plant in Ontario, Canada, that uses PCP and creosote. Sediments were contaminated with hexa-, hepta-, and octa-chlorinated CDD/CDFs. The highest levels observed were: 5.7 ng/g HxCDD, 320 ng/g HpCDD, 980 ng/g OCDD, 6.5 ng/g HxCDF, and 53 ng/g HpCDF for a site 13 meters from the facility's dock and 400 ng/g OCDF for a site 78 meters from the dock. CDD/CDFs have also been found in composts from a yard waste composting facility in the United Kingdom (Harrad et al., 1991). Past use of PCP-based biocides was suggested as the major source of contamination, based on isomer patterns and empirical evidence.

In the mid-1980s, EPA's Office of Solid Waste promulgated land disposal restrictions on wastes (i.e., wastewaters and non-wastewaters) resulting from the manufacture of chlorophenols (40 CFR 268). Table 3-7 lists all solid wastes in which CDDs and CDFs are regulated as hazardous constituents by EPA, including chlorophenol wastes. The regulations prohibit the land disposal of these wastes until they have been treated to a level below the routinely achievable detection limit of 1 ppb in the waste extract for each of the following congener groups: TCDDs, PeCDDs, HxCDDs, TCDFs, PeCDFs, and HxCDFs (standards for waste code F039 apply only to TCDDs and TCDFs). The treatment standard of 1 ppb is based on incineration to 99.9999 percent destruction

and removal efficiency. Section 3.4.3 of this report describes regulatory actions taken by EPA to control the manufacture and use of chlorophenol-based pesticides.

EPA's Office of Water has promulgated effluent limitations for facilities that manufacture chlorinated phenols and discharge treated wastewater (40 CFR 414.70). Although these effluent limitations do not specifically address CDDs and CDFs, the treatment processes required to control the chlorinated phenols that are regulated (2-chlorophenol and 2,4,-dichlorophenol) are expected to control releases of CDDs and CDFs to minimal levels. The effluent limitations for the individual regulated chlorinated phenols are less than or equal to 39 µg/l for facilities that utilize biological end-of-pipe treatment.

Chlorobenzenes are manufactured by electrophilic substitution reactions of gaseous chlorine and benzene (Ree et al., 1988). CDD/CDFs may form during the production of these chemicals, but with less probability than in chlorophenol manufacturing (Hutzinger and Fiedler, 1991a). CDD/CDFs form by nucleophilic substitution and pyrolysis mechanisms (Ree et al., 1988). The factors contributing to the production of CDD/CDFs are: (1) using oxygen as a nuclear substituent; (2) producing or purifying the substance under alkaline conditions; and (3) using reaction temperatures above 150° C (Hutzinger and Fiedler, 1991a).

The concentrations of CDD/CDFs found in single samples of chlorobenzenes by researchers in Germany (Hagenmaier and Brunner, 1987; Hutzinger and Fiedler, 1991a) are listed in Table 3-8. In di-, tri-, tetra-, and penta-chlorobenzene, CDD/CDFs have been detected in the sub-µg/kg range. In hexachlorobenzene, CDD/CDFs have been detected in the µg-mg/kg range. No more recent data on concentrations of CDDs and CDFs in chlorobenzenes could be found in the literature. The limited available published information on CDD/CDF concentrations in chlorobenzene products is not sufficient in quantity (i.e., number of samples) or in detail (i.e., congener-specific results) to enable a reliable estimate to be made of the mass of CDDs/CDFs present in chlorobenzene products even though reliable annual production volume information is available for some products (e.g., 107,526 metric tons of monochlorobenzene and 63,104 metric tons of dichlorobenzene were produced in the United States in 1990) (U.S. ITC, 1991). However, the mono-, di-, and trichlorobenzenes are subject to reporting under the Dioxin/Furan Test rule (Section 3.4.2) and/or the Dioxin/Furan Pesticide Data Call-In (Section 3.4.3).

EPA's Office of Solid Waste has promulgated land disposal restrictions on wastes (i.e., wastewaters and non-wastewaters) resulting from the manufacture of chlorobenzenes (40 CFR 268). Table 3-7 lists all solid wastes in which CDDs and CDFs are regulated as hazardous constituents by EPA, including chlorobenzene wastes. The regulations prohibit the land disposal of these wastes until they have been treated to a level below the routinely achievable detection limit of 1 ppb in the waste extract for each of the following congener groups: TCDDs, PeCDDs, HxCDDs, TCDFs, PeCDFs, and HxCDFs (standards for waste code F039 apply only to TCDDs and TCDFs). The treatment standard of 1 ppb is based on incineration to 99.9999 percent destruction and removal efficiency.

EPA's Office of Water has promulgated effluent limitations for facilities that manufacture chlorinated benzenes and discharge treated wastewater (40 CFR 414.70). Although these effluent limitations do not specifically address CDDs and CDFs, the treatment processes required to control the chlorinated benzenes that are regulated (chlorobenzene; 1,2-dichlorobenzene; 1,3-dichlorobenzene; 1,4-dichlorobenzene; 1,2,4-trichlorobenzene; and hexachlorobenzene) are expected to control releases of CDDs and CDFs to minimal levels. The effluent limitations for the individual regulated chlorinated benzenes are less than or equal to 77 µg/l for facilities that utilize biological end-of-pipe treatment and are less than or equal to 196 µg/l for facilities that do not employ biological end-of-pipe treatment.

PCBs are manufactured by the direct chlorination of biphenyl in the presence of a catalyst. HpCDDs, OCDD, and CDFs, particularly the tetra-, penta-, and hexa-chlorinated CDF congeners, have been detected in commercial PCB formulations (Hagenmaier, 1987) However, the production of PCBs in the United States has been banned under TSCA and the use of in-service PCBs has been dramatically reduced. CDFs can be formed from PCBs under pyrolytic conditions, or by nonpyrolytic conditions via chlorine substitutions on the ortho-positions in the PCB molecule (Ree et al., 1988). Combustion of PCB-containing materials in transformers and capacitors may be a source of PCB-associated CDFs. (See Section 3.5.17.)

Aliphatic chlorine compounds are used as monomers in the production of plastics, as solvents and cleaning agents, and as precursors for chemical synthesis (Hutzinger and Fiedler, 1991a). These compounds are produced in large quantities. In 1990, 13.2 million metric tons of chlorinated aliphatic hydrocarbons were produced (U.S. ITC, 1991). The production of 1,2-dichloroethane and vinyl chloride accounted for 85 percent of this total production. Highly chlorinated CDDs and CDFs (i.e., hexa- to octa-chlorinated congeners) have been found in samples of 1,2-dichloroethane (55 ppb of OCDF), tetrachloroethane (47 ppb of OCDD), and epichlorohydrin (88 ppb of CDDs and 33 ppb of CDFs) (Hutzinger and Fiedler, 1991a). Because no more recent or additional data could be found in the literature to confirm these values, no estimates have been made of the mass of CDDs/CDFs present in these products manufactured annually.

Greenpeace recently issued a report (Greenpeace, 1993) on dioxin emissions associated with the production of ethylene dichloride (EDC) and vinyl chloride monomer (VCM). The Vinyl Institute has responded with a critique of the report (ChemRisk, 1993). Both of these studies are discussed below.

Greenpeace (1993) estimated that plants producing EDC and VCM release 1.8 kg of TEQ/yr to the environment (air, water, and ground combined - possible releases in the final products were not discussed). This estimate was based on an emission factor of 5 to 10 g TEQ/100,000 tons of VCM produced and a worldwide estimate of PVC (and thus VCM) production of 18 million metric tons/yr. This estimate represents the total emissions from all plants in the world but was based on data from only four European plants. Greenpeace (1993) cited some specific information on CDD/CDF formation or releases from a lengthy list of primary references. While most of the specific data came from studies conducted or sponsored by industry, in no case was the information offered by Greenpeace (1993) complete enough to allow calculation of all process or waste stream-specific emission factors to particular environmental media for a given plant.

European PVC manufacturers claim the emission factor is 0.01 to 0.5 g TEQ/100,000 metric tons of VCM, resulting in global emissions from EDC/VCM production as 0.002 to 0.09 kg TEQ/yr (Miller, 1993). There is no apparent dispute between the industry and Greenpeace regarding the formation of CDDs/CDFs during the production process, nor that some CDDs/CDFs are released to various environmental media. However, both European and U.S. manufacturers strongly dispute the total emission factors used in Greenpeace (1993) in arriving at their estimated total of 1.8 kg TEQ/yr emitted world-wide by the PVC industry.

Greenpeace (1993) cites the same specific monitoring information as industry but argues in several case studies that "diffuse emissions" of products and byproducts containing unspecified amounts of CDDs/CDFs constitute a very significant additional source to several environmental media. This appears to be the only rationale presented by Greenpeace (1993) to justify increasing the overall emission factor of 0.01 to 0.5 g TEQ/100,000 metric tons of VCM produced, which is accepted by European manufacturers, to Greenpeace's 5 to 10 g TEQ/100,000 metric tons.

PVC production in the United States is 4.5 million metric tons per year (ChemRisk, 1993). No data could be found on dioxin levels in waste streams or air emissions from PVC plants in the United States. Applying the worldwide emission factors discussed above to the U.S. PVC industry, gives a range of dioxin emissions of 0.45 to 23 g TEQ/yr (based on the industry emission factors) to 230 to 450 g TEQ/yr (based on the Greenpeace emission factors). It is unclear whether EDC/VCM/PVC production and emission control methods are sufficiently similar worldwide to know whether these factors should apply in the United States. Considering this unknown and the lack of measurement data in general and for U.S. facilities in particular, this report does not endorse either of these emission estimates nor is an independent emission estimate presented. Also, insufficient information was provided to indicate how these emissions, if present in the United States, would separate among media. Monitoring efforts to collect these data are highly recommended.

EPA's Office of Water has promulgated effluent limitations for facilities that manufacture chlorinated aliphatic chlorine compounds and discharge treated wastewater (40 CFR 414.70). Although these effluent limitations do not specifically address CDDs and CDFs, the treatment processes required to control the chlorinated aliphatic compounds that are regulated (e.g., 68 µg/l for 1,2-dichloroethane and 22 µg/l for tetrachloroethylene) are expected to control releases of CDDs and CDFs to minimal levels.

CDD/CDF contamination of dioxazine dyes and pigments available in Canada has been observed (Williams et al., 1992). As shown in Table 3-9, OCDD and OCDF concentrations in the µg/g range, and HpCDD, HxCDD, and PeCDD concentrations in the ng/g range were found in Direct Blue 106 dye (3 samples) and Direct Blue 108 dye (1 sample) dyes and Violet 23 pigments (6 samples)(Williams et al., 1992). Dioxazine pigments (e.g., Violet 23 pigment) and dioxazine dyes (e.g., Direct Blue 106 and 108) are derived from chloranil, which has been found to contain high levels of CDD/CDFs and has been suggested as the source of contamination among these dyes (Christmann et al., 1989; Williams et al., 1992; U.S. EPA, 1992b). In May 1990, EPA received test results showing that chloranil was heavily contaminated with dioxins; levels as high as 3,065 ppb TEQ were measured (U.S. EPA, 1992b). (See Section 3.4.2 for analytical results.)

Between 1990 and 1992, EPA learned that dioxin TEQ levels in chloranil could be reduced by more than two orders of magnitude (to less than 20 ppb) through manufacturing feedstock and process changes. EPA's Office of Pollution Prevention and Toxics (OPPT) subsequently began efforts to complete an industry-wide switch from use of the contaminated chloranil to low-dioxin chloranil. Although no chloranil is manufactured in the United States, significant quantities are imported. As of June 1993, EPA had negotiated agreements with all chloranil importers and domestic dye/pigment manufacturers known to EPA who use chloranil in their products to switch to low-dioxin chloranil. EPA will issue a significant new use rule (SNUR) under Section 5 of TSCA when U.S. stocks of chloranil with high levels of CDDs/CDFs are depleted. The SNUR will require industry to notify EPA at least 90 days prior to the manufacture, import, or processing, for any use, of chloranil containing total CDDs/CDFs at a concentration greater than 20 ug/kg (Cash, 1993; U.S. EPA, 1993c).

CDD/CDFs (tetra-, penta-, and hexa-chlorinated congeners) in the ppt range were found in Ni-phthalocyanine when several commercial phthalocyanine dyes were analyzed (Hutzinger and Fiedler, 1991a). Phthalocyanine dyes and diarylide yellow pigments have also been observed to contain PCBs in the ppm range. The PCBs are believed to be generated during manufacture because of the use of high-boiling chlorinated aromatic solvents (Hutzinger and Fiedler, 1991a). EPA, however, has prohibited the processing or distribution in commerce of any diarylide and phthalocyanine pigments that contain 50 ppm or more of PCBs (40 CFR 762.20).

Based on evidence that halogenated dioxins and furans may be formed as by-products during chemical manufacturing processes (Versar, 1985), EPA proposed a rule under Section 4 of the TSCA that would require chemical manufacturers and importers to test for the presence of chlorinated and brominated dioxins and furans in certain commercial organic chemicals (Federal Register, December 19, 1985). The final rule (Federal Register, June 5, 1987) listed 12 manufactured or imported chemicals for which testing was required and 20 chemicals not currently being manufactured or imported that would require testing if manufacture or importation resumed. These chemicals are listed in Table 3-10. The specific dioxin and furan congeners for which quantitation is required and the target limits of quantitation (LOQ) specified in the Rule are listed in Table 3-11. Under Section 8(a) of TSCA, the final rule also required that chemical manufacturers submit data on manufacturing processes and reaction conditions for chemicals produced using any of the 29 precursor chemicals listed in Table 3-12. The rule stated that subsequent to this data gathering effort, testing may be proposed for additional chemicals if any of the manufacturing conditions used favored the production of dioxins and furans.

To date, data have been submitted to the EPA TSCA Docket for 10 of the 12 chemicals requiring testing, however, not every manufacturer/importer has submitted data for every applicable product (Cash, 1993). Manufacture/import of the other two substances have stopped since the test rule was promulgated. [NOTE: All data and reports in the EPA TSCA Docket are available for public review/inspection at EPA Headquarters in Washington, DC.]

The results of analytical testing for dioxins and furans for the eight chemicals for which data are available in the TSCA docket are presented in Table 3-13. Data submitted for pentabromodiphenyloxide and tetra-bromobisphenol A-bisethoxylate are currently under EPA review. Dioxins/furans were found in four of these eight chemicals. The chemicals for which positive results were obtained are: 2,3,5,6-tetrachloro-2,5-cyclohexadiene-1,4-dione (chloranil), octabromodiphenyloxide, decabromodiphenyloxide, and tetrabromobisphenol-A. Table 3-14 presents the quantitative analytical results for the four submitted chloranil samples as well as the results of verification sampling/analysis performed on chloranil by EPA.

It should be noted that although testing conducted under this test rule for 2,4,6-tribromophenol indicated no halogenated dioxins or furans above the LOQs, Thoma and Hutzinger (1989) reported detecting BDDs and BDFs in a technical grade sample of this substance. Total TBDD, TBDF, and PeBDF were found at 84 m g/kg, 12 m g/kg, and 1 m g/kg, respectively. No hexa-, hepta-, or octa-BDFs were detected. Thoma and Hutzinger (1989) also analyzed analytical grade samples of two other brominated flame retardants, pentabromophenol and tetrabromophthalic anhydride; no BDDs or BDFs were detected (detection limits not reported).

In the early 1980s, attention began to focus on pesticides as potential sources of CDDs and CDFs in the environment. Historically, no regulation had been placed on CDD and CDF levels in end-use pesticide products. Certain pesticide active ingredients were known or suspected, however, to be contaminated with CDDs and CDFs (e.g., pentachlorophenol (PCP), Silvex, and 2,4,5-T). During the mid and late 1980s, EPA took several actions to investigate and control CDD/CDF contamination of pesticides. In 1983, the sale of Silvex and 2,4,5-T was canceled for all uses by EPA (Federal Register, October 18, 1983). EPA entered into a Settlement Agreement in 1987 with PCP manufacturers to allow continued registrations for wood uses (Federal Register, January 2, 1987) but which set tolerance levels for HxCDD and 2,3,7,8-TCDD. TCDD levels were not allowed to exceed 1.0 ppb in any product, and after February 2, 1989 (a gradually phased in requirement), any manufacturing-use PCP released for shipment could not contain HxCDD levels that exceeded an average of 2 ppm over a monthly release or a batch level of 4 ppm. EPA then issued a Final Determination and Intent to Cancel and Deny Applications For Registrations of Pesticide Products Containing Pentachlorophenol (Including but not limited to its salts and esters) For Non-Wood Uses which prohibited the registration of PCP for nonwood uses (Federal Register, January 21, 1987).

In addition to these cancellations and product standards, EPA's Office of Pesticide Programs (OPP) issued two Data Call-Ins (DCIs) in June 1987. Pesticide manufacturers are required to register their products with EPA in order to market them commercially in the United States. Through the registration process, mandated by FIFRA (Federal Insecticide, Fungicide and Rodenticide Act), EPA can require that the manufacturer of each active ingredient generate a wide variety of scientific data through several mechanisms. The most common process is the five phase reregistration effort to which the manufacturers (i.e., registrants) of older pesticide products must comply. In most registration activities, registrants must generate data under a series of strict testing guidelines, 40 CFR 158--Pesticide Assessment Guidelines (U.S.EPA, 1988). FIFRA accommodates the fact that some pesticide active ingredients may require additional data, outside of the norm, to adequately develop effective regulatory policies for those products. Therefore, EPA can require additional data, where needed, through various mechanisms as noted above including the DCI process.

The purpose of the first DCI (June 6, 1987), Data Call In Notice For Product Chemistry Relating to Potential Formation of Halogenated Dibenzo-p-dioxin or Dibenzofuran Contaminants in Certain Active Ingredients, was to identify chemicals that may contain halogenated dibenzo-p-dioxin and dibenzofuran contaminants and to quantify and eventually minimize exposure to these contaminants. The requirements made in this DCI parallel requirements established in the Dioxin/Furan Test Rule promulgated under Sections 4 and 8 of TSCA. (See Section 3.4.2.) The list of pesticide active ingredients to which this DCI applied along with their corresponding Shaughnessey and Chemical Abstract code numbers are presented in Table 3-15. [Note: the Shaughnessey code is an internal EPA tracking system--it is of interest because chemicals with similar code numbers are similar in chemical nature (e.g., salts, esters and acid forms of 2,4-D)]. All registrants supporting these chemicals were subject to the requirements of this DCI unless their product qualified for a Generic Data Exemption (i.e., a registrant exclusively used a registered product(s) as the source(s) of the active ingredient(s) identified in Table 3-15 in formulating their product(s)). Registrants whose products did not meet the Generic Data Exemption were required to submit the types of data listed below to assess the formation of tetra- through hepta-halogenated dibenzo-p-dioxin or dibenzofuran contaminants during manufacture. Registrants, however, did have the option to voluntarily cancel their product or "reformulate to remove an active ingredient," described in Table 3-15, to avoid compliance with the DCI.

· Product Identity and Disclosure of Ingredients: EPA required submittal of a Confidential Statement of Formula (CSF) based on the requirements specified in 40 CFR 158.108 and 40 CFR 158.120 - Subdivision D:Product Chemistry. Registrants who had previously submitted still current CSFs were not required to resubmit this information.

· Description of Beginning Materials and Manufacturing Process: Based on the requirements mandated by 40 CFR 158.120 - Subdivision D, EPA required submittal of a manufacturing process description for each step of the manufacturing process, including specification of the range of acceptable conditions of temperature, pressure, or pH at each step.

· Discussion of the Formation of Impurities: Based on the requirements mandated by 40 CFR 158.120 - Subdivision D, EPA required submittal of a detailed discussion/assessment of the possible formation of halogenated dibenzo-p-dioxins and dibenzofurans.

The second DCI (dated June 15, 1987), Data Call-In For Analytical Chemistry Data on Polyhalogenated Dibenzo-p-Dioxins/Dibenzofurans (HDDs and HDFs), was issued for a variety of pesticide active ingredients to the individual manufacturers of each ingredient. (See Table 3-16.) All registrants supporting these pesticides were subject to the requirements of this DCI unless the product qualified for various exemptions or waivers. Pesticides regulated by the second DCI were strongly suspected to be contaminated with detectable levels of HDDs/HDFs.

Under the second DCI, registrants whose products did not qualify for an exemption or waiver were required to generate and submit the following types of data in addition to the data requirements of the first DCI:

· Quantitative Method For Measuring HDDs or HDFs: Registrants were required to develop an analytical method for assessing the HDD/HDF contamination of their products. The DCI established a regimen for defining the precision of the analytical method (i.e., for internal standard--precision within +/- 20 percent and recovery range of 50 to 150 percent, also a signal to noise ratio of at least 10:1 was required). Target quantification limits were established in the DCI for specific HDD and HDF congeners. (See Table 3-11.)

· Certification of Limits of HDDs or HDFs: Registrants were required to submit a "Certification of Limits" in accordance with 40 CFR 158.110 and 40 CFR 158.120 - Subdivision D. Analytical results were required that met the guidelines described above.

Registrants could select one of two options to comply with the second DCI. The first option was to submit relevant existing data, develop new data, or share the cost to develop new data with other registrants. The second option was to alleviate the DCI requirements through several exemption processes including a Generic Data Exemption, voluntary cancellation, reformulation to remove the active ingredient of concern, an assertion that the data requirements do not apply, or the application/award of a low-volume, minor-use waiver.

The data contained in CSFs, as well as any other data generated under Subdivision D, are typically considered Confidential Business Information (CBI) under the guidelines prescribed in FIFRA because they usually contain information regarding proprietary manufacturing processes. In general, all analytical results submitted to EPA in response to both DCIs are considered CBI and cannot be released by EPA into the public domain. Summaries based on the trends identified in that data as well as data made public by EPA are provided below.

To date, more than 100 submissions have been reviewed in response to the two DCIs. The majority have been manufacturing process data in support of waiver requests, analytical method protocols, and sample collection protocols (telephone conversation between S. Funk, EPA - Office of Pesticide Programs (OPP), and J. Dawson, Versar, Inc. on 2/18/93). Analytical results on the levels of tetra- through hepta- HDDs/HDFs have been received and reviewed for 16 distinct pesticide active ingredients (Table 3-17). In general, the analyses have not revealed HDD/HDF concentrations in excess of the LOQs specified in Table 3-11. For those products in which LOQs are exceeded, the identified contamination levels were generally within an order of magnitude of the LOQ and apply only to one or two congeners per product (telephone conversation between S. Funk, EPA/

OPP, and J. Dawson, Versar, Inc. on 2/18/93). Table 3-18 presents a summary of results recently reported by EPA for CDDs and CDFs in eight technical 2,4-D herbicides.

The production and use of chlorine gas has involved processes that result in the generation of CDFs (Rappe, 1992a). Chlorine is commonly produced via electrolysis of brine in mercury cells. High levels of CDFs have been found in the graphite electrode sludge from this chemical process and may have been responsible for occupational exposures among workers who handled these sludges. Svensson et al. (1992) evaluated the relationship between blood CDF levels in chloralkali plant workers and direct exposure of these workers to electrode sludges and to dust and earth contaminated with graphite electrode sludge. Subjects who had been exposed by handling graphite electrode sludge had higher levels of 2,3,7,8-substituted PeCDFs and HxCDFs than reference subjects. Evaluations of congener distribution patterns have demonstrated that the 2,3,7,8-substituted CDFs are the major congeners formed during the chloralkali process (Rappe et al., 1990; Rappe, 1992a).

Until the late 1970s, graphite electrodes were the primary type of anode used in the chloralkali industry (Curlin and Bommaraju, 1991). Since then, metal anodes have been developed to replace graphite electrodes because of production problems associated with their use (U.S. EPA, 1982; Curlin and Bommaraju, 1991). Currently, no U.S. facilities are believed to use graphite electrodes in the production of chlorine gas (telephone conversation between L. Phillips, Versar, Inc., and T. Fielding, U.S. EPA, Office of Water, February 1993). Although the use of graphite electrodes has been eliminated, the potential for CDD/CDF releases from dump sites containing contaminated sludges may still exist (Svensson et al., 1992; Rappe, 1992a).

Catalyst regeneration in the petroleum refinery reforming process has been identified as a source of CDDs and CDFs based on testing conducted in Canada (Thompson et al., 1990). According to Thompson et al. (1990), "catalytic reforming is a refinery process which is used to produce high octane gasoline. The reforming process occurs at high temperature and pressure and requires the use of a catalyst. During the catalytic process, a complex mixture of aromatic compounds known as coke is formed and deposited onto the catalyst. As coke deposits onto the catalyst, its activity is decreased. The high cost of the catalyst necessitates its regeneration. Catalyst regeneration is achieved by removing the coke deposits via burning and activating the catalyst using chlorinated compounds. Burning of the coke produces flue gases which contain CDDs and CDFs along with other combustion products." Thompson et al. (1990) reported total CDD and CDF concentrations of 8.9 ng/m³ and 210 ng/m³, respectively, in stack gas samples from petroleum refinery reforming operations (Table 3-19). It was also found that the CDD and CDF congener distribution patterns observed were similar to those found in municipal waste incinerator ash and stack samples. Because flue gases may be scrubbed with water, internal effluents may also be contaminated with CDD/CDFs. Thompson et al. (1990) observed CDDs and CDFs in the internal wash water from a scrubber of a periodic/cyclic regenerator (Table 3-20).

The Canadian Ministry of the Environment detected concentrations of CDDs in an internal wastestream of spent caustic in a petroleum refinery that ranged from 1.8 to 22.2 ppb, and CDFs ranging from 4.4 to 27.6 ppb. The highest concentration of 2,3,7,8-TCDD was 0.0054 ppb (Maniff and Lewis, 1988). CDDs were also observed in the refinery's biological sludge at a maximum concentration of 74.5 ppb, and CDFs were observed at a

maximum concentration of 125 ppb (Maniff and Lewis, 1988). The concentration of CDD/CDFs in the final combined refinery plant effluent was below the detection limits.

Insufficient data are available to evaluate CDD/CDF releases from these sources in the United States. However, Beard et al. (1993) conducted a series of benchtop experiments to investigate the mechanism(s) of CDD/CDF formation in the catalytic reforming process. A possible pathway for the formation of CDFs was found, but the results could not explain the formation of CDDs. Analyses of the flue gas from burning coked catalysts revealed the presence of unchlorinated dibenzofuran (DBF) produced in quantities of up to 220 ng/g of catalyst. Chlorination experiments indicated that dibenzofuran and possibly biphenyl and similar hydrocarbons act as CDF precursors and can become chlorinated in the catalyst regeneration process. Corrosion products on the steel piping of the process plant seem to be the most likely chlorinating agent. Furthermore, CDFs can form by de novo synthesis from chlorinated hydrocarbons like trichloroethylene, methylene chloride, and carbon tetrachloride in the presence of FeCl₃ and HCl or Cl₂.

Rappe et al. (1989) reported the formation of CDFs (tetra- through octa-chlorinated CDFs) when tap water and double-distilled water were chlorinated using chlorine gas. The CDF levels found in the single samples of tap water and double-distilled water were 35 and 7 pg TEQ/L, respectively. The water samples were chlorinated at a dosage rate of 300 mg of chlorine per liter of water which is considerably higher (by a factor of one to two orders of magnitude) than the range of dosage rates typically used to disinfect drinking water. Rappe et al. (1989) hypothesized that the CDFs or their precursors are present in chlorine gas. It should be noted, however, that although few surveys of finished drinking water for CDD/CDF levels have been conducted, the few that have been published only rarely report the presence of any CDD/CDF even at low pg/L detection limits and in those cases the CDD/CDFs were also present in the untreated water. (See Section 4.3.)

Several recent studies have been conducted to identify the source(s) of CDD/CDFs found in textiles and at dry cleaning facilities. Horstmann and McLachlan (1994) analyzed 35 new textiles and found total CDD/CDF levels generally less than 50 pg/g; however, some items were as high as 290,000 pg/g. The authors conclude that textile finishing processes are not likely to be the source of the high CDD/CDF levels found because of the apparent randomness of the textiles with high CDD/CDF levels. However, the authors hypothesize that the use of pentachlorophenol to preserve cotton, particularly when it is randomly strewed on bales of cotton as a preservative during sea transport, is the likely source of the high levels occasionally observed. As discussed in Section 3.4.3, the use of pentachlorophenol (PCP) for nonwood uses has been prohibited in the United States since 1987. However, Horstmann and McLachlan (1994) comment that PCP is still used in developing countries, especially for purposes of preserving cotton during sea transport. As discussed in Section 3.4.1.5, certain dyes and pigments have also been observed to contain CDD/CDFs and may also contribute to levels found in textiles. Horstmann and McLachlan (1994) also summarize recent research concerning CDD/CDFs in dry cleaning residues and reach the conclusion that new textiles are the source of the CDD/CDFs found.

3.5. MECHANISMS OF FORMATION OF DIOXIN-LIKE COMPOUNDS DURING COMBUSTION OF ORGANIC MATERIALS

The specific molecular mechanisms by which CDDs and CDFs are initially formed and then emitted from combustion sources remain largely unknown and are theoretical. The theoretical basis for conjecture is derived primarily from direct observations in municipal solid waste incinerators and from well conducted laboratory studies. Municipal solid waste incinerators (MSWIs) have been heavily studied from the perspective of eventually finding the specific formation mechanism transpiring within the system, and determining ways to either significantly reduce such opportunities or ultimately hinder the formation kinetics to preclude evolution of these chemicals. Although much has been learned from these studies, it is still not known how to completely block the formation of CDDs/CDFs during the combustion of certain organic materials in the presence of a source of chlorine. Adding to this complexity is the wide variability of organic materials that are incinerated and thermally processed by a wide spectrum of combustion technologies having variable temperatures, residence times, and oxygen requirements. However, it is possible to identify the central chemical events participating in the formation of CDDs and CDFs by evaluating emission test results from MSWIs in combination with laboratory experiments.

The emission of CDDs and CDFs can be explained by three principal theories, which should not be regarded as being mutually exclusive. The first is that CDD/CDFs are present as contaminants in the combusted organic material. This theory is discussed in Section 3.5.1. The second is that CDDs/CDFs are ultimately formed from the thermal breakdown and molecular rearrangement of precursor compounds, which are defined as chlorinated aromatic hydrocarbons having a structural resemblance to the CDD/CDF molecule. This theory is discussed in Section 3.5.2. The third theory, similar to the second and described in Section 3.5.3, is that CDDs/CDFs are synthesized de novo; this means they are formed from organic and inorganic substrates comprised of singular or mixtures of molecules bearing little resemblance to the molecular structure of CDDs or CDFs. Section 3.5.4 discusses the generation of coplanar PCBs. Section 3.5.5 discusses the evaluation of naturally occurring CDDs/CDFs by examinations of sediment core data, and Section 3.5.6 provides a closing summary of the three principal theories of formation.

The first theory states that CDD and CDF compounds present as contaminants in the fuel or waste products that are fed into the combustion chamber are responsible for dioxin and dibenzofuran emissions out the stack of the combustion process. Most work in this area has involved the study of municipal solid waste incineration (MSWI) in which case CDDs and CDFs have been analytically detected in the raw refuse fed into the MSWI. Tosine, et al. (1983) first reported detecting trace amounts of HpCDD and OCDD in the MSW fed into an MSWI in Canada. HpCDD ranged in concentration from 100 ppt to 1 ppb, and OCDD ranged from 400 to 600 ppt. Wilken et al. (1992) separated the various solid waste fractions of MSW collected from municipalities in Germany and analyzed them for the presence of CDDs/CDFs and other organochlorine compounds. Total CDDs/CDFs were detected in all MSW fractions in the following range of concentrations: paper and cardboard = 3.1 to 45.5 ppb; plastics, wood, leather, textiles combined = 9.5 to 109.2 ppb; vegetable matter = 0.9 to 16.9 ppb; and "fine debris" (defined as particles < 8 mm) = 0.8 to 83.8 ppb. Ozvacic (1985) measured CDDs/CDFs in the raw MSW fed into two MSWIs operating in Canada. In one MSWI, CDDs were detected in the refuse in a range of concentration from 10 to 30 ppb, but no CDFs were detected (detection limit: 1 pg/g). In the MSW fed to the second MSWI, CDDs were detected in a range of 75 to 439 ppb, and CDFs were detected only in one of three samples at a total concentration of 11 ppb. EPA has reported on the detection of CDDs/CDFs in refuse derived fuel (RDF) burned in a large, urban MSWI (Federal Register, 1991). From 13 MSW samples taken prior to incineration, CDDs were detected in a range of 1 to 13 ppb, and CDFs were measured in a range of 0 to 0.6 ppb. In these samples, OCDD predominated, and the lower chlorinated congeners were not detected.

Despite these findings, the conditions of thermal stress imposed by the incineration process discounts the likelihood that the total magnitude of CDDs and CDFs, as measured in the raw MSW, can explain the total magnitude of concentration as an emission from the stack of the MSWI (Clement et al., 1990; Commoner, 1990). Contamination, however, may partially contribute to the stack release. Clement and coworkers (1988) performed a mass balance involving an input versus output of dioxin at two operational MSWIs in Canada. These mass balance calculations clearly demonstrated that the mass of CDDs and CDFs emitted at the point of the stack was much greater than the mass in the raw MSW incinerated at the MSWIs, and that the profiles of the distributions of CDD/CDF congeners were strikingly different. Primarily, higher chlorinated congeners were detected as contaminants in the waste, whereas the total array of tetra - octa CDDs/CDFs were emitted from the stack.

Commoner and coworkers (1984; 1985; 1987) evaluated the test data of a mass burn MSW incinerator for the concentration of CDDs and CDFs at multiple sampling points during the combustion process: (1) exit to the furnace; (2) entry to the heat exchanger; (3) inlet to the electrostatic precipitator (ESP); (4) exit to the ESP; and (5) exit to the smokestack. Lowest or nondetectable concentrations of CDDs/CDFs were found at sampling point (1), and highest concentrations were measured at sampling point (5). From these sampling data, Commoner concluded that: CDDs/CDFs were not formed within the furnace region where the waste material was combusted and that usually only OCDD and OCDF were detected in extremely low concentrations at the point of exit to the furnace (if dioxins were detected at all). It was also concluded that the CDDs/CDFs were mostly formed as a synthesis process catalyzed by the properties of fly ash in combination with chlorine, and that this probably transpired within areas downstream of the combustion zone where the combustion offgases had cooled to less than 400° C. Commoner et al. (1984, 1987) ruled out the effectiveness of combustion as a major factor in CDD/CDF emissions from the stack; this would be expected if waste contamination was solely responsible for the emission. This phenomena was independently observed by Environment Canada in a series of tests of a modular MSW incinerator (Hay, et al., 1986; Environment Canada, 1985). On a mass balance basis, the concentration of CDDs and CDFs measured at the stack was approximately two orders of magnitude higher as compared to the inlet to the boiler just after exiting the secondary furnace. The temperatures of the combustion gases at these two points of measurement were 130 and 740° C at the stack and boiler inlet, respectively (Environment Canada, 1985). For the most part, only OCDD was present in the hot gases exiting the furnace, whereas all the congeners were present in the stack emissions, thus giving further evidence that CDDs/CDFs are formed after the combustion zone. Using similar protocols, EPA and Environment Canada (1991) jointly evaluated the emission of CDDs and CDFs from a refuse-derived fuel MSWI operating in the United States. It was found that approximately 5 milligrams of total CDDs and CDFs per metric ton of MSW burned by the facility were measured in the raw MSW prior to combustion, but no CDDs nor CDFs were detected at the point of exit to the furnace prior to the inlet to the economizer (i.e., the heat exchanger used to extract additional heat from the hot gases). Once heat in the combustion gas was extracted for energy purposes and the gases were further cooled to less than 400° C, the total array of tetra- through octa-CDDs and CDFs could be detected.

These series of experiments in which the mass balance of CDD/CDF was estimated within the entire combustor, beginning with the waste and ending with the stack, discount the first theory of dioxin formation (i.e., that dioxin in the feed accounts for all emissions of dioxin from the stack to the air). Moreover, it is expected that the conditions of thermal stress imposed by typical incineration and other combustion sources would destroy and reduce the CDDs and CDFs present as contaminants in the waste to levels that are 0.0001 to 10 percent of the initial concentration, depending on the performance of the combustion source and the level of combustion efficiency. Stehl et al. (1973) demonstrated that the moderate temperature of 800° C enhances the decomposition of CDDs at a rate of about 99.95 percent, but that lower temperatures result in a higher survival rate. Theoretical modeling has shown that unimolecular destruction of CDDs/CDFs at 99.99 percent can occur at the following temperatures and retention times within the combustion zone: 977° C with a retention time of 1 second; 1000° C at a retention time of 1/2 second; 1227° C at a retention time of 4 milliseconds; and 1727° C at a retention time of 5 microseconds (Schaub and Tseng, 1983). Thus, CDDs and CDFs would have to be in parts per million concentration in the feed to the combustor to be found in the part per billion or part per trillion levels in the stack gas emission (Shaub and Tseng, 1983). However, it cannot be ruled out is that CDDs/CDFs in the waste or fuel may contribute (up to some percentage) to the overall concentration leaving the stack.

The second theory states that the production of CDDs and CDFs is a direct result of in-situ thermal degradation of precursor compounds during or after combustion of organic materials. Present theory is mostly derived from laboratory experiments involving the heating of suspect precursors in quartz ampules under starved-air conditions, and in experiments investigating the role that combustion fly ash has in promoting the formation of CDD/CDFs from precursor compounds.

Liberti and Brocco (1982) postulated that the general reaction that may be taking place in a typical combustion process is a thermolytic synthesis and interaction between two families of precursors indicated by A and B. Precursors A are aromatic compounds having a definite phenolic structure (e.g., phenol and polychlorinated phenols), and precursors B are chemical species that can act as a chlorine donor (e.g., PVC and HCl). Esposito et al. (1980) offered a chemical basis for defining a dioxin precursor:

1. The compound is comprised of an ortho-substituted (positions 1 and 2 on the compound) benzene ring in which one of the substituents is an oxygen atom directly attached to the ring, and

2. It then must be possible for the two substituents on the benzene ring to react with each other to form a new and independent compound under the influences of heat and pressure (i.e., dioxin).

Dickson and Karasek (1987) further refined this definition to be consistent with the formation kinetics thought to occur within combustion processes. In their definition, the term "precursor" refers specifically to chlorinated aromatic compounds that are either already present on the surface of combustion fly ash, or are present in the gas phase prior to entering a critical region outside the combustion zone where the gases have cooled and where heterogeneous catalyzed reactions take place that form CDDs/CDFs. Chlorophenols and chlorobenzenes were identified as ideal precursor compounds in these reaction pathways.

Controlled laboratory combustion experiments involving the thermal degradation of aromatic compounds, either singly or in mixtures, have provided useful data in identifying ideal precursor compounds. For example, Jansson and coworkers (1977) generated CDDs through the pyrolysis of wood chips treated with tri-, tetra-, and penta-chlorophenol in a bench-scale furnace operated at 500-600° C. Stehl and Lamparski (1977) combusted grass and paper treated with the herbicide 2,4,5-trichlorophenoxyacetic acid (2,4,5-T) in a bench-scale furnace at 600-800° C and generated ppm_v levels of TCDD. Ahling and Lindskog (1982) have reported on the formation of CDDs during the combustion of tri- and tetrachlorophenol formulations at temperatures of 500-600° C. Decreases in oxygen during combustion generally increased the yield, and the addition of copper salts to the tetrachlorophenol formulation significantly enhanced the yield of CDDs. Combustion of pentachlorophenol resulted in low yields of CDDs except when burned with an insufficient supply of oxygen. In that case, the investigators noted the formation of tetra- through octa-chlorinated congeners. Buser (1979) generated CDDs/CDFs on the order of 0.001-0.08 percent (by weight) by heating tri-, tetra-, and pentachlorobenzenes at 620° C in quartz ampules in the presence of oxygen. It was noted that chlorophenols were formed as combustion by-products, and Buser (1979) speculated that these were acting as reaction intermediates in the formation of CDDs/CDFs.

Recently it has been demonstrated that CDDs and CDFs are formed from aromatic precursor compounds adsorbed onto the reactive surface of fly ash (particulate matter) entrained in the combustion plasma. Moreover, formation occurs outside and downstream of the combustion zone of a furnace to a combustion source in regions where the temperature of the combustion offgases has cooled to between 200 and 400° C (Vogg et al., 1987; Bruce et al., 1991; Cleverly et al., 1991; Gullet et al., 1990a; Commoner et al., 1987; Dickson and Karasek, 1987; Dickson et al., 1992). Vogg and coworkers (1987) have shown that inorganic chloride ions, such as copper chloride, present in the combustion gas may act as a catalyst to promote surface reactions on particulate matter to convert aromatic precursor compounds to chlorinated dioxins and dibenzofurans. After carefully extracting organics from MSWI fly ash, Vogg et al. (1987) added a known concentration of isotopically labeled CDDs/CDFs to the matrix. The MSWI fly ash was then heated in a laboratory furnace at varying temperatures for 2 hours. The treated fly ash was exposed to increasing temperatures in 50° C increments in a temperature range of 200 to 400° C. Table 3-21 summarizes these data.

Because the relative concentration of CDDs/CDFs increased while exposed to varying temperature, Vogg, et al. (1987, 1992) concluded that formation of CDDs and CDFs from precursor compounds on the surface of fly ash transpires during MSW incineration within a specific range of temperature, 250 to 450° C. Within this range, the concentration of CDDs/CDFs increases to some maxima, and outside this range the concentration diminishes. Vogg et al. (1987) proposed an oxidation reaction pathway giving rise to the formation of CDDs and CDFs in the post-furnace regions of the incinerator in the following order: (1) hydrogen chloride gas (HCl) is thermolytically derived as a product of the combustion of heterogeneous fuels containing abundant chlorinated organic chemicals and chlorides; (2) oxidation of HCl, with copper chloride (CuCl₂) as a catalyst, yields free gaseous chlorine; (3) phenolic compounds (present from combustion of lignin in the waste or other sources) entrained in the combustion plasma are substituted on the ring structure by contact with the free chlorine; and (4) the chlorinated precursor to dioxin (e.g., chlorophenol) is further oxidized (with copper chloride as a catalyst) to yield CDDs and CDFs and chlorine.

Gullett and coworkers (1990a; 1990b; 1991a; 1991b; 1992) have studied the formation mechanisms through extensive combustion research at EPA, and have verified the observations of Vogg et al. (1987). It was proven that CDDs and CDFs could be ultimately produced from low temperature reactions (i.e., 350° C) between Cl₂ and a phenolic precursor combining to form a chlorinated precursor, followed by oxidation of the

chlorinated precursors (catalyzed by a copper catalyst such as copper chloride) as in examples (1) and (2), below.

(1) The initial step in the formation of dioxin is the formation of chlorine from HCl in the presence of oxygen (the Deacon process), as follows (Vogg et al., 1987; Bruce et al., 1991):

_

2HCl + 1/2 O₂ ^{——————>} H₂O + Cl₂

(2) Phenolic compounds adsorbed on the surface of fly ash are chlorinated to form the dioxin precursor, and the dioxin is formed as a product from the breakdown and molecular rearrangement of the precursor. The reaction is promoted by the presence of heat and copper chloride acting as a catalyst (Vogg et al., 1987; Gullett et al., 1992):

(a) phenol + Cl₂ ————> chlorophenol (dioxin precursor)

CuCl₂

(b) 2-chlorophenol + 1/2 O₂ ————> dioxin + Cl₂

The major direct source of chlorine available for participating in the formation of CDDs/CDFs is gaseous HCl, which is initially formed as a combustion by-product from the chlorine and chlorinated organic chemicals contained in the MSW (and other fuels) (Vogg et al., 1987; Bruce et al., 1991; Cleverly, 1984; Commoner et al., 1987). MSW contains approximately 0.45-0.90 percent (by weight) chlorine (Domalski et al., 1986). MSW incinerators are a major stationary combustion source of air emissions of HCl, which average between 400 to 600 ppm in the combustion gas (U.S. EPA, 1987). HCl is converted to chlorine vapor by the Deacon process, and the vapor phase chlorine directly chlorinates a dioxin precursor along the aromatic ring structure. Oxidation of the chlorinated precursor in the presence of an inorganic chloride metal catalyst (of which copper chloride was found to be the most active) yields CDDs and CDFs. Increasing the yield of chlorine in vapor phase from the oxidation of HCl generally causes an increase in the rate of formation of CDDs/CDFs. Formation kinetics are most favored at temperatures between 200 to 350° C. Reductions in chlorine production, either by limiting initial HCl concentration or by shortening the residence time in the Deacon process temperature window, should result in decreases in the rate and magnitude of formation of CDDs and CDFs (Bruce et al., 1991; Gullett et al., 1990b; Commoner et al., 1987). Bruce and coworkers (1991) observed a general increase in the formation of CDDs and CDFs with increases in the vapor phase concentration of chlorine. Figure 3-2 shows the apparent dependence of the extent of formation of CDDs and CDFs upon chlorine concentration in

the vapor phase. Bruce et al. (1991) verified a dependence on the concentration and availability of gaseous chlorine in the thermolytic formation of CDDs/CDFs.

In the testing of a variety of industrial stationary combustion sources during the National Dioxin Study in 1987, EPA made a series of qualitative observations on the relationship between total chlorine present in the fuel/waste and the magnitude of emissions of CDDs and CDFs from the stack of the tested facilities (U.S. EPA, 1987). In general, combustion units with the highest CDD emission concentrations had greater quantities of chlorine in the fuel, and, conversely, sites with the lowest CDD emission concentrations contained only trace quantities of chlorine in the feed. The typical chlorine content of various combustion fuels has been reported by Lustenhouwer et al. (1980) as: coal: 1,300 µg/g; MSW: 2,500 µg/g; leaded gasoline: 300-1,600 µg/g; unleaded gasoline: 1-6 µg/g.

The role that temperature plays in the formation kinetics has been investigated by Oberg et al. (1989) on a full-scale hazardous waste incinerator operating in Sweden. Oberg confirmed that the formation of CDDs/CDFs occurs after the furnace. Most of the formation transpired in the boiler used to extract heat for co-generation of energy. In this investigation, significant increases in total concentration of dioxin TEQ occurred between temperatures of 280-400° C, and concentrations declined at temperatures above 400° C. This is in agreement with the experimental evidence of the temperature range defined as the "window of opportunity" for catalytic formation of CDDs/CDFs on the surfaces of fly ash particles.

Dickson and Karasek (1987) have demonstrated that CDDs/CDFs can be directly formed from the thermal conversion and oxidation of chlorinated precursors, in particular chlorophenols, on the surface of MSWI fly ash while heated in a bench-scale furnace. Their experiment was designed to mimic conditions of MSW incineration; to identify the step-wise chemical reactions involved in converting a precursor compound into dioxin, and to determine if MSWI fly ash could promote these reactions. MSWI fly ash was obtained from a facility in Canada and a facility in Japan. The MSWI fly ash was extensively solvent-extracted for any organic constituents prior to initiating the experiment. Twenty grams of fly ash were introduced into a bench-scale oven (consisting of a simple flow-tube combustion apparatus) and heated at 340° C overnight to desorb any remaining organic compounds from the matrix. ¹³C₁₂ -labeled pentachlorophenol (PCP) and two trichlorophenol isotopes (¹³C₁₂- 2,3,5-T and 3,4,5-T) were added to the surface of the clean fly ash matrix, and placed into the oven for 1 hour at 300° C. Pure inert nitrogen gas (flow rate of 10 ml/min) was passed through the flow tube to maintain constant temperatures. Tetra- through octa- CDDs were formed from the labeled pentachlorophenol experiment; over 100 µg/g of total CDDs were produced. The congener pattern was similar to the congener pattern found in MSWI emissions. The 2,4,5-T experiment primarily produced HxCDDs and very small amounts of tetra- and octa-CDD. The 3,4,5-T experiment mainly produced OCDD and 1,2,3,4,6,7,8-HpCDD. Dickson and Karasek (1987) proposed that the chlorinated phenol may undergo molecular rearrangement or isomerization as a result of dechlorination, dehydrogenation, and trans-chlorination before condensation occurs to ultimately form CDDs on the fly ash surface. These reactions ultimately dictate the types and amounts of CDDs that are formed.

Nestrick and coworkers (1987) reported on the thermolytic reaction between benzene (an unsubstituted precursor) and iron (III) chloride on a silicate surface to yield CDDs/CDFs at temperatures ³ 150° C. The experimental protocol was to introduce 100 - 700 mg of native and ¹³C₆-benzene into a macro-reactor system consisting of a benzene volatilization chamber connected to a glass tube furnace. The investigators noted the relevance of this experiment to generalizations about combustion processes because benzene is the usual combustion by-product of organic fuels. Inert nitrogen gas was used to carry the benzene vapor to the furnace area. The exit to the glass tubing to the furnace was plugged with glass wool, and silica gel was introduced from the entrance end to give a bed depth of 7 cm to which the FeCl₃ was added to form a FeCl₃/silica reagent. The thermolytic reaction took place in a temperature range of 150-400° C at a residence time of 20 minutes. Although di- through octa-CDD/CDF were formed by this reaction at all the temperatures studied, the percent yields were extremely small. Table 3-22 summarizes these data.

The third and last theory states that CDDs/CDFs are formed in combustion processes from materials and/or compounds that are not structurally related to CDDs/CDFs on a molecular level. As in Theory 2, synthesis is believed to occur in regions outside of the furnace zone of the combustion process where the combustion plasma has cooled to a range of temperatures considered favorable to formation kinetics. A key component to de novo synthesis is the production of intermediate compounds (either halogenated or non-halogenated) that are precursors to dioxin formation. However, research in this area has produced CDDs/CDFs directly from the heating of carbonaceous fly ash in the presence of

an inorganic ion without the apparent generation of reactive intermediates. Thus, the specific steps involved in the de novo process have not been fully and succinctly delineated. Laboratory experimentation has proven that MSWI fly ash, itself, is not an inert substrate, and the matrix can actually participate in the formation kinetics. Typically the fly ash is composed of an alumina-silicate construct with 5-10 percent concentrations of silicon, chlorine (as inorganic chlorides), sulfur, and potassium (NATO, 1988). Twenty percent of the weight of fly ash particles are carbon, and the particles have specific surface areas in the range of 2-4 m² (NATO, 1988). The distinguishing feature of the de novo synthesis over the precursor synthesis is the thermolytic breakdown and molecular rearrangement of chemical species unrelated to CDDs/CDFs at the start of the process to yield precursor compounds. Theory 2 starts with the precursor compounds already adsorbed onto the surface of fly ash or present in the gas phase (Dickson et al., 1992). By this distinction, however, one could argue that Theory 3 is really an augmentation to Theory 2 because the generation of CDDs/CDFs may still require the formation of a dioxin precursor. Nevertheless, a distinction is presented here for purposes of describing

additional pathways that have been suggested for the thermal formation of these compounds.

To delineate the de novo synthesis of CDDs/CDFs from unrelated matter, Stieglitz and coworkers (1989a) have conducted experiments involving the heating of particulate carbon containing adsorbed mixtures of Mg and Al-silicate in the presence of copper chloride as a catalyst to the reaction. The authors described annealing mixtures of Mg-Al silicate with activated charcoal (4 percent by weight), chloride as potassium chloride (7 percent by weight), and 1 percent copper chloride (CuCl₂) (in water) in a glass tube at 300° C. The retention time was varied at 15 minutes, 30 minutes, and 1, 2, and 4 hours to obtain differences in the amounts of CDDs/CDFs that could be formed. The results are summarized in Table 3-23.

In addition to the CDDs/CDFs formed as primary products of the de novo synthesis, the investigators observed the formation of precursors at the varying retention times of the experiment. In particular, similar yields of tri- though hexa-chlorobenzenes, tri- through hepta-chloronaphthalenes, and tetra- through hepta-chlorobiphenyls, were quantified which were seen as highly suggestive of the role these compounds may play as intermediates in the continued formation of CDDs/CDFs. Table 3-24 summarizes the experimental yields of chlorinated benzenes as a function of the annealing time at 300° C. Stieglitz et al. (1989a) made the following observations:

1. The de novo synthesis of CDDs/CDFs via the reaction of carbonaceous particulate matter exposed to a temperature of 300° C was clearly demonstrated. Additionally, the experiment yielded ppb-ppm concentrations of chlorinated benzenes, chlorinated biphenyls, and chlorinated napthalenes through a similar mechanism. When potassium bromide was substituted for potassium chloride as a source of halogen for the organic compounds in the reaction, polybrominated dibenzo-p-dioxins and dibenzofurans were formed as reaction products.

2. Copper chloride catalyzed the de novo synthesis of CDDs/CDFs on the surface of particulate carbon in the presence of oxygen to yield carbon dioxide and chlorinated/brominated aromatic compounds.

3. Particulate carbon, which is characteristic of combustion processes, may act as the source for the direct formation of CDDs/CDFs as well as other chlorinated organics.

More recently, Stieglitz and coworkers (1991) investigated the role that particulate carbon plays in the de novo formation of CDDs/CDFs from fly ash containing appreciable quantities of organic chlorine. Stieglitz et al. (1991) found that the fly ash contained 900 µg/g of bound organic chlorine. Only 1 percent of the organic chlorine was extractable. Annealing the fly ash at 300-400° C for several hours caused the carbon to oxidize leading to a reduction in the total organic chlorine in the matrix, and a corresponding increase in the total extractable organic chlorine (TOX) (e.g., 5 percent extractable TOX at 300° C and 25-30 percent extractable total organic chlorine at 400° C). From this, Stieglitz et al. (1991) concluded that the oxidation and degradation of carbon in the fly ash are the source for the formation of CDDs/CDFs, and, therefore, are essential in the de novo synthesis of these compounds.

Addink et al. (1991) conducted a series of experiments to observe the de novo synthesis of CDDs/CDFs in a carbon-fly ash system. In this experiment, 4 grams of carbon-free MSWI fly ash were combined with 0.1 gram of activated carbon and placed into a glass tube between two glass wool plugs. The glass tube was then placed into a furnace at a specific temperature in the range 200 to 400° C. This was repeated for a series of retention times and temperatures. The investigators observed that the formation of CDDs/CDFs was optimized at the temperature of 300° C and at the furnace retention times of 4-6 hours. Figure 3-3 displays the relationship between retention time, temperature and the production of CDDs/CDFs from the heating of carbon particulate. Addink et al. (1991) also investigated the relationship between temperature of the furnace

and the production of CDDs/CDFs from the annealing of carbonaceous fly ash. Figure 3-4 displays this relationship. In general, the concentration began to increase at 250° C and crested at 350° C, with a sharp decrease in concentration above 350° C. The authors also noted a relationship between temperature and the CDD/CDF congener profile; at 300° C to 350° C, the lower chlorinated tetra- and penta-CDD/CDF congeners increased in concentration, while hexa-, hepta-, and octa-CDD/CDF congeners either remained the same or decreased in concentration. The congener profile of the original MSWI fly ash (not subject to de novo experimentation) was investigated with respect to changes caused by either temperature or residence time in the furnace. No significant changes occurred, leading the authors to propose an interesting hypothesis for further testing: after formation of CDDs/CDFs occurs on the surface of fly ash, the congener profile remains fixed and insensitive to changes in temperature or residence time indicating some form of equilibrium is reached in the formation kinetics.

Gullett et al. (1994) developed a pilot-scale combustor to study the effect on CDD/CDF formation of varying the combustion-gas composition, temperature, residence time, quench rate, and sorbent (Ca[OH]₂) injection. The fly ash loading was simulated by the injection on fly ash collected from a full-scale MSWI. Sampling and analysis indicated CDD/CDF formation or the injected fly ash at levels representative of those observed at full-scale MSWIs. A statistical analysis of the results showed that, although the effect of combustor operating parameters of CDD/CDF formation is interactive and very complicated, substantial reduction in CDD/CDF formation can be realized with high temperature sorbent injection to reduce HCl or Cl₂ concentrations, control of excess air (also affects ratio of CDDs to CDFs formed), and increased quench rate.

The de novo theory also considers the generation of CDDs/CDFs from the combustion of PVC resin. Key to the de novo synthesis of CDDs/CDFs is the initial formation of HCl from combustion. Paciorek and coworkers (1974) thermally degraded pure PVC resin at 400° C and produced 550 mg/g HCl vapor as a primary thermolysis product, which was observed as being 94 percent of the theoretical amount based on the percent weight chlorine on the molecule. Ahling et. al. (1978) have concluded that HCl can act as a chlorine donor to ultimately yield chlorinated aromatic hydrocarbons from the thermolytic degradation of pure PVC, and that these yields are a function of transit time, percent oxygen, and temperature. The data they observed from 11 separate experiments conducted with a range of temperatures from 570-1130° C indicated that significant quantities of various isomers of dichloro-, trichloro-, tetrachloro-, and hexachlorobenzenes could be produced. Choudhry and Hutzinger (1983) proposed that the radical species Cl× and H× generated in the incineration process may attack the chlorinated benzenes thus formed, and abstract hydrogen atoms to produce ortho-chlorine substituted chlorophenol radicals. These intermediate radical species then react with molecular oxygen to yield ortho-substituted chlorophenols. As a final step, the ortho-substituted chlorophenols act as ideal precursors to yield CDDs/CDFs with heat and oxygen.

Although most of the aforementioned experiments have involved the pyrolysis of anthropogenic substances, the de novo formation of CDDs/CDFs is theoretically proposed to include the combustion of autochthonous (naturally occurring) organic substances (Choudhry and Hutzinger,1983) in the presence of a chlorine donor. This possibility was first advanced by scientists at Dow Chemical Co. in 1978 in a proposed working hypothesis known as "the trace chemistries of fire" (Crummett, 1982). This proposed working hypothesis was based on the following observations:

1. Combustion processes are seldom more than 99.9 percent efficient in converting carbonaceous fuel into carbon dioxide.

2. The remaining 0.1 percent of the fuel is converted into traces of organic species including complex halogenated aromatic hydrocarbons. Most of these compounds have not been identified in combustion emissions.

3. Municipal solid waste and fossil fuels contain complex mixtures of diverse chemical species at variable concentrations.

4. Combustion fuels contain chlorine in a range of 1-5000 parts per million.

5. Particulate matter that is emitted from oil-fired heating and power plants contain vanadium and nickel. Particulates emitted from coal-fired power plants contain vanadium, nickel, iron, and manganese. In combination with silicon and unburned carbon, these species can act as catalysts in the combustion process to form halogenated aromatic hydrocarbons.

6. Chemical reactions that occur in flames include pyrolysis, oxidation, and reduction. Ions, electrons, free radicals, and free atoms interact in a continuously changing environment.

7. Dow scientists found traces of CDDs/CDFs in all particulate matter samples taken from areas that were in close proximity to combustion sources.

8. Precursors for the formation of CDDs have been experimentally proven, and have been identified to be primarily chlorinated phenols and chlorinated benzenes. Because the pyrolysis of polyvinyl chloride (PVC) produces chlorobenzenes, the combustion of PVC may cause the formation of CDDs/CDFs.

Dow Chemical Co. invited the scientific community at large to give advice on ways in which the trace chemistries of fire hypothesis could be tested. The following studies were proposed as a means of testing the hypothesis (Crummett, 1982):

1. Determine if CDDs/CDFs are present in soils (having a relatively high carbon content) taken from drill cores beneath ancient lake beds at depths corresponding to 5, 12, and 35,000 years of sedimentation and deposition.

2. Determine if CDDs/CDFs are present in ice core samples taken from the center of an ancient glacier.

3. Determine if CDDs/CDFs are present in volcanic ash.

4. Determine if CDDs/CDFs are present in sea breezes from remote islands in the South Pacific.

5. Determine if CDDs/CDFs can be formed by the combustion of fossil fuels in the presence of chlorine or inorganic chloride.

6. Determine if CDDs/CDFs can be detected in fish species taken from rivers remote from chemical manufacturing but close to incinerators and fossil-fueled power plants.

Although these studies were proposed in 1978, only items (3), (5) and (6) have even been partially addressed. Thus the "trace chemistries of fire" remains largely a working hypothesis that is in need of further testing and proving through well designed and conducted field sampling and laboratory research programs. Nevertheless, there exists some empirical evidence in this area.

Liberti et al. (1983) showed that CDDs/CDFs could be produced from the combustion of pure vegetable extracts in the presence of chlorine gas and oxygen. Pyrolytic degradation of extracts of chestnut, mimosa, and tannic acid was accomplished in a bench-scale thermal reactor. When combustion proceeded without chlorine gas, phenolic compounds and cresol were formed as primary thermolysis products. When the vegetable extracts were burned in association with chlorine gas or PVC plastic, chlorophenols and CDDs/CDFs were formed. Liberti et al. (1983) postulated that the PVC was acting as a chlorine donor in the formation of CDDs/CDFs from phenolic compounds, and that the chlorine gas directly formed the chlorinated precursor from a phenolic (pre-dioxin) ring structure. Table 3-25 summarizes these experiments.

There is some empirical evidence that the burning of wood, in the presence of chlorine or inorganic chlorides, may form CDDs/CDFs, although the evidence is not conclusive. Few of these experiments had ruled out contamination of the wood fiber by known chlorinated precursors through extraction and chemical analysis. None of the cited experiments attempted to determine if the wood fiber was contaminated by CDDs/CDFs prior to the conduct of the experiment. If the atmosphere serves to widely distribute CDDs/CDFs, and if CDDs/CDFs can exist in the vapor and particle phases in the ambient air, then trees and other biomass can become reservoirs of CDD/CDF contamination by means of particle deposition onto and vapor diffusion into the biomass. Until these possibilities have been addressed and their impacts, if any, are quantified, experiments in which CDDs/CDFs are generated from the combustion of wood must be interpreted with a certain degree of caution, especially with regard to proving that CDDs/CDFs can be formed in nature without human intervention.

Ahling and Lindskog (1982) demonstrated that the combustion of untreated wood in an open fire can generate relatively high concentrations of chlorinated aromatic hydrocarbons in the emissions. These compounds include established dioxin precursors such as di- through hexa-chlorobenzene and tetra- and penta- chlorinated phenols in ppb_v-ppm_v concentrations. In addition, ppm_v levels of benzene were produced. The presence of chlorinated precursors indicates that inorganic chlorides in the plant may be capable of chlorinating unsubstituted aromatic structures. Reaction kinetic experiments involving the formation of HCl vapor (Olie et al., 1983; Choudhry and Hutzinger, 1983) have shown that

HCl can be formed from inorganic chlorides as a result of a reaction between sulfur dioxide and sodium chloride during combustion, as follows:

_

2 NaCl + SO₂ + 1/2 O₂ + H₂O -----> Na₂SO₄ + HCl

or: HCl can be liberated by the catalytic reaction of NaCl and a metal oxide. The general reaction is:

_

NaCl + H₂O + (A) -----> (B) + HCl

where: (A) is a metal oxide, i.e.: Al₂O₃, Fe₂O₃.

Olie et al. (1983) conducted wood burning experiments in a bench-scale combustion unit. Wood treated with pentachlorophenol was incinerated to generate CDDs/CDFs in one experiment, and 60-year-old wood from the demolition of a residence was burned in a separate experiment. The authors alleged that the 60-year-old wood predated the manufacture and use of phenoxy wood preservatives and, therefore, probably was absent any dioxin precursors; however, this was not analytically confirmed. They did not directly monitor the smoke emissions for the presence of CDDs/CDFs, only the collected fly ash. CDDs/CDFs were detected in the fly ash in ppb_w concentrations. However, the authors noted that the quantified CDDs/CDFs could have occurred as a consequence of the previous tests of burning wood treated with pentachlorophenol.

Nestrick and Lamparski (1983) conducted studies on residential wood combustion to evaluate the possibility that CDDs may form. This was accomplished through the evaluation of soot scrapings from the chimneys of wood burning stoves. Samples were taken at random from the eastern, central, and western regions of the United States. Average total CDD levels in the chimney flue scrappings were: 8.3 ppb in the eastern region, 42.5 ppb in the central region, and 9.9 ppb in the west.

EPA tested a freestanding noncatalytic residential wood stove for chimney flue gas emissions of CDDs/CDFs during the combustion of pine and oak (U.S. EPA, 1987d). Through a series of tests, it was determined that the wood fiber was free of known chlorinated precursors (e.g., PCBs, chlorinated benzenes, and chlorinated phenols). The total chloride concentration was found to be 125 ppm for the oak and 49 ppm for the pine wood prior to burning in the wood stove. The combustion of the wood generated ppm levels of aromatic hydrocarbon compounds. This relatively high loading of emissions on the sampling device interfered with any speciation of CDD/CDF compounds in the emissions. However, combustion ash samples provided an alternative matrix for evaluation. The analysis of ash samples from the unit showed that only OCDD was present as a dioxin contaminant at a maximum concentration of 0.09 ppb (by weight). Wipe samples were also taken from inside the chimney flue. OCDD and hepta-CDD were detected in the chimney soot at a maximum concentration of 0.6 ppb and 0.04 ppb, respectively. No lower chlorinated CDDs nor any CDFs were found in the ash or soot wipes.

Choudhry and Hutzinger (1983) have postulated that the complex structure of lignin in wood fiber can pyrolyze to generate CDDs/CDFs if a chlorine donor is present. This theory is based on the experiments of Kirsbaum et al. (1972) in which two lignin preparations (spruce and asp) were thermally degraded in glass tubes at 475° C to yield an array of hydrocarbons, including phenol. If phenol could be formed, and if gaseous forms of inorganic or organic chlorine are available, then the phenol could be chlorinated to form a chlorophenol compound. The latter is then a precursor to the ultimate formation of CDDs/CDFs. In addition, Choudhry and Hutzinger (1983) indicated that continued pyrolysis of other hydrocarbons identified as thermolysis products in the combustion of lignin could ultimately yield benzene. Nestrick et al. (1987) demonstrated that benzene can react with an inorganic chloride in the presence of heat to produce a variety of chlorinated aromatic compounds including CDDs/CDFs. If these thermolytic pathways are operational in lignin pyrolysis, then, in theory, it is possible that forest fires can generate CDDs/CDFs in the smoke, which has been proposed by Clement and Tashiro (1991). Because of the potential importance of lignin pyrolysis as a potential, yet unverified, combustion source of CDDs/CDFs in the environment, additional research should be directed in this area. In the conduct of combustion experiments involving the pyrolysis of lignin, attention should be given to the identification of any CDDs/CDFs or precursor compounds that may exist as contaminants. Only after sample contamination has been completely ruled out can the researcher draw convincing conclusions from the experiment.

Coal is a naturally occurring substance having the potential to form CDDs when combusted. Mahle and Whiting (1980) first reported on the results of high temperature combustion of bituminous coal in a bench scale furnace with the addition of HCl, NaCl, or Cl_2­ and air to yield CDDs. Table 3-26 summarizes these experiments. In experiment III, tetra through octa-CDDs were formed from the oxidation of coal by air which had been bubbled through a solution of hydrochloric acid. In a review of this experiment, Choudhry and Hutzinger (1983) postulated that the hydrochloric acid aided in the chlorination of aromatic hydrocarbons produced as combustion byproducts. This was also the case in experiment IV in which chlorine gas was introduced into the oxidation of coal. Experiment IV produced the highest yield of tetra- through octa-CDDs. When coal was combusted only with air, the exothermic reaction did not generate detectable quantities of tetra- or

hexa- CDDs. Experiment I yielded only hepta- and octa-CDD in quantities close to the detection limit.

The air emission of polychlorinated biphenyls (PCBs) from MSW incinerators is less understood. There are virtually no theories explaining the detection of these compounds in incinerator emissions nor other combustion sources, the exception being the intentional destruction of PCBs in hazardous waste incinerators in which case 99.9999 percent destruction rated efficiency (DRE) must be achiev